Kavousi I: The Archaeological Survey of the Kavousi Region (Prehistory Monographs) [Illustrated] 1931534187, 9781931534185

Table of contents :
Title Page
Table of Contents
Chapter 1 Introduction
Chapter 2 The Physical Landscape
Chapter 3 The Archaeological Survey
Chapter 4 The Pottery
Chapter 5 The History of Settlement
Chapter 6 Gazetteer of Archaeological Sites
Chapter 7 Conclusions
Appendix 1 A Pedological Investigation and Soil Survey of the Kavousi Region
Appendix 2 The Kavousi Fabrics: A Typology for Coarse Pottery in the Mirabello Region of Eastern Crete
Appendix 3 Petrographic Analysis of Some Final Neolithic—Early Minoan II Pottery from the Kavousi Area
Appendiz 4: Concordance of Artifacts
Bibliography
Index
Tables and Charts
Figures
Plates

Citation preview

KAVOUSI I The Archaeological Survey of the Kavousi Region

KAVOUSI The Results of the Excavations at Kavousi in Eastern Crete directed by Geraldine C. Gesell, Leslie Preston Day, and William D.E. Coulson

sponsored by The University of Tennessee under the auspices of The American School of Classical Studies at Athens

PREHISTORY MONOGRAPHS 16

KAVOUSI I The Archaeological Survey of the Kavousi Region by Donald C. Haggis

contributions by John T. Ammons, Peter M. Day, John E. Foss, Louise Joyner, Evangelia Kiriatzi, Margaret S. Mook, Michael W. Morris, Maria Relaki, and Michael E. Timpson

edited by Geraldine C. Gesell and Leslie Preston Day

Published by INSTAP Academic Press Philadelphia, Pennsylvania 2005

Design and Production INSTAP Academic Press Printing CRWGraphics, Pennsauken, New Jersey Binding Hoster Bindery, Inc., Ivyland, Pennsylvania

Library of Congress Cataloging-in-Publication Data Haggis, Donald C. Kavousi I : the archaeological survey of the Kavousi Region / by Donald C. Haggis ; with contributions by John T. Ammons ... [et al.] ; edited by Geraldine C. Gesell and Leslie Preston Day. p. cm. -- (Prehistory monographs ; 16) Includes bibliographical references and index. ISBN 1-931534-18-7 (hardcover : alk. paper) 1. Kavousi Region (Greece)--Antiquities. 2. Excavations (Archaeology)-Greece--Kavousi Region. I. Gesell, Geraldine Cornelia. II. Day, Leslie Preston. III. Title. IV. Series. DF261.K4H34 2005 939'.18--dc22 2005028722

Copyright © 2005 INSTAP Academic Press Philadelphia, Pennsylvania All rights reserved Printed in the United States of America

Table of Contents List of Tables and Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix List of Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Preface to the Kavousi Excavation Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Preface to Kavousi I. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi Acknowledgments for the Kavousi Excavation Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv Acknowledgments for Kavousi I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii Abbreviations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxi Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxiii 1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2. The Physical Landscape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3. The Archaeological Survey. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 4. The Pottery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5. The History of Settlement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 6. Gazetteer of Archaeological Sites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 7. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Appendix 1. A Pedological Investigation and Soil Survey of the Kavousi Region, Michael W. Morris, John T. Ammons, Michael E. Timpson, and John E. Foss. . . . . . . 153 Appendix 2. The Kavousi Fabrics: A Typology for Coarse Pottery in the Mirabello Region of Eastern Crete, Margaret S. Mook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167

vi

KAVOUSI I

Appendix 3. Petrographic Analysis of Some Final Neolithic–Early Minoan II Pottery from the Kavousi Area, Peter M. Day, Louise Joyner, Evangelia Kiriatzi, and Maria Relaki . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 Appendix 4. Concordance of Artifacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Tables and Chart Figures Plates

List of Tables and Chart Table 1.

Chemical properties including carbon fractions, exchangeable cations, and pH of Kavousi 2.

Table 2.

Particle size distribution of Kavousi 2.

Table 3.

Chemical properties including carbon fractions, exchangeable cations, and pH of Kavousi 1.

Table 4.

Particle size distribution of Kavousi 1.

Table 5.

Chemical properties including carbon fractions, exchangeable cations, and pH of Kavousi 3.

Table 6.

Particle size distribution of Kavousi 3.

Table 7.

Chemical properties including carbon fractions, exchangeable cations, and pH of Vronda 4.

Table 8.

Particle size distribution of Vronda 4.

Table 9.

Chemical properties including carbon fractions, exchangeable cations, and pH of Terra Rossa.

Table 10. Particle size distribution of Terra Rossa. Table 11. Chemical properties including carbon fractions, exchangeable cations, and pH of C-T Unit. Table 12. Particle size distribution of C-T Unit. Table 13. Chemical properties including carbon fractions, exchangeable cations, and pH of Vronda 9. Table 14. Particle size distribution of Vronda 9. Table 15. Chemical properties including carbon fractions, exchangeable cations, and pH of Vronda 8. Table 16. Particle size distribution of Vronda 8. Table 17. Chemical properties including carbon fractions, exchangeable cations, and pH of Kastro 3. Table 18. Particle size distribution of Kastro 3.

Chart 1.

Available water-holding capacity for Kavousi Soil Survey pedons.

List of Figures Figure 1.

Map of East Crete, showing location of the Kavousi-Thriphti Survey zone.

Figure 2.

Topographical map of the Kavousi region: north Ierapetra Isthmus and eastern Mirabello coast.

Figure 3A. Geological map of the Isthmus of Ierapetra and environs, showing pre-Neogene rocks, based on Papastamatiou et al. 1959 (P.M. Day). Figure 3B. Kavousi-Thriphti Survey boundaries. Figure 4.

Fields surveyed intensively.

Figure 5.

Sites recovered in the intensive sample, 1989–1991.

Figure 6A. Chordakia and south Kambos survey sites. Figure 6B. Tholos, Hagios Antonios, and north Kambos survey sites. Figure 7.

Survey sites in the area of Kavousi village, Xerambela, and north slope of Mount Papoura.

Figure 8.

Final Neolithic survey sites; FN sites of uncertain size numbered only.

Figure 9.

Early Minoan I–II survey sites; EM sites of uncertain size numbered only.

Figure 10.

Early Minoan III–Middle Minoan IA survey sites; EM III–MM IA sites of uncertain size numbered only.

Figure 11.

Middle Minoan IB–II survey sites.

Figure 12.

Middle Minoan III–Late Minoan IB survey sites.

Figure 13.

Late Minoan IIIA–B survey sites; LM IIIA–B sites of uncertain size numbered only.

Figure 14.

Late Minoan IIIC–Archaic survey sites.

Figure 15.

Orientalizing–Archaic survey sites.

Figure 16.

Hellenistic–Roman survey sites.

Figure 17.

Cataloged sites in extensive and unsurveyed zones.

Figure 18A. Xerambela and north Mount Papoura area: Early Minoan III–Middle Minoan IA survey sites. Figure 18B. Chordakia: Early Minoan III–Middle Minoan IA survey sites.

x

KAVOUSI I

Figure 19A. Chordakia and south Kambos: Middle Minoan IB–II survey sites. Figure 19B. Xerambela and north Papoura: Middle Minoan IB–II sites. Figure 20.

Kavousi region: Middle Minoan IB–II settlement clusters.

Figure 21A. Tholos and Hagios Antonios: Middle Minoan III–Late Minoan I sites. Figure 21B. Chordakia: Middle Minoan III–Late Minoan IB sites. Figure 22.

Xerambela and the north Papoura area: Late Minoan IIIC–Archaic sites.

Figure 23A. Site 1 (loci 16, 24, 38, and 39); Sites 2–4. Figure 23B. Site 1: sketch plan of Minoan and Roman buildings (loci 24, 38, 39) (L.A. Turner). Figure 24.

Site 1: section and plan of Roman warehouse (locus 39) (L.A. Turner).

Figure 25.

Site 1 (loci 16, 38, and 39): Bronze Age pottery and figurine; Roman pottery.

Figure 26.

Site 1 (locus 24): Bronze Age pottery; Site 4: Bronze Age pottery.

Figure 27.

Site 5: sketch plan of Bronze Age settlement (loci 10, 15, 61).

Figure 28.

Site 5: Bronze Age pottery, stone weights.

Figure 29.

Site 5: Bronze Age loomweight, cooking dishes, and figurine; Late Minoan IIIA–B pottery.

Figure 30.

Site 5: Bronze Age pottery.

Figure 31.

Site 6: Bronze Age pottery and obsidian blade fragment; Site 8: Final Neolithic–Early Minoan II pottery (M.S. Mook).

Figure 32.

Site 8: Bronze Age pottery (M.S. Mook).

Figure 33.

Site 8: Bronze Age ground stone tools and door socket

Figure 34A. Site 9: sketch plan. Figure 34B. Bronze Age pottery. Figure 35.

Site 11: Late Minoan IA pithos rim; Site 16: Final Neolithic pottery; Site 17: Bronze Age and Roman pottery.

Figure 36.

Sites 19–20: Bronze Age pottery.

Figure 37.

Site 21: Bronze Age pottery; Site 23: Neopalatial tripod leg; Site 24: Final Neolithic and Early Minoan Pottery (M.S. Mook).

Figure 38.

Site 24: Final Neolithic and Bronze Age pottery (M.S. Mook).

Figure 39.

Site 24: Bronze Age pottery; furnace fragment (24.41); Sites 25 and 27: Bronze Age pottery.

Figure 40.

Site 28: Bronze Age pottery.

Figure 41.

Site 28: Bronze Age pottery.

Figure 42.

Site 32: Bronze Age pottery and furnace fragments.

Figure 43.

Sites 34 and 35: Bronze Age pottery.

Figure 44.

Site 36: Late Minoan IIIA–B pottery and loomweight (M.S. Mook).

Figure 45.

Site 36: Late Minoan IIIA–B pottery (M.S. Mook).

LIST OF FIGURES

xi

Figure 46.

Sites 37 and 39: Roman pottery.

Figure 47.

Sites 40 and 41: Bronze Age and Roman pottery.

Figure 48.

Site 42: Hellenistic and Roman pottery (M.S. Mook).

Figure 49.

Site 44: Early Minoan III–Middle Minoan II pottery.

Figure 50.

Site 45: Bronze Age pottery; Site 48: Middle Minoan IA cup rim; Site 53: Roman pottery.

Figure 51.

Site 57: Bronze Age and Roman pottery; Site 68: Bronze Age pottery.

Figure 52.

Site 69: Bronze Age pottery.

Figure 53.

Site 69: Bronze Age pottery.

Figure 54.

Site 70: Early Iron Age–Archaic pottery.

Figure 55.

Sites 71 and 85: Early Iron Age–Archaic pottery.

Figure 56.

Site 88: Bronze Age pottery; Site 89: Early Iron Age–Archaic pottery.

Figure 57.

Site 90: Early Iron Age–Archaic pottery; Site 92: Early Minoan III–Middle Minoan IA pottery.

Figure 58.

Site 94: Roman pottery; Site 96: Bronze Age pottery.

Figure 59.

Site 96: Bronze Age pottery and figurine fragment.

Figure 60.

Site 97: Bronze Age pottery.

Figure 61.

Site 97: Bronze Age pottery (M.S. Mook).

Figure 62.

Site 97: Bronze Age pottery.

Figure 63.

Site 99: Late Minoan IIIC–Early Protogeometric pottery.

Figure 64.

Site 99: Late Minoan IIIC pottery and terracotta votive plaque.

Figure 65.

Site 99: Late Minoan IIIC pottery.

Figure 66.

Site 100: Final Neolithic, Middle Minoan, and Late Minoan IIIC pottery.

Figure 67.

Site 100: Late Minoan IIIC pottery.

Figure 68.

General topography of the Kavousi-Thriphti Survey area of East Crete (M. Morris).

Figure 69.

Geologic map of the Kavousi-Thriphti Survey area of East Crete. After Papastamatiou et al. 1959 (M.W. Morris).

Figure 70.

Soil survey map of the Kavousi area of East Crete (M.W. Morris).

Figure 71.

A3.2 Geological map of the Isthmus of Ierapetra and environs, showing Neogene and later deposits, based on Fortuin 1977 and Papastamatiou et al. 1959 (P.M. Day).

List of Plates Plate 1A. Kavousi village (center) from northwest: the Kastro (upper center); Mount Kliros (left); Xerambela and Mount Papoura (right); Kambos plain (foreground). Plate 1B. Mount Chomatas (Site 9: center) and Mount Chalepa from north. Plate 2A. Mount Chomatas (Site 9) from south. Plate 2B. Tholos Bay from southwest. Plate 3A. Cha Gorge and north Ierapetra Isthmus from southwest: Monastiraki Katalimata and Chalasmenos (center). Plate 3B. Mount Thriphti from west: the area of Hagia Anna and Argira (Site 97). Plate 4A. Avgo Valley at Pachlitzani Agriada from northeast (Site 82). Plate 4B. Skouriasmenos area (Site 81): recent terrace walls. Plate 5A. Skala (Sites 78–79): terracing southwest of the Kastro (Site 80) (M.S. Mook). Plate 5B. Chordakia: phyllite terrain on east slope (Sites 28, 36–37). Plate 6A. Kambos: irrigation reservoir and greenhouses at Lakkos Skaphes. Plate 6B. Mount Chomatas: limestone terrain on east slope. Plate 7A. Hagios Antonios valley (Sites 5 and 6): intersecting limestone and phyllite terrain. Plate 7B. Hagios Antonios (Site 5): Minoan agricultural terrace wall and eroded limestone terrain. Plate 8A. Argira (Site 97): Minoan house wall and pithos buried under phyllite mudslide. Plate 8B. Site 59: Bronze Age house foundations. Plate 9A. Tholos Bay and the north Kambos from the southeast. Plate 9B. Site 1: Bronze Age building (locus 16). Plate 10A. Site 1: Bronze Age building (locus 16). Plate 10B. Site 1: exterior view of Roman warehouse from northeast (locus 39). Plate 11A. Site 5: Building A from southwest. Plate 11B. Site 5: Building B from northwest.

xiv

KAVOUSI I

Plate 12A. Site 5: agricultural terrace wall. Plate 12B. Site 6: Hagios Antonios rock shelter from east. Plate 13A. Site 7: Early Minoan rock shelter. Plate 13B. Site 7: Early Minoan pottery imbedded in floor surface of the rock shelter. Plate 14A. Site 8: Early Minoan settlement at Alykomouri from southwest. Plate 14B. Site 9: view of Minoan building from north. Plate 15A. Site 9: western wall of Minoan building. Plate 15B. Site 28: Late Roman–Byzantine wall. Plate 16A. Sites 31 and 32: promontory of Chrysokamino from northeast. Plate 16B. Site 31: Theriospilio cave mouth from west. Plate 17A. Site 32: Chrysokamino promontory. Plate 17B. Site 34: Minoan house from southeast. Plate 18A. Site 36: Minoan house walls. Plate 18B. Site 37: millstone (orbis). Plate 19A. Site 37: trapetum mortarium. Plate 19B. Site 59: Minoan house walls. Plate 20A. Site 68: Minoan house wall. Plate 20B. Site 71: view of Azoria South Acropolis from east; Xerambela (center). Plate 21A. Site 78: interior of Aloni Tomb III with larnax in situ (The University of Pennsylvania Museum of Archaeology and Anthropology, neg. # 139725). Plate 21B. Kastro (site 80), Vronda (Xerambela) (site 77), and Azoria (site 71) from the east. Plate 22. Kastro (site 80) and Skouriasmenos (site 81) from the east. Plate 23A. Fabric type I, sherd (K.E. May). Plate 23B. Fabric type I, detail of surface (K.E. May). Plate 24A. Fabric type II, sherd (K.E. May). Plate 24B. Fabric type II, detail of surface (K.E. May). Plate 25A. Fabric type III, sherd (K.E. May). Plate 25B. Fabric type III, detail of surface (K.E. May). Plate 26A. Fabric type IV, sherd (K.E. May). Plate 26B. Fabric type IV, detail of surface (K.E. May). Plate 26C. Fabric type IV, detail of fresh break (K.E. May). Plate 27A. Fabric type VI, sherd (K.E. May). Plate 27B. Fabric type VI, detail of surface (K.E. May). Plate 28A. Fabric type IX, sherd (K.E. May).

LIST OF PLATES

xv

Plate 28B. Fabric type IX, detail of surface (K.E. May). Plate 29A. Fabric type X, sherd (K.E. May). Plate 29B. Fabric type X, detail of sherd (K.E. May). Plate 30A. Fabric type XI, sherd (K.E. May). Plate 30B. Fabric type XI, detail of fresh break (K.E. May). Plate 31A. Fabric type XIII, sherd (K.E. May). Plate 31B. Fabric type XIII, detail of surface (K.E. May). Plate 32A. Fabric type XV, sherd (K.E. May). Plate 32B. Fabric type XV, detail of surface (K.E. May). Plate 32C. Fabric type XV, detail of fresh break (K.E. May). Plate 33A. Fabric type XVI, sherd (K.E. May). Plate 33B. Fabric type XVI, detail of surface (K.E. May). Plate 34A. Fabric type XX, sherd (K.E. May). Plate 34B. Fabric type XX, detail of surface (K.E. May). Plate 34C. Fabric type XX, detail of fresh break (K.E. May). Plate 35A. Fabric type XXI, sherd (K.E. May). Plate 35B. Fabric type XXI, detail of surface (K.E. May). Plate 35C. Fabric type XXI, detail of fresh break (K.E. May). Plate 36A. Fabric type XXV, sherd (K.E. May). Plate 36B. Fabric type XXV, detail of fresh break (K.E. May). Plate 37A. Kavousi 93/79 (Description 1). Photomicrograph showing frequent feldspars and igneous rock fragments. XP, horizontal dimension 4 mm (P.M. Day). Plate 37B. Kavousi 93/73 (Description 1). Photomicrograph showing igneous rock fragments and component minerals, with some metamorphism. Note especially alteration of feldspar in lower left of frame. XP, horizontal dimension 4 mm (P.M. Day). Plate 37C. Kavousi 93/69 (Description 2). Photomicrograph showing similar inclusions that differ in size and sorting from those in Pl. 37A–B. XP, horizontal dimension 4 mm (P.M. Day). Plate 37D. Kavousi 93/80 (Description 3). Photomicrograph showing angular igneous rock fragments and angular calcite (twinning visible). XP, horizontal dimension 4 mm (P.M. Day). Plate 38A. Malia 93/7 (Description 3). Photomicrograph showing same fabric as Kavousi 93/72 and 80 in sample from Malia. Largest angular calcite in top right of frame. XP, horizontal dimension 4 mm (P.M. Day). Plate 38B. Myrtos Phournou Koriphi 93/106. Photomicrograph showing the use of calcite tempering of fabrics in Period II (EM IIA). Cf. Pls. 37D and 38A. XP, horizontal dimension 4 mm (P.M. Day). Plate 38C. Kavousi 93/74 (Description 4). Photomicrograph showing grog-tempered fabric. XP, horizontal dimension 4 mm (P.M. Day).

xvi

KAVOUSI I

Plate 38D. Kavousi 93/70 (Description 4). Photomicrograph showing coarse grog-tempered fabric. Cf. Pl. 38C. PPL, horizontal dimension 4 mm (P.M. Day). Plate 39A. Kavousi 93/75 (Description 5). Photomicrograph showing sand-tempered South Coast fabric, with rounded basic volcanic, biotite/feldspar metamorphic rocks, and serpentine. XP, horizontal dimension 4 mm (P.M. Day). Plate 39B. Kavousi 93/76 (Description 5). Photomicrograph showing sand-tempered South Coast fabric, with rounded basic volcanic rock. XP, horizontal dimension 4 mm (P.M. Day). Plate 39C. Malia 93/55 (Description 5). Photomicrograph showing sand-tempered South Coast fabric, with rounded basic volcanic rock and serpentine. XP, horizontal dimension 4 mm (P.M. Day). Plate 39D. Kalo Chorio 94/12 (Description 5). Photomicrograph showing sand-tempered South Coast fabric, including rounded basic volcanic rock. XP, horizontal dimension 4 mm (P.M. Day). Plate 40A. Kavousi 93/77 (Description 6). Photomicrograph showing fabric with dominant phyllite. XP, horizontal dimension 4 mm (P.M. Day). Plate 40B. Kavousi 93/78 (Description 6). Photomicrograph showing fabric with phyllite and amphibole. XP, horizontal dimension 4 mm (P.M. Day). Plate 40C. Kavousi 93/81 (Description 7). Photomicrograph showing fabric with quartz, metamorphic rocks, red translucent fragments, biotite, and feldspars in groundmass. Note the red tcfs. Vasiliki Ware. XP, horizontal dimension 4 mm (P.M. Day). Plate 40D. Kavousi 93/81 (Description 7). Photomicrograph at higher magnification illustrating fine mono- and polycrystalline quartz, metamorphic rock fragments, and biotite. Vasiliki Ware. XP, horizontal dimension 1 mm (P.M. Day). Plate 41A. Kavousi 93/84 (Description 8). Photomicrograph showing the very fine fabric, with red tcfs and striations from secondary calcite (upper part of frame). Vasiliki Ware. XP, horizontal dimension 4 mm (P.M. Day). Plate 41B. Kavousi 93/82 (Description 9). Photomicrograph showing very fine fabric, with very small igneous rock inclusion in lower left of frame. Vasiliki Ware. XP, horizontal dimension 4 mm (P.M. Day).

Preface to the Kavousi Excavation Series Kavousi I: The Archaeological Survey of the Kavousi Region is the initial volume of the Kavousi Excavation Series, which presents the final report of the Kavousi Project, a program of archaeological investigation around the modern village of Kavousi in eastern Crete. Subsequent volumes will publish the results of the 1987–1992 excavations at the Vronda and Kastro sites in the Siteia Mountains above Kavousi and of the cleaning and new study of the excavations of Harriet Boyd in 1900 and 1901. The first recorded archaeological interest in the Kavousi area is that of Sir Arthur Evans, who passed through the region on trips to the eastern part of the island in 1894 and in 1899, when he purchased pottery and metal objects from a tomb at Plaï tou Kastrou for the Herakleion Museum. He recommended the site to Harriet Boyd, who excavated briefly at a number of sites in the area in 1900. After a week of excavation around the Kavousi Plain, she moved up into mountain sites above the village. She spent one week at Azoria, where she found puzzling remains that included circular structures and pottery of the Archaic and Hellenistic periods, followed by a week each on the Kastro peak and the Vronda ridge. On the Kastro she uncovered what she described as a “chieftain’s house” of thirteen rooms of the Geometric period; not far away at Skouriasmenos was a perfectly preserved, though robbed, tholos tomb with some Geometric and Orientalizing pottery in it. On Vronda she found a building and eight small tholos tombs, which she dated to sub-Mycenaean, a transitional period between the Bronze Age and Iron Age. She returned in 1901 to test other sites, particularly at Aloni and Avgo, and to look for a Bronze Age settlement site. When she was shown the promising Late Minoan site at Gournia, she transferred her excavation there, leaving the Kavousi area permanently. In 1912, however, Edith Hall, one of her colleagues at Gournia, briefly explored burial sites at Kamara tou Tholou just north of Kavousi village and at Hagios Antonios near the Bay of Mirabello. After this active period of excavation, interest in the Kavousi area declined, even though chance finds continued to appear. One such find, discovered during the construction of the water system, led to the excavation of a Geometric shrine at Pachlitzani Agriada by Stylianos Alexiou in 1950. In 1951 a tholos tomb was discovered by a farmer, George Sekadakis, while planting an olive tree. The tomb was never scientifically excavated, but the pottery was taken to the small museum in Ierapetra.

xviii

KAVOUSI I

In 1974 the authors of this preface began a project to update the information available on the early American excavations in the Isthmus of Ierapetra. Because many of the sites were not marked on a map, the first task was to locate them. By 1978 the scope of this project was narrowed to the Kavousi sites and defined as the Kavousi Project. At this time William Coulson joined the project because of his particular interest in studying the pottery found in the Vronda tholos tombs. The first step of the Kavousi Project was to study all the pottery and artifacts from Boyd’s 1900 and 1901 excavations. Some of this material had not been previously published, while the interpretation of the rest of the artifacts needed to be brought in line with recent archaeological discoveries. Permission was also obtained to publish the material from the 1951 tomb discovered by George Sekadakis. Because no map with the sites of the Kavousi area had ever been drawn, the task of the first season in 1978 was to remedy this situation. In the summer of 1979 the vases from the Sekadakis tholos tomb were studied at the Ierapetra Museum. In December of the same year, three weeks were spent in the Herakleion Museum cataloging, drawing, and photographing the material from the Plaï tou Kastrou tomb and the tombs excavated by Boyd at Vronda, Chrondrovolakes, Aloni, and Skouriasmenos. The summer of 1981 saw the first of four cleaning seasons. The Vronda tholoi, both those which had been excavated by Boyd and the one uncovered by Sekadakis as well as a tenth tholos, newly discovered but thoroughly robbed, were cleaned and added to the area map. The Aloni cemetery, southwest of the Kastro, and a tomb identified by Sekadakis at Ridopoulia, east of Kavousi, were also cleaned this season. In 1982 it was arranged that the members of the Kavousi Project would supervise the cleaning of the Kastro for the balloon photograph to be published in the Aerial Atlas of Ancient Crete, and permission was given to publish any finds. New architectural features and considerable pottery were found in the cleaning, and these helped refine the dating of the site. In 1983 and 1984 the Project received permission to clean the building excavated by Boyd on top of the Vronda ridge and to make a plan of the structures she mentioned. Many new architectural features and finds appeared during this cleaning. The results of the cleaning seasons indicated that an excavation of the Vronda and Kastro sites would reveal much new information about the little understood end of the Bronze Age and the beginning of the Early Iron Age in Crete, a time when Crete experienced a major culture change from the prehistoric Minoan and Mycenaean civilizations to the historic Greek civilization. It had become clear that the Vronda “building” was a settlement dated to LM IIIC, the very end of the Late Bronze Age, at that time a rarely encountered period, and that it was important to uncover the settlement and allow it to contribute its part to the history of Crete. It was also evident that the rooms on the Kastro did not belong to a chieftain’s citadel, as Boyd had thought, but to a large settlement of the Geometric period with evidence of earlier use in the Protogeometric era. Because the Kastro was being used as a type site for the Geometric period in Crete, it was particularly important that it be more fully excavated and that corrections be made in its date and function. At the same time, advances in scientific archaeology, particularly in the fields of soil science, faunal analysis, human osteology, and palaeobotany, made it possible to obtain new information for reconstructing the environment and human interaction with it. The development of modern technologies for the study of clays, metals, and stone implements offered promise for reconstructing the social, economic, and political environment of the period.

PREFACE TO THE KAVOUSI EXCAVATION SERIES

In 1985 a conference held in Ierapetra and Kavousi presented the new findings and the evidence for a profitable excavation. The conference initiated interest in reopening work on this neglected period, and in 1986 the Project was granted one of the three excavation permits allotted to the American School of Classical Studies at Athens. Regular excavations of the Vronda and the Kastro sites began in 1987, at which time the Kavousi Project became the Kavousi Excavations under the directorship of Geraldine C. Gesell (Executive Director), Leslie Preston Day (Field Director at Vronda), and William D.E. Coulson (Field Director at Kastro). These excavations continued through 1992, interrupted by a study season in 1991. Since then study seasons were held regularly each summer until 2003. From the beginning, the Kavousi Project was designed as a regional study of the Kavousi area, combining the excavation of two major sites with archaeological and environmental survey. The goal was to reconstruct as thoroughly as possible the culture of the Vronda and Kastro communities at this transitional period between the Minoan and Greek cultures and define their relationship to each other and to the rest of Crete. Donald Haggis was asked to conduct an archaeological survey that would provide the diachronic pattern of settlement in the Kavousi area. This would be the basis for the placement of the two excavated sites, Kastro and Vronda, within the complex history of the Kavousi area. The specific goals that were defined for the excavation will be addressed in the following volumes. These include the chronological sequences of pottery, necessary in working out the chronology of the sites, study of the architecture and artifacts used in the different periods, a study of the life styles of the people, their religion, funerary customs, level of technology, agriculture, and contacts with the rest of Crete and the eastern Mediterranean. This volume, The Archaeological Survey of the Kavousi Region, provides a comprehensive look at the topography of the area, its natural resources, and the way in which the local people interacted with them over time, as shown in the changing pattern of settlement. It sets the stage for the report on the excavations and provides an introduction to the local soils and to the pottery classification used by the excavators. Geraldine C. Gesell Knoxville, TN, 2005 Leslie Preston Day Crawfordsville, IN, 2005

xix

Preface to Kavousi I This book presents the results of an archaeological survey that was conducted in the area of the modern village of Kavousi in northeastern Crete during three successive field seasons from 1988 to 1990, followed by study seasons in 1991, 1992, and 1996. The fieldwork, the Kavousi-Thriphti Survey (KTS), was carried out in conjunction with the Kavousi Project’s excavations of the Early Iron Age sites of Vronda and Kastro (Day and Snyder 2004; Gesell, Day, and Coulson 2000; 1995; Coulson et al. 1997) and formed the basis for my Ph.D. dissertation at the University of Minnesota (Haggis 1996a; 1992). The aims of the survey were to provide a broad archaeological and environmental context for these excavations, to explore the wider region, and to recover and examine sites contemporary with Vronda and Kastro. The project was thus initially conceived as a complement to ongoing excavation, a means of augmenting the data used in the interpretation of the excavation context. The scope of the survey included the examination of a number of different periods of human activity in the Kavousi region, while the reconstruction of the history of settlement benefited from the results of a number of neighboring projects (Gesell 2004). In addition to the excavations at Vronda and Kastro, work at Pseira (Betancourt and Davaras 2003; 2002; 1999; 1995; 1988; McEnroe 2001; Floyd 1998), Mochlos (Soles 2004; 2003; Soles et al. 2004; Barnard and Brogan 2003), Chalasmenos (Tsipopoulou 2004; 2001; Coulson et al. 1995), Katalimata (Nowicki 2003; 2000b; Coulson et al. 1995), Vasiliki (Eliopoulos 2004; 1998), Kalo Chorio (Haggis 1996c), Evraïka, and Chrysokamino (Betancourt and Floyd 2000–2001; Muhly and Betancourt 2000; Betancourt 2000a; Floyd 2000; Betancourt et al. 1999), provided stratified contexts for establishing ceramic sequences and detailed information on site histories and functions, ceramic distribution, metallurgy, trade, and staple production. It remains to integrate these data into discussions of the wider regional contexts. The Vrokastro Archaeological Survey Project (Hayden 2004a; 2004b) and the Gournia Project’s survey of the north Ierapetra Isthmus (Watrous et al. 2000; Watrous and Blitzer 1999; 1995) have documented contiguous areas of the Mirabello region, and their results will provide the broad regional perspective necessary in understanding settlement patterns at Kavousi. The full application of these data from both excavation and regional survey must, however, await the final publication of projects in the Mirabello area. The aim here is not to anticipate results but to integrate such data where available and immediately relevant. The core of the present text and site gazetteer were written in residence at the American School of Classical Studies at Athens in 1991–1992 and revised on Crete

xxii

KAVOUSI I

in 1996, allowing me to consult preliminary reports of recent and ongoing excavation and survey in the Kavousi and Mirabello areas, which were published between those dates. While my approach and perspective have changed considerably since 1992, the chapters incorporate little new information. The data are essentially as they were recorded from 1989 to 1992. The objectives of the project were to provide: 1. A site catalog and documentation of the archaeology of the Kavousi region 2. A broad overview of the data and summary of results 3. An interpretation of the settlement patterns and land use toward a reconstruction of the history of human activity in the sample area The chronological framework was from Final Neolithic (4000 B.C.) to Late Roman (4th–7th centuries A.D.) with a specific focus on the Bronze Age and Early Iron Age (3200–700 B.C.). The fieldwork was conducted as a dissertation project under a topographical survey permit; inherent limitations of the permit from the Greek Ministry of Culture, as well as of time, funding, and expertise of the primary investigator precluded systematic collection of stone tools, and the study of Byzantine, Venetian, Ottoman, and 20th century material. Topics forming focused or detailed synthetic studies on Early Iron Age settlement patterns (Haggis 2001; 1993a), Early Bronze Age chronology and settlement structure (Haggis 2002; 1999a; 1993b), and the Roman remains at Tholos (Haggis 1996b) have been published elsewhere and will thus be treated summarily here. My own changing perceptions of the data have required reconsideration of old ideas and, indeed, the identification of new problems. Not only has new evidence come to light but also, as projects in East Crete are brought to the publication stage, entirely new issues and avenues of research have developed. Furthermore the methods, approaches, and technology of intensive archaeological survey have changed radically in the past 15 years since the initial design and implementation of the KTS (cf. Alcock and Cherry 2004a). The goal of this present book, however, is neither to engage this rich and problem-oriented literature, nor to present a discourse on survey methodology. The purpose here is simply to present the data that were collected during the Kavousi-Thriphti Survey, to introduce various archaeological contexts that have a relevance to current issues and problems in the archaeology of Crete, and to explain how the idiosyncratic process of conducting the survey has affected the results. In the course of the fieldwork, an ongoing dialogue between excavation and survey was complemented by, and indeed in many cases generated by, scientific methods and the parallel research interests of zoologists (Snyder and Klippel 1999; 1996; Klippel and Snyder 2000; 1991), botanists (Flint-Hamilton 2000; 1999), geologists and soil scientists (M. Morris 2002; 1994; Timpson 1992; Watrous et al. 2000), and physical anthropologists (Liston 1993). Ceramic petrography (P. Day 1997), for example, employed in regional ceramic provenience studies has greatly enhanced and in many ways utterly changed the way we perceive economic and cultural spheres of influence in the Bronze Age Aegean. The discussion of the pottery (Chapter 4 and Appendices 2 and 3) has been revised to include some information from recent excavations. The original pottery sherd descriptions are retained and appended to the site entries in the gazetteer (Chapter 6), providing a catalog for each site where collections were made. The original KTS (Kavousi-Thriphti Survey) pottery catalog numbers are included for cross-referencing with the dissertation (see Appendix 4). Since the initial fieldwork was carried out between 1988 and 1991, an unprecedented number of projects have been conducted within the survey zone. The catalog

PREFACE TO KAVOUSI I

xxiii

entries in the site gazetteer make use of published and unpublished information, but they have been greatly abbreviated because detailed preliminary reports have been published by the excavators since the original records of these sites were made. Most important are the excavations in the area of Monastiraki at Katalimata and Chalasmenos (by Metaxia Tsipopoulou, William Coulson, and Krzysztof Nowicki) and the survey of the north Isthmus of Ierapetra by L. Vance Watrous and Harriet Blitzer. Such work now replaces my and Nowicki’s topographical study of these sites and areas (Haggis and Nowicki 1993). Nowicki’s work at Katalimata has recovered Neolithic, Protopalatial, and Neopalatial phases, while Watrous’s study of the nearby plain of Kamina and Monastiraki will elucidate its broader context. Excavations at the Chrysokamino (habitation and metallurgical) sites by Philip Betancourt, Cheryl Floyd, and James Muhly have significantly augmented the data, changing our understanding of their chronology and function. I am grateful to these excavators for their permission to use information from these sites. With these exceptions, the catalog entries in the gazetteer are essentially as they appeared in my dissertation. The local Bronze Age coarse-ware chronology has, since its initial publication, become refined through the efforts of numerous researchers on various Mirabello projects, and represents the continuing work of Margaret Mook, co-author of the original study of Kavousi material, and the petrographic work by Peter Day (Appendices 2 and 3). The details of this complex chronology will be refined and elaborated in the publications of the ceramic sequences of the Kavousi Excavations (Mook 2004; 1999; Mook and Coulson 1997; 1993), the Mochlos Excavations (Barnard and Brogan 2003), and the Vrokastro Archaeological Survey Project (Hayden 2004a; 2004b; 2003a; 2003b). These areas represent two vitally important ceramic production centers in the north Isthmus area, which have informed our understanding of material patterns and exchange systems in the broader region and have helped to shape the research design of the Kavousi-Thriphti Survey (Haggis 2000). While I have made every effort to update and utilize relevant bibliography, reflecting the vast amount of excavation, intensive survey work, and synthetic studies published since this manuscript was revised in the fall of 1996, the primary aim of this book is to present the data as originally compiled and examined (1989–1992)—the gazetteer of sites, the physical context, and the settlement history—as a contribution to the archaeological record of the region. Thus, this study reflects on the most relevant of this recent work, which has indeed already been effectively integrated into a wider discourse on regional settlement patterns in Crete and the Aegean (e.g., Hayden 2004a; Cunningham and Driessen 2004; Borgna 2003; Wallace 2003; Driessen 2001; Cunningham 2001; Nowicki 2000a; Farenga 1998). Donald C. Haggis Koutsouras, Crete, 1996 Chapel Hill, NC, 2004

Acknowledgments for the Kavousi Excavation Series The Kavousi Excavations were sponsored by the University of Tennessee under the auspices of the American School of Classical Studies at Athens with the permission of the Greek Ministry of Culture. The major supporters of the excavation in addition to the University of Tennessee have been the Institute for Aegean Prehistory, the National Endowment for the Humanities (an independent federal agency), the National Geographic Society, and Mr. Richard L. Sias and Mrs. Jeannette F. Sias. In particular, the University of Tennessee has supported the excavations through the Office of Research, Faculty Development Grants from the Graduate School, and funds from the College of Arts and Sciences, the Department of Classics, the Department of Anthropology, the Agricultural Experiment Station, and the College of Agriculture and Natural Resources. Special thanks must be given to Sheadrick A. Tillman (former Assistant Vice Provost for Research) and Kenneth R. Walker (former Assistant Vice President of Research), C.W. Minkel (former Dean of the Graduate School), Anne Mayhew (Vice Chancellor for Academic Affairs and Dean of the Graduate School), William Stuart Riggsby (former Dean of the College of Arts and Sciences), and three Heads of the Department of Classics: Harry C. Rutledge, who gave the project his complete support from the beginning and whose enthusiasm for the project was instrumental in securing private funding, Susan Martin, and David Tandy, both of whom readily continued this support. Supporting foundations include the American Philosophical Society, the David A. Packard Foundation, the David and Lucile Packard Foundation, the Samuel H. Kress Foundation, and the Joullian Foundation. Support for faculty and students was provided by the Wabash College Faculty Development Fund and Student Intern Program and the University of Minnesota Graduate School and Office of International Programs. Other colleges providing support include the College of Wooster, Gustavus Adolphus College, College of St. Catherine, and Randolph-Macon College at Ashland, Va. Major contributors include Mr. James T. Bradbury and Mrs. Louise Bradbury, Mr. Donald S. Kennedy, Mrs. Katherine J. Nordsieck, Dr. Harry C. Rutledge, Mrs. Doreen C. Spitzer, and members of Harriet Boyd Hawes’s family: Mr. Alexander B. Hawes, Mr. Alexander B. Hawes, Jr. and Mrs. Jane G. Hawes, and Ms. Sue Hawes. Many faithful donors are or have been members of the East Tennessee Society of the Archaeological Institute of America, faculty members, or alumni of the Department of

xxvi

KAVOUSI I

Classics at the University of Tennessee: John M. Armistead, Esq., and Mrs. Julia B. Armistead, Dr. Paul Barrette and Dr. Susan D. Martin, Mr. Richard S. Bagwell and Mrs. Laura J. Bagwell, Mr. Charles K. Bayne and Mrs. Pauline S. Bayne, Mr. Richard M. Berry, Mr. Bernard S. Borie, Mr. James D. Cape, Dr. Jefferson Chapman and Mrs. Vicki Chapman, Mr. Arnold G. Cohen and Mrs. Linda M. Cohen, Dr. Christopher P. Craig and Mrs. Ann E. Robinson-Craig, Dr. E. Charles Crume, Jr., Dr. Kenneth Curry, Dr. John H. Fisher and Ms. Audrey A. Duncan, Dr. Scott E. Goins, Dr. John M. Googin and Mrs. Janet Googin, Mrs. Susan Neas Hankins, Mr. Charles P. Jones and Mrs. Janelle O. Jones, Mr. Richard B. Korsmeyer and Mrs. Lynn Korsmeyer, Mr. Steven D. Kramer and Mrs. Phyllis A. Kramer, Dr. Henri A. Levy and Dr. Bettie J. Levy, Dr. Herbert G. MacPherson and Mrs. Janet W. MacPherson, Mr. Raymond M. McMillan, Mr. Arthur G. Mitchell and Marsha K. Mitchell, Esq., Mr. Don G. Mitchell and Mrs. Judy A. Mitchell, Mr. Peter G. Poulos and Dr. Paula Nassen Poulos, Mrs. Thelma Present, Dr. J. Reece Roth and Mrs. Helen M. Roth, and Mrs. Gail Smelcer. Other regular donors include Mrs. Mary H. Barnes, Mr. Lloyd E. Beebe, Ms. Barbara Bell, Mr. Donald A. Coulson and Mrs. Catherine T. Coulson, Dr. Panos G. Gregoriou and Mrs. Lilia P. Gregoriou, Mr. and Mrs. Edward C. Joullian III, Dr. George R. Martin and Mrs. Ruth G. Martin, Mrs. Betty E. Matthew, Mr. George Seminoff and Mrs. Sharon Seminoff, Dr. Morris M. Weiss and Dr. Terry Weiss, Mr. S. Lynn Williams and Mrs. Noriko Williams. The directors wish to express their gratitude to all those who assisted with the project and the excavations including the following: Yannis Tzedakis, Ios Zervoudaki, and Katerina Romiopoulou (Directors for Prehistoric and Classical Antiquities, Ministry of Culture of Greece); Costis Davaras, the late Nikos Papadakis, Metaxia Tsipopoulou, and Stavroula Apostalakou (Directors of the East Crete Ephoria). The directors are grateful to Stylianos Alexiou, Yannis Sakellarakis, Charalambos Kritsas, Alexandra Karetsou, Eva Grammatikaki, and Nota Demopoulou (Directors of the Herakleion Museum) for their assistance in the study of material from Boyd’s excavations at Kavousi and the study of comparanda from various other sites. The directors are also grateful to Athanasia Kanta, Vassilis Dougalis, and Evangelis Sachperoglou for special assistance. The American School of Classical Studies at Athens helped in all stages of the work of the Kavousi Project. Sincere thanks are extended to Henry S. Immerwahr, Stephen G. Miller, the late William D.E. Coulson, James D. Muhly, Stephen V. Tracy (Directors of the School) and to their administrative assistant Maria Pilali. Sincere thanks are also extended to Thomas M. Brogan (Director of the INSTAP Study Center for East Crete) for assistance during the study seasons. The directors wish to express their special appreciation to Philip Betancourt (Executive Director of the Institute for Aegean Prehistory) for his constant support. The directors would like to express their appreciation for all the assistance and goodwill of the people of Kavousi, Pacheia Ammos, and Ierapetra throughout the years that the team of the Kavousi Excavations worked in their area. Finally, the directors are particularly thankful to Malcolm H. Wiener, the founder of the Institute for Aegean Prehistory, who has funded the Wiener Lab at the American School of Classical Studies and the INSTAP Study Center for East Crete, providing ideal study conditions and technical services for the Kavousi Excavations as well as other projects both in East Crete and elsewhere in the Aegean area.

Acknowledgments for Kavousi I Permission to conduct the Kavousi-Thriphti Survey was granted by the Greek Ministry of Culture and the 24th Ephorate of Prehistoric and Classical Antiquities. I am grateful for the help of the staff of the East Cretan Ephoreia and Archaeological Museums of Siteia and Hagios Nikolaos, and would like to thank Costis Davaras, the late Nikos Papadakis, Stavroula Apostolakou, and especially Metaxia Tsipopoulou, without whose support the fieldwork could never have been completed. The survey was conducted under the auspices of the American School of Classical Studies at Athens, and I thank the staff of the school and especially its director during the period of the fieldwork, the late William D.E. Coulson, for his support of the project at every stage. The survey was a component of the Kavousi Project, and I am indebted to the project directors, Geraldine C. Gesell, Leslie P. Day, and William Coulson for encouraging and facilitating the fieldwork. Gesell and Day generously let me use their unpublished site catalog and field notes of the extensive survey of the Kavousi and north Ierapetra Isthmus areas. Their knowledge of the sites in the Kavousi region was a constant guide during fieldwalking and preparation of the dissertation that formed the basis for this book. The preliminary environmental study of the Kavousi area was conducted by Richard Hebda (1986–1988), Curator of Botany and Earth History at the Royal British Columbia Museum; the soil, plant, and geomorphological study was carried out by John Ammons, John Foss, Michael Timpson, and Michael Morris of the University of Tennessee (1989–1992). I am indebted to my collaborator throughout the project, Margaret Mook, who was responsible for much of the ceramic study in the early phases of the project, designed the coarse-ware chronology used in the survey, and collaborated in the architectural and depositional study of the abandoned traditional settlements in the Kavousi area. Claudia Honeywell helped in the arduous and frequently tedious task of fieldwalking the Kambos in 1989. Lee Ann Turner drew the plan and section of the Roman warehouse at Tholos and many of the plans of the traditional settlements in the Avgo valley. I owe special thanks to Krzysztof Nowicki, whose contribution to the study of the Early Iron Age sites, especially Katalimata and Chalasmeno, was fundamental to my understanding of these settlements and the regional historical context in the Dark Age. Gerald Cadogan patiently introduced me to the idiosyncrasies of East Cretan and Mirabello area wares from Myrtos Pyrgos. The readers of my dissertation, William Coulson, Frederick Cooper, Joseph Alchermes, and Thomas Kelly, provided

xxviii

KAVOUSI I

numerous insights that have helped to improve this final report. Vance Watrous and Harriet Blitzer, while in residence at the American School in 1991, graciously gave of their time to read the initial chapters of the dissertation, were invaluable sources of advice on practical and methodological issues, and provided constant support and inspiration. Leslie Day and Geraldine Gesell patiently read and corrected errors in numerous versions and drafts of the present manuscript, and I have benefited greatly from the resultant discussion. The manuscript has been significantly improved by the input of Philip Betancourt at the INSTAP Academic Press, and the anonymous referees of both the Publication Committee of the American School of Classical Studies at Athens and the INSTAP Academic Press. I take full responsibility however for any errors in the text, the field methods, and methodology employed. Thanks are owed to Jeffrey Soles and Thomas Brogan for permitting me on a number of occasions to examine the Mochlos pottery deposits; I have learned much from the discussions of the Mochlos ceramic sequences with Brogan and Evangelia Sikla. The directors of the Vrokastro Survey Project, Barbara Hayden and Jennifer Moody, encouraged me to examine the pottery recovered in the Kalo Chorio region, and Moody generously made available to me the results of her detailed fabric study. I am grateful, too, to Philip Betancourt for permitting the study of Pseira excavation deposits, and to Cheryl Floyd for sharing her knowledge of the ceramic sequence and guiding us through the stratigraphy of that site. My own participation in the Gournia Project (1992–1996) has enhanced my understanding of the ceramic sequences, settlement systems, the meaning and function of site forms and hierarchies in the region, and cultural landscapes in general. In this I owe special thanks to Vance Watrous, director of the Gournia Project, for encouraging my participation in that survey and for countless hours of discussion on Cretan archaeology. I am extremely grateful to Philip Betancourt for allowing me to examine the Early Minoan (EM) III pottery from Chrysokamino. In 1996, Evangelia Sikla showed me substantial contemporary EM IIB–III deposits from Mochlos, which she is studying for publication. I am indebted to her for taking the time on a number of occasions to let me examine this very important pottery and for discussing its archaeological context. Special thanks are owed to Leslie Day for making available for study the pottery from Pre- and Protopalatial levels at Vronda. Peter Day, David Wilson, and Todd Whitelaw offered invaluable insights on the Prepalatial pottery from the survey samples and aspects of regional ceramic production and exchange (cf. Appendix 3). There are many others whose contribution took the form of advice, encouragement, and stimulating discussion of practical, methodological, or theoretical aspects of the fieldwork: Jack Davis, the Douloumis family, Eleni Georgoulaki, Kevin Glowacki, Georgos Grammatikakis, Richard Hope Simpson, Manolis Kasotakis, Walter Klippel, Harold Koster, Sylvie Müller, Jean-Claude Poursat, Oliver Rackham, Georgos Sekadakis, Lynn Snyder, Nikolis Spiliarotis, Thomas Strasser, and Eberhard Zangger. The photographs and drawings throughout are the work of the author, unless otherwise indicated in the lists of figures and plates. Briece Edwards, Margaret Mook, Sondra Jarvis, Margaret Reid, and Lee Ann Turner assisted in the inking of the illustrations. The fieldwork on Crete was supported by a Fulbright Collaborative Research Grant (1989/1990), a University of Minnesota Doctoral Dissertation Fellowship and Special Grant (1989), a Wenner-Gren Pre-Doctoral Small Grant-in-Aid for Anthropological Research (1990), the Doreen C. Spitzer Fellowship from the American School of Classical Studies at Athens (1990/1991), and the Olivia James Traveling

ACKNOWLEDGMENTS FOR KAVOUSI I

Fellowship of the Archaeological Institute of America (1991/1992). The University Research Council of the University of North Carolina at Chapel Hill (Research and Publication Grants) partially supported the preparation of the preliminary report and several of the illustrations. A research leave and Junior Faculty Development Award, granted by the University of North Carolina at Chapel Hill, provided the means to prepare this manuscript on Crete and in Athens in the fall of 1996. Finally, I thank my parents, Alex J. Haggis and Pauline P. Haggis, without whose practical help and moral support the fieldwork and this report would not have been possible. The authors of Appendix 3 are indebted to a great many people for facilitating this petrographic study. For permission to sample pottery from a number of sites and for facilitating that process, we wish to thank: the Greek Archaeological Service and Directorate of Conservation, 24th Ephoreia, P.P. Betancourt, T. Brogan, G. Cadogan, the late W.D.E. Coulson, P. Darcque, C. Davaras, L.P. Day, G.C. Gesell, D.C. Haggis, B. Hayden, K. Manteli, J. Moody, M.S. Mook, A. Nikakis, N. Papadakis, O. Pelon, J.-C. Poursat, E. Sikla, J. Soles, M. Tsipopoulou, L.V. Watrous, and T.M. Whitelaw. We are especially grateful to G. Cadogan and P.S. Quinn for permission to refer to their unpublished work. Efi Kartsonaki produced the maps with some style. This work would not have been possible without the financial support of a number of sponsors: the Institute for Aegean Prehistory, the National Endowment for the Humanities, the Natural Environment Research Council of the U.K, and a grant given to the GEOPRO TMR Network, funded by the European Commission (ERB FMRX CT 98–0165).

xxix

Abbreviations The following chronological abbreviations are used (Final Neolithic and Bronze Age dates are based on Hankey and Warren 1989 and Nowicki 2003): FN EM MM LM SM EPG PG G LG EO O

Final Neolithic (ca. 4000–3200 B.C.) Early Minoan (ca. 3200–2100 B.C.) Middle Minoan (ca. 2100–1600 B.C.) Late Minoan (ca. 1600–1100 B.C.) Subminoan (ca. 1100–1000 B.C.) Early Protogeometric (ca. 1000–950 B.C.) Protogeometric (ca. 1000–900 B.C.) Geometric (ca. 900–700 B.C.) Late Geometric (ca. 760–700 B.C.) Early Orientalizing (ca. 700–660 B.C.) Orientalizing (ca. 700–600 B.C.)

A C H R ER LR B V Ott. Mod. B.P.

Archaic (ca. 600–480 B.C.) Classical (ca. 480–323 B.C.) Hellenistic (ca. 323–69 B.C.) Roman Early Roman (69 B.C.–4th c. A.D.) Late Roman (4th–7th c. A.D.) Byzantine (8th–12th c. A.D.) Venetian (12th–17th c. A.D.) Ottoman (17th c.–1900 A.D.) Modern (1900–present) before present

Max. PPL Pres. r sa sr tcf Th. W. XP

maximum plane polarized light preserved rounded subangular subrounded textural concentration feature thickness width cross-polars

The following other abbreviations are used: a AMSL c:f:v D. E frag. H. ha. KTS L. m

angular above mean sea level coarse:fine:void ratio diameter Early fragment height hectare Kavousi-Thriphti Survey length meter

xxxii

KAVOUSI I

The following soil science abbreviations are used: A Ap B BC Bss Bt Bw C Cb Cr C-T E O R

Mineral master horizon formed at the surface or below an O horizon that exhibits mixing of mineral and organic materials An A-horizon that exhibits disturbance by plowing or other agricultural uses Mineral master horizon that forms below an A, E, or O horizon that exhibits obliteration of much or all of the original rock structure A transitional horizon with characteristics of both an overyling B horizon and an underlying C horizon, but is more like B than C A B-horizon that exhibits the presence of slickensides A B-horizon that exhibits accumulation of silicate clays that either formed in place or were translocated from the overlying horizon A B-horizon that exhibits development of color or structure with little or no apparent accumulation of alluvial materials Mineral master horizon, excluding hard bedrock, that exhibits little effects of soil forming processes A C-horizon that has been buried Weathered or soft bedrock sometimes referred to as a paralithic contact Cretaceous-Tertiary Mineral master horizon exhibiting loss of clays, sesquioxides, or organic matter Master horizon dominated by organic material Hard bedrock

Glossary The text uses the following terms, which are specialized terminology or do not have exact English equivalents: akrotiri aloni argillic basic cutans geotrese kalderimi mandra mass-wasting metochi ped pedon perivolia plateia polypedon residuum revma sideropetra slickensides soil profile thermokepion trapetum type pedon ultrabasic

headland; promontory; peninsula threshing floor, usually with a stone-built border a soil horizon that is characterized by the movement of phyllosilicate clays in the horizon from a horizon above refers to an igneous rock with a low silica content modifications to soil peds due to processes such as coatings of clay on ped surfaces as the result of argillic horizon formation deep well cobbled path (paved or partially paved mountain road) sheep or goat pen downslope movement of soil or rock material as a direct result of gravity rural hamlet or field house a natural unit of soil structure (similar to a clod except that a clod is formed artificially) a three dimensional body of soil that adequately expresses the horizons developed from soil formation processes household or in-field gardens town square; village center a group of contiguous or similar pedons, usually considered the primary unit for soil mapping soil formed by the weathering of bedrock in situ torrent; gully; seasonal river gray crystalline limestone stress surfaces that are polished and striated produced by one mass sliding against another a vertical section of soil through all its horizons and extending into the C horizon impermanent greenhouse olive crushing bed/press a pedon used to characterize a soil mapping unit refers to an igneous rock with a lower silica content than basic

1

Introduction

This chapter gives a brief account of the history of archaeological research in the Kavousi area and describes the theoretical and methodological background of the Kavousi-Thriphti Survey. It aims to bring together what was known of the area’s archaeology and history before fieldwork began and to describe the reasons for conducting the survey. In addition, the chapter outlines the archaeological problems and issues that had a bearing on the research design. The area of the modern village of Kavousi and its surrounding district is composed of three converging environments in northeastern Crete: the Isthmus of Ierapetra (the narrow, twelve-kilometer-long lowland running north to south between the Bay of Mirabello and the Libyan Sea), the east coast of the Bay of Mirabello, and the Kavousi mountains (west Siteia Massif or Thriphti Range) (Figs. 1, 2; Pl. 1A). These distinct geophysical features are not only important factors in past and recent land use and communication patterns; their particular manifestation in this spot suggests that topography should be of paramount consideration in our interpretation and reconstruction of ancient human activity in this

region. Although topography has been a central concern in early archaeological reconnaissance since the turn of the 20th century, and often a focal point in discussions of causal factors related to historical discontinuities, today it is only one of many environmental features recorded by regional projects toward reconstructing ancient environments and assessing human-landscape interaction. For Harriet Boyd Hawes who first explored the Kavousi area in 1900, it was the correlation of diachronic changes and regional variability in settlement patterns with physical terrain that led her to conduct her extensive survey of the north Isthmus and Kavousi areas, perhaps one of the first such surveys in the Aegean (Boyd 1904–1905; 1904; 1901). Topography was a guide for examining new areas and for discovering and interpreting new site types. It was a context for understanding periodic variation in settlement patterns in a specific defined region. The reinvestigation of the Kavousi region was guided by many of the same problems—generated by a past century of research (cf. Haggis 2004; Gesell 2004, 15). Most important was that the region itself is a crossroads and gateway, a link

2

KAVOUSI I

between two major natural routes connecting the north and south coasts, the Aegean and Mediterranean, and between eastern and central Crete. The physical situation of the survey zone suggested two premises: first, the def inition of patterns within the survey zone would have to take into account a wider scale region, beyond the boundaries of the predetermined study universe. And second, human activity and settlement patterns, while necessarily

affected by external stimuli, are unique to their environment and could change periodically in a variety of ways depending on local social systems as well as ecological conditions. It is however the past century of intensive archaeological research in the north Isthmus of Ierapetra and the regions contiguous to the Bay of Mirabello, that has helped to shape the research design of the archaeological survey at Kavousi.

History of Research Archaeological sites in the Kavousi area were first examined in 1896 and 1899 by Sir Arthur Evans, who identified a number of sites, including the settlement on the Kastro (site 80), a tholos tomb at Plaï tou Kastrou (site 79), the Roman village at Chordakia (site 37), and the Minoan house on Chomatas at Chrysokamino (site 34; Fig. 2) (Gesell 2004; Evans 1896, 513; Frothingham 1896, 455–459; Hogarth and Bosanquet 1899, 321). The first systematic exploration of Kavousi, however, was carried out by Harriet Boyd, who surveyed on foot and horseback, identifying and excavating some fifteen sites in 1900 and 1901 (Gesell 2004; Hawes et al. 1908; Boyd 1904; 1901). The Early Iron Age sites of Vronda (site 77), Kastro (site 80), and their associated cemeteries were the primary focus of Boyd’s expedition, and the best recorded and published of the sites that she examined. In addition to the full-scale excavations of the Vronda settlement and cemetery at Xerambela, the Kastro peak, and cemeteries at Skala (or Aloni; site 78) and Skouriasmenos (site 81), Boyd explored the neighboring hilltop of Azoria (site 71)—a site that she believed was of some importance as an Early Iron Age center continuing into later periods—and a nearby cemetery at Chondrovolakes with Orientalizing burials (Fig. 2). In the Avgo Valley, Boyd excavated a Minoan house at Panagia (site 87) and identified the location of an LM III larnax burial nearby (site 86), as well as a tholos tomb on the Trapeza Ridge above Panagia (Hastings 1905). In exploring the Kambos plain in search of a major Minoan site to excavate, Boyd dug a test trench in the Roman warehouse at Tholos Bay (site 1) and identified Bronze Age buildings on the farthest point north of the Roman site. These walls were the remains of the Minoan

town that Boyd was seeking, but the shallowness of the deposition and the evidence of Roman-period disturbance of the early levels discouraged excavation. At Hagios Antonios, she explored the Minoan town (site 5) but excavated only a portion of a terrace wall and, in the valley below, a Hellenistic– Roman grave and cult site. Most important was Boyd’s identification of the Prepalatial site of Alykomouri (site 8) on the hill above the Hagios Antonios valley. Farther south, in Chordakia, she excavated Roman rock-cut cist tombs (site 30) and Late Roman–Byzantine walls and a cistern at Katsoprinos (sites 28 and 29). Boyd’s detailed observations of the landscape are indispensable in reconstructing archaeological sites today and periods of occupation that are now lost or disturbed by modern land use and natural formation processes. Furthermore, her descriptive accounts of contemporary agricultural practice, water supplies, vegetation, soil types and conditions, and traditional land use are data that have been used in visualizing the preWorld War I environment and settlement structure— important sources of analogy in reconstructing ancient land use patterns and the occupational history of the region. Boyd left Kavousi in 1901 to begin excavating Gournia (Boyd 1904–1905). Edith Hall (Dohan) and Richard Seager joined Boyd at Gournia but also conducted a number of small excavations in the Kavousi area. Seager excavated LM I–III burials at Ridopoulia, immediately east of Kavousi village (Gesell, Day, and Coulson 1983, 413–420). In 1910, Hall explored the area of the Metallurgical Site of Chrysokamino (site 32), which had been known to Boyd; she also excavated within the EM I–II cave of Theriospilio (Kolonospilio; site 31) (Foster 1978,

INTRODUCTION

6–9; Hall 1912; Mosso 1910, 289–290). Hall returned to Kavousi in 1912 to excavate LM III burials at Kamara tou Tholou (site 26) and the EM I–II burial cave at Hagios Antonios (site 6) (Hall 1914, 183–185). Subsequent work in the Kavousi area was largely topographical in nature, although in 1950, Stylianos Alexiou excavated the important Geometric–Archaic hearth temple at Pachlitzani Agriada (Makellos; site 82) (Alexiou 1956). The first systematic regional study of the Kavousi area was an extensive survey conducted by Geraldine Gesell and Leslie Preston Day, who discovered and cataloged several new sites as part of an exploration of the entire north Ierapetra Isthmus and east Mirabello regions (Gesell 2004, 14; Gesell and Day 1976). Their aim was to bring together the disparate pieces of information from past research in the Mirabello area into a coherent account of regional settlement patterns, by locating and documenting the incompletely published and unpublished sites that had been recovered during the reconnaissance and sporadic excavations since the turn of the 20th century. The scope and results of Gesell and Day’s work—which identified some thirty-three sites in the Kavousi area alone—demonstrated the complex patterns and density of Bronze Age, Early Iron Age, and Roman settlement in the Kavousi and north Isthmus regions. Gesell and Day were joined by Coulson in 1978 to form the Kavousi Project, and efforts were focused on the excavation and study of the sites of the Xerambela or northern Papoura drainage south of the modern village, including Vronda, Kastro, and neighboring cemeteries (Gesell 1990; 1986; Gesell, Day, and Coulson 2000; 1995; 1990; 1988; 1985; 1983; Coulson et al. 1997; Gesell, Coulson, and Day 1991). The Kavousi Project was, however, only one of many research endeavors in the Mirabello region, many of which are now in various stages of study and publication. Between 1971 and 1985, Jeffrey Soles and Costis Davaras conducted extensive topographical studies and excavation at Gournia and Mochlos (Soles 1992; 1991; 1988; 1979; 1978), beginning full-scale excavations at the latter site in 1989 (Soles and Davaras 1996; 1994; 1992; Soles 2004; 2003; Soles et al. 2004; Barnard and Brogan 2003). At Vrokastro, lying to the west of Kavousi across the Ierapetra Isthmus, where Barbara Hayden had been studying the architecture of the Early Iron Age settlement (Hayden 2003b; 1983; 1981), a

3

regional survey was begun in 1986 by Hayden and Jennifer Moody, including the important area of the Classical polis of Olerous and the Minoan town of Priniatikos Pyrgos (Hayden 2004a; 2004b; 2003a; 2003b; 2001; 1997a; Hayden and Moody 1990; Hayden, Moody, and Rackham 1992). In 1985, Philip Betancourt and Costis Davaras began excavations on the island of Pseira, subsequent to a number of important studies of pottery and artifacts from the Kavousi and Mirabello areas (Betancourt and Silverman 1991; Betancourt 1984; 1983; 1977; Betancourt and Davaras 2003; 2002; 1999; 1998; 1995; 1988; Betancourt et al. 1979; Betancourt, Davaras, and Hope Simpson 2004; 2005; Floyd 1998; McEnroe 2001). More recently, excavation has been conducted by Tsipopoulou, Coulson, and Nowicki at two Early Iron Age sites known to Boyd, Katalimata and Chalasmenos (Tsipopoulou 2004; 2001; Tsipopoulou and Coulson 2000; Nowicki 2003; 2002; 2000b; Coulson et al. 1995), and survey in the Gournia and north Isthmus areas, begun in 1992 by L.V. Watrous, has closed the spatial gap between the Kavousi and Vrokastro survey zones (Watrous et al. 2000; Watrous and Blitzer 1999; 1995). Excavations in 1993 by the author in Kalo Chorio have produced the first stratified habitation deposits of EM I date in eastern Crete (Haggis 1996c). In 1995–1997, Philip Betancourt, James Muhly, and Cheryl Floyd conducted survey and excavation at the Early Minoan III Metallurgical Site of Chrysokamino and the Minoan house on Mount Chomatas, sites important in reconstructing the history of settlement in the Kavousi plain and the coastal hills bordering the Bay of Mirabello (Muhly 2004; Betancourt 2000a; 1998; Muhly and Betancourt 2000; Betancourt and Floyd 2000–2001; Floyd 2000; Betancourt et al. 1999). The continuing reinvestigation of Vasiliki by Antonis Zois (Betancourt 2000b; Zois 1993; 1992; 1982; 1976) and the recent excavations on the neighboring hilltop of Kephala by Theodoros Eliopoulos (Eliopoulos 2004; 1998; Rehak and Younger 1998, 169) have changed our perception of settlement structure in the north Isthmus. This fieldwork in the Mirabello area has created an extraordinary database and research environment, where the results of survey and excavation can be combined to determine local ceramic sequences and spatial distribution, site types and functions, and settlement patterns.

4

KAVOUSI I

The Kavousi-Thriphti survey was a component of the Kavousi Project and thus a single part of this larger research effort in the Mirabello area. The complementary interrelationships of these various

projects and the effects of their diverse aims, perspectives, and periods of interest have helped to form both the methodology and the interpretation of results presented in this volume.

The Goals and Background of the Kavousi-Thriphti Survey The Kavousi-Thriphti Survey provides a broad history of settlement from Neolithic to Roman times. The principal objectives were, however, to explore the physical setting of Vronda and the Kastro and to reconstruct the settlement patterns in the region in the Early Iron Age and the periods of transition between the Bronze and Early Iron Ages. Given the large sizes and complex and overlapping chronology of these two settlements and their associated cemeteries, an immediate concern was to determine their agricultural and pastoral catchment areas by examining the sites’ physical surroundings, available natural resources, and the botanical and zoological data derived from the excavations. An important consideration was the functional relationship between the excavation sites and other contemporary settlements in the Kavousi area, the north Isthmus of Ierapetra, and the Kavousi mountains (Figs. 1, 2). Another objective was to construct from the archaeological evidence a local model of settlement that might be applied to the exploration of Early Iron Age cultural environments elsewhere in Crete and the Aegean. Important in the model were the combined dimensions of space and time derived from results of survey and excavation (Haggis 2001; 1999b; 1993a). The overlapping catchment areas and chronology of these settlements and their cemeteries suggested interesting patterns of socioeconomic interaction as well as diachronic changes in settlement patterns that might have been related to changing social and economic systems. We began to visualize Early Iron Age society on a regional scale, that is, with occupation and local identity linked to an area wider than the spatially focused realm afforded by excavation alone. Control of water supplies, access to resources for staple production, pastoral boundaries, and social ties across the landscape were seen as driving forces of territorial definition and regional organization, and ultimately the

behavioral systems potentially underlying the material patterns in the archaeological record. Although considerable effort has been expended in recent years toward reconstructing important regional Early Iron Age ceramic chronologies (Coulson 1986), studies of Early Iron Age society have relied almost exclusively on incomplete and unstratified burial assemblages (Whitley 1991a; 1991b), or the results of single-site excavations (cf. Nowicki 1999a; 1987a). Survey in the Kavousi region was an opportunity to explore in detail an Early Iron Age cultural landscape and to reconstruct the scope of its economy and resources. The settlement history integrates information from sites of various periods and functions, in conjunction with an unparalleled continuously stratified ceramic sequence (Mook 2004) and accompanying faunal and botanical data (FlintHamilton 2000; 1999; Klippel and Snyder 2000; 1991; Snyder and Klippel 1999; 1996). The regional information from the survey will eventually supplement the publication of data from the excavations of Vronda and Kastro and nearby settlements of Katalimata and Chalasmenos, permitting a detailed reconstruction of Early Iron Age communities in East Crete, and contributing to on-going and future synthetic studies across the island (Borgna 2003; Wallace 2003; 2001; 1997; Nowicki 2000a; 1999a). This transition from Bronze Age to Iron Age is an important focus of this study, especially because historical models and broad cultural generalizations for the Aegean have traditionally stressed the significant, but not unproblematic, evidence for demographic change; military conflict; foreign invasions; internecine strife and interregional conflicts; and systems collapse (Karageorghis and Morris 2001; Hayden 2001; 1988; Nowicki 2001; 2000a; 1999a; 1996; 1995; 1994; 1990; 1987c; 1987b; Rutter 1992; Snodgrass 1991; 1990; 1987; 1971; Schachermeyr 1979; Desborough 1973; 1972; 1964;

INTRODUCTION

Pendlebury, Pendlebury, and Money-Coutts 1937– 1938b, 136–141). Linguistic, ethnic, and cultural transformations involved not only the decline and disappearance of palatial economic and political systems, but a myriad of complex causal factors related to historically documented catastrophes. Disturbances in the eastern Mediterranean manifested themselves in changes and discontinuities in the archaeological record (Nowicki 2001; 2000a). Recent archaeological approaches to culture change in this period have, however, also tended to see cultural processes as responses to a changing Mediterranean economy and as positive adaptations to economic, social, and political conditions after the dissolution of the Bronze Age palace-based systems (Borgna 2003; Haggis 2001; 1999b; Foxhall 1995; cf. Snodgrass 1987; Betancourt 1976). These approaches tend toward explanations of how staple-based economies developed and functioned in regional environments. What, they ask, was the relationship of Early Iron Age settlements to neighboring settlement systems, territories, and the coastal mercantile centers that were based not on state-controlled exchange but on local merchant interests in expanding trade routes in the Mediterranean? (Sherratt and Sherratt 1993). The Kavousi-Thriphti Survey takes this same approach to the Early Iron Age settlements at Kavousi, addressing issues of adaptation to the local environment that provided its resource base (cf. Borgna 2003; Wallace 2003; 2001). We examine the nature of intersite relationships on various spatial scales within the survey zone and ask how these relationships might reflect the long-term development of social, political, and economic systems following the dissolution of Bronze Age palatial organization and leading to the rise of poleis in eastern Crete. Analysis of culture change in the Early Iron Age was thus at the core of the research design, as much as the interest in exploring the nature of land use and the sociopolitical dimension of the territory of the Early Iron Age settlements. The small Early Iron Age communities formed a physical context and test case for evaluating the discontinuities in the use of settlements and cemeteries, the diverse forms and functions of cult places, and the idea of regional identity in early Greek society on Crete (cf. Foxhall 1995; Morgan 1993; Coulson 1986). The Early Iron Age is seen as a formative period in the emergence of regionally integrated social and economic systems.

5

Similarly, Prepalatial Crete has long been the focus of discussion of incipient complexity leading to palace societies (Day and Wilson 2002; 1994; Driessen 2001, 53; Schoep 1999; Watrous 1994; Wilson 1994; Wilson and Day 2000; 1999; 1994; Branigan 1991; 1988a; 1988b; 1984; 1968; Cherry 1984; 1983a; Warren 1987). Thus a series of problems, parallel to that of the Early Iron Age, took shape in our analysis of the Bronze Age settlement patterns. Among these issues is the question of emerging complexity and the processes involved in early state formation in the late Prepalatial and early Protopalatial periods. The KTS recovered late Prepalatial settlement patterns that might indicate aspects of organizational structure in EM III–MM IA and the early Protopalatial period (Haggis 2002; 1999a). Recognition and recovery of late Prepalatial remains on the surface of sites proved, however, to be complicated by the density of EM I–II or Protopalatial occupational phases at these sites. The difficulty of this task was compounded by lack of well-published, substantial, and clearly stratified EM III–MM IA pottery in Crete, especially in the eastern part of the island. Nevertheless, the recovery of this phase in intensive survey (and excavation), and in diverse contexts (ritual, mortuary, and habitation), is absolutely essential to our understanding of the processes of state formation, or evaluating evidence for changing social and political systems around 2000 B.C. (Haggis 2002; Driessen 2001; Schoep 1999; Manning 1994; Watrous 1994; Tsipopoulou 1992; 1988; Dabney and Wright 1990; Cadogan 1988; 1986; Peatfield 1987; Cherry 1983a). The essential problem in most approaches to the emergence of palatial society is perhaps less in the richly varied theoretical sphere, than in the realm of data and chronology. Past studies drew primarily on EM II material as representing the formative cultural phase, or extrapolated from the Protopalatial period or worse yet from LM I palatial society and organizational structure (cf. Day and Wilson 2002; 1998; Knappett and Schoep 2000). The purpose of our work at Kavousi was to examine this crucial transitional phase in late Prepalatial Crete on a regional scale. Another related focus of the survey was on the development of Neopalatial settlements. Does the evidence suggest a continuous and uninterrupted expansion of sites and growth of population through the Protopalatial period into the LM I phase, or a discontinuity implying

6

KAVOUSI I

abruptly changing organizational structure that might be the result of extraregional systemic changes (Driessen 2001)? The study of the emergence of palace centers and territories and the nature of the geopolitical organization in eastern Crete requires examining the landscape on multiple spatial scales. This region of the island presents interesting problems in deriving a meaningful structure from apparent material patterns. Traditional notions regarding site hierarchies, center-periphery relationships, and palace-hinterland integration may not be uniformly applicable in all areas of Crete (Bennet 1990; 1987; 1985; Branigan 1972) or in all periods of the palaces; monolithic views of palatial organization and state structure are being eroded by new approaches (Cunningham and Driessen 2004; Schoep 2002b; 2001; Vansteenhuyse 2002; Driessen 2001; Cunningham 2001; Knappett and Schoep 2000). Neopalatial palace centers in eastern Crete such as Zakros, Petras, and Gournia occupy coastal locations and insubstantial if not marginal agricultural zones. Furthermore, they have idiosyncratic plans and relationships to urban centers. The distribution of urban centers in the Mirabello region is on the whole very dense, while territorial relationships between such sites are ambiguously reflected in the archaeological evidence. Finally, defining the components of site hierarchies is highly problematic, as is the lack of easily definable topographical boundaries and catchment areas, or demonstrable territories for palatial centers (Tsipopoulou 2002; 1997; Tsipopoulou and Papacostopoulou 1997; Bennet 1990; Cherry 1986). In this context, the survey at Kavousi is a contribution to the evaluation of issues of palatial organization in East Crete. The role of the palace at Gournia, still controversial in its attribution, is central to our assessment of the sociopolitical or economic organization in the north Isthmus, Mirabello Bay area, and East Crete (Soles 2002; 1991; Watrous and Blitzer 1999; Branigan 1972). The town centers of Mochlos, Pseira, and Priniatikos Pyrgos must also be explained in terms of their territories, extraregional connections, and relationship to each other and to the palace at Gournia. In the Kavousi area, the Neopalatial towns of Tholos and Hagios Antonios were probably linked to these same Mirabelloarea centers, as well as to the overland Isthmus route, and thus to broader regional sociopolitical systems, economic spheres, and site hierarchies. The

meaning of local patterns and our interpretation of organizational structure in the landscape, however, will ultimately be affected by the results of surveys in the Mirabello area and in Crete in general (Hayden 2004a; 2004b; 2003a; Branigan 1998; Moody, Rackham, and Rapp 1996; Moody 2004; 2000; 1987; Haggis 1996a; Watrous 1982; Watrous et al. 2000; 1993; Hayden, Moody, and Rackham 1992; Müller 1998; 1996; 1992; 1991a; 1991b; 1990; Tsipopoulou 1989; Nixon, Moody, and Rackham 1988; Blackman and Branigan 1977; Nowicki 2004; 2000a; Schlager 1991; Whitley et al. 1999; 1995). Nevertheless, the patterns recovered in Kavousi can tell us about changes in land use preferences; exploitation of local resources of water, land, and physical terrain and natural routes; and site types, functions, and hierarchies coinciding with broader regional and island-wide changes. The survey looked at the long-term history of the region as settlement patterns on various geographic scales. The “region”, as defined by the survey boundaries, is presumed to encompass a realm of human activity shaped by local environment and organizational structure, as well as extraregional economic and political systems. Thus, the concept of hierarchical structure in settlement patterns—as an indication of social or economic organization—was examined on multiple spatial scales, necessarily encompassing the wider region of Mirabello and eastern Crete. The results suggested that settlement patterns might be the manifestation of two primary agents: resource availability and site functions. The examination of contrasting settlement patterns in different environments, and evidence for site interaction between and within topographically and environmentally distinct areas of the survey zone, led us to design probable organizational models. The importance of these small-scale patterns of activity is that they might help explain society on a local level—a local structure that is potentially complex, informing the broader regional picture. The survey focused on distinctly nonhierarchical patterns of settlement that could be meaningful in reconstructing this local social structure and economic system. The issue of scale was recognized as essential in interpreting settlement patterns of different periods (cf. Alcock and Cherry, eds. 2004; Cherry 1999). Microregional differences in terrain, topography, and natural resources, and changes in site functions

INTRODUCTION

through time are seen as variables affecting settlement patterns. If material patterns of human activity are culturally and environmentally dictated, then the functions of sites, their immediate environments, and their potential interrelationships will have an effect on the larger regional pattern—the spatial structure that is used to define aspects of societal complexity. A site-size hierarchy may then exist in the archaeological record on a regional scale precisely because of cultural and environmental conditions on the local level. The meaning of settlement patterns lies not merely in the evidence of complexity by virtue of an evident hierarchy, but in the specific functions of the various components of that hierarchy (or other relational configuration) that are discernible at various scales (Crumley 1987; 1979; 1976; Crumley and Marquardt 1987; Ehrenreich, Crumley, and Levy 1995; cf. Keswani 1996; Schoep 2002a). Thus, one problem of the Kavousi survey was to consider the meaning of patterns on the local level that affect our interpretation of the settlement history of the broader Mirabello and north Isthmus regions. In this respect the present project contributes to the cumulative picture to be derived ultimately from the results of the contiguous Vrokastro and Gournia surveys, as well as further afield (Hayden 2004a; Driessen 2001). The eventual aim of work in the Mirabello area should be not merely to compare the results of various surveys but to integrate the detailed patterns of settlement toward an understanding of cultural systems and organizational structure in a number of environments. The survey considered the effects of external, primarily economic, influences on settlement in the survey zone. The position of Kavousi (on the Bay of Mirabello, at the farthest northeastern point on the Isthmus of Ierapetra, and at the main mountain passage into the Siteia Massif and eastern Crete) is at a confluence of natural communication and transportation routes, and is therefore susceptible to a series of external variables and economic influences that could have affected both settlement development and site interaction. The chief evidence available in the archaeological record for exchange activity between regions is pottery. Bronze Age ceramic production and distribution in the Kavousi and Mirabello areas are important indications of economic activity. Fragments of large, coarse utilitarian vessels used for storage, transport, cooking, and other processing

7

functions are ubiquitous in survey samples in Kavousi, and their shapes, fabrics, and surface treatments are chronologically diagnostic and regionally distinctive. While the vessels themselves may have been exchangeable commodities, they may also be containers, standing proxy for the movement and distribution of produce in the region. The goals of the ceramic study conducted during the survey were four: to record the distribution of coarse-fabric types as a means of generating a complete site chronology within the survey zone; to discern differences in production and distribution between regions in different periods; to trace the extent of ceramic distribution in the north Isthmus area; and to suggest economic or political motives or structures behind the evident distribution. The project explored the possibility of determining social or political structure in the landscape by comparing presumed or hypothetical spheres of economic influence with various observable site hierarchies, and with discernible spheres of ceramic distribution. The survey of the Kavousi area continues the research begun by Harriet Boyd in 1900. While methods have changed and knowledge of artifacts and contexts has improved, the aims of her and our studies are much the same. Indeed, it could be argued that the work conducted in the Mirabello region from 1900 to 1915 provided a conceptual focus for a new era of research. These early excavations and surveys have essentially defined the research universe, the sample area, and have drawn the notional boundaries of the regions that we try to define in cultural terms. In many ways this extensive previous knowledge has been as much a hindrance as a help, biasing the approach to the region by presenting such a detailed picture of the archaeological landscape. The preeminence and obtrusiveness of excavated sites—especially those examined at the turn of the century—have caused a scholarly predilection for a certain view of the landscape, one codified by publication and conservation. Indeed, the vagaries inherent in early excavations, often random responses to impressions of the archaeological landscape, have helped to create a new systemic context of the “published site” and its associated region. The Kavousi-Thriphti Survey has had to struggle with the evidence produced by past archaeologists, a stratigraphy of research that is as convincing in its clarity of description and

8

KAVOUSI I

simplicity of results as it is biasing in its definition of political centers and cultural regions. The survey provided a new context for assessing evidence for human activity in the Bronze Age and Early Iron Age with a view to understanding intersite relationships on various spatial scales. The goal of the survey was to understand the exploitation of the local environment and the effects of broad regional and island-wide political, economic, and cultural systems on local settlement patterns. While conforming to the specific research aims of the investigation of the LM IIIC and Early Iron Age sites of Vronda and Kastro, the survey encompassed a much broader chronological and spatial scope than that provided by the excavations. This included the northeast Isthmus of Ierapetra and the eastern coast

of the Bay of Mirabello. The project documented archaeological sites in the region in order to reconstruct the history of settlement from the Final Neolithic period (c. 4000 B.C.) until the Late Roman period (6th–7th century A.D.). While the spatial focus of the survey is the study area, the “region of Kavousi,” this is clearly neither a complete cultural region nor political territory in all periods. As a unit of landscape with its own resources, it is as diverse economically as it is exposed and receptive to a myriad of external influences. To reach an understanding of the sites and their interrelationships we will eventually need to connect this material to local stratified ceramic sequences, site functions, and broader spatial patterns in the Mirabello and north Ierapetra Isthmus region.

2

The Physical Landscape

This chapter describes the topography, terrain, hydrology, geology, and soil of the Kavousi area and serves as an introduction to the detailed scientific study of geomorphology, vegetation, and sediment development given by Michael Morris in Appendix 1 (“A Pedological Investigation and Soil Survey of the Kavousi Region”), a component of a larger study of soils in eastern Crete (M. Morris 2002). Analysis of local vegetation, modern land use, and recent and modern settlements has an important place in archaeological studies, in that it contributes to the overall picture of the physical landscape and provides a basis for reconstructing past human activity and responses to environmental conditions

(Moody 2004; Moody and Rackham 2004b; M. Morris 2002; Timpson 1992; Moody, Rackham, and Rapp 1996). In the Kavousi region, the physical landscape has been dramatically transformed as a result of political and technological changes since the turn of the 20th century and economic changes since World War II. These changes in the local environment have altered patterns of land use, the nature of settlement, and human-landscape interaction. Nevertheless, by examining the recent and current landscape and modern agricultural and pastoral practices, we can begin to construct an account of human activity in the region, while reconstructing aspects of the ancient environment.

Situation and Features The survey zone of 33 square km is comprised of the modern koinoteta (parish or district) of Kavousi (Figs. 1–2). It is situated on the eastern edge of the Bay of Mirabello in northeastern Crete.

While regional elections in 1999 dissolved the political function of the koinoteta, placing the area and its villages within a larger geopolitical division of the demos of Ierapetra, the village community

10

KAVOUSI I

nevertheless retains its geographic identity and many aspects of its local government. The area includes the northeastern extension of the Isthmus of Ierapetra, which connects the peninsula of East Crete to Central and West Crete, and the northwestern edge of the Siteia Massif (the Kavousi Mountains or Thriphti Range) (Figs. 1–2). The island of Pseira lies about 2.5 km off the coast in the Bay of Pseira, to the north of the study area, and the promontory of Mochlos is about 8 km to the northeast (Fig. 2). The Early Minoan–Middle Minoan site of Chamaizi is about 18 km to the east of Kavousi; Vasiliki is some 6 km southwest; and Pacheia Ammos (6 km), Gournia (8 km), Vrokastro (10 km), and Priniatikos Pyrgos (12 km) all lie to the west along the Bay of Mirabello. The lowlands are dominated by a string of coastal hills—Mounts Schinias, Chomatas, and Chalepa—running roughly north-south along the eastern edge of the Bay of Mirabello and ranging from 138 to 225 m in height (Fig. 2; Pls. 1B, 2A). Between these hills and Mount Kapsas is the main Kavousi plain or “Kambos,” which is composed of the fields of Tholos, Kambos, Hagios Antonios, and Chordakia (Fig. 2). At the northern edge of the plain is the small alluviated valley and bay at Tholos (Pl. 2B). The inlet is used today as a roadstead for fishing boats. Southwest of the modern village are the fields of Evraïka and Kamina, which are a mixed alluvial formation, bordered on the east by the high cliffs and talus slope of Mount Papoura, and on the west by the southernmost of the coastal hills, Mount Chalepa. The Kambos, beginning in the north at Tholos, terminates in the south at Kamina, where it merges with the northeastern Isthmus of Ierapetra, thus creating a northward extension of the isthmus. The village of Kavousi sits on a high (100–150 m), stable alluvial ridge rising above and south of the Kambos at the confluence of two main drainage basins in the Kavousi Mountains: the Thriphti watershed to the south, and the Avgo-Bebonas watershed to the east (Fig. 2; Pl. 1A). The position thus also commands two important river catchment areas in the west Siteia Massif, and the most frequently used transportation routes into eastern Crete (Fig. 1). Furthermore, the village sits at a natural transition between the Isthmus of Ierapetra and the Kambos Plain, and between the Kambos and highlands of Papoura and Avgo. Some three-and-a-half kilometers southwest of Kavousi is the Cha Gorge, the

precipitous outlet of a third river that drains from the Thriphti watershed southwest into the Plain of Kamina and the northern part of the Isthmus of Ierapetra (Pl. 3A). The Kavousi Mountains rise sharply south and east of the modern village (Fig. 2; Pls. 1A, 3A–B). They consist of Thriphti, with a peak of 1,476 m; the Bebonas Range, rising to the peak at Kliros (1,334 m), and the Orno Range, northwest of the village, with peaks at Kapsas (1,002 m) and Askordalia (1,238 m). These highlands form the western edge and beginning of the Siteia Massif, which extends across the entire east end of the island (Figs. 1–2; Pl. 3B). Kavousi is centrally located within the koinoteta at the intersection of three major traditional transportation routes. The first and most important passage into East Crete is the Avgo Valley, which bisects the Thriphti Mountains at their north end. The route climbs up the deep gorge east of Kavousi, rising to the Bebonas watershed immediately above Avgo (Fig. 2; Pl. 4A), and then continues in an eastward direction, south of the Orno Mountains to the villages of Bebonas and Chrysopigi (Roukaka); here it joins two other routes, one branching south of Chrysopigi, by way of Lapithos and Stavrohori to the sea at Koutsouras, and the other traveling eastward to Siteia and the east Diktaian Range (Fig. 1). The Avgo Valley is the main passage to East Crete from the Isthmus of Ierapetra. The second route climbs immediately south of Kavousi village up the northeast edge of Mount Papoura and along the Thriphti drainage to the villages comprising the region of Thriphti-Aori: Mylonia, Drakalevrion, Tsamanti, and Platania (Aori). From the village of Platania a route traverses the valley between Mount Thriphti and Bebonas and extends east to the mountain villages of Orino and Stavrohori. A third route (Figs. 1–2) climbs east-northeast of the village along the west contours of Kapsas and continues east, high above the coast, on the northern spurs of Orno to the villages of Lastro, Sfaka, Tourloti, and Mouliana, and on to Chamaizi and Siteia. It is important to emphasize that the main natural routes accommodated by the topography went through Avgo and Papoura; the routes were paved in Venetian or Ottoman times with a kalderimi (paved or partially paved mountain road; cf. Rackham and Moody 1996, 156–157). The third route, around Kapsa to the north coast, is the path of the modern

THE PHYSICAL LANDSCAPE

asphalt highway. It is neither a traditional road nor a likely ancient passage. The routes described above are like spokes on a wheel extending out from a hub at Kavousi. The position of the village is of considerable geographic importance, lying as it does in the northeastern extension of the Isthmus of Ierapetra and at the northwest edge of the Siteia Mountains, at the gateway to the mountainous regions of Avgo and Papoura. While several major transportation routes converge in precisely this area, two deserve discussion here. One is the north-south passage down the Ierapetra Isthmus from Pacheia Ammos to Ierapetra (Fig. 1). Only 12 km in length, the passage is the shortest distance between the north and south coasts of Crete and has the additional advantage of being relatively level (the high point is the watershed at Episkopi). A western route (Fig. 1) runs along the southern edge of the Bay of Mirabello, then crosses the Isthmus route and continues westward to Gournia and Istron. This east-west coastal route connects the Mirabello

11

Bay with the main passages into East Crete from Central Crete, at modern Vrahasion (from the Malia Plain and the north coast) and at Kritsa-Lato (from the Lasithi Plain). In addition to internal communication routes, the Kavousi area has two ports providing direct access to the Bay of Mirabello. One is at the north edge of the Kambos at Tholos and the Bay of Pseira (Fig. 2; Pl. 2B); the other is Agriomandra (Hagiomandra or Hagios Ioannis), which is a small natural roadstead situated northwest of Kavousi, directly on the Bay of Mirabello between the hills of Chalepa and Chomatas (Fig. 2) and directly south of Chrysokamino. Agriomandra was used as a commercial port and access to the Aegean before World War II. Submerged deposits of amphorae testify to its use as a harbor during the Roman period, and the cavechapel in the cliff above the cobbled path leading to the port suggests its continued use during Byzantine and Venetian times.

Bedrock Geology and Soil The geology of the Kavousi area (Fig. 3) consists primarily of two broad zones (Morris 2002, 3–5; 48–50; 61–62): (1) the late Paleozoic and early Mesozoic mountains of the northwest Siteia Massif (Kavousi Mountains or Thriphti Range) and coastal hills along the west side of the Kambos plain, and (2) the Quaternary or Neogene alluvial sediments of the lowlands (Morris 2002, 61–62; Rackham and Moody 1996, 15–18; P. Day 1991; Durkin and Lister 1983, 91–96; I. Papadakis 1985; Papastamatiou et al. 1959). The uplands immediately surrounding Kavousi village are mixed bands of Permian-Triassic phyllite and dolomite. Phyllite is also exposed below Xerambela and along the ridges to the east of Vronda as far as the Kastro and Skouriasmenos (Pls. 4B, 5A). Vronda itself is composed of dark gray dolomite limestone overlying yellow, gray, and green schist. The Thriphti Range, south and southeast of the Vronda area, is composed principally of late Paleozoic carbonate rock. Northeast of Kavousi and also farther south, the large west front of the range at Kapsas and Papoura is composed of bluish (gray-blue) crystalline limestone (sideropetra) of

Permian date. The Thriphti Valley to the east of Papoura consists of Permo-Triassic phyllite, which varies in color from gray, dark gray, to green and maroon; these phyllite rocks are mixed with sandstone, quartzite, conglomerate, gray limestone, and dolomite (Durkin and Lister 1983, 91–96; Papastamatiou et al. 1959). The Cha and Platys river drainages (Pl. 3A), that originate in the Thriphti watershed, south of Kavousi near Aori (Fig. 2; Pl. 3B), are deeply incised valleys. Slopes on both sides of the Thriphti watershed are steep, and streambeds are in many places nearly vertical eroded cliffs. This is especially the case southwest of Thriphti village (Platania), where the main valley feeds into the Cha Gorge, a steep, 600 m deep, V-shaped slice into the mountain fronts of Papoura and Lamia (Pl. 3A; Rackham and Moody 1996, 23, 25–27). On the northern side of the watershed, the valley descends from Thriphti and rapidly becomes a precipitous gorge on the eastern side of the Kastro and Skouriasmenos (Pls. 4A–B). The Avgo Valley, just east of Kavousi, joins the north Thriphti drainage, at a point called Makellos or Pachlitzani Agriada (Pl. 4A); water draining from

12

KAVOUSI I

these two valleys feeds the Platys River, which runs northward through Kambos to Tholos Bay (Fig. 2; Pl. 2B). At the watershed above Avgo (Fig. 2), the steep valley is not deeply cut, permitting the alluvial sediment to be transformed into agricultural terraces. These terraces are built into the hillsides surrounding the valley on all sides and climb eastward straight up the steep slope to the Bebonas watershed. In the Thriphti drainage (Pls. 3B–4B, 5A), the terraces cling precariously to the thin veneer of even steeper hill slopes of 40–45 degrees (Rackham and Moody 1996, 141–142). Alluvial lowlands lie north of Kavousi village in the Kambos (Pls. 1A, 5B, 6A) and southwest of the village in the Isthmus area at Evraïka and Kamina (Morris 2002; 1994; Rackham and Moody 1996, 31). A long fault running northeast to southwest through Kavousi and forming the west faces of Mounts Kapsas and Papoura divides the plain from the mountains (Rackham and Moody 1996, 23, 25–27). To the west, on the west border of the plain of Kambos is the line of hills that form the steep (30–45 degrees) east coast of Mirabello Bay (Fig. 2). These hills (Mounts Schinias, Chomatas, and Chalepa) are another thrust front running northeast-southwest into the north Isthmus below Kavousi (Pls. 1B, 2A). The southernmost hill, Mount Chalepa, slopes gradually into the Isthmus lowland, where the Cha River drains from Thriphti into the bay at the east end of Pacheia Ammos. The bedrock of the hills is crystalline limestone of Permian date that alternates with Triassic gray

dolomite and Permo-Triassic phyllite (Fig. 3; Pls. 1B, 2A, 6B, 7A). The east slopes of the hills are gradual inclines, the steepest being less than 30 degrees (Pl. 7A). The west side of the front, however, facing the Bay of Mirabello, consists of very steep talus slopes and vertical cliff faces (Pl. 2A). The soils and sediments of the Kavousi area are of several types: bedrock soils, alluvium, stream deposits, and anthropogenic deposits (man-made terraces and fields) (Morris 2002, 7–21; cf. Butzer 1982, 43–66; Vita-Finzi 1978, 71–82). Bedrock soils are thinly layered on the peaks and ridges of the coastal hills and the upland slopes of Thriphti and Papoura; the soil depth is usually less than 0.10 m. Alluvial material is unconsolidated soil and rock transported to its present position by means of gravity; within this category are also soils that have moved downslope by water and wind. The present condition of the alluvial veneer is a product of both human activity and natural processes: the human processes include agriculture and pastoral activity and involve both slope degradation and slope conservation; the natural conditions and processes that contribute to the formation of alluvial soil include bedrock composition, slope percentage and distance, slope aspect, rainfall, and drainage (Rackham and Moody 1996, pp. 18–25). Stream deposits are sediments generated by river action and deposited in valley bottoms and river beds; included here is the material deposited in the bed and on the banks of the Platys River.

Bedrock Terrain and Erosion Exposed bedrock terrain is most obvious in the uplands south and east of Kavousi. Cliffs are exposed along the west faces of Mounts Kapsas and Kliros, in the gorge on the eastern side of the Kastro, and along the west and north faces of Mount Papoura. West of Kavousi, on the coastal hills of Schinias, Chomatas, and Chalepa, intensive browsing by sheep and goats on the dolomite and crystalline limestone areas has created environments where the terrain is either exposed or thinly covered bedrock (Pls. 1B, 6B, 7A–B). The soil veneer on the mountain ridges and peaks throughout the region is usually very thin, and pockets of

stony soil, only a few centimeters thick, are interspersed among barren outcrops of bedrock. On steep slopes, 30 degrees and greater, small amounts of colluvial material may gather to support sparse grass and garigue plant communities. The west faces of the coastal hills are essentially cliff terrain, with steep slopes (> 25 degrees) or nearly vertical inclines that hold loosely consolidated talus debris. Where the talus material contains some silts, patches of garigue plants may take hold and even flourish. On the east slopes of the hills, the incline can be more gentle (15–25 degrees). The ridges and shoulders of the hills are nevertheless

THE PHYSICAL LANDSCAPE

exposed bedrock or contain only thin surficial deposits. On the slopes of the limestone and dolomite hills, colluvial sediments are too thin to support any agriculture and show no evidence of recent terracing or slope preservation. These hills are used today for winter pasture, which hastens the erosion process on these soils. Where the parent bedrock is phyllite (schist), however, on the north and south ends of Mount Chomatas, the pattern of erosion and land use is very different. Modern terracing in these areas is extensive, and the terrain is stepped, with and without retaining walls (Pls. 5B, 7A). On older terraces that are now in disuse, no walls are evident but the colluvium remains surprisingly stable. New terraces at Hagios Antonios, on the northeast side of Chomatas, are built for olives, and only a few stone walls support the stepped landscape (Pl. 7A). The depth of the sediments on the phyllite hills varies from one to three meters, and the shoulders and slopes collect substantial and deep silty clays with few rock inclusions. Thus, in both terraced and unmodified areas of the schist outcrops, colluvium is deeper and more stable than on equivalent slopes of limestone and dolomite. This is a result of two concurrent processes. The first process is land use; the second and perhaps more important process is the pattern of erosion of the parent bedrock. Traditionally, phyllite slopes are favored for rain-fed agriculture because of the water-retaining capacity of the silts. Limestone terra rossa soils, on the other hand, are stony and have a higher evaporation rate, so they are favored for herding rather than agriculture. The nature of the parent bedrock compounds the effect of land use (Rackham and Moody 1996, 18–24). The limestone and dolomite hills (given equal slope, aspect, environment, and vegetation) erode faster than the phyllite hills and form less stable slopes and, thus, a greater area of exposed bedrock (cf. Pls. 5B, 7A). On very gradual slopes and on hill shoulders (at Mounts Chalepa, Schinias, Vronda, and Azoria) the colluvium is brecciated material of large to middle-sized rocks (cobbles and small boulders) in a silty clay matrix. On more steep slopes, water action washes out the sediment, leaving cobble- to boulder-sized rocks, often supported by patches of scrub. On slopes greater than 20 degrees or so, angular stones collect on the backslopes and create massive talus slides or individual

13

gullies. This is the case on the west face of the two fault fronts at Mounts Kapsas and Papoura (Pls. 3A, 6A) and along the west face of the coastal hills; a similar colluvial development exists on the northeast side of Mount Papoura and above Kavousi village at Panagia Skali and Azoria. At the two latter locations the hills are extensively terraced, but the slopes are today very unstable because of prolonged disuse. The steep west slopes of Mounts Kapsas and Papoura show the most extensive talus formation (Pl. 3A). This scree consists of large angular stones of various sizes, mostly loose and uncemented. In the fields of the north Isthmus, the deposited stone debris is deeply stratified and boulders (house-size) have been noted as far as one kilometer from the Papoura mountain front. At the foot of Mount Kapsas, the loose talus debris appears to alternate with stones in a silty matrix supported by patchy scrub plants. Near Tholos and in the area of Lakkos Skaphes, the Kapsas debris is cemented and even consolidated into a conglomerate rock face, indicating a slow depositional rate rather than rapid and recent sheet erosion (Pls. 2B, 6A). In general, in the Kavousi area, the limestoneand dolomite-generated sediments are difficult to cultivate because of their high rock content, shallowness, and instability due to sheet erosion. Terracing in these areas is especially labor-intensive (Pl. 7B). Conversely, the phyllite hills of Mounts Thriphti and Chomatas are extensively terraced, and the patterns of erosion, vegetation, and land use are very different (Rackham and Moody 1996, 29–30). The phyllite schists are soft, fractured rocks that allow rainwater to percolate and weather the rock gradually in situ. As a result, these slopes have less run-off, and less of their veneer has been lost. Phyllite sediments are deep and generally more stable than in similar situations on limestone hills. The contrast is especially evident in areas where there are no obvious signs of recent terracing. The phyllite-generated soils show evidence for the greatest constancy and density in human settlement and land use. Deep-weathered phyllite sediments are evident at Hagios Antonios, Mount Chomatas, and on the Kastro, where deposition on the backslopes is as much as 3 m deep (Pls. 4B–5B, 7A). At Thriphti Argira (site 97), a bulldozer revealed Minoan house walls and an upright pithos covered by a slope-wash layer of phyllite soil (over 1.5 m deep), superimposed on a 0.5 m layer of terra rossa (Pl. 8A).

14

KAVOUSI I

The fact that phyllite-derived soils and sediments are easier to terrace and cultivate than limestone soils is, as noted above, not the only reason they might be preferred for agriculture: probably more important, in this region, is the fact that phyllite bears silty soils, which hold more plant-available moisture than equivalent limestone-generated sediments (cf. Rackham 1972, 286; M. Morris 2002, 7). Climate, groundwater level, stone content, and aridity are the overriding factors determining the longterm use of phyllite soils, even if in other respects the terra rossa soils of the plain would seem to be a better choice. The cultural evidence—the modern and ancient preference for phyllites—thus contrasts with the results of the soil study (Appendix 1), which indicate that the fertile limestone-based alluvium in the plain would be of optimal value for agriculture, if irrigated.

The alternating limestone and phyllite barriers imply environments highly conducive to spring formation (Rackham and Moody 1996, 30, 42). Each of the phyllite-limestone divisions in the survey zone (Hagios Antonios, Chordakia, Kavousi village, Vronda, and Avgo; see Pls. 1A, 5B, and 7A) has seasonal springs, permanent water supplies, or evidence of once-perennial springs or cisterns. Although modern agricultural techniques, new irrigation practices, changing agricultural interests, and an increased population have lowered the groundwater level and affected the number of perennial springs currently in the Kavousi area, it is possible to reconstruct where a number of perennial springs were located.

River Deposits and Alluvium The lowlands of Kavousi are made up of fluvial sediments and mixed alluvial deposits. The sediments are variable, and different areas of the plain show varying patterns of development, land use, and vegetation. Well-developed flood plain sediments are not obvious in the Kambos, except perhaps in the northern part at Tholos at the mouth of the Platys River (Pls. 2B, 6A). This area may have considerable deposition, sufficient to obscure archaeological sites. The composition of the plain as a whole however, is the result of a combination of factors. Colluvial sediments from the coastal hills and Mount Kapsas, moving downhill and merging with streambed deposits from small gullies and torrents, appear to have covered older fluvial deposits of the Platys River (Pls. 5B, 7A). The down-cutting of the river shows the bedding to be thick (about 10 m) at the modern village of Kavousi, where there are stable alluvial deposits and steep gradients. This consolidated alluvial ridge at the village was the site of the first Neolithic settlement (site 24) in the region (Pl. 1A). The Platys river channel in the Kambos drops to about 2 m deep and levels out at Tholos, where sand and pebble deposits dominate the flood plain. In the central plain there is little evidence of substantial flood deposition. Sand, gravel, and rounded

water-worn cobbles, however, define a narrow strip (about 100 m wide) on either side of the modern stream channel. This stream deposit extends from the middle of the Kambos plain to the deltaic deposit at Tholos Bay. Sherds in this area are sparse and almost always water-eroded, indicating that fluvial deposits may have obscured sites and even redeposited material in the river channel. Study of alluvial deposition in the north Kambos has demonstrated conclusively that there were two distinct sequences separated by a gravel deposit (Appendix 1; M. Morris 2002, 61–63; 1994). An exposed section of a river scarp near sites 11 and 5, southeast of Hagios Antonios (Fig. 1), shows two intact depositional sequences (“Kavousi 2 Pedon”): an upper, young soil (gravelly sandy loam) void of artifacts and an earlier sequence, separated from the upper alluvial event by a lag gravel deposit. The earlier sequence begins about 0.61 m below the ground surface and shows evidence for soil development (M. Morris 2002, 67–72; 1994, 242–261); human activity is suggested by the presence of Neopalatial potsherds (not water-worn). A radiocarbon date of 3040 +/- 90 y. B.P. (1407–1209 B.C. after calibration) provides a good corroborating terminus ante quem for the sequence in the Late Bronze Age. While the ceramic

THE PHYSICAL LANDSCAPE

evidence provides only a post quem date for the use of the surface, the radiocarbon date indicates that the first alluvial event should have occurred in the Late Bronze Age or earlier. Michael Morris speculates on the cause of the Bronze Age aggradation, postulating a period of destabilization of the slopes above Kavousi and in the Avgo Valley, perhaps a result of deforestation by human agency (M. Morris 2002, 73; cf. M. Morris 1994, 259). Indeed, the Middle Bronze Age settlement pattern shows the most extensive exploitation of the upper elevations and a continuous growth of settlement in the Avgo and north Papoura areas. Could overbrowsing and deforestation in the areas of Avgo, Papoura, and Xerambela have caused such destabilization of the slopes and encouraged sheet erosion in the upper catchment areas and watersheds of the Platys river drainage? One wonders whether the observable change in settlement patterns in the MM III through LM IA periods, with their decided decrease in the use of the upper elevations, may have coincided with the erosive event that caused the deposition of this earlier alluvium. The early soil was eventually buried, probably sometime after the Early Iron Age, by another alluvial episode. Morris suggests that the Late Geometric/ Early Orientalizing period abandonment of the upper elevations—and, indeed, the abandonment of agricultural terraces of Avgo, Xerambela, and Papoura—could have created conditions that led to another phase of destabilization, erosion, and deposition (M. Morris 2002, 74–75; 1994, 259–260). After the mid–5th century B.C., the Kavousi area experiences an almost complete abandonment until Hellenistic times (3rd century B.C.). This 5th–3rd century B.C. gap in habitation and land use likely is the phase of terrace and slope destabilization, and most probably the period in which the younger soil was deposited in the first alluvial sequence of the Kavousi 2 soil formation. While such evidence of depositional episodes in the Bronze Age and Iron Age might suggest that sediments of the later event now obscure Bronze Age sites in areas of the flood plain, the full spatial extent of such deep alluviation remains problematic (cf. M. Morris 1994, 257). Middle Minoan and LM I pottery has been found on the surface of sites even in the central area (sites 15–18) and at the fringes of the Kambos plain (sites 5, 11–12). Neopalatial architectural remains (house walls)

15

were found at site 27 in the south Kambos, and site 16, in the north, contained even a Final Neolithic component. Furthermore, the condition of many sites in the Kambos is poor, frequently lacking subsurface remains. It is evident that they are “surface sites”—sites unprotected by recent alluviation, which have been reduced to surface scatters by centuries of exposure to agricultural activities and natural transformational processes. The archaeological evidence would indicate that the Bronze Age and Iron Age/Classical alluvial sequences, while significant, did not obscure sites in an area much wider than that of the actual flood plain of the Platys river. Indeed, the swath formed by the overbank deposit from Hagios Antonios (between sites 11 and 15; and 5 and 17) to Tholos Bay, contained little or no prehistoric ceramic material. In general, this overbank stony sediment of the river in Kambos supports only sporadic garigue and maquis growth, but in recent years, attempts have been made to exploit the sands and silts for growing olives and planting in greenhouses. Such attempts, however, are usually ineffective without direct pipeline irrigation. A large part of the plain is composed of alluvium from the surrounding hills, and there is little evidence of recent river deposition extending beyond the overbank deposits mentioned above. The central plain consists of limestone terra rossa sediments mixed with gully deposits and colluvium from the hills on the east and west sides of the plain (Pls. 5B, 6A, 7A). The terra rossa sediments are extremely rocky with angular stones ranging in size from pebbles to boulders. The primary reasons for the high stone content of the plain sediments are the steep incline and rapid erosion rate of the bluish-gray limestone on the west face of Mount Kapsas. Cultivation is generally difficult in these lowlands and involves extensive field modification. Stones are removed from the soil during successive plowing periods and piled on field walls or gathered in rock piles on the field fringes. On gradual slopes at the east edge of the plain, the terrace walls are commonly linear accumulations of rubble or even massive stone heaps that have been built up during successive periods of field cleaning (Pls. 2B, 6A). Examination of the angular stony sediments of the plain shows them to be debris from adjacent hills, not fluvial material. The geological map calls these sediments “recent alluvium,” and while the rates of erosion and the

16

KAVOUSI I

extent of deposition are difficult to determine chronologically, there is no indication that there were events causing major sedimentation that might have completely buried or affected the visibility of most Bronze Age sites in the plain, with the exception of main river deposits mentioned above. Evidence in support of this is twofold. First, there is no indication, in the Kavousi area, of systematic deep plowing or soil removal or transportation that might have revealed ancient soils or alluvial events, thus exposing the existing remains of prehistoric settlement. Second, Bronze Age pottery and architecture is evident on the east side of the plain (even at the edge of the Kapsas scree) and throughout the Kambos (Pl. 6A). Significant in this respect is the MM–LM I site 25 in the middle of the Kambos area, and sites 17, 18, and 25, lying at the very base of the Mount Kapsas scree line. Here, wall segments and artifacts are situated on the surface of the site. Of considerable chronological importance is site 16, already mentioned, which produced Final Neolithic pottery. This site is only 200–300 m from the edge of the scree from Mount Kapsas, suggesting fairly gradual and very ancient deposition, at least on the east side of the plain. It is only on the west footslopes of Kapsas and Papoura and in the flood plain between Hagios Antonios and Tholos where sites might have been buried. The soil development of the Kavousi 2 alluvial sequence, mentioned above, is a case of such an occurrence (cf. Appendix 1). On the west side of the plain, colluvial sediments merge with the Platys river bed. These soils alternate according to the variations in the parent bedrock of the coastal hills. From north to south, the sediments are terra rossa (dolomite) at Tholos, merging with phyllite silts at Hagios Antonios. The phyllites alternate with limestone and terra rossa across the plain on the east side of Chomatas and Chalepa. The small pockets of phyllite sediments at Hagios Antonios and Chordakia are notable (Figs. 2, 3; Pl. 5B).

These are stable white-gray to yellow silty clays with few stone inclusions; in the marginal areas they are mixed with the terra rossa. These localized sediments (ca. 15 ha.) show evidence of continuous land use from the Bronze Age until present day. The high silt content and concomitant high water-retaining capacity of these soils, as noted above, make them quantitatively more valuable land for dry agriculture than the wide expanse of terra rossa in the central plain. While the terra rossa soils have equivalent or better nutrient values, their capacity for plant-available moisture is lower because of their high rock content (cf. M. Morris 2002, 72). In the plain of Kavousi, a direct correlation can be made between settlement and land use activity and sediment composition. The phyllite-generated sediments have greater numbers of associated sites and higher sherd densities than the terra rossa. The only exception is the red soil in the sinkhole at the center of Lakkos Ambeliou (Fig. 2). The Kavousi Project’s soil study has demonstrated that this is probably the best agricultural land in the Kavousi area for both its nutrient value and ability to retain plant-available moisture (Appendix 1). The density of Bronze Age habitation in Lakkos Ambeliou is no doubt related to the agricultural viability of this soil. The difference, however, in artifact counts between the white and red soils is important; the mean off-site density in phyllite areas such as Chordakia is some twenty times greater than in Kambos, while the on-site average is somewhat less, but still more than ten times greater. Furthermore, these areas are today favored for olive cultivation and small gardens. They have traditionally supplied the best barley and wheat yields in the Kavousi lowlands. The variable qualities of these different soils and sediments and their contexts have directly affected the history of land use and agriculture in the Kavousi area.

Terracing and Slope Conservation The extent to which terracing (ancient and modern) has affected the now-existing colluvial mantles is not entirely known. That ancient terraces existed is certain, and it is very likely that destabilized terraces in the coastal hills and upper elevations were

responsible for periods of alluviation in areas of the Kambos plain (M. Morris 1994, 257–260). Recent field work on the adjacent island of Pseira by Betancourt and Hope Simpson has revealed evidence for Bronze Age terracing as well as the damming of

THE PHYSICAL LANDSCAPE

torrent beds in order to gather silts for relocation on built terraces; a number of ancient terrace walls have been located, and excavation has shown use in the Bronze Age, and a complex system of land management (Betancourt and Hope Simpson 1992). In Kavousi, the most extensive terracing exists in the Vronda and Kastro areas (Pls. 1A, 4B, 5A). On the coastal hills, terracing is limited to the phyllite and mixed limestone-phyllite deposits (Mount Chomatas). In Hagios Antonios, on the southeast slope of Mount Schinias, there are several walls built of local dolomite blocks (Pls. 7A–B), and associated with surface pottery of LM I, LM III, and Roman date. One “Cyclopean” wall at this site was actually excavated by Harriet Boyd in 1900 and revealed no indication of return walls or floors. Boyd correctly called it an “ancient agricultural terrace wall” (Boyd 1901, 130–131). The walls no doubt belonged to a complex field system situated downslope from the Neopalatial houses. The associated pottery, although never published, was most likely Neopalatial, considering the extant surface remains and Boyd’s description of dark-on-buff spiral and floral decorations. While frequently difficult to identify, ancient terrace walls have been recovered during surveys in Gournia and Mesara (Watrous et al. 2000; Hope Simpson et al. 1995, 374, 398). Other walls in the Tholos plain, where the river undercuts Hagios Antonios and Schinias, may have been constructed to dam the Platys River to adjust water flow, volume, and direction, and perhaps silt deposition. Nearby are the large nucleated Neopalatial sites (5, 11, 15, and 17) in the north Kambos, which were situated on the fringes of fields directly on either side of the river, perhaps to maximize the potential of the streambed for irrigation. Although no walls have been recovered to justify such a reconstruction of water management, it is most likely that the Neopalatial period was the most active phase of terracing, river use, and irrigation in the lowlands. (cf. Hope Simpson et al. 1995, 355, 374–375, 396). Extensive terracing exists above the modern village in the Vronda and Kastro areas and east of the village at Avgo (Pls. 4B, 5A). While several areas of phyllite sediments at Mounts Thriphti and Chomatas are terraced without supporting walls, stone-built terraces dominate the steep slopes of Mount Papoura, and the areas around the Kastro, Aloni, Azoria, and Xerambela. The walls are of various

17

sizes (1–2 m) and support terra rossa and mixed phyllite/terra rossa sediments. The villagers of Kavousi have no recollection of transportation and deposition of soils behind these terrace walls. The terraces are built into colluvial mantles, often on very steep slopes of 20–45 degrees (Rackham and Moody 1996, 141). The morphology and composition of the terraces vary but usually consist of terra rossa and mixed phyllite/terra rossa silty matrix with dense angular rock inclusions; their form is either stepped or braided, although precarious “pocket” terraces exist as well (Rackham and Moody 1996, 141). Virtually every hill slope in the Kavousi and Thriphti uplands shows evidence of some terracing. The most extensively terraced areas are the mountainous limestone environments, probably as a result of the instability of their slopes. Terraces retain fine silty sediments and have a high water-retaining value (Butzer 1982, 150–151; Rackham and Moody 1996, 142; 1992; Appendix 1). Slope preservation for agriculture is the primary reason for terracing in the highlands, especially on the steep slopes of the Vronda-Kastro area. Some terrace walls are built into talus slopes for the purpose of stabilizing active slides and reclaiming agricultural land. On the footslopes of Mount Kapsas this has been accomplished with some success. Terraces examined at the base of Kapsas and above the Vronda hill appear to be composed of angular stones and loose talus debris; however, it is likely that these terraces have a thick rock pile front that supports lenses of sands and silty clays. Agricultural terraces have been excavated at Vronda and Kastro, and much is known of their construction. On the Kastro (east and west slopes), the terrace walls are built directly upon ancient house walls and architectural terraces. Excavation revealed moist silty clay sediments at the back of the terrace and against the bedrock face. Near the inner face of the terrace wall, the fill is gravel and sandy loam with large rocks ranging from cobbles to boulders in size abutting the wall’s inner face. The agricultural terraces were constructed directly on top of PG–EO house walls, while the pottery from the terrace fill was found to be predominantly Late Geometric and Early Orientalizing in date—probably debris from the hill top. On the west side of the Kastro, nearly 2 m of colluvial material was excavated. On the east slope, about 3 m of slope-wash debris filled the LG–EO houses after abandonment. While there is

18

KAVOUSI I

little evidence of terracing, the slopes (30–40 degrees) have remained surprisingly stable and cultivable. The date of the original agricultural terraces at Kavousi is unknown, but there is some evidence for construction or use in the Bronze Age, especially south and east of the modern village of Kavousi (sites 63–69) and in coastal hills at Hagios Antonios (site 5). At Chrondrovolakes (sites 68–69), Middle Minoan house foundations sit directly on top of agricultural terraces in use today, perhaps indicating early slope management for agriculture (Pl. 8B). On terraces in the higher elevations, Minoan sherds have been found densely distributed and in a condition that shows little of the wear characteristic of water erosion or slope friction. In situ sherd scatters such as sites 65–67 are perhaps remains of rural houses, while at Vronda (sites 73–76) the Bronze Age pottery might be evidence for early use or construction of the terraces for agriculture. The sherds on the Vronda terraces are generally smaller in size. This is a result of the annual working (hand-tilling) of the terrace sediments for vine maintenance. Further evidence of ancient terrace use is at Skala and Aloni (sites 78–79), where the tholos tombs of

the cemetery were built directly upon what are now modern vine terraces. In Avgo, multiple terraces of various periods indicate a long history of landscape modification. This is especially significant if one considers the rapid erosion rate of the steep limestone hills. Hillsides without built terrace walls, even with a gradual slope (such as Mount Chalepa), have lost much of their colluvial veneer (Pls. 1B, 6B). The decay of terrace walls would create, over time, major sheet erosion, requiring continued maintenance of the slopes. Given the recent evidence from Pseira of Bronze Age and Byzantine landscape modification, it is reasonable to think that the ancient practice of slope conservation in this region has determined the stability of the terraces in modern times (Van Andel and Runnels 1988, 147). The lack of extensive Holocene alluvial deposits in the Kambos plain suggests a fairly continuous history of land use and slope maintenance since the arrival of the first inhabitants of Kavousi in the Neolithic period. The alluvial sequences—mentioned above in the context of the study of soil formation from the north Kambos plain—might, however, indicate Bronze Age and Iron Age– Classical episodes of slope destabilization.

Vegetation and Recent Land Use The vegetation of the Kavousi area may be divided into four categories: garigue and grasses, pine woodland, woodland of streambeds and gullies, and cultivated plants. The vegetation existing today in the survey zone is a result of large-scale land modification involving both intentional and unintentional changes in the landscape as a result of changing agricultural and pastoral practices. The most prevalent type of uncultivated vegetation is a low scrub or garigue with intermittent maquis, especially on terra rossa soils in the lower elevations (M. Morris 2002, 5–7; 1994, 163; Rackham and Moody 1996, 111–115; Zohary and Orshan 1965). These plant types are well-adapted to drought, the low soil depths of eroded hill sides, and high alkaline pH (M. Timpson, personal communication; Rackham 1972, 286). The garigue plants are of various sizes and most often less than one meter in height, the average being about half

a meter. The most prevalent types are faskomilo or sage, thyme (Thymus capitatus), thorny burnet (Sarcopoterium spinosum), and spiny broom (callicotome). Plants such as dense grasses and asphodel (Asphodelus microcampus) dominate many hillsides and the cultivated lowland, especially in areas unbrowsed and cleared for agriculture. The presence of dense grasses, particularly in phyllite sediments, is evidence of reduced grazing pressure in these areas. While maquis vegetation is variously interspersed with garigue and grass, there are few concentrations of evergreen shrubs of any height. On the south-facing slopes of Mount Chalepa, dense clusters of juniper and Pistacia lentiscus grow in abundance in the bedrock soil mantles; this is also the case along the west slope of Azoria and Panagia Skali where, in addition to juniper and Pistacia lentiscus, large quantities of wild oak (kermes oak) also flourish.

THE PHYSICAL LANDSCAPE

In gully bottoms and streambeds, deciduous and evergreen maquis plants grow undisturbed by agricultural or pastoral activities. In the gullies of the Avgo Valley, Mylonia (north Thriphti Valley), and at Vronda, there are scattered pines, oak, and wild olive. Avgo is a remnant woodland where plane trees, oak, olive, blackberry, pine, cedar, and cypress thrive both in streambeds and on alluvial terraces (cf. Patten 1908, 58). The southern part of the Thriphti Valley is dominated by pine forest, which is particularly abundant on the north- and northwest-facing slopes of Thriphti, south and west of the Thriphti-Aori watershed (Pl. 3B). Evidence for landscape modification in the Thriphti uplands is considerable, and the present pine forest is likely to be a relatively recent occurrence (Rackham 1972, 294; cf. M. Morris 2002, 12–13). The pastoral agriculture of the Kavousi area has traditionally required the cutting and burning of forest. This has ceased only gradually in the past three decades, with the irrigation of Kambos and the shifting of agricultural interest to the lowland plain. Even today, with reduced pastoral and agricultural pressure on the Thriphti Valley (Pl. 3B), the area remains mostly an alternating environment of agricultural terraces (vines and small gardens) and bare scrub land (pastoral). A brush fire in 1987, for example, destroyed much of the pine forest in the southwest Thriphti Valley and on the slopes of Mount Lamia where, later, in 1989, terraces were constructed for vine planting. This reuse of the highland forest areas (500–600 m) for agriculture is one phase of the cyclical processes of land use, extending back as early as EM III–MM I. On the northwest slopes of Thriphti, above Monastiraki and south of the Cha Gorge, agricultural terraces of uncertain date are now covered with pine forest. Intensive agricultural exploitation of the plain for tree and garden crops is a phenomenon beginning in the 1950s but rapidly expanding in the 1970s. The Kambos and Tholos areas are today densely planted with olives, small gardens, and some greenhouses (bananas, tomatoes, cucumbers); olives now dominate the lowland as the primary crop, and as in many areas of Greece, it is the basis of the local economy. Olive trees are cultivated intensively throughout the plain and in the upland terraces to an elevation of about 400 m and sporadically to about 700 m. Small gardens and vines are also dispersed throughout these groves. This intensive cultivation

19

of the lowland is a direct result of irrigation projects begun in the late 1960s. Deep wells or geotreseis located in the Avgo Valley, Kambos, Kamina, Petras, and three reservoirs at Kamina, Lakkos Skaphes, and Kapsas (west slope) now provide ample water for irrigation in the plain (Pl. 6A). The geotreseis are not ordinary wells. Constructed during the immediate postwar years as part of the Marshall Plan, and again during the period of the Junta in the late 1960s and early 1970s as part of irrigation and agricultural land reclamation projects, these wells provide access to groundwater and aquifers in environments and at a depth ordinarily inaccessible (cf. Allbaugh 1953, 258–262). The Kambos irrigation system has also been upgraded in the last fifteen years with the implementation of an automatic direct-pipeline system that carries water directly to fields, plots, individual gardens, and single trees. The system conserves water, reduces watering time and labor, and allows irrigation of any plot of land in any location of the plain or surrounding hills. Thus the physiognomy of the lowland plain has been entirely transformed since 1970, most notably in the 1980s, when direct pipeline irrigation allowed exploitation of the “marginal lands,” which, at Kavousi, are the stony terra rossa soils of the central Kambos plain. Barley, wheat, carob, and some olives were grown in the plain before the construction of the first lowland geotrese in 1967. In the years before 1960, less than 20 percent of the coastal lowlands was used for olive production, and yields were poor or inadequate even for subsistence needs of the village. The figure of 20 percent is derived from interviews with local farmers and an estimate, made during field walking, of the age of the olive trees in each field of the Kambos and Tholos plains. Of course, the specific age of a tree is often impossible to determine by mere observation, as irrigation, pruning, soil quality, and type of cultivation are all factors that might have affected tree size. By using a general size index, however, for each of the soil areas (either phyllite or terra rossa), it was possible to arrive at a rough estimate of age categories. Because most of the trees in the phyllite areas of Hagios Antonios and Chordakia were older than fifty years, and in the central plain younger than ten to fifteen, the estimate in these extreme groups was considered fairly accurate. More difficult, however, was to determine the age of trees in areas north and

20

KAVOUSI I

south of the central Kambos; these ranged from young trees (ca. 1970–present), to older trees (preWorld War II); and several recently grafted wild trees at field fringes. The age estimate was roughly justified (judging the size of trees and density of foliage) by examination and comparison of aerial photographs (1968), Land-Sat satellite image (March 1987), and a photograph of the plain taken by Harriet Boyd from the Kastro in 1900 (Boyd 1901, 128, fig. 1). The photograph is very interesting as it clearly shows a narrow zone of trees and dense vegetation extending along the extreme western edge of the Kambos plain from Tholos in the north to Chordakia in the south. The principal areas for olive cultivation were clearly the phyllite sediments of Chordakia and Hagios Antonios, where the oldest trees of the plain are located. In all of the aerial images, the central Kambos shows consistent low foliage or barren ground. This remaining 80 percent of the plain—particularly the areas of Kambos and Kamina—is known to have produced inconsistent yields of barley, always low in comparison with the upland terraces, and always inadequate for the needs of the village (Boyd 1901, 137–139; 1904, 10, 30). In the plain, the phyllite soils in Chordakia and Hagios Antonios were frequently planted in wheat, although barley was the chief staple. Barley and olives were grown on all the upland terraces, where today vines are prevalent (Boyd 1901, 137, 149), and it is clear that all kinds of crops could be supported by a variety of terraces (cf. Rackham and Moody 1996, 141–142). Boyd’s description of the Kavousi region in 1900, the land use history derived from inhabitants of Kavousi and Aori villages, coupled with the tremendous disparity in agricultural potential between highland terraces and lowland plain, would argue against a rigid terrace-form typology, based on the current modern use of terraces. According to villagers of Kavousi and Platania (Aori-Thriphti), the Kavousi and Thriphti area stepped (and even pocket-shaped) terraces were built initially for growing barley, even though construction, use, and maintenance were difficult (Pls. 4B, 5A; contra Rackham and Moody 1992). Very simply, the Kambos and Kamina regions of the plain could not produce sufficient barley yields to support the local population. Use of the upland terraces for grain was a necessity. The lowland fields, while consistently planted, provided yields that were only supplemental to the harvests

derived from the mountainous regions of Thriphti, Papoura, Avgo, and Vronda. This disparity between plain and mountain in agricultural potential is striking. Olives that yielded fruit in the plain had only one-fifth the value of trees in the uplands. Even in periods of aboveaverage rainfall, the trees in the Kambos produced consistently less oil and of poorer quality than those at Vronda, Azoria, and at the village. Wheat or barley production in a year of good to average rainfall would yield about 200–250 kilos of grain per stremma (0.10 ha.) at Vronda, compared to only 100 or less in Kambos. Prior to the irrigation projects of the late 1950s, at least 30 percent of the village’s flour had to be imported. Grape vines, which now dominate the highland terraces at Avgo, Kastro, Mylonia, and Vronda, have replaced the barley and wheat. Some vines were grown in Kambos, particularly at Petras, in the plain west of the modern village, but with no substantial yields. Moisture accumulating in the valley between Mount Chalepa and the village has destroyed many of the vines, and today yields are negligible. The small gardens that today alternate with young olive in the plain are a variety of types. Potatoes, squash, beans, onions, okra, lettuce, and other greens are grown, as well as fruit trees, of which oranges, tangerines, loquat, pomegranate, and bananas are the most prevalent (Pl. 6A). The use of the plain for these scattered gardens is an occurrence of the last two decades, with the exception of Tholos, where wells and a nearby spring provided irrigation. The wells of Tholos are numerous, but of limited value in periods of low rainfall and nonfunctional in periods of drought. The period of use is also seasonal, even in years of plentiful rainfall. The Tholos wells were abandoned after the construction of geotreseis. The agricultural terraces in the area of the modern village, south of the village at Chondrovolakes and Xerambela, and in the uplands of Thriphti and Avgo have traditionally been favored for agriculture (Pl. 5A). The factors that have probably dictated settlement locations in the modern period (as in the Bronze Age) are water supply and arable land. The lowland plain lacks sufficient water for irrigation and cultivable land for rain-fed agriculture. Before World War II, water for village consumption was either transported in stamnoi (water jars) from the spring at Avgo or supplied from a single well in the center of the village. This latter well is

THE PHYSICAL LANDSCAPE

reported to have permitted only one stamnos (about 15 liters) of water per 2 people per day. The wells at Tholos are reported to have provided little water of drinking quality; however, a spring in the southwest corner of the bay apparently supplied drinking water for Seager’s workmen excavating at Pseira in 1906 and 1907 (Seager 1910, 5). In 1946/1947, the lower of two springs at Avgo was tapped and channeled to the village, providing, for the first time, sufficient water for drinking, as well as irrigation of small gardens and olives near the village. In 1955, the first substantial irrigation system was constructed. The main spring at Thriphti was tapped (at Hagios Niketas), and irrigation troughs were built along the precipitous north Thriphti Valley; but this brought water again only to the village proper for irrigation of nearby olives and household gardens. This system, although largely obsolete, still functions and supplies supplemental irrigation water to the upland valleys and the terraces around the village. Year-round extensive agricultural activity in the Kambos and dependence on the central settlement of Kavousi village were both stymied by lack of water. This situation was changed only with the implementation of the irrigation systems in the 1950s and 1970s. The history of land use in Kavousi—inextricably linked to variable water supplies and diverse environmental resources— has only emphasized the great disparity in food staple production, soil potential, and water availability in the mountains and lowland plain. Until World War II, and to a large extent, until the Junta era, nearly 85 percent of the population of the village of Kavousi occupied the mountain hamlets for all but three to four months of the rainy season (December–February). Activity in the plain consisted of olive harvests, grain planting, and running of flocks on fallow in the fall and winter months. The major focus of agricultural activity throughout the year was in the upland valleys. At Avgo, five separate settlements existed, which continue in use today as seasonal harvest facilities, storage areas, and even as vacation homes. The alluvial terraces at Avgo extend in an east-west direction from the Bebonas watershed and provide ample and well-watered gardens of apples, pears, walnuts, potatoes, tomatoes, beans, and squash. Boyd’s “Avgo Notebook” of 1901 records a variety of trees in the valley, including fig, carob, walnut, mulberry,

21

olive, pomegranate, plane, cedar, pear, lemon, almond, hazelnut, and apple. No fewer than three springs supply drinking and irrigation water all year round. To the southwest of Avgo are the villages of Aori in the upper Thriphti Valley: Mylonia, Drakalevrion, Tsamanti, and Platania (Thriphti). Here, wheat and barley were grown on virtually every hillside and upland plain. Today, vines dominate the slopes, as well as gardens of apples, pears, almonds, walnuts, and potatoes. Water is supplied from springs at Hagios Niketas and Thriphti. The site of Mylonia had one of the four major mills in the Ierapetra-Siteia area; the others are located at Monastiraki (Cha), Stavrohori, and Hagios Ioannis. Xerambela, on the northern slopes of Papoura and overlooking the Kambos plain, is the last modern settlement cluster that should be mentioned. This is smaller than the others, consisting of three house groups and a spring. The terraces in the area support almonds, pears, grapes, carob, figs, and olives. Nearby is the large area of the Vronda hill, the site of the LM IIIC settlement and Geometric cemetery. In this area and on adjacent terraces today, villagers plant vines where they once cultivated wheat and barley crops. These mountain seasonal settlements are not metochia (rural farmstead, or tenant farmstead), although that term is sometimes used today; they are substantial settlements originally inhabited continuously from April to November and sporadically from November to March. The pastoral environment of the highlands, even above 700 m is adequate for year-round browsing, and the mandra and mitato (shepherd’s pen) at the Bebonas watershed is constructed for winter use (H. Koster, personal communication). The pattern of herding today usually requires the movement of flocks to the coastal hills in November and December. This is a result of the recent depletion of grazing land and the extensive cultivation of the upland terraces with grapes and other fruit. The mountain settlements of Xerambela, Avgo, Aori, and Thriphti, are today “satellite” or “seasonal settlements.” Prior to the last world war, they were essentially year-round habitation locations, while occupation of the “main village” of Kavousi was, for many people, primarily during barley planting in the rainy season (November–January) and spring harvests. Kavousi village was also the site for centralized storage facilities for export crops (such as

22

KAVOUSI I

wine, carob, and cheese), and the location of the school house. Architectural features within the mountain hamlets themselves illustrate a major economic change in their last period of use (1950–1960): houses were reduced in size and changed in function by means of building of screen walls, blocking doorways, and converting storerooms and living/cooking areas into winepresses (Mook 2000; Mook and Haggis 1994). All these renovations involved partial abandonment of the settlements and are clear indications of the transition from large-scale barley production to wine production and the change from year-round to seasonal habitation. This period of postwar agricultural transformation corresponds as well to the conversion of upland barley terraces to vine terraces. The hamlets are, on average, 0.50 ha. in size and contain roughly 10–24 house units built in an agglomerative fashion; the population of each hamlet was probably 20–100 occupants. These hamlets were arranged in groups of usually 2 to 5. A cluster of hamlets is topographically isolated from other clusters, contains its own spring or springs, and controls its own arable and grazing land. The manifestation of the cluster pattern, and the configuration of the individual hamlets with aggregate plans are the result of a combination of factors: topography and terrain, scarcity of arable land, kinship groups and patterns of land ownership, and available transportation routes. Topography, terrain, and proximity of natural resources are obviously important conditions determining the traditional settlement pattern at Kavousi. The hamlets in each settlement cluster are situated for easy access to gardens and springs. The terrain may dictate their size and orientation. They are located on exposed bedrock or marginally productive soil so as to minimize waste of arable land. Each group or cluster is separate from the others topographically and is composed of extended family groups and associated ancestral land holdings. Families lived in contiguous architectural units, or they spread across various hamlets of a cluster. The spatial dimension of the highland settlement pattern is the most important in understanding the cultural factors that determine the cluster arrangement. Within the region of Kavousi, the basic social unit is the household, but the family identifies itself initially

not with a hamlet, specific house, or architectural unit, but with the area or region that is comprised of the cluster and its associated land. The cluster pattern suggests an orientation of settlement toward the highland terraces of Xerambela, Avgo, and Thriphti, primarily for barley cultivation and proximity to water and flocks. The impetus for the change in settlement patterns from the clusters of dispersed hamlets to the nucleated central village of Kavousi was primarily economic. An agrarian household economy was rapidly replaced by a market economy in which the lowland plain was converted into an olive-producing environment, and the upland terraces replaced barley and wheat with vines. Pipelines from the Avgo geotrese in 1950 provided water for Kavousi village, and importation of flour secured the economic transition. Fishing and trade have played an important role in the local economy before 1960. The bays of Agriomandra and Tholos were frequent roadsteads, if not active ports. Tholos has probably always been used for fishing, and there is traditional memory of Ottoman-period trade activity. Agriomandra, situated at the mouth of a narrow gorge west of Kavousi, on the Bay of Mirabello, was, until World War II, an active shipping point for goods transported through the Kavousi Mountains and the Avgo Pass. Oil, cheese, and meat, from Orino, Chrysopigi, Lapithos, and Stavrohori as well as the settlements of Kavousi, would be put on boats at Agriomandra for shipment to Herakleion, Hagios Nikolaos, and Athens. Kavousi was also known for its aromatic oils, produced from the leaves of cedar, sage, and laurel (Spanakis 1991, 322). The settlement pattern has changed somewhat since 1950 and drastically since 1970, as a result of a combination of technological, economic, and social changes: the extensive irrigation of the plain, the development of the olive as a cash crop, the expansion and formalization of local government, and the federal requirement of school attendance. With direct pipeline irrigation since the late 1980s, settlement has started moving into the Kambos proper. Several new houses, rental rooms, and hotels have been constructed in Kavousi since fieldwork began in 1988. This is a result of population pressure in the village, tourism, and the movement away from the village center to areas of agricultural and touristic interest.

3

The Archaeological Survey

This chapter offers an overview of the methodology, field techniques, and results of the KavousiThriphti Survey. It includes a discussion of problems encountered in the field and interpretive issues that affect the description of the sites in the site gazetteer (Chapter 6) and the final synthesis in Chapter 5, which presents the history of settlement in the Kavousi area. The objectives are to explain how the data were recovered and interpreted and to discuss the criteria that were used to establish site

sizes, as well as the nature of artifacts, features, and surface conditions that were used to define site functions. The chapter also presents some of the evidence for formation processes that have affected the size of existing archaeological sites and the process of artifact recovery, as well as the sampling methods that have an impact on the interpretation of site chronology and function. Environmental variability and site interrelationships are reserved for discussion in Chapter 5.

General Considerations The Kavousi region is characterized by extreme variations in soil, terrain, topography, and water resources. These variations condition the form of wild and cultivated vegetation, modern and ancient land use and choices of settlement, and formation processes—and thus, site conditions. Such environmental diversity, while common throughout Crete, was a primary concern in the choice of survey

methods. Two other factors, affecting both field methods and results, were considerations in the research design: first, fieldwalking was conducted as topographical survey, precluding the systematic collection, storage, and study of artifacts and the detailed analysis of off-site material. And second, the fieldwork was conducted by no more than three but usually only a single fieldwalker (the author)

24

KAVOUSI I

with some obvious implications for the level of intensity and recovery of data (Cherry 1994; 1983b). The number of fieldwalkers affected speed, extent, and intensity of coverage, the range of material that could be physically collected and properly studied, and ultimately the sampling methods employed. Stone tools and pottery from the Byzantine period through the 20th century were not recorded systematically or regularly collected, although observations on such material are included in the site catalog (Chapter 6). Fieldwalking was conducted primarily to record pottery sherd densities across the landscape in order to locate relatively high concentrations of artifacts, or “loci.” Loci were defined as sites, as components of sites, or as “off-site” material during the revisitation of the locus. Off-site material was thus not collected or studied, although notebooks were kept in which general observations were made on the chronology, ceramic fabrics, and types of vessels recognized while walking. Such material was counted in order to define potential site locations and to evaluate relative densities across the landscape that could provide correlative information on formation processes and land use potential of specific areas of the survey zone. The variety of soil types, environmental zones, and known water sources in the region invited close comparison of sherd densities and physical conditions. The Bronze Age and Early Iron Age coarse-ware ceramics—the most prevalent surface artifacts— were not understood in sufficient chronological detail at the start of the survey to make refined chronological and functional designations for offsite material during the fieldwalking stage, although broad distinctions such as “Bronze Age,” “Early Iron Age,” “Hellenistic,” “Roman,” “Byzantine–Venetian,” and “Ottoman–Modern” were made while walking the fields. Narrow definitions were, however, possible during the site-recording and study phases of the fieldwork. This kind of chronological and spatial resolution was not possible in the early stages of fieldwalking at Kavousi, and field methods were designed accordingly. The coarse-ware chronology eventually developed for the region was the result of a continuous process of study, recognition, and definition of coarse fabric

types, and association of those types with identifiable fine wares on specific single-period sites and with material from well-stratified deposits from nearby excavations at Kavousi, Mochlos, Pseira, Vasiliki, Kalo Chorio, and farther afield at Malia and Myrtos (see Appendix 2; Haggis and Mook 1993). It was during the locus revisitation and the site-recording stage of fieldwork that the coarseware chronology could actually be applied. Utilizing the chronology designed by the KavousiThriphti Survey, the recent intensive survey in the neighboring Gournia region has succeeded in defining nearly all Bronze Age coarse wares collected in one-hectare fields during primary fieldwalking (Haggis and Mook 1993; Watrous and Blitzer 1999; 1995; Haggis 2000; Watrous et al. 2000). The survey zone is made up of widely varying physical and geological environments and converging communication routes in East Crete: the Ierapetra Isthmus, the Bay of Mirabello, the Avgo Valley, and the fault front of Mounts Papoura and Kapsas (Fig. 1). These factors suggest that the archaeological landscape is potentially complex—formed as much from outside influences as from local environmental conditions and cultural dynamics. The situation of the study area thus called into question the extent to which settlement development, economy, and social organization are linked externally to Mirabello and the Aegean, and internally to the south coast and the Siteia Massif and valleys. The coarse-ware fabric typology and chronology was developed as a means not only of dating the Bronze Age sites discovered in the survey zone but also of quantifying and evaluating the nature of external influences, interregional contacts, and economic spheres of influence. Thus, an aim of the fieldwalking was to draw the boundaries of spheres of ceramic distribution in various periods of the Bronze Age and Early Iron Age, with a view to defining potential zones of economic exchange or redistribution. Such definition is a first step in equating spheres of influence and sociopolitical or cultural regions. The meaning of the directionality of the exchange, the processes of distribution, consumption, and use of the vessels, or what they might have carried is treated in Chapter 5.

THE ARCHAEOLOGICAL SURVEY

25

The Survey Boundaries The survey boundaries were initially drawn to encompass approximately the geopolitical district of the traditional and recent community or koinoteta of Kavousi. The broad limits of the survey zone therefore conform to an existing geopolitical region: 33 sq. km that include the village of Kavousi and its catchment area—the lands traditionally exploited by its inhabitants (Figs. 1–4). It is interesting that while the koinoteta of Kavousi was officially brought into the political district (demos) of Ierapetra in 1999, patterns of land use and social identity within the village remained firmly linked to this traditional hinterland. The area is fairly evenly stratified into mountains, lowland plain, and coastal hills. Bordered by the sea on the north and west and by the mountains of Kapsas, Kliros, and Thriphti on the east, only in the southwest did the survey zone lack definitive topographical boundaries. Here the borders were initially defined on the west by the national highway connecting Pacheia Ammos with Ierapetra, and on the south, the field road extending roughly east-west from Vasiliki to Monastiraki. This corner of the northeast Isthmus of Ierapetra (south and southeast of Mount Chalepa) is the area of Evraïka, Kamina, and Monastiraki. Lying outside the Papoura and Avgo drainages, the region belongs topographically to the north Isthmus of Ierapetra and the Cha River catchment. It was thus decided in 1990 not to survey this important area intensively but to incorporate it into the neighboring Gournia Project, which was in the planning stages at that time, beginning intensive coverage of the north Isthmus in 1992 (Fig. 3). This area of the northeast Isthmus was eventually surveyed in 1993, and the results will be published as part of the Gournia Project (Watrous et al. 2000; Watrous 1993; Watrous and Blitzer 1999; 1995). The total area surveyed intensively by the Kavousi-Thriphti Survey consisted of some 17–18 square kilometers, encompassing the entire Papoura, Avgo, and northern Thriphti catchment areas that drain north from watersheds at Bebonas and Aori-Thriphti and feed into the Kambos plain and Bay of Tholos (Fig. 2). The eastern boundary was conveniently the mountain fronts of Kapsas and Kliros, with the Bebonas watershed in between; the southern boundary was the watershed at Thriphti village (Aori or Platania),

the cliffs of Papoura, and the national highway between Kavousi and Pacheia Ammos; and western and northern boundaries were formed by the Bay of Mirabello (Figs. 2, 3). The area is neatly divided into separate topographical zones that are environmentally distinct. The mountain zone (areas 100–800 m above mean sea level) consists of some 8 sq. km in the area of Mount Papoura and the north Papoura drainage and the Avgo Valley. These mountainous areas are extensively and densely terraced. Although traditionally planted in barley and wheat, they now support primarily olives up to 400 m, and sporadically up to 700 m above sea level. Gardens and vines are grown throughout the upper elevations. The expansive upland plain and valleys of Mount Papoura and the steep slopes between Avgo and the Bebonas watershed are used today principally for herding. Barren scrub (maquis and phrygana) has returned to these vast areas that were once oak and pine forest and barley fields; terraced hillsides now support sporadic vines, fruit and nut orchards, and small gardens. The Kambos plain alluvium (less than 100 m in altitude) is some 7 sq. km in area. Before World War II, it was exploited principally for barley; since 1970, the plain has become dominated by olives with intermittent gardens and vines. The string of coastal hills, Mounts Schinias, Chomatas, and Chalepa (100–230 m above sea level) make up about 6 sq. km, and the topography and environment are complex. Barren bedrock terrain, used exclusively for winter browsing—or more recently as rock quarries—dominates the limestone environments. These alternate with outcrops of phyllite, whose soils and deep sediments now accommodate gardens and terraces with olive groves. The survey boundaries enclose a topographically and geopolitically defined area within a single river catchment. The area also provides samples of topographical and environmental zones of roughly equal size in which to compare types of settlement and patterns of land use (cf. Moody and Rackham 2004a, 6–13). We did not assume, however, that these boundaries delineated the limits of a cultural or political territory in any period, except modern times; the dynamic character of the diachronic cultural landscape remains a methodological problem and analytical variable in archaeological survey (cf.

26

KAVOUSI I

Wright 2004, 115; Cosmopoulos 2001, 23; Sutton 2000, 3; Watrous et al. 1993, 215; Cherry et al. 1991, 16; Cherry 1994, 93; 1983b, 385–389). The nature of the physical setting and the distinctive topography emphasize external links and extraregional connections. The obvious access to the Bay of Mirabello and the Aegean, and the topographical link to the Isthmus route across the island, anticipate multiple streams of external influences in a number of periods, suggesting that the sample area is but one spatial component of larger-scale social, economic, and political systems. But taking into consideration the impact of exogenous influences on local settlements is as much an a priori basis for interpreting the patterns as envisioning the sample area as a closed universe, a bounded zone encompassing closed cultural systems. In defining any sample area (in both excavation and survey) there is a tendency to telescope the focus of investigation, visualizing the archaeological context as a closed system of exclusively interacting parts. On the scale of regional survey, the area may be selected to accommodate a definition of a “territory.” This commonly takes the form of a delineation of presumed spatial boundaries of political spheres of central places such as cities and palaces. Such cultural boundaries help us to form useful hypotheses in the initial research design, but an important aim of survey should be to work toward defining such cultural regions, not

merely to presume their existence. Ultimately the spatial scale of analysis is of paramount importance, not only in designing recovery methods, but in visualizing variable and dynamic scales of cultural interaction across and between regions (Wright 2004). In efforts to reconstruct regional settlement hierarchies or any other components of complex societies, there is an internal tension between analytical frameworks and research goals—the desire to use small-scale internal dynamics of settlement that may result from local environmental or cultural variables to interpret broader regional systemics (cf. Wright 2004, 115; Terrenato 2004). The survey boundaries at Kavousi were drawn with few preconceptions of the size of a cultural region. They do, however, conform to the area of a known geopolitical district and an actual community and its territory (that of the modern village of Kavousi), and to the hypothetical catchment areas of Early Iron Age settlements. The area was chosen to be large enough to include an entire settlement system but small enough to permit us to recognize the idiosyncrasies of local land use and cultural responses to environment. At the same time it was sufficiently large to recover patterns of economic or political dependence on, and interaction with, larger-scale regional systems of settlement in the north Isthmus, the Mirabello coast, and eastern Crete.

Fieldwalking The fieldwalking methods varied between the plain and mountain areas of the survey zone. The aims were achieved by techniques tailored to facilitate complete and efficient ground coverage of contiguous fields by usually no more than two fieldwalkers and to account for differing terrain, topography, density of vegetation, and levels of accessibility and ground visibility in various environments. While practicality (variables of time and funding) was a primary concern in the choice of fieldwalking methods at Kavousi, it is important to emphasize here that the lack of uniformity is considered a significant concern in rendering comparable results on the local scale and between regions (Alcock and Cherry 2004; 1994, 95–105;

Hayden 2004c, 3–6; Alcock, Cherry, and Davis 1994, 137–143; Cherry 1994; Kardulias 1994, 10– 17; Watrous et al. 1993, 214–222; Cherry, Davis, and Mantzourani 1991b, 457–462; Wright et al. 1990, 603–608; Cherry et al. 1991, 13–36; Runnels and van Andel 1987, 304–309; Bintliff and Snodgrass 1985). The use of different methods at Kavousi might indicate that the off-site sherd densities, and therefore the quantifiable aspect of site definition, may not be completely comparable between lowland and mountain areas of the survey zone. This point will be discussed below. The aim of fieldwork was to recover archaeological sites, that is, to find loci or concentrations of artifacts, usually pottery, stone tools, or architectural constructions in the

THE ARCHAEOLOGICAL SURVEY

landscape. The primary purpose of fieldwalking was to measure the variability of the density of potsherds across the landscape as a means of defining sites, but also in order to draw a correlation between the evidence of human activity and types of soil and sediment, and other environmental factors. In the Kambos plain area (Fig. 2: Tholos, Schinias, Hagios Antonios, Kambos, Chordakia, and Lakkos Ambeliou), in areas around Kavousi village, the method of fieldwalking involved initially the definition of surveyable areas or “fields,” usually indicated by natural features—ravines, gullies, torrents, cliff edges—and man-made features such as roads, fieldwalls, irrigation lines, and buildings (Fig. 4). The fields varied greatly in size and shape as they were not intended to be uniform diagnostic units, only convenient and identifiable borders for walking uniform passes. Passes were individual walking lines, bisecting the fields at intervals of 20 to 25 m. A “pass” refers to a directional walking line rather than a transect (cf. Hayden 2004c, 3–4; Jameson, Runnels, and van Andel 1994, 219; Watrous et al. 1993, 218; Cherry et al. 1988, 162; Cherry et al. 1991, 20–28). These were walked in a “boustrophedon” fashion; that is, a single line was walked on a predetermined compass bearing, then after completion of the line at the edge of the field, the fieldwalker would turn at a right angle, pace off 25 m, and then return across the field on the alternate bearing. While walking, sherds were counted continuously along the pass, and totals were tallied in units of 50 m. For each 50-meter unit notes were kept describing soil, vegetation density and ground visibility, and land use; general observations were made on the nature of archaeological material present. Since paces were counted along each 50-meter unit of the pass, it was possible to note the specific location of individual artifacts, scatters, and anomalous increases in sherd densities. Given that background sherd densities were quite low for all but the area immediately surrounding Kavousi village, it was possible to stop along passes, diverge from the compass bearing, and investigate and make notes on unusual artifacts or increases in sherd density. During the fieldwalking stage, field notes on chronology and function of vessels became more detailed as the local Kavousi and Mirabello area coarse fabric sequence was understood. In order to maintain bearings in long transects, biodegradable flagging tape was used periodically along the line

27

and at the edge of the field. This technique was especially effective in walking the huge expanses of the Kambos plain, where the dense olive groves not only precluded locating small fields but also obscured the fieldwalker’s view of bearing points. At the end of each day, the transects were drawn directly on Greek Army topographical maps (1: 5000), and the transect units and sherd counts were plotted, making reference to information in the field notes that might indicate artifact chronology, problems of visibility, and possible “locus” locations. The locus—usually an anomalous or relatively high density of artifacts—was drawn onto the topographical map using the unit densities as rough 25 x 50 m guides to outline the possible maximum size of the site. Customarily, however, smaller loci were plotted directly on the map during fieldwalking, and sherd counts and descriptions at individual paces provided very specific locational information. Once a locus was defined by sherd densities and notebook information, it was revisited and described using a locus record sheet, noting the specific location, size, topography, chronology, vegetation, land use, architecture, and artifacts. The field map and locus records were then used as guides for returning to all of the loci in 1991 to make the final site records. In the hills along the west side of the Kambos plain, in the mountainous areas south of Kavousi village, and in the Avgo Valley, the precipitous or uneven terrain, dense vegetation, unstable slopes, and extensive terracing precluded the use of regular linear passes that could be managed easily by one field-walker (Figs. 2–4). The method employed in these environments was to walk closely spaced contours—usually in between natural features such as gullies or ravines—in a “boustrophedon” or zigzag pattern from the top of the hill or mountain slope to its base. Sometimes isolated peaks and hillocks could be walked from peak to base in a spiral pattern. In this mountainous terrain, measuring sherd densities in regularly paced units proved to be neither effective nor feasible. In attempts to adhere to 50 m units, accurately locating transects and units was difficult and more time-consuming than simply plotting directly on the map any and all traces of visible artifact material or features. “Background noise” (off-site sherds) per se, did not really pose a problem, since where artifacts were visible, the material was considered a locus and later revisited. It was advantageous to measure anomalous high densities or

28

KAVOUSI I

homogeneous scatters in the terraces immediately south and east of Kavousi village, where the fields are densely strewn with Byzantine to modern-era pottery. Frequently, however, even when terrain in the mountains permitted using measured transects and units, the density of vegetation, recent agricultural terracing, and modern buildings precluded recovery of densities that could be compared with the Kambos plain off-site densities. Even so, the actual coverage in the mountains was more intensive than in the plain, because when walking contours, the spaces between accessible ledges or terraces might be no more than five to ten meters. The intensity of coverage in the mountain zones, notwithstanding, the number of sites recovered in these areas was similar to that of the Kambos plain. Across the survey zone the average number of sites per square kilometer was 4.26. In the Kambos, 5.33 sites per square kilometer were recovered, while in the coast hills the average was 3.23 sites. In the

mountain areas south and east of Kavousi (Avgo and North Papoura), the average was close to that of the plain (4.66). A third method of fieldwalking was employed in the areas south of Kavousi, in the Kamina plain and the highlands of Mount Papoura (Figs. 2, 4). These areas, marked “extensive” on the map (Fig. 3), represent the southern half of the Kavousi koinoteta. While they were included in the original plan of work, time did not permit intensive coverage. Furthermore, it was realized during fieldwalking that these areas are topographically separate from the intensive zone (Figs. 2, 3) forming the natural northern extensions of both the Isthmus of Ierapetra and the Thriphti mountains. The extensive survey of this area consisted of single long transects through the zones (Kamina plain and Mount Papoura) and site prospecting in areas of known and excavated sites. I include the sites recovered in these areas in the gazetteer as a matter of record.

Locus Definition and Site Recording A locus is any concentration of artifacts (such as sherds or stone tools) or features (such as rock cuttings or architecture) that was recorded during the primary fieldwalking stage. All loci were revisited and reexamined in order to make a complete record of the find spot and to determine whether or not the locus could be defined as a site—a definable context and distinct locale of past human activity. At this stage the locus was reinvestigated, artifacts and features were recorded, and the locus’s relationship to nearby loci was defined. In the final record, a locus may be defined as a site, usually represented as an artifact scatter with definable limits; a number of adjacent or contiguous loci (or transect units) may be understood to form a single site, that is, a continuous scatter of artifacts warranting study and recording as a single site; a single locus may be found to consist of two or more sites; or finally, in very rare instances, the locus may be ultimately indistinguishable from the artifact densities in the surrounding fields, and therefore not recorded as a site but defined as background noise. Some loci are built features or constructions such as cisterns, rock-cut tombs, kilns, or spring houses with no

associated artifacts, requiring only verbal description, a site drawing, and photograph. The aim of site recording was to describe the findspot and reach some conclusions on the periods of occupation and nature of its use. The process comprised determining the size, density, shape, and chronology of the artifact scatter and recovering a sample of artifacts that might be representative of its date range and function. Site recording was as flexible as fieldwalking, requiring modifications in the method depending on site size, terrain, and ground visibility. The usual site-recording technique consisted of laying out two lines, one oriented north to south, the other east to west, so they bisected at the approximate or assumed center of the locus. From this center point, a tape measure was extended as a radius along the established lines and sherds were counted in units, usually 2 m wide and 5 m long, until two sherds or fewer were counted in two consecutive units. Sometimes smaller units were used, depending on the size site and density of the sherds. This permitted an assessment of the overall density of the site and defined its limits on four points. The area between

THE ARCHAEOLOGICAL SURVEY

each of the points on the site’s periphery was examined carefully in order to detect irregularities in shape or gross misinterpretation of size, so we could more accurately sketch the site’s approximate form. The mean on-site density derived from sherd counts in the 50 m units, while fieldwalking was 0.457 2 sherds per 100 m for the Kambos plain, and 5.58 2 sherds per 100 m for Chordakia. The extremely low on-site densities derived from fieldwalking was in all likelihood a result of the methods used—essentially an individual walking at intervals of 20 to 25 m—which is considerably less intensive than large teams walking at intervals of 10 to 15 m with full artifact collection capabilities. In the Vrokastro area fieldwalking was conducted at intervals of about 12 to 16 m (Hayden 2004c), while in the South Argolid (Jameson, Runnels, and van Andel 1994, 219), walking was conducted at intervals of 5 to 15 m, a level of intensity significantly greater than that at Kavousi (cf. Watrous et al. 1993, 218–221; Cherry et al. 1991, 22; Wright et al. 1990, 604–605; Bintliff and Snodgrass 1988, 510; Cherry, Davis, and Mantzourani 1991a, 47, table 3.4; Watrous et al. 1993, 220–221). The implication of this level of intensity is that, in the Kambos, sites smaller than 2 about 225 m (ca. 15 m in any dimension) lying squarely between the passes may have been overlooked; this suggests that any number of the smallest order of sites (Class 1; see designation below) might be missing from the data. In the Kambos however, the generally dispersed distribution of artifacts—conforming to patterns of deep plowing (cf. Terrenato 2004; Ammerman 1985)—and the general absence of background noise in the region, indicate that the number of small sites actually missed should be negligible. On site, all of the sherds within a 2 m wide swath along a transect were collected and separated into diagnostic fabrics and shapes and then counted and recorded in the site notebook. Finally, a random collection of fine ware and diagnostic

29

fabrics was made in each of the quadrants formed by the bisecting lines. This method was an efficient way of determining site boundaries and onsite densities, while recovering a sample of pottery representative of the full range of occupational phases. A sketch plan was then made of the site defining surface features and topography. The method just described emphasizes obtaining a broad and unbiased sample of ceramic fabrics at the expense of provenience data. This emphasis was decided at the start of the project and retained consistently through the fieldwork. The purpose of sampling on Bronze Age and Early Iron Age sites was, first, to define the borders of the site, thus establishing its maximum size; and second, to recover the full range of coarse-ware fabrics present on the site surface that were taken to be parallel to the range of occupational periods. Coarse wares in the Kavousi area were found to be chronologically diagnostic with a degree of resolution useful for dating sites in survey, and the sampling procedure emphasized their recovery. The method proved ineffective, however, in establishing the changing size and shape of sites through time with degrees of accuracy attained by collection of multiple small-scale units, such as circles on multiple radii or squares on an orthogonal grid. Thus the sizes and shapes of the blackened areas on the site map (Fig. 5) represent the sites’ maximum dimensions, irrespective of occupational phase and based on a combination of sherd densities acquired through fieldwalking and site recording. Relative site sizes by period were nevertheless determined through various methods: (1) by recording the densities and chronology of a number of loci within a single larger site; (2) by means of separate collections according to natural features on sites where sampling in swaths along bisecting transects proved to be ineffective or impossible; and (3) through observations during “grab” pick-ups of the quadrants.

Ceramic Study and Depositional Processes The objective of the site-sampling method—to collect broad swaths of pottery from bisecting transects through the site—was to obtain a substantial,

unbiased, and representative sample of coarse wares on the site surface. While a number of smaller collection units would have provided more detailed

30

KAVOUSI I

provenience information (Plog, Plog, and Wait 1978, 407), it would have been necessary to locate and collect a large number of such units in order to avoid biasing the coarse-ware sample (such as sherds collected from single vessels) and to obtain a cross-section that was representative of both the range of fabrics present and their relative quantity. The site-sampling technique employed was both efficient in terms of time spent per site and effective in recovering the full range of fabrics. Since the use of standard sampling transects biased the quantity of sherds in favor of small sites, the large sites were collected within separate loci or spatial subdivisions indicated by variable sherd density, topography, or features (such as modern buildings, fences, and field boundaries). This means that the actual percentages of coarse-ware types are properly speaking not quantifiably comparable between sites—since the sites and loci were not exactly the same size and the sample units were not proportionate to site size. The relative volume of different wares on the same site did however provide useful information on periodicity, longevity, and intensity of use and patterns of occupation. The majority of Bronze Age and Early Iron Age coarse-ware fabrics found in the Kavousi area are products of two neighboring but distinct geological environments and areas of production: the phyllitequartzite zone that includes the Kavousi and Mochlos areas and the northeast Isthmus; and the zone of granitic and dioritic rocks that encompasses the area of Kalo Chorio and Gournia, west of the Isthmus along the foothills bordering the Bay of Mirabello (see Appendices 2 and 3; Hayden 2004a, 23; Moody et al. 2003; Haggis 2000; P. Day 1997; 1995; 1991; Haggis and Mook 1993). These Bronze Age and Iron Age production zones are discussed in greater detail in Chapter 4 and Appendices 2 and 3). The bedrock geology and sediments of these neighboring regions are the sources of distinct inclusions that are macroscopically distinguishable components of a number of local coarse and medium coarse ceramic vessels that were produced throughout the Bronze Age and Early Iron Age. While the inclusion types and perhaps production traditions continued or were revived in later periods, the focus of the Kavousi-Thriphti Survey study was on eighteen Bronze Age coarse fabrics, thirteen of which were determined to be chronologically diagnostic for one or more periods (EM I–LM IIIC). The

period range for each fabric was determined by systematic examination of stratified pottery assemblages from neighboring excavations in the Kavousi, Mirabello, and north Isthmus areas. The fabrics (see Chapter 4 and Appendices 2 and 3) were defined by differentiating features that were discernible to the naked eye or in 10x magnification (using a hand lens). Distinguishing features were color (core; surface or slip), surface treatment, presence or absence of organic temper, and the quantity and types of rock and mineral inclusions. It was important that the fabrics could be distinguished from each other macroscopically so that they could be identified in the field by means of examination by the naked eye or a quick glance through a hand lens and thus sorted efficiently during site recording. The fabrics recorded during the survey do not represent the total number of coarse wares produced or used in the Kavousi area in the Bronze Age and Early Iron Age. The eighteen fabrics that were defined are only those that could be unambiguously identified during the site recording (Haggis and Mook 1993). The thirteen diagnostic types are fabrics that were correlated with certainty to stratified excavation deposits and were thus assigned specific chronological parameters. A complete chronological definition, petrographic analysis and definition, and determination of provenience of the fabrics will be among the results of the numerous excavations in the north Isthmus area (e.g., Barnard 2003; Barnard and Brogan 2003; Day, Joyner, and Relaki 2003; Floyd 1998). The aim of our survey’s fabric study was to define coarse wares that could be used as evidence for the chronology and function of sites and for drawing preliminary conclusions about economic interaction between two identifiable production zones in the Mirabello area. Eventually the results of complementary ceramic studies from both excavation and survey will permit the detailed definition of regional ceramic spheres of distribution, their meaning, and changes through time (cf. Haggis 2000). Such distributional zones might support models of socioeconomic systems, once the fabric types and ware groups are correlated to vessel shapes and integrated into discussion of site types and functions (e.g., Watrous and Blitzer 1999). Our study of coarse-ware pottery from the Kavousi region began with two observations made during the fieldwalking stage: first, the predominant artifact type recovered on the ground surface was

THE ARCHAEOLOGICAL SURVEY

coarse-ware pottery, usually fragments of large, coarse-tempered vessels used for storage, transport, cooking, and food preparation, or other industrial processing. Most common were sherds of pithoi, jars, amphorae, jugs, tripods, trays, dishes, basins, and lekanai. Larnakes, beehives, lamps, strainers, scuttles, loom weights, figurines, among many other types of coarse-ware artifacts were also recovered, but less frequently. Second, there was considerable diversity in the observable fabrics. The sherds displayed remarkable differences in color, surface treatment, and inclusions, characteristics that were assumed to be related to changes in production over time, or to differences in production locales, reflecting either different sources of raw materials or individual production preferences and traditions. The predominance of the coarse wares and their apparently diverse range of identifiable types led to further considerations of the meaning of the surface sample and the probable impact of site formation processes on this sample. An important question was whether the surface ceramic assemblages derived from survey would provide a complete picture of a site’s chronology. The apparent predominance, ubiquity, and ultimately the obtrusiveness and visibility of coarseware pottery on the surface of archaeological sites are the result primarily of two phenomena related as much to recovery techniques as to depositional processes. The first observable phenomenon is the “size effect.” Coarse-ware pots tend to be large—often cumbersome, unwieldy, and heavy, breaking into sherds that tend to be larger than breakage of fineware vessels. The “size effect hypothesis” suggests that the relative proportion of large artifacts to small artifacts will be greater on the surface of a site than at any subsurface level (Schiffer 1987; Baker 1978). While the size effect is a quantifiable phenomenon in itself, we can assume that natural formation processes, vegetation cover, and recent site disturbance are generally less likely to affect the visibility of larger artifacts than smaller ones. Pithos sherds and large tripod legs, for example, are obtrusive survey finds, frequently found in field walls and rock piles at the fringes of fields containing sites, having been selected from the site surface along with cobbles and boulders of similar size. The color of the terracotta against the gray limestone makes them easy to spot even in dense vegetation. The second phenomenon concerns depositional processes.

31

Coarse-ware vessels tend to be well represented in de facto refuse contexts (in situ primary use and discard contexts) but especially in post de facto refuse deposits (contexts in which material is the result of secondary disturbance or deposition at a site during abandonment). Vessels such as cooking wares and pithoi are frequently left behind at a site during abandonment, are slow to decompose, and do so in proportionately larger individual pieces than fine wares. Furthermore, because their preservation potential is high, they are likely to be reused or recycled if the site is reoccupied. Although the circumstances of site abandonment, such as rate of abandonment and the distance to next settlement, may directly influence the kinds of objects left at a site, on the whole, object size, weight, ease of transport, and replaceability, are important variables in the processes of deposition at the time of site abandonment (Schiffer 1987). A premise of the coarse-ware study at Kavousi was that such vessels are not likely to have been saved and removed during site abandonment but could easily have been reused during subsequent phases of reoccupation. Recycling is evidently a common practice at Azoria, where recent excavations have shown that large coarse Geometric kraters and LM IIIC pithoi were reused into the Archaic period (Haggis et al. 2004). Therefore, pottery sherds from these vessels—and thus their use phases—would be well-represented on the site surface. Coarse wares form a ubiquitous class of objects that are more likely than fine wares to be left on a site during the abandonment stage, discarded on site during the use phase, and reused if the site is reinhabited. Their obtrusiveness and apparent lack of susceptibility to the effects of human abandonment or secondary natural processes on the site surface imply that they would represent a range of occupational periods potentially more complete than fine wares, if their chronology could be established. The coarse-ware study was thus implemented as a means of mitigating the inherently distorting variables of site formation and depth, sherd diagnosticity, and site function (Terrenato 2004, 41), all of which might be of greater concern than fieldwalking or sampling intensity in deriving usable comparative results. Site-sampling methods also play a significant role in the recovery of certain types of artifacts. The likelihood of the presence of coarse wares on

32

KAVOUSI I

a site surface and their large size would account for their high recoverability. Such vessels are more likely to be visible to a fieldwalker and thus to be selected during site recording. These factors—vessel size, weight and function, the process of abandonment, and the conditions of recovery—point to

the viability of the coarse-ware sample as indicative of all phases at a site in which such vessels were used. The chronological and regional significance of the coarse-ware pottery will be considered in Chapters 4 and 5.

Site Definition The density of settlement and the extent of human activity and land use in a given area of the survey zone were gauged by means of the relative density of potsherds counted while fieldwalking and the number of periods recognized during site recording. This correlation between activity and sherd density does not presume a particular cultural variable, land use practice, or discard behavior. It merely suggests that the more sherds you find, the greater the likelihood that people lived nearby or used the place for some purpose. Corroborating evidence of agricultural potential of specific soils, sediments, water resources, and topography did, however, lead to hypotheses of land use but not of specific mechanisms of discard. Surface conditions, density of vegetation, and natural formation processes (such as alluvial sediment formation or “slope wash” conditions) that might have moved cultural material or obscured sites were considered carefully in assessing the meaning of relative sherd densities across the landscape. During the fieldwalking, in instances in which surface conditions did not seem to have significantly affected the sherd counts, the apparent density of pottery was related either to the existence of an actual site or the number of periods represented there—that is, the greater the number of periods, the denser the sherds. The varying sherd densities, therefore, were assumed to correlate to levels of intensity of land use, the extent of repeated use, or longevity of occupation for a particular area or site in the region. These densities were then compared to other areas throughout the survey zone. When linked to environmental information, high sherd counts suggested, a priori, a long-term interest in specific areas that were the most productive or useful agriculturally. The multiplicity of periods represented in such high-density fields was important information, if clouding the determination of site limits or interpretation of patterns of deposition in

specific periods. A date range parallel to that of the high-density fields could be found on contiguous or nearby sites. Dense concentrations of artifacts were called sites; the map of sites (Fig. 5) illustrates the locales with the highest observable sherd densities, relative to their immediately surrounding area and with discernible limits. The area surrounding the site is the “field,” not a diagnostic unit in itself, but a conveniently defined zone for fieldwalking (Fig. 4). These fields were designated “high-density fields,” such as Hagios Antonios, Chordakia, and Lakkos Ambeliou 2 (mean of 0.46 sherds/100 m ), and “low-density fields” such as Kambos and Papoura (mean of 0.028 2 sherds/100 m ). In high-density areas—usually phyllite soils—the high on-site averages required considerable adjustment of criteria for determining site borders. While the “2 sherds or less/unit” rule for establishing site borders worked in some instances, in others it had to be increased proportionate to the surrounding off-site mean. On rare occasions, other criteria, such as surface condition of the pottery, the chronology of the material in the sample unit, and topography, played a part in determining site limits. Sometimes the physical shape of the site could be determined from the topography and correlated to both a drop-off in sherd density and a change in the chronology and condition of the sherd material. Surface-worn sherds, for example could be found at site fringes and downslope of the site’s center. In high-density fields such as Chordakia (Fig. 4: Chor A; Fig. 6), the volume of background noise is also a clear condition of intensive and continuous agricultural use of the area, especially in Venetian–Ottoman periods. High-density fields such as in Hagios Antonios and Chordakia (Fig. 4: Chor A, Ant; Fig. 6), while occasionally causing difficulties in maintaining tidy sherd density criteria for site definition, provided

THE ARCHAEOLOGICAL SURVEY

the most important information on patterns of land use in the Kavousi region. The site map (Fig. 5) shows five separate locales with concentrations of activity suggested either by particularly dense onsite and off-site sherd scatters, such as in Hagios Antonios (sites 5–12) and Chordakia (sites 28–46), mentioned above, or noticeably dense concentrations of identifiable sites as in Tholos (1–4), the north Kambos (15–18), the north Papoura drainage (59–80), and Avgo (83–91) (Figs. 2, 5). The archaeology of each of these areas shows a remarkably long-term aggregation of sites and considerable evidence for continuity of settlement and exploitation of land and water resources. The distinguishing characteristic of Hagios Antonios, Chordakia, Xerambela-Chondrovolakes, and Avgo (Fig. 2) is the combination of perennial water resources and arable land for dry agriculture. The phyllite sediments in these areas are discrete locales of arable land with particularly dense ceramic material. The oldest olive tree in the Kavousi area, with a diameter exceeding three meters, is located in such an environment, at Chondrovolakes (Figs. 5 and 7: site 68). These ancient trees and their associated terraces on the slopes of Papoura—between Vronda, Kastro, and the modern village—are evidence of the importance of these specific soils to ancient as well as recent farmers. Furthermore, springs at Avgo and Xerambela flow all year round. While the aquifer at Hagios Antonios now flows only in the winter and spring, it was perennial in years before the disruption of groundwater levels in the plain with the construction of geotreseis (deep wells). Chordakia lacks a spring today, although like Hagios Antonios and Avgo, it is situated at the juncture of gray carbonate (Mount Chalepa) and phyllite (Mount Chomatas) outcrops, an ideal physical location for springs. A Late Roman or Early Byzantine cistern was excavated by Boyd in 1900 on the northern edge of Chordakia (site 29) (Boyd 1901, 156)— possibly archaeological evidence for a flowing aquifer at this location. But a Roman well, discovered at the northeastern edge of site 37 on the southern edge of Chordakia, suggests dependence on high groundwater in addition to free-flowing springs. The soil quality appears to have been a primary factor in choosing settlement locations, and the correlation between high sherd densities and phyllite sediments is a strong indication of the productivity of the latter for rain-fed agriculture.

33

Tholos is different. Here the soil is sandy terra rossa alluvium and stream-bed and overbank deposits from the Platys River. While the soil is of secondary value to the phyllite sediments, fresh water is plentiful. A perennial spring is located in the southwest corner of the bay, and prior to the geotreseis of the 1960s, the Tholos plain was literally filled with wells for the irrigation of small gardens at the mouth of the river. The location of the Minoan town and Roman buildings (site 1) on the bay itself, however, emphasizes maritime activities and an important economic role that had less to do with local farming than with fishing, trade, and transshipment. There is a striking difference in both land quality and settlement between the central Kambos plain and the phyllite zones mentioned above. On the whole, sherd densities across the central and northern plain are very low, frequently devoid of even Venetian, Ottoman, and 20th-century material. The very rocky terra rossa soil is densely packed with angular talus debris from Mount Kapsas, and the lack of known springs means that the area had minimal value for dry agriculture. In recent times, fields are watered exclusively from geotrese-fed reservoirs on and above the hill of Lakkos Skaphes (Pl. 6A). Minoan and Roman settlements preferred the edges of the plain, usually the slopes or peaks of foothills rising above the lowland (Figs. 5 and 6: sites 11–14, 20–25). One important exception is north of the hill of Lakkos Skaphes (site 20) where a series of five sites (Figs. 5 and 6: 15–19) with Bronze Age and Roman remains occupy a total area of nearly 0.25 sq. km on the east side of the Platys River. It is possible that this cluster of sites, like the settlements (sites 5 and 11) on the west bank of the river at Hagios Antonios, exploited the Platys River for irrigation of fields in Neopalatial and Roman periods. Socioeconomic factors associated with a growth of settlement in Tholos and Hagios Antonios in LM I and again by the 2nd century A.D. might be sought for explaining nucleation of settlement and the anomalous high density of activity in the north Kambos plain. The area of Kavousi village itself is problematic. Venetian, Ottoman, and 20th-century architecture now covers an area of over half a square kilometer as the village has expanded into the Kambos plain. Fieldwalking in the vicinity of the village was impeded by modern buildings and fenced gardens,

34

KAVOUSI I

and despite the graciousness and patience of modern inhabitants in their willingness to allow exploration of their private gardens, the ground surface was obscured by vegetation and modern debris and installations. A nimbus of predominantly Byzantine, Venetian–Ottoman, and 20th-century pottery surrounds the village. While the concentrations are variable on the north and west, a field westsouthwest of the village produced an anomalous rise in sherd densities. This increase in the pottery sherd count is represented as “site 56” on the map (Figs. 5, 7); it is nearly four hectares in size, containing material dating from prehistoric through modern times. The sherds are distributed fairly evenly throughout the field as if spread by a plow or through regular manuring. The site is also bisected by a gully that originates as a drainage channel in the southwest corner of Kavousi village. Sherd densities increase in regular fanlike patterns as one approaches the overbank deposit of this torrent. The particular pattern of sherd distribution, the eroded condition of much of the pottery, and thoroughly mixed chronology of the sherds are taken as evidence of displaced material or secondary deposition. The site is likely to be composed of a

combination of: (1) slope wash debris from Kavousi village and the neighboring Roman site 57; (2) deposition from the torrent that drains into the field from Kavousi village (site 24); and (3) material from both the village and site 57 in the form of intentional discard (waste and manuring) during agricultural use of the area (cf. Alcock, Cherry, and Davis 1994). The bulldozing of the path for the modern highway no doubt contributed to the movement of sherds into the plain from surrounding sites 24 and 57 (Figs. 5, 7). The complexity of site 56 indicates the long and varied history of the Kavousi village area. The consolidated alluvial shelf upon which the village sits is an optimal site location at the confluence of Papoura and Avgo drainages, and at the natural passage into the Siteia Mountains. In the center of the village there is a spring, and nearby, a modern building operation brought to light a deposit of pottery spanning the Neolithic–Late Roman periods (site 24). Such tantalizing evidence underscores both the importance of the location throughout the history of the region, as well as the difficulties in interpreting regional settlement patterns without knowing the size and function of Kavousi village.

Site Classification A total of 110 loci were identified during fieldwalking and then reexamined. Of these, 101 were eventually defined, and four broad site size classifications were used: classes 1–4 in ascending order. Class 1 sites consist of potsherd scatters that are 0.01–0.10 ha., the average covering 0.042 ha. This classification accords well with the lowest size ranking of EM, MM, and Roman sites in the neighboring Vrokastro area (Hayden 2004a). Where architecture is visible, it usually consists of the foundations of a single building, such as at Minoan sites 9, 25, 35, 59, and 68 (Fig. 5). The finds (storage, cooking, processing implements) suggest common domestic functions, and the terms “farmhouse” and “farmstead” are used for convenience, although it is recognized that such buildings may have had diverse forms and various economic, political, and perhaps even military roles in larger settlement systems (cf. Bevan 2002). Minoan houses vary in size: the house

excavated by Boyd at Avgo (site 87) measures about 20 x 22 m (0.044 ha.), and the area of the artifact scatter is essentially confined to the terrace on which the building was constructed (ca. 0.05 ha.). Another typical Minoan house is site 59 (Pl. 8B), where the nearly square cyclopean foundations (11 x 10.5 m; 0.011 ha.) have an associated scatter of 0.03 ha. Slightly larger is the megalithic building at site 68 (16.5 x 15 m; 0.024 ha.), whose scatter conforms to the average 0.04 ha. The Minoan house at site 34 on Mount Chomatas (Chrysokamino Habitation Site) represents the larger end of the scale, measuring 25 x 20 m (0.05 ha.) (Betancourt and Floyd 2000–2001; Floyd 2000; Betancourt et al. 1999). Ancillary architectural and ceramic remains cover a significantly larger area (0.14 ha.), suggesting a larger class site, although it is likely to have been only a single farmstead. The sherd scatter actually extends downslope covering an area of almost 0.60

THE ARCHAEOLOGICAL SURVEY

ha., thus removing it from the category of class 1 sites, even if much of this material might be slope wash debris from the house itself. The problem of strictly classifying multiperiod sites such as site 34 was occasionally insurmountable, since insufficient provenience data were available to estimate accurately site sizes in all periods. The lowest level, class 1 category, thus includes a variety of types of buildings and surface scatters. Most are evidently single structures with and without ancillary buildings, and if domestic, probably housing 1–2 families. Class 2 (“hamlet” or “farmhouse cluster”) is defined by scatters that are 0.10–0.60 ha. in size, averaging 0.25 ha. The corresponding ranking in the neighboring Vrokastro area (Hayden 2004a) includes middle-size sites in EM, two size classes in MM and Roman, and the smallest of the lowest level of LM I sites. Most sites in this category at Kavousi are multiperiod, causing problems in determining periodic changes in size. Site 69 at Panagia Skali, for example, is a continuous scatter, 0.375 ha., and there is no spatial differentiation of periods on the site surface (Figs. 5, 7). Groups of pottery recovered from two separate road scarp exposures, however, were MM I–II and MM III– LM IA respectively, suggesting concentrations of activity in different areas of the site in different periods. The earliest identifiable site in this category is site 8 at Alykomouri (FN–EM III), whose minimum 0.12 ha. size suggests a small hamlet. The Early Iron Age site at Melisses (site 85) in the Avgo Valley (0.25–0.30 ha.), and Panagia Skali (site 70, 0.13 ha.) represent the most common forms of this apparently long-lived type of site. The Roman hamlet in Chordakia (site 37) is roughly 0.42 ha. in size, with a discrete concentration of tile, querns, pottery, a well, and olive-processing equipment (Fig. 6). While the actual sherd density makes this site appear as the larger, class 3, category on the map (Fig. 6), the smaller estimate is based on the distribution of tile and the overall topography of the site. The Minoan equivalent of site 37 in Chordakia is just across the valley at site 28 (Katsoprinos); here the Roman and Byzantine remains have partially obscured the MM I–LM I scatter of 0.35 ha. The number of houses in this class 2 category will of course vary depending on period and site function. Unfortunately, architecture is rarely exposed or sufficiently preserved to make even a

35

rough estimate of the number of houses possible. Furthermore, the functions of small class 2 sites will have differed very little from those in the class 1 category, while the largest class 2 sites were essentially small villages—that is, class 3 in size. There is no attempt here to construct a strict size-function correlation, only to provide a general typological framework for interpreting the data. Excavated examples in the class 2 category are few. Myrtos Phournou Koriphi (0.18–0.24 ha.) had probably five or six households, according to Whitelaw’s (1983) interpretation, while class 3 sites in the survey zone, such as LM IIIC Vronda (0.60–0.70 ha.) and Chalasmenos (0.64 ha.), contained twelve to sixteen house units (cf. Nowicki 1999a). As a rule of thumb, the modern hamlet of Trapeza in the Avgo Valley, occupying 0.375 ha., contained fifteen to twenty families, although the houses themselves are, on average, half the size of EM II Myrtos or LM IIIC Vronda. Class 2 sites, therefore, should probably have contained between six and eight households. Class 3 sites (“large hamlets” or “villages”) have a range of 0.60 ha. to 1.0 ha. and include the large Early Iron Age settlements of Vronda (site 71), Kastro (site 80), and the Roman sites 53 and 57 (Figs. 5, 7). It is interesting that relatively few Bronze Age sites fit this category. Notable are sites 11, 15, 17, and 27 in the Kambos plain, which might represent small villages in the Neopalatial period (Fig. 6). Architecture is evident only at sites 17 and 27, but the remains are not sufficiently extensive or clearly preserved to make house-size designations. Sites in Vrokastro corresponding to the Kavousi class 3 category would include the largest EM and LM IIIC sites and intermediate size MM and LM sites (Hayden 2004a; cf. Nowicki 2000a). Class 4 sites are any scatters larger 1.0 ha. Included in this category are Tholos (site 1), which is a minimum of 2.59 ha.; site 5 at Hagios Antonios (3.0 ha.) in LM I (Fig. 6); and the Azoria hilltop (site 71), which occupies 1.14–1.5 ha. from LM IIIC–LG (Fig. 7); in the seventh century the site was as large as 15 ha. (Haggis et al. 2004). The sites in this category are clearly large village or small townsized settlements, and architectural remains are sufficiently preserved to justify the classification. The expansive concentration of pottery southwest of Kavousi village (site 56) is technically a scatter in this class, but, as mentioned above, local conditions

36

KAVOUSI I

suggest that the material is not evidence of habitation but of slope wash and discard debris from surrounding sites at Kavousi village and Sta Lenika. The functional distinction between the various size classes is not easily determined in all cases, even if site sizes are accurate. Topographical conditions, variable house sizes, external use areas, chronology of use, building function, and formation processes will ultimately have affected site sizes, helping to blur the distinction. The purpose of the classification is only to provide a rough means of assessing, or perhaps visualizing, possible site functions and interrelationships in different periods. The largest order, large class 4 sites, are clearly substantial nucleated centers. Similarly clear in morphology are the smallest class 1 sites. Although some may have specialized economic or political functions within a site hierarchy, the evidence suggests single farmsteads. Class 2 and 3 sites are hamlets and villages of various sizes from roughly 1,000 2 to 10,000 m , and this diversity in size indicates variable populations from three or four households in the smallest to fifteen or twenty in the largest. Obviously the distinction between small class 2 and large class 1—as between small class 4 and large class 3—will have been negligible. The attribution of functions to the site size classes is dependent on physical context (environment, natural resources, and chronology) and interrelationships suggested by broader regional patterns on various scales—the presumed structure formed by the hierarchical links between parts of the settlement system. The economic functions of the lowest and highest-order settlements, represented by class 1 and 4 sites respectively, may be indicated by their specific sizes and locations. Class 1 sites are small, inland, isolated buildings that should have depended economically and perhaps also politically on a higherorder site or sites within a complex settlement system. It is unlikely that they functioned independent of a larger system of settlement. Similarly, class 4 sites, because of their large size (and nucleated population) and either central (e.g., Azoria [site 71]) or coastal (e.g., Tholos [site 1]) locations, had functions involving regional and extraregional economic spheres and arrangements of settlement (Figs. 6, 7). The role of class 4 sites is necessarily complex, as their size alone suggests the existence of a dependent territory and other smaller sites in the vicinity. The interdependent relationship

between the countryside and the center required organization of production, exploitation of available resources, and extraregional contacts to maintain the population of both the center and its territory. If we assume the basic social unit of class 1 sites to be the household, then class 4 sites must have required an internal organization to structure the relationship between households, and between the town and outlying sites. By definition then, if class 4 sites—being the highest-order centers in the study area—are central places, then two further premises may be permitted: (1) they are economically connected in some way with neighboring central places outside the survey zone; and (2) the resident population is likely to have been supplied its needs by means of exchange with the lower-order settlements in the region and with neighboring higher-order settlements outside the area. Class 1 sites, therefore, as the lowest order, should have interacted with class 4 sites interdependently. Given the small size of the sample area, site sizes, and the relatively small scale of the necessary organizational structure, little distinction should be made between “urban” and “rural” settlements. Even in an essentially rural landscape such as Kavousi, a complementary relationship seems likely to have existed between contemporaneous sites. Large class 2 and class 3 sites are more difficult to interpret because they indicate complex site interrelationships that may have changed drastically from period to period. These hamlets and villages are, however, of primary importance in reconstructing the settlement history in the Kavousi region and Mirabello area, if only a few generalizations can be made about their form. Class 2 and 3 sites existed as unranked clusters in the Early Iron Age (LM IIIC–Geometric) or as dependents of first-order centers in the Roman period. In the Early Bronze Age they were isolated and essentially selfsufficient, if not autonomous villages. In the Neopalatial period, they were second-order sites, perhaps connected hierarchically to class 1 and 4 settlements. It is interesting that there are no certain sites of class 3 size in the Protopalatial period, even though this is a well-defined classification at Vrokastro in the Middle Bronze Age (Hayden 2004a). Sites 28, 34, and 69 are possible examples, but each contains a substantial MM III, Neopalatial, or LM III component that has obscured the configuration of the MM I–II material.

THE ARCHAEOLOGICAL SURVEY

There is probably great variation in the function of class 2 and 3 sites depending on period, environment, and the function of the broader settlement system—that is, the distinct expressions of economic, social, religious, or political relationships among sites. And there is no strong reason to believe that they were uniformly or even primarily “secondorder” sites or intermediate economic functionaries between farmhouses and towns. On a social level, such sites might be very important as expressions of local identity in a regional context. On the one hand, class 4 sites are considered large villages or towns, consisting of a nucleated population whose identity must be linked to a community that transcends or bisects familial ties, in contrast to class 1 and small class 2 sites—the farmsteads—which are conceivably composed of single families, necessarily linked to larger-scale economic and social hierarchies. On the other hand, the class 2 and 3 hamlets and villages, because of their small size, are likely to be lineage based, but large enough to remain strongly connected to both the land and a kinship group. They represent semiautonomous corporate groups whose link to the landscape permits expression of their local identity irrespective of a larger-scale economic systems or regional political hierarchies. Consideration of the environmental context and potential interrelationships between the class 2 sites is essential in defining the settlement patterns in MM, LM, and Roman periods. While individually they appear to be economically dependent on and in turn support a higher-order center, such as a class 3 or 4 village or town, their distribution suggests a complex regional structure. In the late Prepalatial and Protopalatial periods (Figs. 10, 11), for example, farmsteads and small hamlets (class 2) are clustered together in close groups around water supplies and isolated areas of arable land. The sites in such clusters are probably interdependent economically because of their number, close proximity, and shared resources. Their spatial interrelationships on this microregional scale could be culturally determined, representing important aspects of the social structure in the region. This pattern, which is formative in EM III–MM IA and fully developed in MM I–II, is particularly interesting, since there is no higherorder center in the study area that would fill out a hierarchy, demonstrating the economic dependence of these sites on a village or town. While such a center may have existed at the location of the modern

37

village of Kavousi, or somewhere outside the survey area, the existence of a site hierarchy does not in itself explain this dispersed pattern of clusters of small sites, their function in the region, or the cultural and environmental variables responsible for their formation. Could these distinct groups of farmsteads and hamlets have functioned like the nucleated villages of the Neopalatial (class 4) and Roman (class 3) periods? In the Neopalatial period, single class 1 farmhouses exist in the same locations as the Protopalatial clusters. Their isolation in the countryside suggests close ties to a class 4 center. Recognition of the MM I–II cluster pattern reinforces the notion that determination of site size categories, even if consistent from period to period, has little value in and of itself in determining sociopolitical relationships across the landscape. The interpretation and attribution of site functions within the region is context-related, while the determination of context (environment, chronology, and culture) is dependent on the analysis of patterns of material culture on variable spatial scales. The categories of site sizes described above derive their meaning less from the surface assemblage than from settlement patterns, and given the diversity of settlement patterns through time, the individual functions of the various site size classes must also change diachronically. There is, nevertheless, a tendency to attribute meaning—such as the “farmhouse,” “hamlet,” “village,” and “town” designations used here—to site size classifications; the nomenclature presumes the nature of intersite relationships on a regional scale, often without defining the parameters of the organizational structure in the region. Site functions, based on size criteria, are thus only really meaningful on a regional scale and in the specific chronological context of a defined settlement system. The relationship between sites defined on size criteria is the essence of ranking in a spatial context, and such rank-size designations anticipate the structure in the region by presuming the region’s parameters in physiographic (topography) or cultural (artifacts) terms. The definition of the region is however, time-dependent, expanding and contracting on spatial scales. The “scale” is the specific area in which all components of a system are operable, and in which a range of intersite relationships are expressed in the archaeological record. It is arguable that, while the present survey zone adheres to boundaries of Kavousi in the 19th

38

KAVOUSI I

and 20th centuries, in the Roman period, and more recently in 1999, this same region became part of the political territory of Hierapytna (modern Ierapetra). In the Classical and Hellenistic periods those areas of the Kavousi region that did not belong to the city-state of Larisa were disputed territory (Haggis et al. 2004; Guizzi 2001; Bowsky 1994; Bennet 1990; Guarducci 1942). In the Bronze Age, the Kavousi “region” is cloudier still, even if a definition is attempted by first locating a functional center, and then determining the sphere of a dependent region. In all likelihood the region itself might fit better into the Periphery Model outlined for the Nemea Valley and Southern Argolid (Wright 2004), where the understanding of sociopolitical configurations will require more subtle material patterns derived from consideration of contingent areas and supra-local scales of analysis. The Kavousi survey zone lacks a palace in the Neopalatial period, a usual sign that it must have been part of a larger territory controlled by Gournia to the west (cf. Soles 2004; 2002; 1991; Watrous et al. 2000; Watrous and Blitzer 1999; Warren 1985; 1984), Petras or Zakros to the east (Tsipopoulou 2002; 1997), or by a palace, as yet undiscovered, underlying modern Ierapetra to the south (Watrous et al. 2000, 477; Watrous and Blitzer 1999; 1995). The latter is an intriguing possibility, all the more striking in light of analogous Classical, Roman, and modern settlement patterns in the Isthmus. Yet the multiplicity of various centers around the Mirabello Bay only compounds the problem of interpreting the Neopalatial pattern in terms of unilateral hierarchical structure (cf. Hamilakis 2002; Cunningham 2001). And the language of economic and political power relationships, such as “control,” “influence,” and “territory,” needs to be redefined within specific

cultural contexts and by means of a new theoretical framework (cf. Soles 2002; Knappet and Schoep 2000; Cunningham 2001; Knappett 1999). What were the interrelationships among sites such as Hagios Antonios and Tholos, and coastal towns of Priniatikos Pyrgos, Pseira, and Mochlos (Figs. 1, 2)? How did production and distribution in these areas operate within overlapping political and economic spheres? As Carole Crumley suggests, “The more complex the system of settlement (and the higher the level of sociocultural integration), the more likely it is that the system’s functional lattices overlap” (1976, 68). The broader region itself— Cherry’s highest level autonomous political unit (1984, 40)—is diachronically variable, and so are the form and function of operative centers. Furthermore, if the functional lattices of a complex system overlap, it is possible that they operate on variable spatial scales. That is to say, in Kavousi, economic systems might not manifest themselves in material patterns similar to that of political, religious, or social systems. The meaning, therefore, of the site size ranking in any period requires a framework to distinguish the function and character of the relationship between sites. The rank-size classification of sites here is only one way of initially ordering the data. As an interpretive tool, allowing us to establish regional configurations of settlement and interrelational hierarchies, it does not predict the meaning of regional systems. The categories provide formal definitions and spatial dimensions only within specific environmental and cultural contexts. The aim of the synthesis in Chapter 5 is to interpret the contexts toward defining various functional relationships between sites of various sizes, through time and in a variety of environments.

Results Six sites were discovered with Final Neolithic pottery (sites 6, 8, 16, 24, 32, and 71); there are possible remains at sites 31 and 67, and excavations at the Chrysokamino Habitation Site (site 34) and Vronda (site 77) revealed FN sherds in the lowest levels (Fig. 8). The sizes of Neolithic sites were generally difficult to determine. The largest quantity of

Neolithic pottery came from the excavation debris from the Kavousi village site (secondary deposit site 24), so the actual size of the site is unknown. In the Kambos plain, site 16 produced a concentration of Neolithic sherds in a 15 x 2.0 m swath along the site’s eastern edge, but this strip is also the bulldozer furrow of the asphalt road that connects Kavousi

THE ARCHAEOLOGICAL SURVEY

and Tholos. No pottery was recovered on the east side of the road, but the concentration on the embankment suggests that the discovery of the site, however large it may originally have been, was dependent on the exposure of subsurface material by the 0.80–1.00 m deep cutting for the road. Neolithic pottery was recognized at EM I–II sites 6, 8, 31, and 67, but the amount of material was too small to make accurate determinations of site sizes and functions. The remains from site 16 suggest the location of a single isolated house. Since material here and at several other sites was produced by subsurface exposure or actual excavation, it is possible that other remains of this period have been buried by recent alluvium or post-Neolithic occupation. Recent excavation at Azoria (site 71) has demonstrated widely dispersed distribution of Final Neolithic remains across an area of about 1.0 ha. of the South Acropolis, suggesting a site of considerable size (Haggis et al. 2004). Ten sites are known that span the periods of Early Minoan I–II (Fig. 9), indicating no significant growth in the occupation of the Kavousi area. The evidence points to three main settlements in the region—probably not larger than class 2—at Alykomouri, Kavousi village, and Vronda. Two special function sites were found at Hagios Antonios and Theriospelio, while four other sites had only traces of EM I–II activity. Alykomouri (site 8) produced sufficient evidence to determine the size of the site. The exposed surface area is under 0.20 ha., and like Myrtos Phournou Koriphi, probably represents a hamlet of five or six households. The associated rock shelter at Hagios Antonios (site 6) was no doubt its cemetery. The other main EM I–II site in the region is Kavousi village (site 24), but, as in the Neolithic, the size is unknown. EM I–II remains at Tholos (1), Vronda (77), and possible evidence at sites 62 and 87 are difficult to interpret since postEM occupation has obscured site sizes and functions in these periods. Much EM I–II material was recovered by E.H. Hall in her excavations at Theriospelio (site 31), a date range that is confirmed by A.A. Zois, who visited the cave and recovered several diagnostic sherds that are now in the Hagios Nikolaos Museum. The cave itself is large enough for habitation or cult use, but its location, on the precarious west slope of Mount Chomatas, and its entrance, which is barely 1.0 m high, are hardly suitable conditions for regular habitation. It was

39

surely a temporary, burial, or cult site. The recent excavations at the neighboring sites at Chrysokamino (sites 32 and 34) also produced sherds of FN–EM II in date (Betancourt 2000a; Floyd 2000; Betancourt et al. 1999). The EM II remains at Vronda are still under study, but their context suggests considerable disturbance by subsequent EM III–MM II and LM IIIC occupation. The area is likely to have been smaller than the size of the LM IIIC settlement. At Tholos, on the coast, EM I–II coarse wares were recovered and taken to be evidence of early exploitation of this important harbor. Sixteen sites had a phase or phases of occupation during EM III–MM IA (Fig. 10). During the initial process of locus recording, only a few diagnostic features of EM III–MM IA were recognized. The full range of local wares and fabrics has not been published in sufficient detail to facilitate identifying these phases in survey samples. Certain diagnostic fine ware types such as handmade deep round cups, thick-walled carinated cups with a pronounced base, conical cups with a distinctive rim band, and White-on-Dark Ware were rarely preserved well enough to be visible on the site surface. This problem was made worse by the fact that EM III–MM IA pottery was uniformly represented on multiperiod sites, requiring separation of the phase from EM II and MM IB–II material. The only exception to this pattern is Chrysokamino, which is a special-function site. Examination of recently excavated EM III–MM IA material from deposits at Mochlos, Vronda, and Chrysokamino has helped to confirm the EM III– MM IA date of Mirabello fabrics XX and XXI (Haggis and Mook 1993). The refinement in the dating of fabrics XX and XXI—common for Whiteon-Dark Wares and a range of shapes including hole mouth jars, bridge-pouted jars, lekanai, and pithoi— has permitted an adjustment in the dating of sites previously thought to be MM I–II (Haggis 1992). Pending the publication of the results of recent excavations in the Mirabello area, it is not possible to distinguish separate EM III and MM IA phases. Late Prepalatial remains are prevalent in the Hagios Antonios and Chordakia areas (Fig. 10). In the former location, specific site sizes cannot be determined, but White-on-Dark Ware at Alykomouri (site 8) and its cemetery (site 6) suggests continuity in the Hagios Antonios region into EM III–MM IA. Site 5, on the slopes of Schinias below

40

KAVOUSI I

the Hagios Antonios rock shelter, was established in EM III–MM IA, but the actual size of the settlement is unknown. In Chordakia, the evidence is clearer. Here the EM III–MM IA remains are evidently class 1 and small class 2 sites: individual houses or small groups of buildings (one to three households). Where architecture is evident, for example at site 44, foundations of a single house are categorically class 1. Remains at Vronda (site 77), Chondrovolakes (site 68), and Azoria (site 71) in the north Papoura catchment area are probably sites of similar size. This dispersal of small sites is an obvious break from the nucleated pattern of EM I–II, and new site locations occur, especially in the mountainous regions of Papoura (site 77), Bebonas (site 92), and Thriphti Argira (site 97). Sixty Protopalatial sites were recovered in all, a remarkable increase from the late Prepalatial period (Fig. 11). Only seven of these are of the “possible site” category, that is, producing such negligible amounts of MM IB–II pottery so as to preclude certain definition of size or function. The Protopalatial sites may be characterized by their small size (0.01–0.60 ha.) and the prevalence of imported “Mirabello fabrics,” coarse wares with granodiorite temper that were produced in the Kalo Chorio region. Forty-nine of the sites are unequivocally class 1 or small class 2 sizes, probably consisting of one to three households each. Some eleven sites produced MM IB–II remains but fall into larger class 2 and even class 3 and 4 categories. In each of these instances, however, the largest size of the site represents its maximum dimensions in the Neopalatial period. Sites 1 and 5, for example, are class 4, but their individual loci produced not only greater quantities of LM I pottery (and a continuous scatter across the site) but also localized and discontinuous concentrations of Protopalatial sherds. This might suggest the existence of individual, disconnected class 1–2 sites. At class 3 sites such as 11, 15, 17, 23, and 27 in the Kambos, the Protopalatial pottery was found discontinuously in small amounts all over the site. While the size in MM IB–II cannot be determined for certain in these cases, the relatively consistent high densities of Neopalatial pottery are taken to indicate the maximum spatial extent in LM I. The Protopalatial size is presumed to be smaller than the class 3 rank. The large number and small sizes of Protopalatial sites—consistently falling into class 1 and 2

categories—along with their dispersed pattern in the landscape exhibit a striking increase in population from the preceding EM III–MM IA. Even Tholos (1) and Hagios Antonios (5), in MM IB–II, appear to have been composed of no fewer than three class 1–2 sites, perhaps individual farmsteads or household units, exploiting the land around the houses themselves and in the vicinity. The most likely candidate for a higher-order center on which these households could have depended is Kavousi village, which produced substantial evidence for MM IB–II phases. The size of the site in MM I–II cannot of course be determined because of later occupation. Vronda (site 77) as well may have been a Protopalatial site exceeding the class 2 category, but like Kavousi village (site 24), its actual size is not known. On the whole, there are three characteristics of the settlement pattern in MM I–II: (1) the sites are small, and there is no evidence of site size differentiation suggesting a functional or organizational hierarchy; (2) the habitation sites previously occupied in EM III–MM IA uniformly continue in use into MM IB–II; and (3) some fifty new sites are founded during MM IB–II, but generally in areas previously occupied in late Prepalatial times. Continuity from the late Prepalatial period is the tendency, but EM I–II sites such as Alykomouri and the Hagios Antonios rock shelter, and special-function sites such as Chrysokamino and Theriospelio, all seem to go out of use in EM III or MM IA. In general, the results of the survey at Kavousi show remarkable changes between Prepalatial and Protopalatial periods, with the EM III–MM IA phase comprising an important transition. It is interesting that Kavousi village (site 24), which shows continuous occupation from Final Neolithic, lacks evidence for EM III–MM IA occupation in the excavation sample. The predominance of Protopalatial pottery at that site, evidence of reoccupation by MM IB–II, emphasizes this period of discontinuity at the end of the Early Bronze Age. There is a reduction in the total number of recorded sites by LM I (Fig. 12). Although MM III pottery was recovered in substantial quantities from sites such as 69, it could not be distinguished from LM IA in a sufficient variety of wares and fabrics to determine the details of its distribution. A total of forty-one sites were found with MM III–LM I pottery, while an additional seven sites were called “possible.” This drop in site numbers is

THE ARCHAEOLOGICAL SURVEY

less remarkable than the changes in the overall pattern. The MM III–LM I sites show variation in size, a marked contrast to the regular dispersed farmhouses of the Protopalatial period. Two class 4 sites, 1 and 5, were clearly large settlements (Figs. 6, 11). While class 1 and 2 sites continued in use from MM I–II or were new foundations, several large class 2 and class 3 sites emerged by LM I. The latter—sites 11, 15, 17, and 23—must have been hamlets of as many as fifteen houses. These new settlements, neither small farmsteads nor full-fledged villages, are obviously linked to a spatially discernible hierarchical order. On the whole, they seem to have reached their maximum size in LM I. The evidence for earlier Protopalatial occupation on such sites is slight or spatially discontinuous. In marked contrast, the class 4 sites have well defined and substantial MM I–II remains, and several Protopalatial class 1–2 sites continue in use or are reoccupied in the Neopalatial period. Some class 2 sites do display new Neopalatial building. For examples, Site 28 on Mount Chomatas at Katsoprinos above Chordakia, and the Habitation Site at Chrysokamino (site 34) are large MM I–LM I structures whose Neopalatial pottery was found concentrated in the area of visible architecture. Recent excavations at Chrysokamino (Betancourt and Floyd 2000–2001; Floyd 2000) have confirmed and expanded the known date range and occupational history of the site. Excavations at Vronda (site 77) also recovered evidence for Neopalatial activity (L.P. Day, personal communication), and while the spatial extent cannot be defined with certainty, it is clear that this site remained a choice settlement location throughout the Bronze Age. A total of ten sites produced evidence of LM IIIA–B occupation (Fig. 13). Three sites are burial locations, discovered through excavation. At the Habitation Sites on Mount Chomatas and in Avgo, the LM IIIA–B material (Chrysokamino Habitation Site 34 and site 87) material was brought to light through excavation. In neither instance was the Postpalatial material prominent on the surface during survey. These sites represent single Neopalatial houses, reoccupied in LM III. The best evidence for LM III activity, possibly reoccupation, comes from Kavousi village (site 24), Hagios Antonios (site 5), and Chordakia-Kephalolimnos (site 36). The latter, site 36, is a single house, class 1–2 in size, similar to sites 34 and 87 mentioned above. These houses were

41

founded in Proto- or Neopalatial periods and then were reoccupied in LM IIIA–B. The only evidence of larger-scale Postpalatial activity is in the class 4 settlement at Hagios Antonios (site 5), where LM IIIA–B remains were recovered from at least two separate areas of the Neopalatial town, suggesting not uniform or widespread occupation, but characteristically sporadic resettlement following the destruction in LM IB–II. LM III activity is probably underrepresented in the sample across the survey zone, since identification of sherds of this period depended on recognition of fine-ware shapes and decoration rather than coarse fabrics. While there are distinctive cooking-ware fabrics of this period, they are frequently difficult to distinguish from local phyllite-quartzite tempered Protopalatial wares. Although a number of very distinctive Mirabello-area LM IIIC fabrics do appear to begin as early as LM IIIB, they were not apparent in the survey samples. On the whole, LM IIIA–B pottery was hard to recognize on LM I sites, and caution is recommended in predicting the extent of depopulation in the Kavousi area. Nine settlement sites are Early Iron Age in date (LM IIIC–Archaic), while four cemeteries and two shrines were known from previous reconnaissance and excavation (Fig. 14). Class 2 (70 and 85) and class 3 (77 and 80) sites are large settlements, although Azoria (site 71) is clearly a class 4 site (1.5–3.0 ha.), and by its Orientalizing and Archaic phases it exceeds 14 ha. in size (Fig. 15). In the Avgo Valley, site 85 is a large class 2 site, while sites 89–91 represent smaller loci of Early Iron Age– Archaic activity. The latter sites on the Trapeza Ridge are partially obscured by the phyllite alluvium, dense vegetation, and stone debris on this steep slope. At site 89, for example, an Archaic deposit (pithos, cooking ware, fine ware) was discovered in a road scarp, about half a meter below the ground surface. Such evidence could mean that sites 89–91 are simply exposed areas of one or two larger sites (class 2 or 3). In the Monastiraki area on either side of the Cha Gorge are the sites Chalasmenos (site 99) and Katalimata (site 100), which fit the class 3 category (Tsipopoulou 2004; 2001; Coulson et al. 1995; Haggis and Nowicki 1993). No Classical sites were found, and only trace remains of the Hellenistic period (sites 42 and 71) exist in the survey area. The region has 23 Early and Late Roman sites (Fig. 16). The most striking

42

KAVOUSI I

feature of the Roman-period settlement pattern is the emergence of hamlets or small villages of large class 2 or class 3 size (sites 17, 28, 37, 53, and 57), which are dispersed at regular intervals at the edges of the Kambos. While farmsteads do exist (sites 40–42), the sites tend to be agglomerations of probably several houses. Special-function sites such as Tholos (1) and Sta Lenika (57), have very large, well-built warehouses, and site 39 at Agriomandra was, like Tholos, a port facility. Cemetery sites

were also recovered at Chordakia (30) and Hagios Antonios (5). Tile is ubiquitous and olive oil-processing equipment is evident at the larger sites, suggesting economic specialization in arable zones. Two sites in the north Isthmus and Kamina areas, but lying outside the intensive zone (Fig. 17: 94 and 96), produced olive-oil processing equipment; both had a trapetum mortarium, and site 94, a plaster-lined oil-water separation tank.

4

The Pottery

This chapter presents an overview of the pottery recovered and documented during the site-recording stage of the survey and an outline of the Neolithic, Bronze Age, and Early Iron Age coarse fabric typology, which is published in greater detail in Appendices 2 and 3 below. The bold-faced numbers used throughout this chapter are references to the object catalog numbers located in the site gazetteer (Chapter 6): “6.4,” for example, refers to site 6, sherd number 4, in the object catalog for that site. The artifact is usually the smallest unit of cultural data recoverable from the landscape in surface survey. Pottery sherds are durable artifacts, frequently surviving both natural and human postdepositional formation processes. They are ubiquitous and obtrusive across the Aegean landscape. The form and composition of pottery can change through space and time as a result of varying local environments and resources, technology, sociopolitical systems, as well as cultural contexts of production, distribution, and consumption. Identifying evidence for production and deposition of pottery is a primary concern in ceramic provenience studies. Intensive archaeological survey, measuring the

density and distribution of potsherds on the ground surface, can be of help in identifying production locations and spheres of distribution. Several factors account for a sherd’s present location and condition: deposition in a primary use or discard context, secondary deposition, and a number of human or natural postdepositional processes—such as erosion, fluvial action, and various agricultural activities involving landscape modification. The complementary stratigraphic scopes of both excavation and survey offer the opportunity to define patterns of chronologically, regionally, and functionally diagnostic ceramic types across the landscape with a finer spatial resolution than that derived from a few excavation sites deposits. Ceramic fabric areas are locales in which raw materials can be linked through petrographic analysis to specific ceramic fabrics, and production zones are likely locales within the fabric area that comprised the manufacturing centers (P. Day 1997, 225–227; 1995; 1991, 171–173). As clay and inclusion sources are identified in the field, and as actual kiln sites are discovered through excavation or survey, production centers can then be determined with reasonable

44

KAVOUSI I

accuracy. The size, form, and ultimately the borders of the distributional zones, as well as their chronological variability are discovered through intensive survey (Haggis 2000). Determining the spatial distribution of similar artifact types, such as pottery, is also the traditional means of establishing hypothetical “culture regions”—areas comprised of homogeneous contemporary artifacts that are distinctly different from that of neighboring culture regions. A region defined by the distribution of similar ceramic types can also be called a “ceramic sphere.” The map of such spheres would be variable in qualifying components: these can be wares, shapes and forms, painting styles, fabrics, or a combination of any of these features. The production area is spatially narrower than the ceramic sphere, comprising the area of the actual distribution of specific vessels of certain provenience, that is, from a def inite place of manufacture. Information derived from surveys may eventually demonstrate the extent of distribution and relative densities of fabrics from different production centers in order to refine our understanding of ceramic spheres and material culture regions. These are potentially different from those simply comprising stylistically similar ceramic features (such as shape or surface decoration) and the actual distributional zones are complex and changing through time. A spatial model might predict a pattern of changing densities of specific fabric types, the highest level being at the locale or center of production, decreasing outward from that center. This pattern, however, is dependent on a myriad of social, political, and economic variables that define the cultural contexts of exchange, transport, use, and ultimately deposition—the locale of consumption (Day and Wilson 2002; 1998). Production zones may also be comprised of one or more competing or complementary production centers. It is assumed that the scope of the distribution of the pottery itself represents a region in which social, political, or economic systems affect, if not control production and exchange. More complicated still, and potentially more complex systemically, is a pattern of distribution and exchange involving two overlapping ceramic spheres of distribution, each with identifiably distinct production locales and traditions. The Kavousi survey zone is one such area whose coarse wares, in the Bronze Age and Early Iron Age,

illustrate overlapping and potentially competing spheres of production and distribution. The chronological variability of the coarse-ware ceramics, derived from two distinctly different production zones and geological environments, offered an opportunity for evaluating small-scale patterns of ceramic exchange in the Mirabello and north Isthmus region. The movement of pottery in and around the north Isthmus area is the result of trade or exchange of the pots themselves, such as storage and processing vessels (pithoi, larnakes, lamps, tripods, scuttles, trays, and basins) or containers for the transport, trade, or redistribution of food or other products (hole-mouth jars, pithoid jars, oval-mouthed amphorae, jugs, and piriform jars). Finally, some vessels may have been moved between areas or regions, not as products for exchange, but as personal possessions or goods carried by migrating populations or individuals moving only short distances. In certain cultural contexts, pots are moved for specific activities away from the locale or region of manufacture—beehives for honey production, larnakes for burial, amphorae and jars for transport of various products, and even cooking vessels as equipment and supplies for agricultural, pastoral, or cult activities. The pattern of ceramic consumption must therefore be determined archaeologically from the use context—the depositional context of the artifacts themselves. This de facto refuse is of course best recovered and understood at the level of excavation. The broad distribution of objects recovered in survey may provide a framework for addressing questions of the cultural context of microregional exchange systems. Study of the Kavousi Bronze Age and Early Iron Age ceramics was, however, a means of dating sites in the survey. The coarse wares provided a consistently plentiful range of ceramic artifacts with which to determine chronological variability in fabric types, and also to explore the cultural or economic meaning of that same variability. During the initial stages of fieldwork, the Bronze Age, Early Iron Age, and Archaic and Roman ceramic sequences were not sufficiently understood or adequately published for the Mirabello and Isthmus regions to provide a local pottery chronology for survey. This is surprising given the intensity of fieldwork in East Crete during the decades before World War II. On the whole, early publications are highly selective and inconsistent in the range of material presented: potsherds and coarse wares were

THE POTTERY

infrequently published in favor of showing whole fine-ware vessels or attractive and unusual pieces, and stratigraphy and provenience were secondary concerns. Nevertheless, attempts were made at reconstructing local and regional Bronze Age sequences with important results (Hall 1904–1905; Hawes et al. 1908; Seager 1906–1907)—probably in direct response to parallel stratigraphic concerns at Knossos. But even these local chronologies remained problematic or controversial (e.g., Haggis 1993b; Watrous 1994; Zois 1976). The practice of publishing material that was excavated at the turn of the 20th century has produced useful results if the specific provenience is well known (e.g., Hayden 2003b; Betancourt and Silverman 1991; Betancourt 1983; 1977; Gesell, Day, and Coulson 1983; Foster 1978). But generally the archaeological context of such material is not understood well enough, and the sample selected is much too limited to reconstruct detailed stratigraphic ceramic sequences, which can give the kinds of details required for an intensive survey. An important aim of the reexploration of the Mirabello area has been to reexamine old excavation sites and to define and refine local and regional Bronze Age and Early Iron Age chronologies. Concurrent excavations at Pseira, Mochlos, and Kavousi produced stratified contexts with comparative wares and fabrics that were used to define the Kavousi-area fabric types for the Bronze Age and Early Iron Age (Betancourt and Davaras 2003; 2002; 1998; 1995; Brogan, Smith, and Soles 2003; Barnard and Brogan 2003; Floyd 1998; Coulson et al. 1997; Mook 2004; 1999; Mook and Coulson 1997; 1993; Soles and Davaras 1996; 1994; 1992). Recent excavations at Kalo Chorio, Evraïka, Chalasmenos, and Katalimata, and finally Chrysokamino have greatly augmented the range of stratified comparanda for survey material at Kavousi (Tsipopoulou 2004; Tsipopoulou and Coulson 2000; Nowicki 2003; 2002; Betancourt 2000a; Betancourt and Floyd 2000–2001; Floyd 2000; Betancourt et al. 1999; Haggis 1997; 1996c; Haggis et al. 1993; Coulson et al. 1995). Myrtos Phournou Koriphi, Myrtos Pyrgos, Malia, and Petras also provided useful ceramic comparisons, although these sources are far from the Mirabello area (Tsipopoulou 2002; Cadogan 1990; 1977– 1978; Pelon 1970; Warren 1972). The purpose of examining pottery from recently excavated sites was to define features and fabrics

45

that could help to date sherds found during survey. Furthermore, the similarity of Bronze Age pottery from different areas of East Crete—especially the area between the west Siteia Mountains and Lasithi—presented interesting questions of ceramic regionalism and the definition and meaning of spheres of economic and cultural influence. Common Prepalatial and Protopalatial forms and fabrics from the Isthmus, Mirabello, and Lasithi regions have instigated discussions of “culture regions,” regional and interregional connections, patterns exchange and spheres of political influence (cf. Haggis 2000; Knappett 1999; Cadogan 1995). Ceramic regionalism in Bronze Age Crete—the diversity of forms, wares, fabrics, and their spatial distribution—presents problems in establishing relative chronologies, especially because we rely on central Cretan or Knossian sequences. It is however, this same regionalism that permits the definition of ceramic spheres and production zones. The study of regionalism facilitates the reconstruction of the economic or social behavior that underlies apparent patterns of ceramic distribution. Petrographically based provenience studies have begun to assess ceramic regionalism in prehistoric Crete (Day and Wilson 1998; P. Day 1997; 1995; 1991; Whitelaw et al. 1997; Wilson and Day 1994). This work is radically changing the potential value of ceramics as archaeological evidence for complex systems with overlapping social, economic, and political spheres. Studies have focused primarily on the definition of production zones and workshops, the centers of fabric areas, and broad interregional patterns that are based on samples from individual stratified assemblages. Peter Day defined two distinct fabric areas—perhaps representing two main workshops in the north Isthmus and Mirabello areas—that have a significant bearing on the Kavousi region in the Bronze Age and Early Iron Age (Day 1997; 1995; 1991). On Bronze Age sites in the survey zone, there are fabrics with granitic-dioritic inclusions. Because the source material comes from the area around Gournia and Kalo Chorio-Istron, it is assumed that the production center and the pots themselves originated on the western side of the north Isthmus, that is, near the clay and temper sources (Hayden 2004a, 23; Watrous et al. 2000). The second fabric area is east of the Isthmus, including the Mochlos and Kavousi areas. Here siltstone and phyllite fabrics, from the phyllite-quartzite

46

KAVOUSI I

series, originate (Barnard 2003; Haggis 2000; Myer, McIntosh, and Betancourt 1995; P. Day 1991). Understanding the relationship between ceramic spheres, culture regions, and areas of political or economic influence is dependent, in the first instance, on mapping of the distribution of the pottery. Pottery survives archaeologically and may indicate various activities such as the movement of produce or the exchange of the vessels themselves. Looking at Bronze Age ceramic regionalism inevitably invites questions of organizational structure. Where and in what conditions was the pottery produced and distributed? Considering the organization of ceramic manufacturing and exchange leads, in turn, to the difficulties in drawing survey boundaries around preconceived notions of complex forms of organization or state societies. Does the culture region, defined primarily through ceramics, correspond to other analytical frameworks—such as site size hierarchies or architectural styles—for defining systems of settlement? Gerald Cadogan’s arguments for a Lasithi culture region—a political and economic territory of an Old Palace period state—is striking in that it defies traditional topographical and locational analyses of potential boundaries (Cadogan 1995; 1994; 1990; Poursat 1987). What is compelling in Cadogan’s arguments for a Protopalatial Lasithi “state” is the complexity of overlapping functional lattices—ceramic systems, metallurgy, administration, religion, and agricultural and pastoral production—and the potential of pottery production and consumption to provide key evidence for sociopolitical interaction. While the homogeneity of the pottery in the defined region was Cadogan’s primary evidence, Knappett’s recent reanalysis of MM II pottery from Malia and Myrtos suggests that there is no evidence of direct economic interaction or Malia-based regional control of production or exchange (Knappett 1999; 1997). Knappett’s nuanced study does emphasize similarities in table wares that might indicate reproduction of elite consumption patterns across the region— evidence he says for a decentralized or segmentary state based on ritual interaction reinforcing political power through dominant ideology.

This example drawn from MM II eastern Crete might serve to demonstrate that the study of pottery sherds and their location and distribution is only an initial stage in reconstructing patterns of intra- and interregional social interaction and economic organization in the landscape. Ceramic distributional zones are—like systems of settlement—scale and time dependent; we should neither anticipate the location of a Bronze Age production center because of the proximity of a palace, nor evaluate too quickly the organizational role of a palace because of its ceramic sphere. In the north Isthmus and Mirabello regions for the Bronze Age and Early Iron Age, ceramic production zones have been identified. Survey has begun to show the potential chronological and spatial variability of the distribution of wares, especially coarse wares, with known provenience. Since the Kavousi region is located precisely in between two identifiable production zones (Day’s “fabric areas”) the chronological variability and spatial distribution of these distinctive fabrics on the ground surface are contexts for analyzing small-scale patterns of exchange between neighboring areas along the southern and eastern Mirabello coast line. The overlapping ceramic spheres, varying in spatial dimensions through time, reflect economic and perhaps sociopolitical changes throughout the Bronze Age and Early Iron Age. These patterns of coarse-ware distribution are discussed in Chapter 5. For the immediate aims of the survey, the chronological variability in coarse-ware fabrics permitted the definition of index wares, diagnostic fabrics that could be used to substantiate a date range that was uniformly wider than that indicated by the exclusive sample of fine ware on the site surface. Of the eighteen coarse fabric types defined during the survey, thirteen were found to be datable to one or more periods for sites in the Kavousi region (Haggis and Mook 1993; Appendix 2). These wares are discussed below in the context of shape and ware descriptions for each period.

THE POTTERY

47

Final Neolithic Neolithic pottery was usually, but not exclusively recovered from sites that also had an Early Minoan I–II phase. This suggested either an early discontinuous phase of occupation or continuity from the Neolithic period into EM I. The pottery conforms generally to what is called Final Neolithic (Manteli 1992; Vagnetti, Christopoulou, and Tzedakis 1989; Vagnetti and Belli 1978), and while the phase is best understood or defined in central and western Crete (Manteli and Evely 1995, 9–11; Betancourt 1990), especially at Phaistos (Vagnetti 1975), comparanda are available in closer proximity to the study area (Hayden 2003a; Nowicki 2003; Manteli 1992; cf. Vagnetti and Belli 1978; Pendlebury, Pendlebury, and Money-Coutts 1937–1938a, 16–20); recent excavations at Kephala Petras (Papadatos 2004) and Azoria (Haggis et al. 2004) should provide significant stratified habitation deposits of FN material. Final Neolithic pottery in Kavousi is distinctly different from EM I. We have avoided using the term “FN/EM I,” which assumes either a synchronous use of FN and EM I ceramic features in an early EM I chronological phase or a distinct transitional phase, during which the adoption or development of EM I styles was occurring but FN ceramic traditions were retained. Both definitions of FN/EM I presuppose that EM I ceramic features are derived from central Crete—therefore producing a lag time in their acceptance in the east—or that there was a local cultural resilience and resistance to exogenous ceramic traditions (cf. Betancourt 1999; Hood 1990). Such assumptions lack supporting archaeological evidence. While recent excavations by Yiannis Papadatos at Kephala Petras near Siteia are finally producing assemblages of contiguously stratified FN–early EM I pottery (2004), it remains to examine a complete sequence from LN to EM I in both central and eastern Crete (Vagnetti 1996, 36), in order to understand regional idiosyncrasies and shared characteristics of assemblages that are potentially diverse in chronology and function. In any case, it is important to maintain a conservative stance in defining the chronology of surface remains. The features of FN and EM I ceramic phases, however, have been distinguished in East Crete, and the characteristics of one excavated EM I assemblage are well documented (Haggis 1997;

1996c; 1995; 1993b). In spite of the stylistic variation in Final Neolithic wares across the island (Vagnetti and Belli 1978), the pottery is clearly different and generally distinguishable from that of EM I. There is yet no compelling reason to combine phases of the periods, or to assume that East Crete adopted or developed EM I ceramic features significantly later than Knossos (cf. Nowicki 2003, 14; Betancourt 1999). As Lucia Vagnetti (1996, 36) has pointed out, “We still have to decide whether, when we combine the evidence from various sites, the outstanding typological differences are due to chronological seriation or rather to regional varieties.” Final Neolithic sherds in the Kavousi area are taken to represent a pre–EM I chronological phase (ca. 4000– 3300 B.C.). In the area of Hagios Antonios, FN sherds were found at the FN and EM I–III sites of Alykomouri (site 8) and the Hagios Antonios rock shelter (site 6). The concave base (6.4) from the terrace of the rock shelter is similar to spoutless jugs and carinated bowls from Katalimata (Nowicki 2003, 17), Kastelli Phournis (Manteli 1992, 119), and bowls from Nerokourou (Vagnetti, Christopoulou, and Tzedakis 1989, 24, 32, figs. 15:6, 19:46). The exterior has a thick red exterior slip, smoothed or lightly burnished, and a gray-brown burnished interior. The fabric is a dark gray core with very dense, tiny white and gray quartz, granodiorite, and sometimes calcite inclusions. The bowl handle (8.4) from neighboring Alykomouri (site 8) has a smoothed gray-brown slip with phyllite and quartz inclusions that suggest local manufacture. The closest local parallel for this type of bowl handle is from the EM I cemetery at Hagia Photia (Davaras and Betancourt 2004, passim; the shape is called a bowl with tab handle). In the fabrics of these two pieces from the Hagios Antonios area, we see possibly early signs of two separate locales of production on either side of the Isthmus of Ierapetra: (1) the gritty granodiorite fabric of the bowl base has an origin west of Kavousi in the area of Gournia–Kalo Chorio, and (2) the phyllite-tempered handle is of local or east Isthmus origin. The area is called here “local,” because Kavousi sits squarely in the phyllite-quartzite zone. At Kavousi village (site 24), carinated bowl fragments 24.4 and 24.7 have a high polish and, like the

48

KAVOUSI I

bowl base 6.4, are probably of west Isthmus manufacture. The shape is common in Final Neolithic, and the example 24.4 has the distinctive wide vertical strap handle depressed into a wide saddle-like groove toward the top preserved edge (Nowicki 2003, 56, fig. 27: 1–2; Manteli 1992, 119, fig. 11:B). Petrographic analysis of the fabric of handle 24.4 has shown surprising similarities with examples from EN Knossos (Tomkins and Day 2001). Bowl rims 24.23, 24.24, and 24.40 are similar in fabric and surface treatment (Pendlebury, Pendlebury, and Money-Coutts 1937–1938a, 18, fig. 7). Bowls with pierced rims (6.1, 24.9–24.11), the socalled “cheese pots,” are also taken to be Final

Neolithic (Nowicki 2003, 54–62; Vagnetti, Christopoulou, and Tzedakis 1989, 24, fig. 15:9; 1989, 38, fig. 22:95–101). Of local manufacture are the phyllite-tempered wares from site 16, an open-air settlement in the Kambos plain. The cylindrical pierced lug handle (16.1) is a common type, found at Azoria and throughout the Gournia survey zone; pierced lugs and the internal handle or lug of a bowl (“baking pan” or “cheese pot”) (16.3) have parallels from Nerokourou and elsewhere (Nowicki 2003; Vagnetti 1996, 32; Vagnetti, Christopoulou, and Tzedakis 1989, p. 40, fig. 23: 102–105).

Early Minoan I–II The most prevalent pottery on EM I–II sites are coarse wares, either imports containing granodiorite inclusions from the neighboring Kalo Chorio region (“Mirabello fabrics”), or presumably locally produced wares with phyllite and quartz temper (Barnard 2003, 3–11; Floyd 1998, 179; Myer, McIntosh, and Betancourt 1995; Haggis and Mook 1993; P. Day 1995; 1991). Admittedly, the region encompassing the phyllite-quartzite geology and sediments is considerably larger than the Kavousi survey area itself, and the term “local” merely indicates the likelihood of production in the vicinity of the eastern Isthmus, eastern coast of the Bay of Mirabello, and northwest Kavousi mountains. Phyllite-quartz-tempered wares from Neopalatial Pseira, for example, originate, according to Day, in the neighboring valley and Bay of Mochlos (P. Day 1991). While there is considerable diversity in both wares and fabrics during EM I–II, two fabric types were defined as diagnostic for the period. The most distinctive fabric of EM I and possibly EM IIA is KTS type IX (Haggis and Mook 1993; cf. Barnard 2003, 10). This fabric is red-brown to orange in color with dense, sharply angular, white inclusions. They appear to be metamorphosed limestone, chert, or quartz, and the regular size, distribution, and angularity might mean that they were prepared temper. Day has suggested a provenance in the region of the granodiorites of the west Isthmus and Kalo Chorio. Bowl rims (24.16, 24.17) from Kavousi village (site 24) are examples of EM I shapes in the fabric. Rim 24.17 has a wiped interior, a common feature of

wares from Kalo Chorio (Haggis 1996c). Another EM I fabric that continues into EM II–III is KTS type II. This is a dark red-brown “cooking ware” with very fine, dense, microscopic white and yellow-white quartz particles, and occasional pieces of calcite and granodiorite. The bowl rim 24.22 is of fabric type II, and like 24.17 mentioned above, is burnished on its exterior and wiped on the interior. The pottery of EM I is better represented, and on the whole, more clearly understood than that of Final Neolithic, because of recent excavations at Kalo Chorio (Haggis 1996c). The commonest fine and medium-coarse wares are dark gray burnished (including “Pyrgos Ware”), dark-on-light painted (“Hagios Onouphrios Ware”), and orange-buff burnished wares. Avoiding the imprecise and regionally biased ware group designations, the ware nomenclature used here is derived from a current discourse on Early Minoan ware groups that have been established by Peter Day and David Wilson for Knossos and applied to the EM I–IIA assemblage at Kalo Chorio and EM sites in the Vrokastro and Gournia regions (Wilson and Day 2000; 1999; 1994; Wilson 1985; cf. Hayden 2004a; Haggis 1996c). It is surprising that dark gray burnished wares are not common in survey assemblages, given their overwhelming representation in excavation deposits (Haggis 1996c; Hood 1990; Wilson 1985). The rarity of the ware might be a condition of the pottery itself, which tends to be very fine, often thin walled and easily breakable; surfaces erode easily, frequently losing their burnished or

THE POTTERY

pattern-burnished surfaces. Some examples were recovered. The goblet or chalice rim (8.2) from Alykomouri (site 8) has a dark gray core and distinctive pattern-burnished exterior and burnished interior typical of EM I chalices. A goblet or chalice rim 24.12 from Kavousi village is also of dark gray ware, and its brick-red core is similar to EM I fabrics from Knossos (Wilson 1994, 30; 1985, 360; Hood 1990). Typical examples of dark gray burnished ware are chalice fragments from the same site that are dark gray to gray-brown (24.39) or tending toward orange-buff burnished ware (24.27). EM I–II dark gray ware sherds were also recovered from the cave of Theriospelio (site 31) in Edith Hall’s excavations in 1910 (Foster 1978). Among the pottery from the cave in the Mount Holyoke College Collection, two pieces are EM I. While Karen Foster dates a rim fragment to EM II (Foster 1978, 6), the fabric, the burnished rim band, and the wide diameter of the bowl (ca. 20 cm) indicate that the piece is, in fact, an EM I chalice; similarly, the “goblet fragment” is an EM I chalice stem and base, although the horizontal grooves on the exterior are features spanning EM I and IIA (Haggis 1996c, 671; Foster 1978, 9; Warren 1972, 157). Dark gray burnished and painted wares continue into EM IIA. Goblet bases 24.1 and 24.2 have distinctly thick cylindrical stems and vertical pattern burnishing, similar to EM IIA examples from Knossos (Warren 1972, 157, fig. 41: P39–41; Wilson 1985, 302, fig. 11: P37–41). The light gray or “fine gray” ware pyxis fragment (6.3) from the Hagios Antonios rock shelter is no later than EM IIA. A typical EM IIA light gray ware bowl rim 24.38 from Kavousi village has good parallels from Myrtos and Knossos (Warren 1972, 155, fig. 39: P24–27; Wilson 1985, 296–297, fig. 8: P1–3). This has the distinctive internally thickened rim, gray core, and gray-brown to brown polished surface; the fabric is coarser than most gray wares, with dense white quartz and black-and-white granodiorite inclusions that suggest a “Mirabello” or Kalo Chorio/Gournia area of manufacture. Dark-on-light painted wares, like the Fine Gray Ware sherds, are rare in surface assemblages. Examples are the jar rim 8.1 from Alykomouri, and jug fragment 6.2 from the Hagios Antonios rock shelter. The fabric of these pieces is distinctly EM I: a hard medium-coarse or semifine core that is pink-buff or orange-salmon in color, a buff slip and

49

orange, reddish brown, or dark brown paint (Haggis 1996c, 668). Such examples are somewhat different from EM IIA–B painted wares such as the jug fragment 8.3, which is a soft buff fabric, with a worn buff slip and dark brownish black paint. Both EM I and EM II varieties of dark-on-light painted ware are perhaps difficult to recognize in survey samples as they tend to lose their slip and painted decoration when exposed to the elements. Vasiliki Ware, although produced nearby in the Isthmus in EM IIB, is rare in surface assemblages despite the ware’s resilient fabrics and highly polished surfaces (Watrous et al. 2000; Betancourt et al. 1979). This is perhaps all the more striking when one considers the significant quantities exported to neighboring Priniatikos Pyrgos, Mochlos, and Myrtos Phournou Koriphi (Whitelaw et al. 1997). The goblet rim 24.21 with a pinched rim is a type common at Myrtos, as is the goblet 24.8. Among granodiorite-tempered fabrics are the deep-flaring bowl rim 8.19, the hammer-rim bowl 8.23, and the lekanis rim 24.20. This Mirabello fabric is a variant of KTS type II (Haggis and Mook 1993) with a green to red-brown core, and dense microscopic quartz, calcite, biotite, and granodiorite inclusions. During site recording, this fabric was not distinguished from type II, which is used commonly for larger cooking vessels such as baking dishes or plates (6.6, 8.9, 8.16, 8.20). Phyllite-tempered fabrics, such as the gray burnished goblet stems 24.1 and 24.2 mentioned above, are presumed to be of local manufacture, rather than imported from neighboring Gournia-Kalo Chorio, and are equally distinctive in their shape range. A spouted jar rim (8.8) and several bowl bases (8.5, 8.6, 24.5) have close parallels at Myrtos Phournou Koriphi. There are also rim fragments (8.13, 8.14) belonging to EM II tripod and flat-based cooking pots (Warren 1972, 178). The presence of west Isthmus or Mirabello fabrics on EM I–II sites in the Kavousi area is evidence of an active and dynamic movement of coarse-ware vessels around the north Isthmus region. Indeed, recent studies of ceramic regionalism in EM II eastern Crete have shown a dramatic increase in Mirabello imports into Phournou Koriphi by EM IIB (Whitelaw et al. 1997). In Myrtos Period I (EM IIA), south central or Mesara imports dominated the exogenous wares found on the site, and these are generally low-bulk, high-value items such as Fine

50

KAVOUSI I

Gray Wares and dark-on-light painted wares. In EM IIB, Myrtos was apparently importing not only low bulk pieces, such as Vasiliki Ware, but also significant amounts of large, coarse Mirabello fabric items from the north coast, including cooking pots, trays, deep bowls, amphorae, and pithoi. Evidence for this wide distribution of Mirabello fabrics down the Isthmus corridor augments significantly our understanding of interregional exchange. At the very least, it points to an economic organization or sociopolitical structure that developed and maintained efficient, sophisticated production and distribution systems, as well as effective means of transport. It is obvious that the Kavousi area, much nearer than

Myrtos to the center or centers of production, was squarely in the Mirabello distributional zone, receiving the same variety of vessels, but plainly in larger quantities and from an earlier date than the south coast site. The obvious availability of the local phyllite-quartz fabrics—which are found alongside the Mirabello fabrics—does not seem to have greatly affected the Kavousi area’s receptivity to the west Isthmus pots. It is important to emphasize that these two apparently competing traditions had already existed by EM II and formed a pattern of interregional interaction that was to continue until the end of the Protopalatial period.

Early Minoan III–Middle Minoan IA Pottery of the late Prepalatial period (EM III–MM IA), from well-stratified deposits in East Crete, was not sufficiently published to provide more than a skeletal picture of the ceramic phases in the Kavousi area. Pottery from Prepalatial levels at Mochlos, Vronda, and recently at Chrysokamino will eventually form the basis for the local sequence. The pottery from Period II at Myrtos-Pyrgos should also contribute significantly to a broad regional ceramic picture. Phyllite-tempered wares continue into EM III– MM IA at Kavousi, but they could be distinguished from MM I–II phyllite fabrics only by diagnostic shapes or surface treatment. Easier to recognize were the Mirabello imports, granodiorite- and quartz-tempered fabric types II, XX, and XXI. Fabric type II, the typical Mirabello fabric or “cooking ware” of EM I–II continued into EM III– MM IA, but in an apparently wider range of vessel types, including deep-flaring bowls, larnakes, and pithoi, as well as tripod cooking pots (8.12). Generally, the fabric is coarser and sandier than EM II varieties, with more obtrusive granitic-dioritic particles. Another important change is in the firing of EM III–MM IA type II, which produces a hard surface and core, and an even red or black color. The other EM III–MM IA fabric types are XX and XXI. Like type II, they are tempered with quartz and granodiorite originating in the west Isthmus and Gournia and Kalo Chorio areas. While

originally defined as two separate types (Haggis and Mook 1993), these fabrics are essentially variations of a single type with varying quantities of granodiorite or quartz inclusions. For convenience, they were recorded as a “type XX/XXI.” Among the samples, however, proportionately greater quantities of quartz would tend to indicate a type XX, while greater quantities of granodiorite defined the type XXI. The examples recovered seemed more like variations on a theme within the same production tradition than distinctly different fabrics. The fabric has a buff, pink, orange, or pink-buff core (only occasionally gray), diffusing to a buff exterior or slip. The inclusions, which are of various sizes, include granodiorite, quartz, biotite, calcite, and (rarely) red or brown mudstones or phyllites. The most common shapes are jars, pithoi, pithoid and hole-mouth jars, larnakes, basins, jugs, and amphorae. Types XX and XXI have a gritty or sandy appearance, especially on surfaces where the slip has eroded leaving the biotite and white quartz exposed. Common shapes in the fabric are small pithoi (44.4, 44.8, 44.12, 44.13), basins (44.10, 97.13), amphorae (28.16, 68.2), lekanes (68.4), jugs (97.15), and larnakes. Hole-mouth and bridgespouted jars (5.16, 28.11) are pots that are typical of EM III assemblages from the area of Gournia, Mochlos, Vronda, and Chrysokamino. Examples are burnished, red-washed, dark-washed or in White-on-Dark Ware. Isthmus area White-on-Dark

THE POTTERY

Ware jars occur in several fabrics including XX, XXI, and even cooking-ware fabric II. Thin flattened tripod feet (41.1) and flaring bowls (32.1, 32.2) in fabric type II are taken to be EM III–MM IA in date, as are cooking-ware shapes such as jar rims 1.4 and 8.12. A characteristic of bases of large jars (28.21, 32.6, 92.3) is a slight depression or very shallow groove where the wall of the vessel meets the base, caused by dragging a finger or modeling tool along the base of the pot while turning it. Locally produced examples 28.21 and 92.3 are made with phyllite and quartz temper. Phyllite-quartz-tempered fabrics are common in EM III assemblages in Mochlos, but their fabrics could not be distinguished from MM I–II varieties in the Kavousi area. Fine-ware cups 44.5, 44.14, 92.2 are late Prepalatial (cf. Momigliano and Wilson 1996, 51, fig. 30: 179, 181–182), but on the whole, fine wares of EM III–MM IA date were not easily recognized— that is, not readily distinguishable in surface samples from MM IB–IIA. Thus it was the imported Mirabello coarse wares (fabrics II, XX, and XXI) that permitted the identification of EM III–MM IA occupational phases. White-on-Dark Ware, typical of the north trench at Gournia, was recovered in the survey only at Hagios Antonios-Alykomouri; it is known from early excavations at the Hagios Antonios rock shelter and the Theriospelio cave, and in recent excavations at Chrysokamino itself. Local phyllite-tempered fabrics of late Prepalatial date are well represented in EM III deposits

51

at Mochlos and Chrysokamino. They are, however, only occasionally recognizable in survey assemblages. Mirabello coarse fabrics, however, tend to dominate both survey and excavation samples. Fabric types II, XX, and XXI might eventually be better represented in Myrtos Pyrgos II deposits than in subsequent Pyrgos III (MM II), a fact that serves to emphasize the longevity of the Prepalatial ceramic tradition in Mirabello—and exchange with the south coast—continuing past the EM IIB destruction and into EM III–MM IA (Cadogan 1995, 99–100). Mirabello imports at Petras also appear to be more prevalent in EM III–MM IA than in MM IB–IIA. But the late Prepalatial fabrics and shapes are not the same as those of EM I–II. A sudden trend toward destandardization among wares appears as a marked increase in both fabric types and variety of wares, accompanied by greater diversity in shapes. The most important change in the Mirabello and north Isthmus traditions is the fabric shape “crossover” phenomenon. That is, some shapes were no longer obviously ware- or fabric-specific, showing remarkable variability and perhaps experimentation (Haggis 2000). Cooking and storage vessel fabrics, for example, appear to have become interchangeable. White-on-Dark ware is produced in a diverse array of similar fabrics, but the shape range—cups, bowls, jugs, and especially hole-mouth and bridgespouted jars—overlaps with countless other wares and fabrics.

Middle Minoan IB–IIB By the Protopalatial period there is a return to standardization in both fabrics and shape repertoires. While Mirabello imports are found less frequently on sites of MM IB–II than EM II–MM IA outside of the north Isthmus and Mirabello area— such as at Myrtos Pyrgos and Petras (Knappett 1999, 632)—within the Kavousi region in the Protopalatial period, such imports are the common coarse-ware types. Mirabello fabrics III and VI emerge alongside the local phyllite-tempered type IV as the main coarse wares for large utilitarian vessels in MM I–II.

Fabric type VI is used primarily for cooking vessels—trays, dishes, and tripods. It has a red or reddish brown color, usually a dark red or brown core, and very obvious and equidimensional black and white granodiorites; quartz and calcite bits are present as are gold biotite particles (Haggis and Mook 1993). This fabric is evidently the same as the typical “Mirabello Cooking Class” found on Pseira and Mochlos (Barnard 2003, 7; Floyd 1998, 179). Fabric type III has a pink-buff or orange-buff core, diffusing to gray, with a surface slip that is usually buff or tan. Regularly sized and shaped granodiorite

52

KAVOUSI I

particles are the abundant inclusions. This is the standard fabric for a large variety of storage and processing vessels: especially pithoi, basins and lekanes, larnakes, jugs, jars (bridge-spouted and hole-mouth), and amphorae. These Mirabello types are the dominant fabrics found on Protopalatial sites in the northeast Isthmus and Kavousi regions (Barnard 2003, 8). The local phyllite-tempered fabric (type IV), used primarily for cooking vessels, jugs, and amphorae, is red-brown in color with red or gray phyllites, white or gray quartz, quartzite, and some calcite inclusions. Type IV examples usually have a gritty or sandy texture, and the slip is not very well preserved. The composition of these three fabric types suggests continuity of ceramic traditions from the preceding EM III–MM IA. Diversity and variation characterize the late Prepalatial fabrics, while regularity and standardization mark the appearance of the Protopalatial wares. Protopalatial type VI—the primary cooking-ware fabric—is a development of earlier type II. The primary differences between them are the larger size, quantity, and regularity of the added granodiorite temper in the later type VI. Similarly, late Prepalatial fabric types XX and XXI seem to be precursors of MM I–II type III. Here again the regularity of size, shape, quantity, and distribution of the granodiorite temper in type III suggests a tendency toward standardization of production. What sets the Protopalatial wares apart from the late Prepalatial examples is the consistency of the fabrics: biscuit color, size, dimension, and distribution of both temper and inclusions. They appear to conform strictly to a standard. MM I–II ceramic production surely involved using constant sources of materials, regular techniques, and consistent preparations of temper. By way of contrast, a number of samples of late Prepalatial fabric types II or XX appear to be variations on a theme. This standardization of both the fabrics and ranges of shapes is discernible among the Mirabello fabrics in MM I–II, suggesting perhaps greater centralization of the organization of production. In spite of the dominance of the Mirabello ceramics in the Kavousi region, the local phyllite fabric, type IV, was a consistent part of the coarse-ware assemblages. The nature of the long-standing interaction, and perhaps competition, between the west and east Isthmus remains unclear. Why was the eastern Isthmus and Kavousi region so receptive to pottery

from the production zones of Kalo Chorio-Istron and Gournia? What is the significance of this Mirabello ceramic sphere? Does it represent a pattern of exchange, economic influence, or social and political influence from west to east in the region? While local phyllite wares of EM III–MM IA are not known well enough to assess chronological changes in the transition from Prepalatial to Protopalatial, it is possible that the common (type IV) phyllite cooking ware represents a reorganized or burgeoning east Isthmus tradition in MM I–II. Are these apparently competing traditions in contiguous areas around the Bay of Mirabello symptomatic of economic competition, social interaction, or even conflicting centers? The system that encouraged the area of distribution of Mirabello coarse wares remains unclear. The fine-ware pottery is certainly within the formal parameters that have come to characterize the culture region of “Lasithi,” as described by Gerald Cadogan and more fully elaborated by Carl Knappett: an assemblage distinctly different and separate from that of the Knossos and Phaistos regions of central Crete and again from the Zakros and Palaikastro areas (Knappett 1999; Cadogan 1995). This region is marked by Malia on the north, Pyrgos on the south, Kastelli Pediada on the west, and Hagia Photia or Chamaizi on the east—probably a bit larger than Carl Knappett’s version, which does not extend east of the Isthmus (Knappett 1999, 625, fig. 6). While Cadogan’s characterization of the region as Protopalatial state territory remains problematic, the identification of such a culture zone has a resonance that is certainly amplified by the degree of emulation of certain fine-ware types suggesting links in elite consumption practices (Knappett 1999). Some common Protopalatial shapes in Mirabello fabric type III include pithoi (88.4), jars (27.3, 28.29), basins (35.2), amphorae (88.3), and lekanes (28.22, 28.31). Common pithos shapes such as 44.4 and 68.8 (MM IA or MM IB) have sharply rolled-back rims. Hole-mouth jars with a baggy, internally thickened rim and bridge-spouted jars are EM III–MM IA shapes that continue to be used into the Protopalatial period (5.16, 28.11). Thin oval tripod legs (1.12, 28.26) and bases (28.23) are common shapes in fabric type VI in addition to cooking pot rims (1.11, 28.25, 34.13, 41.6, 41.7) with distinctly thickened or triangular sections,

THE POTTERY

cooking trays, and dishes (5.24), as well as an assortment of jars (1.36, 4.2, 28.28, 45.1, 68.7) and lekanes (68.6). The dating of sherds in the local phyllite-quartztempered type IV is generally more difficult as the fabric continues into MM III (e.g., 5.18), and a version appears again in LM III (5.37) and the Early Iron Age (99.46). Thus, the type IV fabrics are generally not good indicators of date without periodspecific shapes. Very thin oval-section tripod legs (1.10, 40.7, 41.8), dishes (5.22, 5.23, 34.11), ovalmouth amphorae (35.1), and pithoi (4.4, 44.3, 44.9) are recognizable MM I–II shapes in this phyllitetempered fabric. The pithos fragment 92.1, for example, has trickle decoration that is common in Protopalatial. Jars (28.27), lekanes and basins (28.21, 28.24, 68.5), lamps (28.17), loom weights (5.21), and figurines (5.25, 5.26) are some examples that illustrate the varied local production tradition in the phyllite temper. Fine wares in MM I–II Kavousi are most like assemblages at Myrtos Pyrgos (Pyrgos II and III), Quartier E and M at Malia, and the Protopalatial houses at Mochlos and Vasiliki (Stürmer 1993; Andreou 1978; Pelon 1970). Conical cups (28.10, 40.1, 44.2, 100.11,) are the most common indicators of MM I–II date in surface survey assemblages.

53

Frequently such cups have a dark brown, black, or red wash on the interior surface (1.22, 28.5), and either straight or flaring walls (28.1, 28.3) (Andreou 1978, fig. 11; Betancourt 1990, fig. 18: 225; Haggis 1996c, 654, fig. 9: HT 11–13; Pelon 1970, pl. 35:9). The tumblers recovered (1.7, 28.7, 100.12), are most likely to be MM IB, while straight-sided cups, large wide-flaring conical cups (tumblers with sharply concave sides) (4.1, 44.5, 45.3, 45.4, 69.4, 69.6), and most varieties of carinated cups (24.33, 24.34, 68.1, 97.6) are probably not earlier than MM IIA (Andreou 1978, 76–77, fig. 11; Haggis 1996c, 654, fig. 9: HT 11–17; cf. MacGillivray 1998). Deep round and semiglobular (hemispherical) cups have a low, flat, thin base that very gently curves up to the wall. Varieties can be handmade or wheel made, and some may belong to MM IB–IIA carinated cups (Andreou 1978, figs. 12:13; 15:19). Examples were found at nearly all MM I–II sites in the Kavousi area (35.3, 40.2, 41.5, 44.6, 44.7). Shallow, wide-flaring bowls or saucers (e.g., 24.28) are common in Protopalatial Malia, Myrtos Pyrgos, and Kalo Chorio (Andreou 1978, figs. 1:12, 12:16; Haggis 1996c, 653–655, fig. 10: HT 18–20; Pelon 1970, pl. 11:1–2). Fine collared jars such as 28.4 and 69.7 are also diagnostic for MM I–II (e.g., Momigliano and Wilson 1996, 16, fig. 9, 24–30).

Middle Minoan III–Late Minoan I In Middle Minoan III, the Kavousi area pottery changes drastically. The Mirabello fabrics (II and VI), derived from the Kalo Chorio/Gournia region, are not present at MM III–LM IA sites and seem to have been replaced entirely by locally produced phyllite-tempered fabrics for the full range of cooking, storage, transport, and processing vessels. While the local phyllite-quartz-tempered type IV continues in use for some time into MM III, a new phyllite fabric, type I, is gradually favored for nearly the entire range of coarse utilitarian vessels. It is the standard coarse ware in the Neopalatial period (Barnard 2003, 5–7). This fabric was the first type to be observed and recorded because of its prevalence throughout the area and its use for nearly all of the coarse-ware shapes. Fabric type I is, on the whole, similar to type IV in its color

(brown, red, orange-red) and use of local schist (purple, red, and gray and green) as the primary macroscopically identified inclusion material. But unlike type IV, type I examples have little or none of the white and gray, gritty quartz sand, quartzite, and calcite particles. The phyllite inclusions look as if they have been added as prepared temper. They are large in size (3.0–5.0 mm) and often appear densely packed, regular in size and shape, and consistent in color. Overall, the Neopalatial fabric seems to be a development of or even improvement on the local Protopalatial type IV. Examples are harder, more consistent in firing, with fewer spalling cavities, and better preserved surfaces than the usual type IV. The surfaces of sherds are very hard, and the inclusions are well imbedded in the body of the pot, with foliated phyllites lying

54

KAVOUSI I

parallel with the surface, rarely protruding beyond the slip. The slip is very well preserved in type I, something that can rarely be said for examples of type IV in any period. What is striking about Neopalatial sites in the Kavousi area is the complete dependence on this purely local (eastern isthmus, Mochlos, and Kavousi Mountains) pottery, to the exclusion of granodiorite-tempered fabric from across the Isthmus in the Kalo Chorio region. Considering both the duration and predominance of the Mirabello fabrics in the Kavousi region—from as early as EM II—the break in the importation of these wares at the end of the Protopalatial period signals a significant organizational or economic change in the LM I landscape, affecting the relationship between these neighboring areas of the north Isthmus region (Cunningham and Driessen 2004, 108; Haggis 2000). The full range of cooking and storage vessels are produced in fabric type I. Tripod cooking pots have distinctive profiles. The body can be globular with a straight (1.35) or pinched-out rim (9.5, 20.12, 20.13), and some varieties have a straight wall (5.17) with thickened and everted rims (9.7, 96.8). Tripod legs with thick-oval sections are diagnostic shapes used for dating rural Neopalatial sites in the Kavousi region. The most distinctive profile (97.16) has a fairly flat and smooth back or underside, and a rounded or arching, convex front side (20.16, 21.2, 97.16–97.18). The tip of the leg, where preserved, is tapered and neatly squared (21.2, 69.20). In some examples— especially short or squat varieties—the tip of the leg is rounded (17.6) and the oval section is rounded on both the front and back. Cooking dishes (1.5, 1.33, 1.34, 19.3, 20.8–20.11) and trays (1.3) are common, but they are not as chronologically distinctive as the tripods. The shape of Neopalatial pithoi (9.8, 34.10) is different from that of the Protopalatial kinds. The commonest form has a tall, straight neck and a pinched-out or rounded rim (34.10) that is distinctly thickened (11.1). Protopalatial types, on the other hand, have a shorter neck and usually a sharply everted rim that may even be rolled back over the shoulder of the vessel. Neopalatial jars (69.16, 69.18), lekanes (19.5), and even strainers or incense burners (1.6) are made in fabric type I. Middle Minoan III—as a phase that could be distinguished easily from LM IA—is represented primarily by the material from loci 63 and 86 (site 69).

The pottery from these loci can be compared to the MM III examples from Quartier E at Malia, the acropolis at Knossos, and recent excavations at Mochlos (Catling et al. 1979; Pelon 1970; MacGillivray 1998). The fine wares of MM III are often of poorer quality than those of MM II or LM IA (Betancourt 1985, 103; Popham 1967, 337). Fabrics are soft red or buff, sometimes without the good burnish of LM IA, and frequently heavier-tempered semifine or even medium-coarse fabrics are used for shapes that in MM II would have been fine ware. Conical and straight-sided cups with pronounced ribbed or rilled walls are found at Malia and Knossos (5.41, 24.31, 69.2, 69.5, 69.9, 69.11, 69.13) (Andreou 1978, 157; Catling et al. 1979, figs. 17 and 18: deposit A cups; cf. Popham 1974, 188–189). Conical cups with a retracted base, neatly convex wall, and in-turning rim are found in MM III deposits at Knossos, Kommos, and Malia (Betancourt 1990, 40; cf. Popham 1974, 188–189). Kavousi-area examples are 5.6, 69.1, and 69.9. Straight-sided cup bases with a beveled edge immediately above the base (5.13, 69.4, 69.12) begin no earlier than MM IIB and are very common in MM III (Betancourt 1990, 40). The oval-mouth amphora is the typical transport vessel of Middle Minoan Crete, and distinguishing between MM I–II and MM III and LM IA types was difficult. Examples in fabrics XX and XXI are MM IA (28.16, 68.2), while mediumcoarse type IV varieties (20.7, 35.1, 69.17) are MM IB–II and MM III. LM IA is well represented at three sites, 5, 20, and 34. Tortoise shell ripple (34.5), foliate bands (34.6), solid circles, dots (97.1), and running spirals (5.1, 17.5, 20.15, 28.15, 97.4) are motifs that become fully developed in LM IA (Betancourt 1990, 41; Betancourt 1985, 113; Popham 1967, 337–339). While most commonly painted as a red or dark brown slip applied to the yellow-buff surface of the vessel, such designs can also be added in white paint on a dark ground, or in white on the plain clay surface of the vase (97.3). Straight-sided cups (5.13, 34.4), deep round cups, hemispherical cups (34.3, 34.9), conical cups—with both convex and straight walls (5.42, 5.43, 5.49, 34.7)—and oval-mouth amphorae (20.7) continue from MM III into LM IA. Sites 1, 19, and 96 produced useful samples of LM IA–B pottery; that is to say, these are sites that very likely continue into LM IB. The most easily

THE POTTERY

recognized vessel type is the hemispherical (semiglobular or rounded) cup with a rounded profile, pronounced base (1.25, 5.50, 19.1, 19.2, 96.1), and articulated, everted, or S-shaped rim (1.27, 1.28, 5.7, 5.47, 21.1, 96.2, 97.8; see Barnard and Brogan 2003, figs. 4–8; Betancourt 1985, 131: fig. 99:B, E; Catling et al. 1979, 26, 31; Floyd 1998, 181–182; Sackett and Popham 1970, 221–222). Sharply everted rims begin in MM III, but are canonical for LM IB (Floyd 1998, 181–182; Watrous 1992, 112–113). Although harder to identify (or less prevalent), straight-sided cups and bell cups (1.30, 19.7, 20.2) were present on LM IA–B sites, while stemmed shapes (1.21) were rarely identified (Sackett and Popham 1970, NP54; Watrous 1992, 113). Unfortunately, it was not possible to separate LM IA and LM IB settlements, distinguishing Neopalatial sites

55

abandoned in LM IA, thus demonstrating the potential effects of the Thera eruption and LM IA destruction across the landscape (Watrous et al. 2000, 476–477; cf. Driessen and Macdonald 1997). The evidence for LM IB occupation at Mochlos (Brogan, Smith, and Soles 2003) is significant, and the excavators have postulated a thriving population, increased settlement complexity, and regional connections with the palace at Gournia (cf. Soles 2004; 2002). This general LM IB prosperity they extend to the entire Mirabello region, including Kavousi (Soles 2004; Brogan, Smith, and Soles 2003, 116). While their arguments are compelling, it is nevertheless likely that sites 5, 20, 28, and 69 at Kavousi were not reinhabited after LM IA, while sites 1, 19, and 69 and the Chrysokamino Habitation Site (site 34) surely continue into LM IB (Fig. 14).

Late Minoan II and Late Minoan IIIA–B LM II and IIIA–B ceramic phases are probably underrepresented in the survey data for lack of recognizable and chronologically specific East Cretan shapes and coarse-ware fabrics. It should be emphasized, however, that: (1) evidence for LM II–III occupation is similarly sparse in the results of neighboring Gournia and Vrokastro surveys as well as in the Chania area of west Crete (Hayden 2004a, 123– 124; Moody 2004, 255; Watrous et al. 2000, 477) and (2) excavations at Gournia and Mochlos have shown sharply retracted settlement or a significant reduction in the areas of reoccupation in LM III (Brogan, Smith, and Soles 2003). Extrapolating from the excavated sample, the sharply reduced LM III settlement pattern at Kavousi is well within reasonable expectations of depopulation. No certain evidence for LM II was found in the Kavousi area. But this gap in the archaeological record is less likely to indicate an actual period of abandonment than the lack of local receptivity to new and exogenous styles. LM II is known stratigraphically primarily from Knossos, Kommos, and in eastern Crete from Palaikastro; Mochlos provides the most detailed and compelling publication of an LM II phase in eastern Crete (Brogan, Smith, and Soles 2003, 101–105; MacGillivray et al. 1992; Popham 1984, 159–160; Watrous 1992; 1981, 75–77). Problems still abound, however, in the identification

of regionally distinct LM II assemblages (Betancourt 1985, 150). In the Lasithi area of eastern Crete—that is in the region from Pedhiada to Chamaizi—the definition of an actual stratigraphic phase and assemblage has proved more difficult than the identification of individual characteristics of shape and decoration. Even in Kommos, where the phase is stratified neatly under an LM IIIA:1 deposit, the definition is based on the identification of styles that are separative—that is, not part of LM IB or LM IIIA:1 assemblages: the Ephyraean goblet and krater (among others) are new, while the LM IIIA:1 ledge-rim cup is absent (Watrous 1992, 20; 1981, 75–77). The characterization of the LM II assemblage as a whole emphasizes affinities with both LM IB and LM IIIA:1. A similar situation exists at both Knossos and Palaikastro (MacGillivray et al. 1992, 138–139; Momigliano and Hood 1994, 117–118; Popham 1978). At Mochlos where an LM II phase has been firmly established, the evidence consists of individual sherds of the Knossian LM II style (Brogan, Smith, and Soles 2003, 104). In the sample from surface survey in the Kavousi area, fine-ware shapes were generally diagnostic for LM II–III:A or LM IIIA:2–IIIB; decoration, other than monochrome, was very rare. Among the coarse wares, type IV returns as the predominant fabric for cooking and storage vessels.

56

KAVOUSI I

Cooking-pot rims (5.34, 5.35, 36.4), trays (36.8), and cooking dishes (baking plates) (36.7) in fabric type IV are diagnostic for LM III, but it is the shape criteria rather than the fabric that indicates the date. Tripod cooking-pot legs with a distinctly round section (5.37, 57.10) are good indicators of LM III but provide little resolution within the period. Similarly, the kalathos base 36.10, rim 5.36, and basin 5.38 are generally diagnostic for LM III. The pithos wall fragment (34.1) is in fabric type IV, and its decoration of incised chevrons is a very common LM III pithos motif—although not unknown in LM I. The primary evidence used for assigning LM III dates mostly consisted of diagnostic fine wares. And the lack of definitive decorative motifs and scarcity of diagnostic shape features (such as goblet and kylix stems, squat pyxides, ledge-rim cups, stirrup jars) made it difficult to make important chronological distinctions. Distinguishing LM IIIA:1 from LM IIIA:2–B thus proved impossible at most sites. Site 5 however, did produce a concentration of LM IIIA:1 pottery, while site 36 had material that could be assigned to LM IIIA:2–B. LM IIIA conical cups

(5.32, 5.33, 5.40) with a flaring profile and wide rim diameter (10 cm) and small base (3–3.5 cm) are distinguished from the heavier and squat LM IA–B varieties (Watrous 1992, fig. 24:523, 524, 531; fig. 27:615–616; Popham 1970, 79). Kylix and bowl rims (5.27–5.31) date to LM II–LM IIIA, and are similar to LM IIIA:1 examples from Mochlos (Brogan, Smith, and Soles 2003, 106, fig. 7; Popham 1970, 78–79). Monochrome painted exteriors predominate, although a painted band on the interior of the rim (5.27, 5.29, 5.31, 36.12) is a characteristic of both cups and kylikes in LM IIIA at Kommos and Mochlos (Brogan, Smith, and Soles 2003, 106, fig. 7: P1135; Watrous 1992, 127–132). They appear decidedly rarer in examples published from the Unexplored Mansion at Knossos (Popham 1984). Deep bowls (24.29, 24.30, 36.1–36.3) were good indicators of LM IIIA:2–B as were krater rims such as 36.5. Rarer diagnostic profiles included a stirrup jar (24.37), kylix rim (36.12), and pyxis base (36.14). The amphoroid krater fragment 36.9 with the debased octopus (a rare example of painted decoration) is of a common type in LM IIIB.

Late Minoan IIIC In LM IIIC, the range of coarse-ware fabrics and shapes is remarkably wide and varied, and different fabric types are very distinctive, permitting unequivocal differentiation from earlier LM I–III wares. The changes suggest a firm break with traditions established in LM I. There is not only a great variety of fabric types local to the area—that is, containing phyllite and quartz inclusions—but for the first time since the Protopalatial period, Mirabello imports with granodiorite inclusions are again easily discernible. The phyllite-tempered type IV, the prevalent coarse-ware fabric of LM IIIA–B, appears to have continued in use, although primarily for cooking pots, trays, and plates. The other five LM IIIC fabrics are distinctive (see Appendix 2). Fabric type X (and its finer version, type XI) has a pink-buff or orange core and a thick bright yellow or yellow-buff slip. Red and gray phyllite inclusions are prevalent along with white and gray quartz and quartzite bits. In most examples, the

inclusions protrude through the slip to the surface of the pot. Fabric type XIII is similar to X in clay color, but the yellow-buff slip is thickly applied, giving the pot a very smooth regular surface. The inclusions are densely packed red or purple subrounded to subangular phyllites (mudstones or siltstones) and occasional lumps of terracotta. These three fabric types—X, XI, and XIII—are most commonly used for large vessels, including pithoi, jars, and basins with everted or ledge rims. The medium-coarse type XI is used for an array of vessels including jugs, pyxides, amphorae, large kylikes, kraters, and stands. Type XV is a west Isthmus or “Mirabello” import. Sherds are characterized by the even tan or buff clay and slip, dense white granitic inclusions (from the granodiorite series), and very obvious chaff voids. The inclusions appear as prepared and added temper as they are very regularly spaced and equidimensional (cf. Hayden, Moody, and Rackham 1992, 325–328).

THE POTTERY

Finally, type XVI is unusual as it contains silver mica (muscovite) and white to gray inclusions, which are probably quartz and calcite. The fabric color is red to orange, and the surfaces are usually gritty and sandy to the touch. It is commonly used for cooking pots and dishes, as well as pithoi and jars. Several coarse-ware shapes are diagnostic for LM IIIC in the Kavousi area. The basin with the distinctive rib or carination below an out-turned or articulated ledge rim is a shape common throughout the survey zone especially in fabrics XI and XV (70.8, 90.4, 99.33–99.35). Parallels are known from excavated contexts at Vronda, Kastro, Kastri, and Karphi, as well as from surface material collected at numerous sites throughout the Siteia Mountains. This vessel type is diagnostic for LM IIIC (Tsipopoulou 2004, 117; Nowicki 2000a, pls. 12–13; Mook and Coulson 1997, 337–365; Nowicki 1996, 273– 274; 1994, 259–263; L. Day, Coulson, and Gesell 1986, pl. 82b; Sackett, Popham, and Warren 1965, 290, fig. 11; Seiradaki 1960, 8, fig. 5:1–4). Pithoi are usually made in fabrics X, XI, XIII, or XV, and they have incised chevrons (85.1, 100.43, 100.44), hatched bands (85.4, 100.42), and stamped circles (85.3, 90.1). While they may be prevalent in LM IIIC, these forms of surface treatment do continue throughout the Early Iron Age (Tsipopoulou

57

2004, 108–109; L. Day and Snyder 2004, 65–66; Nowicki 2000a, 268–269, pls. XXVI–XXXIII; Mook and Coulson 1997, 337–365; Nowicki 1996, 273–274; 1994, 259–263; L. Day, Coulson, and Gesell 1986). Tripod feet in fabric type IV (99.45–99.47, 100.34, 100.45) with a round section and often a finger impression at the base are diagnostic for LM IIIC–PG in the Kavousi area (Tsipopoulou 2004, 113; Nowicki 2000a, 268). The most common diagnostic fine-ware shape for LM IIIC is the deep bowl (99.2, 99.3, 99.9, 100.5, 100.6) (Mook 2004, 164–170; Mook and Coulson 1997, 337–365; Coulson et al. 1997, 360–361; Gesell, Day, Coulson 1995, 104–106). Bowl bases were easily recognized (70.2, 70.5, 99.21–99.25, 100.21) in surface assemblages. While a reserved disk (99.21, 99.25, 100.21) on the bottom interior of bowls may be LM IIIB, the style occurs at LM IIIC sites such as Kastri, Karphi, Vronda, and Kastro (Mook 2004, 167–170; Popham 1965, 321; Sackett, Popham, and Warren 1965, 282). Other fine-ware shapes observed included cylindrical pyxides (99.1, 99.26, 99.27), kalathoi (99.8), kraters (100.2), and stirrup jars (100.23, 100.41) (Mook 2004; Tsipopoulou 2004; Kanta 1980, 282–283; Seiradaki 1960, 18–19, fig. 12).

Protogeometric–Archaic Protogeometric pottery is well known from stratified deposits on the Kastro (80) (Mook 2004) and from the graves on Vronda (77) (Gesell, Day, and Coulson 1990). It was not regularly recovered in survey, but it should be said that certain coarse-ware fabrics (IV and XIII), shapes (pithoi, basins, and tripods), and surface decoration (incised chevrons, circles, spirals, hatching) continue with modifications from LM IIIC into PG and even Geometric (Coulson et al. 1997; Gesell, Day, Coulson 1995; Nowicki 1996, 273–274). The most easily recognized fine-ware shapes, however—skyphoi, kraters, and amphorae—were not encountered. A reason for the paucity of PG pottery is that nucleation of population had probably occurred by PG. Vronda was abandoned as a settlement site (though PG pottery is recognizable in the graves), and the Kastro (site 80) had reached its maximum size. Azoria and the

Avgo sites show signs of PG activity, but visibility during survey work was greatly hindered by postPG occupation, vegetation, and modern land use. Recent excavation at Azoria in 2002 has confirmed a complete LM IIIC–Archaic sequence that was not readily apparent during survey (Haggis et al. 2004). Incised and stamped pithoi are common in PG and later, although forms do begin as early as LM IIIC (e.g., 85.1–85.4). Impressed spirals (85.2) and circles (85.3, 90.1), however, are more common in PG than in LM IIIC. Basins (70.8, 90.4) are also LM IIIC forms that continue into PG–G, but their profiles tend to be sharper and more angular than the Bronze Age varieties. Late Geometric, Orientalizing, and Archaic periods are better represented than PG. LG Skyphoi (70.3) and deep cups have flat or slightly concave bases. Lids 71.3 and 89.6 are LG–O, as are basins (71.6), and cooking pots

58

KAVOUSI I

(89.4) with horizontal ribs or grooves (Coldstream 1992, pl. 32:GD 32, GE 25). The latter begin in LG and become standard types in the Orientalizing and Archaic periods (Callaghan 1992, pl. 75:H3.8). Relief pithoi with chain, guilloche, and wave patterns (70.6, 71.7, 89.1, 89.3) are Orientalizing– Archaic (Haggis et al. 2004; Whitley et al. 1999,

248; Boardman 1961, 57–59). Of Archaic date are bowls with ribbed exterior surfaces (70.1, 71.5; Callaghan 1992, 91, pl. 74:H2.4.), cups (89.5, 90.5; Haggis et al. 2004; Callaghan 1992, pls. 74:H1B.7, 75:H3.7), and lekanes (71.1–71.2; Coldstream 1992, pl. 57:GG9; pl. 59:GH49).

Roman Early Roman (2nd century B.C. to 3rd century fine wares were not readily apparent under the veneer of fourth–sixth century red slipped wares, jars, amphorae and cooking pots. Two pieces of Sigillata A were found—one from Tholos (site 1) (1.19) and one from the Kavousi village deposit (site 24)—suggesting activity at these sites as early as the 1st century B.C. Also of the 1st century B.C. to 1st century A.D. are the ring bases 94.1 and 57.9. Red slip wares of the 1st to 2nd century A.D. were fairly common. Bowl fragments 1.17, 1.20, and 53.6 are Eastern Sigillata B, possibly Hadrianic in date (Boardman 1989, 88–89, fig. 26:15; Sackett 1992, 156–159, pls. 169:T1.5, 172:D1.2–4, 174:D3.7, 175:D4.1–4). The plain-ware basin (53.4), in an orange or pink-buff porous fabric, and with a distinctive ridge just below the rim, is also probably Hadrianic (Sackett 1992, pl. 179:D4.45). Out-turned rims (17.1, 37.2, 40.4, 57.1–57.3) from small ribbed or unribbed jars could be as early as the Severan period, but are found consistently with late Roman fine wares. The fabric is hard reddish orange, and it usually has a distinct gray core and dense fine white and dark inclusions (Sackett 1992, 250–252, pl. S1.4–5, 192:U.40–44, 51). A.D.)

By far the most abundant and easily recognizable fine-ware type on Roman sites in the survey zone was “Late Roman C” or Phocaean Red Slip Ware (Hayes 1997, 62–64). Hayes form 3 (17.2), form 6 (37.3, 37.4), and form 10 bowl rims (1.14, 53.2, 94.2) were very common as were bowl bases, the latter showing considerable variety (53.1, 53.5, 94.4, 94.7, 94.8) (Boardman 1989, 90–93; Hayes 1972). Late Roman cooking pots (40.5, 40.6, 53.3, 94.3) and Hayes Form 107 large bowls (94.6) were ubiquitous (Hayes 1997, 76–80; G. Sanders 1999, 470–471, figs. 13–14). Less common were Hayes Forms 104–106 in African red slip ware (42.3). Other bowls (37.1) have a distinct ledge rim that usually hangs slightly below the rim line; parallel grooves decorate the top of the rim (Boardman 1989, 98–99, fig. 31:166–169). Parallels are among African Red Slip examples (Hayes Form 93) dating to the fifth and early sixth centuries. Early Roman (1.15, 39.1, 57.11) and Late Roman (17.3, 40.8, 57.5, 57.7–57.8) amphorae were very common finds especially at the ports of Tholos (site 1) and Agriomandra (site 39), and at the warehouses at Sta Lenika (site 57).

5

The History of Settlement

This chapter describes the character of settlement and land use in the Kavousi area. It outlines the history of settlement from Neolithic to Late Roman times, integrating the site typology and environmental data presented in Chapters 2 and 3 into a discussion of the meaning of settlement patterns through time. A reconstruction of settlement

development and site interaction in various periods provides a picture of changing responses to both local environment, as well as external social, political, and economic influences. Site catalog numbers are given in bold face throughout, in order to facilitate cross-referencing with the pottery discussion (Chapter 4) and the gazetteer (Chapter 6).

Neolithic Neolithic pottery was originally identified at a total of six sites during the survey (Fig. 8). Evidence for latest or Final Neolithic was recovered from the Hagios Antonios area, at the FN–EM II site of Alykomouri (8) and the rock shelter burial (6). Final Neolithic has also been identified in the cave of Theriospelio (31) and above Kavousi village at Chondrovolakes (67), Azoria (71), and Vronda (77) (Figs. 8, 9). These six sites, while producing small amounts of material, suggest the

diversity of possible environments and topography that were exploited here in the fourth millennium B.C. Since the fieldwalking stage of the survey, Neolithic pottery has been recovered from the excavations of the Chrysokamino Project—at both the Minoan Habitation Site (34) and the Metallurgical Site (32)—suggesting the extensive exploitation of the Chordakia and south Mount Chomatas areas in the Final Neolithic (Muhly 2004; Betancourt 2000a; Betancourt and Floyd

60

KAVOUSI I

2000–2001; Floyd 2000; Betancourt et al. 1999). While the determination of the extent and functions of these sites in the FN period should await the final publication of the Chrysokamino project, it is likely that the Habitation Site (34) was a settlement in FN that probably used the cave at Theriospilio for burials (31) and the surrounding slopes of Mount Chomatas for agriculture (Betancourt et al. 1999, 351). Finally, excavation by the author at Azoria (71) in 2003 has demonstrated that the Final Neolithic occupation there may have been as large as 0.36 ha., covering a significant part of the hilltop of the South Acropolis (Haggis, Mook, and Snyder 2005; Haggis et al. 2004). FN Azoria apparently had an open-settlement plan, with distinctly separate architectural units (cf. Nowicki 2003, 69). While the settlement exploited mostly local sources of chert for blades and arrowheads, the ceramic fabrics demonstrate the use of both local phyllite-quartzite sources as well as production centers in the granodiorite zone across the isthmus in the Gournia and Kalo Chorio area. Should this site turn out to be a substantial village, then it may have been one of the main Neolithic settlements in the region. The Hagios Antonios area, occupied continuously throughout the subsequent Bronze Age, is a likely choice for Neolithic habitation. The small valley, sheltered from the coast on the north and west by the hills of Mounts Schinias and Chomatas, had both good agricultural land and a perennial spring. The silty phyllite soil was ideal for rain-fed agriculture. A similar environment exists above and behind Kavousi village. A single Neolithic sherd from site 67 on the north-facing slope of Mount Papoura, and the FN material from excavations at Vronda (77) and Azoria (71) are evidence that the plentiful springs and rich soils of the area had attracted Neolithic settlers. Like the Hagios Antonios Valley, the area between Xerambela and Kavousi village was densely settled throughout the Bronze Age. The preference for Theriospelio (31), however, as an occupation location is problematic for topographical reasons. The site is a cave, on a secluded and inaccessible promontory—essentially a cliff ledge—situated over 60 m above the sea, on the steep-sloping western face of Mount Chomatas. The mouth itself is only 1.20 m high and 3.0 m wide, the ceiling lowering to an uncomfortable 0.80 m high

before one passes the 10 m long entrance to the main chamber. Habitation would have been difficult since the surrounding steep slopes of gray limestone are prone to erosion, providing little open space for domestic activities. It is unlikely, therefore, that the adjacent terrain would have been normally selected for agricultural use. The cave seems to have been better suited to burial, cult, or perhaps refuge functions. By far the best preserved Neolithic remains recovered during survey come from Kavousi village (secondary deposit 24) from deep excavations near the central square (Pl. 1A; Fig. 8). The ridge on which Kavousi village is located consists of stable consolidated alluvium and is the optimal site location in the Kavousi region. Sitting about 100 m above sea level, and some 20–50 m above the Kambos plain, the site, like Alykomouri (8) in the Hagios Antonios Valley, is on a defensible rise and contains a perennial water source. Furthermore, the village is situated at the confluence of the main drainage basins feeding the Platys River. Thus a defensible position, multiple water sources, and accessibility to both the lowland plain and the rich agricultural land on the slopes of Mount Papoura were important considerations of the first settlers in the Kavousi region. Indeed, the remains of Neolithic at sites 67 and 77 indicate the use of these mountainous areas by a population perhaps dependent on Kavousi village (24), or more likely the large settlement at Azoria (71) to the east. The sizes of sites 67 and 77 are, however, unknown, and no actual evidence of a site hierarchy in Neolithic is readily apparent. At site 16 in the north Kambos plain (Fig. 8), the Neolithic component consists of the remains of probably not more than a single house. Its location, directly on the terra rossa of the Kambos, suggests climatic conditions that were perhaps wetter than today, permitting exploitation of these poor rocky soils for agriculture. The small size of the site—probably a simple rural installation—is evidence of expansion of agricultural activity into the Kambos from the nearest known Neolithic site at Alykomouri (8), and perhaps the earliest known exploitation of the Platys River alluvium. It is tempting to see three main centers of Neolithic occupation in the Kavousi area; one at Alykomouri (8), one at Kavousi village (24), and another at Azoria (71). The sites occupy naturally defensible locations, have permanent water supplies,

THE HISTORY OF SETTLEMENT

and are situated directly above or adjacent to good arable land. Unfortunately, it is impossible to determine the actual sizes of these sites in the Neolithic because of subsequent occupation and modern activity. In the vicinity of both Alykomouri and Azoria are the clearly small Neolithic sites 16 and 67. Could each main habitation center have had smaller rural dependent sites? It is safe to say that these small loci of Neolithic activity in the Kambos plain and on the slopes of Mount Papoura are indications of agricultural interests and the exploitation of diverse terrain and environments. The Neolithic settlement pattern thus remains unclear. It is important to keep in mind that Neolithic pottery sherds recovered from sites 16 and 67 were found in bulldozer furrows, suggesting some depth beneath the current alluviated ground surface, and the discovery of pottery from Kavousi village, Azoria, and Chrysokamino was the result of actual excavation. Thus it is probably safe to assume that other Neolithic sites might exist in both the mountains and the plain, and that survey has produced only a skeletal form of the actual pattern (cf. Hayden 2004a, 47; Nowicki 2003, 69–71). Another problem is chronology. The Neolithic is not sufficiently well known in East Crete, nor is the present survey sample substantial enough to assess the possible phases within a period lasting most of the fourth millennium B.C. Nowicki’s observations on Final Neolithic sites in a number of areas (most recently in his excavations at Katalimata) is neither contradicted nor negated by the apparent lack of refuge sites in the neighboring Kavousi area (Nowicki 2003; 2000b; 1999b). His model of settlement for the latter part of FN (FN II) involves the late foundation of defensible sites, particularly on the coast, followed by dispersal inland at the end of the period or in an early stage of EM I (Nowicki 2003, 1999b; cf. Hayden 2003a, 382). FN sites in Kavousi occupy both coastal and inland, both defensible as well as open lowland locations, suggesting that an understanding of the chronology of sites— phases within FN—is a prerequisite to applying Nowicki’s model. If for example Azoria represents an early (FN I) site, which seems to be indicated by its multiple stratigraphic phases (and long tenure of use) and dependence on local chert (rather than imported obsidian), then it could well be unrelated to Nowicki’s pattern of short-term defensible coastal sites, which might more aptly accommodate the

61

Alykomouri (8) settlement. Whether Final Neolithic sites in the area pre-dated or postdated others, or were contemporaneous, is a problem that remains unresolved. Our chronology and understanding of regional variability in this period need to be refined by more problem-oriented excavation (e.g., Haggis et al. 2004; Papadatos 2004; Nowicki 2003; Vagnetti 1996; Vagnetti, Christopoulou, and Tzedakis 1989). What can be said is that the first inhabitants of the Kavousi region chose inland sites above the alluvial plain but in close proximity to it, while their water supplies were located at or near the sites themselves and were easily protected. With the exception of the Theriospelio cave, the sites look inward and border the plain; they are essentially excluded from but within reach of the sea. The FN pattern as a whole demonstrates the occupation of hilltops above alluvial plains within reach of the sea, while inland sites are positioned along communication routes linking mountain regions with the coastal plain. This general view of the fourth millennium, albeit chronologically compressed, is reflected in the results of survey elsewhere, such as at Ziro, Vrokastro, and Lasithi in eastern Crete (Hayden 2004a, 42–53; 2003a, 384–386; Branigan 1998, 57–61; Watrous 1982, 10), as well as in the Mesara (Watrous et al. 1993, 223–224), where sites are found on hills above and flanking arable zones. Connections to inland routes (Hayden 2003a, 385) and the dispersal of sites into mountainous regions (Moody 1987, 292–293) perhaps to exploit pastoral resources are also material patterns observable across the island. The farmstead sites in the Kambos, Chordakia, and above the village are evidence of expanding agricultural interests or possible seasonal activities. While it is hard to trace a clear pattern of actual expansion from the coast to the mountain regions (Hayden 2003a, 391; Moody 1987, 93, 292), it does appear that each of the main ecological zones was being exploited during FN. On the whole the area appears to have been thinly settled, with ample agricultural resources in the plain and grazing no doubt on the coastal hills and mountains south of Kavousi village. Three locales were originally occupied precisely where the best springs are located. It is possible that the putative village sites—Hagios Antonios-Alykomouri (8), Kavousi village (24), and Azoria (71)—represent separate, autonomous, largely self-sufficient, and

62

KAVOUSI I

perhaps lineage-based groups. Smaller sites, such as Kampos (16), Chrysokamino (34), Vronda (77), and Chondrovolakes (67) were probably farmhouses, related to and dependent on the larger villages, filling out a weak two level hierarchy of settlement in FN. Each of the villages and their dependencies

exploited a local environment, and each probably controlled its own resources. If these neighboring groups coexisted, their relationship was no doubt peaceful and possibly cooperative within the basin of the Kambos.

Early Minoan I–II In the Early Bronze Age (Fig. 9), the total number of sites does not increase, and the Kavousi village and Hagios Antonios areas continue to be the focal points of habitation in the region. The sites of Kavousi village (24) and Alykomouri (8) show substantial evidence of continuous occupation from EM I through EM IIB, and while the size of the Kavousi village site remains unknown, Alykomouri was apparently small, perhaps no larger than about 0.18 ha. The site itself is, however, a very erosive phyllite ridge (Fig. 9), and an exposed scarp of a recently bulldozed terrace below the site has revealed sherds at a depth of over 1 m. Thus it is likely that mud slides above and below the exposed area of the visible sherd scatter may be concealing portions of a much larger settlement. The presence of the burial cave at Hagios Antonios, across the valley from Alykomouri (Pl. 7A), would indicate that the rough limestone terrain of Mount Schinias was the choice cemetery location for the settlement. Rock shelters are densely distributed along the southern edge of Mount Schinias and extend in a line from Hagios Antonios to the saddle at Alykomouri. A large rock shelter at Alykomouri saddle (site 7) is used today as a mandra, but Prepalatial sherds were found packed into the cave’s floor. The spring and good arable phyllite soil suited the EM I–II inhabitants of Alykomouri as they had in Neolithic times. Fragments of marine shells demonstrate that the inhabitants used the sea, and obsidian blade fragments point to links to the Aegean. Large querns and a variety of hand stones, derived from the nearby river bed of the Platys River (in the Kambos), are evidence of household agricultural activities. The Chrysokamino Habitation Site (34), as well as Theriospelio (31), on the west side of Mount Chomatas, continued into EM I–II. While this hard-to-reach cave site is technically a coastal location, the cave terrace extends for

only 3 m before dropping off to a cliff-like descent (ca. 40–45 degrees) into the sea. It is important that there is evidence for EM I–II at Tholos Bay at the very north end of site 1 (Fig. 9). This is the first indication of permanent occupation at this important harbor location. The EM I–II pottery (coarse fabric types II, IX) is concentrated in the northern area of locus 16, but its distribution has been affected by the large complex of MM I–II and Neopalatial buildings that eventually dominated the site. It is likely, however, that site 1 by EM IIA was an active port, with connections to the Bay of Mirabello, Pseira, Mochlos, and the Aegean (not unlike Branigan’s characterization of Mochlos itself for the same period) (Branigan 1991). Indeed, the Hagios Antonios burial cave, rich in imported metal artifacts, also contained stone vases, perhaps obtained from the neighboring islands of Mochlos or Pseira. An incised fine gray bottle or juglet—similar to one in the Hagios Nikolaos Museum from the cemetery at Hagia Photia—is an EM I–IIA Cycladic import (Haggis 1993b). The establishment of the foundation of Tholos in EM I–II is perhaps part of an overall process that characterizes the beginning of the Early Bronze Age in the Aegean: growing interregional trade, increased off-island contacts, and the burgeoning maritime economy (Wilson 1994; Broodbank 1993; Haggis 1996c, 680–681; Cosmopoulos 1991; Warren 1981). An obsidian core and blade fragment from site 1 (observed, not collected) suggest that the people in this area were importing raw materials and perhaps manufacturing blades and other implements for sites throughout the Kambos basin. Kavousi village, far to the south of Tholos and Hagios Antonios, is clearly another major site, showing continuity of habitation from FN through EM I into EM IIB (Figs. 8, 9). South of the village,

THE HISTORY OF SETTLEMENT

in the northern Mount Papoura drainage, a small site, 62, probably the remains of a single house, produced EM I–II remains, and there are for the first time traces of Prepalatial material in the Avgo Valley, west of Kavousi at site 87. On the Xerambela ridge at a height of 400 m is the predominantly LM IIIC site of Vronda (77), which has produced EM IIB pottery in recent excavations (Fig. 9). Thus, with a total of only ten sites in EM I–II, the region appears to have remained thinly settled. Nevertheless, expansion of settlement into the Xerambela and Avgo areas is a good indication that new areas were being exploited and perhaps new water sources were being sought. Both Avgo and Xerambela have perennial springs, and new settlements in these areas may be connected to efforts to secure water and territorial boundaries, as well as a desire to exploit the upper elevations for herding and grain agriculture. An increase in population or climate change by EM IIB could have required movement into higher elevations where springs and arable land are more plentiful and water supplies constant. Rain-fed agriculture is suited to the silty phyllite and mixed terra rossa alluvium in these mountainous areas. Exploitation of the Xerambela and north Mount Papoura drainage by the site of Vronda would, however, have required terracing and considerable efforts in land management. While traces of Neolithic were found by the Kavousi-Thriphti Survey below Xerambela at Chondrovolakes (67), the EM II material from the Vronda settlement probably represents an expansion of settlement from Kavousi village, which shows continuous habitation from FN, with distinct EM I, IIA, and IIB phases. The neighboring site of Azoria (71) is occupied in FN and then again, as at Vronda, in the Late Prepalatial period (Haggis, Mook, and Snyder 2005; Haggis et al. 2004). Expansion of settlement in EM I–II with the foundation of important sites at Tholos and in the Avgo valley indicates interests in the sea for trade and cultivation of entirely new areas. Significant changes probably occurred in EM II, as the evidence of a new settlement at Vronda would suggest—a period of dynamic growth of settlement accompanying economic changes. While the Kavousi area was certainly importing ceramic vessels such as darkon-light pottery (Hagios Onouphrios Ware) from the Mesara as early as EM I, by EM II Mirabello fabrics increase remarkably, and their wide distribution by EM IIB suggests a highly complex and dynamic

63

economy in the region (Whitelaw et al. 1997, 267). EM IIB is also the period when Myrtos on the south coast was importing a large percentage—as much as half—of its total ceramic assemblage from the Mirabello area. This trans-Isthmus trade in a wide variety of ceramic vessels, including coarse wares such as amphorae, pithoi, and cooking pots, is a sign of extensive economic interaction not only between areas of the north Isthmus but interregionally between the north and south coasts. In the wider region of eastern Crete, Cycladic imports increase in EM II as does evidence for interregional trade. Perhaps the central Cretan influence that we see in EM I and EM IIA pottery in eastern Crete, and the Cycladic character of the EM I–IIA cemetery at Hagia Photia are indications of the expansion of trading interests and interregional communication, as well as an increase in organizational complexity as new mechanisms were required to maintain production and exchange systems. These mechanisms permitted the transport and distribution of not only luxury items, exotics, and small high-value fine vessels, but a fairly wide range of cooking, transport, and storage vessels (Haggis 2000; P. Day, Wilson, and Kiriatzi 1997; Whitelaw et al. 1997; Haggis 1999a; 1997; 1996c; Watrous 1994; Wilson and Day 1994; Wilson 1994). Is there evidence in the settlement patterns to suggest a settlement hierarchy in EM I–II? Unfortunately site sizes at Vronda, Kavousi village, and Tholos are impossible to determine accurately because of subsequent occupation. Alykomouri, the only well-preserved EM I–II site, appears to have been small, perhaps no bigger than Vasiliki or Myrtos, that is, a hamlet of five to ten houses, spanning the entirety of the Early Bronze Age. But as indicated above, the erosive characteristics of the phyllite soils make accurate estimation of site size difficult; much of the settlement may have been buried. In the neighboring Vrokastro area, a complex threepart hierarchy is suggested for FN–EM I with a surprisingly large average settlement size of 0.354 ha. and a doubling of sites in the region by EM III (Hayden 2004a, 68; Hayden, Moody, and Rackham 1992). A pattern of growth is also detected at Gournia where there is a continuous expansion of settlement, obvious coastal trading interests, and changes in the subsistence base (Watrous et al. 2000, 475). Similarly in the Mesara there is a doubling of site numbers evident between FN and EM I and a slowed pace of growth suggesting a plateau by EM

64

KAVOUSI I

II (Watrous et al. 1993, 223; Watrous 1994, 705). In Hagiofarango, there is a dispersed pattern of houses and hamlets associated with distinct patches of arable land (Blackman and Branigan 1975, 17–36), a settlement system related to settlement expansion in the Mesara. While the growth trend is not as pronounced in western Crete as in other areas, the sites in the Chania region are on the whole large in size, exceeding one ha., and there is evidence for the extensive clearing of the Akrotiri peninsula and a coastal emphasis of settlement location (Moody 1987, 297–300). In the Kavousi area there is little evidence of significant settlement expansion and even less to suggest a meaningful regional hierarchy, although it is possible that Kavousi village could have been a large site in this period, controlling both the use of the Platys River and transportation routes into East Crete through the Avgo Valley. Furthermore, Kavousi village is situated at precisely the geophysical transition between the lowland plain and the Kavousi Mountains, a position that permitted it to control the crucial passes into the Kavousi Mountains and Thriphti Range, that is, the principal pastoral environment in the region. Thus, the importance of the physical location of Kavousi village should not be underestimated if the precise size of its EM I–II site is indeterminable. Each of the main EM I–II sites in the Kavousi area—Tholos, Alykomouri, Kavousi village, and Vronda—are situated in environments suitable for rain-fed agriculture, and each controls a perennial spring. The Tholos area has many springs, but the alluvium is mostly rocky terra rossa. The site was chosen for its natural port, not its water supply. The other sites were self-sufficient communities with populations small enough to be supported from local resources. The relationship between these sites must have been cooperative. Alykomouri-Hagios Antonios no doubt exploited and controlled the northern plain and probably Tholos, and thus the main route to the bay of Pseira and the Aegean. Kavousi village and Vronda, on the other hand, controlled land routes south to the isthmus and southeast to the Kavousi mountains. The interdependence of these communities, if not originally based on lineage ties, was reinforced by economic necessity. By EM II, the communities exploited the advantages of a growing regional economy and burgeoning interregional and off-island connections. If,

for example, Mochlos had become a gateway to the Aegean, as Branigan has argued (Branigan 1991), the most efficient communication route between the plain of Mochlos—which is completely enclosed by mountains—and Mirabello and the north Isthmus was perhaps through the Bay of Pseira to Tholos and then south through the Kambos to the plain of Kamina and the Isthmus of Ierapetra. Could Tholos have been an early transshipment point by EM II? And could the EM population of the Kavousi area have thrived precisely in response to an economic system that nurtured a new Mirabello-area and trans-Isthmus trade? Such exchange patterns are already evident in the distribution of ceramics (P. Day, Wilson, and Kiriatzi 1997; Whitelaw et al. 1997). The pattern of settlement in EM I–II is, however, highly nucleated. The Kavousi sites were probably comprised of several families each exploiting local arable land, a picture reminiscent of Whitelaw’s (1983) original characterization of Myrtos Phournou Koriphi. Nevertheless, this small-scale arrangement of nucleated hamlets or villages functioned within the wider regional context of the Bay of Mirabello and eastern Crete. The economic systems within this region were clearly diverse and dynamic, requiring a level of organizational structure that may be reflected in the appearance of large buildings at Vasiliki and Mochlos and farther afield at Malia and Palaikastro (Driessen 2001, 59; Schoep 1999, 270– 271; Haggis 1999a, 61–62; Soles and Davaras 1996, 180; MacGillivray and Driessen 1990). While there is yet little direct indication that these EM II buildings represent a political component whose control spread much beyond the village or local agricultural catchment area, regional ceramic studies have demonstrated considerable economic integration— long-distance ties between communities, which might reflect both economic integration and complex sociopolitical interrelationships (e.g. P. Day and Wilson 2002; Wilson and Day 2000; Whitelaw et al. 1997). Whatever social systems existed in the Kavousi area by EM II, the complexity implied by the coastal location of sites, the presence of obsidian and Cycladic imports, long-distance exchange of ceramics, and large central buildings, is not readily apparent as a hierarchical ordering of sites whose sizes could reflect economic or sociopolitical ranking in the region.

THE HISTORY OF SETTLEMENT

65

Early Minoan III–Middle Minoan IA In the late Prepalatial period, there are changes in settlement patterns suggesting an increase in population and restructuring of settlement. Although evidence of EM III–MM IA occupation is recognized as important in reconstructing the development of complex state-level systems of settlement in Minoan Crete, there remain significant problems in the recovery and interpretation of material patterns attributable to this period (Haggis 2002; 1999a; Manning 1994, 231–234; Watrous 1994, 717–718; Watrous et al. 1993, 224; Branigan 1988b; Warren 1987; Cadogan 1986, 158–164; Cherry 1983a, 41). Not least of the problems is the definition of ceramic phases (Andreou 1978, 164–172; Cadogan 1986, 158; Momigliano 1991; 1990; Manning 1994, 231; Watrous 1994, 718–720), and our ability to recognize them in survey samples (Hayden 2004a, 76; Watrous et al. 1993, 224; Blackman and Branigan 1977, 68). At Kavousi, our ability to recover late Prepalatial material was greatly improved by our understanding of local coarse-ware fabrics and shapes derived from excavation samples in the region at Chrysokamino, Vronda, Mochlos, Vasiliki, and Gournia (Hayden 2004a, 80; Watrous et al. 2000; Betancourt et al. 1999; Soles and Davaras 1996; Betancourt and Silverman 1991; Betancourt 1984). The recognition of certain wares and fabrics thus allowed us to reexamine the pottery recovered during the site-recording stage in order to disaggregate and identify the late Prepalatial phase on sites with otherwise dominant EM I–II or MM II components (Haggis 2002; 2000; 1999). In the survey zone, the total number of sites increases to sixteen, with the exploitation of areas previously unoccupied (Fig. 10). In the Hagios Antonios area, habitation continued into EM III at both the Alykomouri settlement (8) and in the burial cave (6), but a new settlement was established southeast of the rock shelter at the foot of Mount Schinias (5). Site 5 is located above the plain on the gradually descending southeast slope of the limestone hill (Pl. 7). Although the exact size of the site in EM III is unknown, the position is preferable to Alykomouri. While the latter is inaccessible from the plain on the high spur of Mount Chomatas, the new site 5 with its southeastern aspect is both protected from the north winds and enjoys a strategic

view of the Kambos plain and the north-south land route to Tholos Bay. Furthermore, the placement of the site on the rocky and denuded hill slope of Mount Schinias and on the terra rossa colluvium at the base of the hill was a conscious decision. Avoidance of the silty phyllite soils of the Hagios Antonios Valley for habitation would have maximized their agricultural use potential. Alykomouri, on the other hand, is located directly in the phyllite alluvium; its position, while defensible, is both inaccessible from the plain and occupies too narrow a space to permit significant expansion of settlement. If the EM III–MM IA foundation of site 5 is the result of local population growth in Hagios Antonios, it is possible to imagine that the new population expanded not into the valley itself—and into the areas of best agricultural land—but moved to the edge of the valley where both fields and water supply were accessible but also where communication with the Kambos plain was easier. The growing population might as well have needed to optimize the use of the area’s limited resources leading to an eventual abandonment of the habitation site (site 8) in the phyllite alluvium. In the area of Hagios Antonios, continuity of settlement and land use from EM II into EM III is apparent. This is not, however, the case in other areas of the survey zone. It is striking that two important EM I–II sites, Tholos (1) and Kavousi village (24), show no evidence of EM III–MM IA occupation. It seems unlikely, however, that this lack of recognizable material is a meaningful indication of discontinuity. While the apparent uniformity of EM IIB destructions should permit us to look for discontinuity on a regional scale, this important transitional period of Aegean prehistory is yet unclear even in the excavation context. In the broader region, sites such as Myrtos Pyrgos, Vasiliki, Gournia, Priniatikos Pyrgos, Mochlos, and Vronda continued from EM II into EM III. It is perhaps the poorly understood stratigraphic position of EM III and MM IA pottery, and the extreme regional diversity of ceramic forms that have led to controversy over this important transitional phase in the development of Minoan culture (Schoep 1999, 270; Watrous 1994, 718–720; Manning 1994, 231; Momigliano 1991; Cadogan 1986, 158; Andreou 1978, 164–172).

66

KAVOUSI I

The condition of the evidence from the well deposit of Vasiliki and the north trench at Gournia is perhaps echoed at Myrtos Pyrgos, Mochlos, and Kavousi, where EM III and often MM IA material is clearly post de facto refuse, that is, secondary dumped deposits that are usually the result of late MM IA or even Protopalatial activity on the site. Furthermore, the deposits are frequently localized, that is, physically moved from their original position and usually obscured completely by subsequent habitation. Therefore, since the Kavousi village site represents a single excavated area, lack of remains of the period is perhaps not surprising given the small size of the sample and the longevity of the use of the site. Tholos, on the other hand, is bedrock terrain, where the fullest extent of the site is exposed on the ground surface. In this latter instance, the lack of EM III–MM IA remains is probably a sign of abandonment of this coastal site after the EM IIB destructions. Perhaps the growth of settlement in the neighboring Hagios Antonios Valley is a result of population movement inland from Tholos. In the north Mount Papoura catchment area (Fig. 18), Vronda (77) continued into EM III–MM IA, but new sites were founded at Chondrovolakes (68), evidently a single building, and at Azoria (71). The use of the hilly terrain south of Kavousi village is not surprising. The Vronda spring provided a yearround water source, and the mixed phyllite and terra rossa hillsides were ideal for rain-fed agriculture. This area of Chondrovolakes and Xerambela is also a natural route from the Kavousi plain inland to the highland pastures of Mount Papoura and Thriphti; another route extends east, past Azoria, to the Avgo Valley. Evidence of a growing population, expansion of settlement, and exploration of new environments in EM III–EM IA is apparent in these remote mountain regions as well as in the plain and coastal hills. A farmhouse or hamlet (92) was established east of Avgo at the Bebonas watershed (Fig. 10). This site, lying some 800 m above sea level, was founded in EM III along with site 97, which is a hamlet or village (ca. 600 m) on the lower western slope of Thriphti, high above the Cha Gorge (Fig. 17). These new sites were established by settlers seeking new agricultural land, pasturage, and stable water supplies. Their existence is evidence of a new impetus for settlement siting, controlling new pasturage, and bringing areas previously unoccupied or unexploited under cultivation. In the plain there are

signs that the limited arable was used to its fullest potential; sites are located on essentially uncultivable terrain, leaving the arable free for agricuture. This apparent agricultural resource management accords well with the evidence for the first signs of activity in the mountain zones. One area in the Kavousi region has not only the most dense EM III–MM IA remains of occupation but also evidence for the foundation of no fewer than six separate new sites. This area, Chordakia (Fig. 18), is located at the western edge of the Kambos and at the southeastern slope of Mount Chomatas. The region consists of three separate locales: Katsoprinos (site 28), Kephalolimnos (site 36), and Lakkos Ambeliou (sites 35, 41, 44, 45). Chordakia, like Hagios Antonios, is a small valley of silty phyllite colluvium, which is the most productive arable land in the plain. While no spring is known today in the vicinity, the juncture of limestone and phyllite bedrock in this area is conducive to the formation of natural aquifers. Perennial springs in similar environments in the Hagios Antonios and Xerambela areas have dried up in recent times—a reminder that climatic changes and modern land use have affected the hydrology of the region. In Chordakia, a Byzantine cistern at site 29 could mark the position of a spring, now dry, much in the same way that modern cisterns are located at spring apertures to collect the slower flow in the summer months and to control the volume and rate of flow for irrigation. Such cisterns are located at springs at Avgo and Hagios Antonios. A well (of Roman date) is located in the plain at site 37. The EM III–MM IA sites in Chordakia are probably separate houses, since the architectural remains at sites 44 and 36 are demonstrably foundations of individual buildings. The sites are clustered together on bedrock terrain at the fringes of arable fields. As at Hagios Antonios, occupation of the phyllite alluvium was strictly avoided in order to maximize the total area of cultivation. The overall patterns at Chordakia and Hagios Antonios are thus similar, and the concentration of settlement in these two locales is not surprising in light of the relative agricultural potential of the various soils of the Kambos. These two isolated outcrops of phyllite at Hagios Antonios and Chordakia have produced alluvium that is preferable historically to the rocky, arid terra rossa of the central area of the plain. These two areas are also small valleys, protected on the north

THE HISTORY OF SETTLEMENT

and west by the mountains of Mounts Schinias and Chomatas, but each enjoys easy access to the sea and plain. Finally, Hagios Antonios and Chordakia controlled their own water supplies. What is interesting about the dispersal of houses in Chordakia is the lack of evidence for a substantial EM I–II settlement. At Theriospelio (31), located on the western side of Mount Chomatas, the inaccessible location, topography, and aspect do not favor a permanent settlement function. Although recent excavations at the Chrysokamino Metallurgical Site (32) and Habitation Site (34) have recovered FN and EM pottery, the material is too insubstantial to be the remains of settlements of significant size (Betancourt 2000a; Betancourt and Floyd 2000– 2001; Floyd 2000; Betancourt et al. 1999). Thus, while the development of settlement in Hagios Antonios was a continuous process of growth from Neolithic or EM I culminating in the foundation of site 5, the sites in the Chordakia area appear to have been mostly new foundations, and the origin of their population remains somewhat obscure. Is this settlement expansion into the Chordakia area in EM III– MM IA simply part of the same phenomenon of population growth and the exploitation of new areas by a local population? Indeed, new sites are founded in Hagios Antonios, Chondrovolakes, and high in the mountains at Bebonas and Thriphti (Fig. 10). It is possible that this cluster of houses at Chordakia represents settlers spreading south from Hagios Antonios or northwest from Kavousi or Xerambela, joining a small local population in exploiting the rich agricultural potential of the Lakkos Ambeliou and the phyllite soils of west Kambos. If so, this farmhouse cluster represents a group with lineage ties to the settlements at Hagios Antonios or Kavousi-Xerambela. By MM IA, it had established its own subsistence base, and no doubt its own social identity within the region by virtue of its association with the Chordakia Valley. What is apparent with the dispersal of settlement in EM III–MM IA is the beginning of the foundation of distinct and separate clusters of class 1–2 sites within a 0.5 km radius of springs and within close proximity to definable locales of arable land. Hagios Antonios, Chordakia, and Xerambela are these late Prepalatial site clusters, which, we will see, continue in use, grow more densely settled, and take their final form in the subsequent Protopalatial period. Dispersed sites clustering in

67

locales of concentrated arable land could be the result of a process that began in EM III–MM IA, and there is no particularly strong indication of a new or intrusive population element into the region, only a response to local environment in the context of a gradual population increase or perhaps climate changes at the end of EM II. Such a gradualist model is, of course, a positivist reading of the necessarily compressed chronology of the survey data. Not only is it difficult to recognize EM III–MM IA sherds in survey samples, but stratigraphic and ceramic phases are not sufficiently understood from excavation. Nevertheless, the discontinuity apparent at Tholos (1) and Kavousi village (24) may point to actual abandonment, while at Hagios Antonios, sites 8 and 6 continued in use after EM IIB long enough to see the use of Whiteon-Dark Ware. These sites were abandoned only with the foundation of site 5, nearby, probably in MM I. Such a scenario stresses the ambivalence of the survey data in forming inductive arguments for cultural continuity by means of a necessarily telescoped chronology. Can an argument be made for an intrusive population element that might be responsible for the new dispersed settlement pattern following the breakup of the old nucleated EM II hamlets? Given both the proximity of the new sites to the plain, the lack of a defensible character of the sites themselves, and the extensive use of diverse environments especially in the hinterland, if the newcomers were involved in the changes, either they arrived slowly and peacefully or they were successful in completely supplanting the EM II village structure. What can be said is that the Kavousi area experienced a significant restructuring of settlement in EM III, a change that marks a clear break from a pattern that had remained fairly constant since the Final Neolithic. One interesting site associated with the Chordakia settlement cluster in EM III is of course the Chrysokamino Metallurgical Site (32), located on the small saddle of the headland of Mount Chomatas jutting out into the Bay of Mirabello (Fig. 18) (Betancourt 2000a; Betancourt et al. 1999). While EM III–MM II pottery was recovered on the surface of the site, the most prevalent artifacts were chafftempered terra cotta fragments with holes piercing their sides. Black slag silicate was also found among the terra cotta fragments. While a number of hypotheses have been put forward on the function of

68

KAVOUSI I

Chrysokamino, thermoluminescence dates derived from analysis of the terra cotta fragments confirm a Prepalatial date; Stos-Gale has argued that the pieces were fragments of terra cotta furnaces for a copper-smelting operation (Stos-Gale 1993; Nakou 1995). Both the date and the function of the site were reconfirmed and refined by Betancourt, who excavated the site from 1996 to 1998 (Muhly 2004; Betancourt 2000a; Betancourt et al. 1999). Could the foundation of new settlements in Chordakia in EM III have been the result of the establishment of the smelting site at nearby Chrysokamino? No evidence for an EM III–MM IA site exists on the west side of Mount Chomatas, although Hazzidakis, visiting Theriospelio, reported finding White-on-Dark Ware (Foster 1978; Mosso 1910, 219; Hawes et al. 1908, 33). The pottery recovered from the cave by Hall and later by Zois is Neolithic–EM II (Zois 1993, 340). The cave itself, even if it was unsuitable for habitation, could have functioned as a burial cave, if not a temporary camp or shelter. What was the relationship between the Chordakia sites and nearby Chrysokamino (Fig. 18)? Two valleys, Chordakia and Lakkos Ambeliou, are situated about 600 meters southeast of the Mount Chomatas headland, on which the metallurgical operation is situated. These valleys provide the nearest and best agricultural environment for the workers of the smelting site. Furthermore, the density of settlement—no fewer than six sites with evidence of EM III–MM IA occupation—is a strong indication that the population working the smelting operation at Chrysokamino was derived from the Chordakia cluster. The nearby port of Agriomandra, situated south of Chrysokamino, is a natural roadstead at the mouth of the torrent that feeds directly west from the Lakkos Ambeliou-Chordakia watershed. The latter is precisely the location of the most dense concentration of late Prepalatial habitation in the area. A well-built kalderimi (road) is a reminder of the use of Agriomandra as a port in Venetian, Ottoman, and recent times, and deep offshore deposits of amphorae are evidence of its function as a Roman harbor. Before World War II, the roadstead here was a transshipment point for produce such as cheese, carob, wine, and oil, exported out of the Kavousi region and into the Bay of Mirabello. While no EM III–MM IA pottery was recovered on the shore of the alluvial bay, it is likely that it was a port for the

Chordakia cluster, and more importantly, for the importation of copper ore and the export of bulk copper from Chrysokamino. The difficulties in recovering and recognizing phases of EM III–MM IA notwithstanding, the distribution of sites seems to indicate dispersal of population and adherence to arable land and water supplies. The population seems also to have grown significantly from EM I to III. The large number of smaller sites—at least nine of them class 1–2— especially in non-defensible locations on the fringe of the Kambos plain, would suggest an integrated socioeconomic system. While the port of Tholos may have been abandoned in this late Prepalatial phase, perhaps a sign of the vulnerability of the littoral zones, the sites in the hinterland seem to have thrived. Sites are in the open, and on the whole, easily accessible from the plain. The Kambos was used for agriculture, with the choice soils of Hagios Antonios and Chordakia drawing the densest habitation. The existence of sites on the upland slopes of Mount Papoura (Xerambela), Avgo, and Thriphti illustrate that the upper elevations, even as high as 800 meters, were intensively exploited (Fig. 18). Could expanded agricultural exploitation of the coastal hills, foothills, and plain have required shepherds to seek new pasture in the upper elevations? Finally, the special function site of Chrysokamino and its port at Agriomandra are indications of an active economy involving off-island and interregional exchange. Such a dynamic economic picture is reflected in the continuation and perhaps proliferation of pottery made of Mirabello (Kalo ChorioGournia) fabrics, not only imported by the Kavousi settlements, but in this period reaching distant ports such as Malia, Myrtos Pyrgos, and Petras-Siteia. While vessels made of the distinctive Mirabello fabrics are found as far away as Knossos in Early Neolithic (Tomkins and Day 2001) and play a significant role in an intensive network of exchange throughout EM II (Whitelaw et al. 1997), the distribution of a wide variety of wares and vessel types in such fabrics is only intensified in EM III–MM IA (Haggis 2000). The late Prepalatial period at Kavousi is one of considerable change. Interregional communication and exchange must have accompanied the expansion of settlement. Such a pattern seems to look decidedly forward to the Protopalatial period (cf. Hayden 2004a, 76) rather than back to the nucleated insular pattern of EM II.

THE HISTORY OF SETTLEMENT

Our ability to reconcile the pattern of change in late Prepalatial Kavousi with other regions of Crete is probably hindered by our limited understanding of the stratigraphic divisions separating EM III and MM IA and the regional variations of the pottery in use during these chronological phases; certainly the idiosyncratic definition of ceramic and occupational phases in various surveys has perhaps compounded the problem (Hayden 2004a, 76; Driessen 2001, 53; Watrous 1994, 718–720). While Hayden has attributed the hazy situation of late Prepalatial at Vrokastro to our lack of understanding of EM III ceramic forms, she admits that a retraction of settlement may account for the drop in site numbers (Hayden 2004a, 76, 99). In the neighboring region of Gournia, Watrous detects a sharp reduction in settlement in EM III, only to be followed by resettlement in MM IA (Watrous et al. 2000, 475). Admitting that the lack of stratified sequences in the Mesara region may be clouding the picture, Watrous also sees a similar reduction in settlement in south central Crete in EM III (Watrous et al. 1993, 224), where both the Mesara and Hagiofarango (Blackman and Branigan 1977) show a reduction in site numbers. Thus, for Watrous (1994, 720–736) MM IA becomes a watershed of culture change across the island, anticipating the rise of palatial complexity. Such a transformation of the cultural landscape involved the establishment of completely new sites, the development of new types of sites (such as peak sanctuaries), the reestablishment of elite burials, the revitalization and intensification of exchange with the eastern Mediterranean, and the proliferation of seals (including

69

hieroglyphic types), following a veritable hiatus subsequent to the EM IIB destructions. While some would place many of these changes firmly in the final stages of the late Prepalatial period, sharply separating them from the architectural foundation of palaces in MM IB (e.g., Schoep 2002b, 19–20; 1999; Haggis 2002; 1999a; Branigan 1993), the implications of Watrous’s approach for survey—and indeed the development of Minoan palatial society in general—are important: a real discontinuity is followed by a significant restructuring of the landscape in the period following the EM IIB destruction horizon encompassing a period of some 350 years (ca. 2250–1900 B.C.). On various regional scales, surveys in Crete could well be recovering evidence for Watrous’s historical discontinuity—the absence of clearly or easily identifiable EM III or MM IA sites, or alternatively the identification of new EM III or MM IA sites in new locations. While this discontinuity is clear enough, it will be more difficult to identify an actual gap in habitation contexts, on both the level of the site and the region. Whether the new sites in the Kavousi area are EM III or MM IA—and this is not unimportant if we are to visualize a real hiatus in the regional occupational history—what is important in the material pattern is the remarkable change from a fairly stable and constant if not static settlement structure in EM II. The new pattern established by the end of this late Prepalatial phase reflects a major reorganization of the Cretan cultural landscape, one that sets the stage for the Protopalatial settlement expansion which is apparent across the island of Crete (Driessen 2001).

Middle Minoan IB–II The transition from EM II to EM III was apparently marked by change. A nucleated pattern is replaced by an emergent dispersed pattern. While the same regions were exploited, some important sites were abandoned, and new sites were founded in previously uninhabited areas. These discontinuities perhaps signal climatic change, the breakdown or reorganization of EM II social structure, or perhaps even the arrival of newcomers in various areas of the island. Unfortunately, too little is known of the archaeology and scale of the EM IIB destructions.

While this late Prepalatial discontinuity is significant, a preponderance of evidence points to regional continuity of land use, ceramic production, and exchange. The period of transition in EM III–MM IA is characterized by growth of population and changes in economic systems, highlighting a new sociopolitical restructuring of the landscape. The changes provide the basis for developments in the subsequent Protopalatial period at Kavousi, foreshadowing the fourfold increase in the number of sites (Figs. 11, 19, 20). The Protopalatial population

70

KAVOUSI I

increase and settlement expansion evident in the Kavousi area are echoed dramatically in other areas of Crete (Driessen 2001, 61–63; MacGillivray 1997, 23; Warren 1987, 53–54). At Malia there is an astounding jump in site numbers from six to 80 (Müller 1996, 922; cf. Schoep 2002a, 122); at Vrokastro, the number of sites doubles (Hayden 2004a, 99–100; Hayden, Moody, and Rackham 1992, 323– 324), demonstrating a pattern of growth roughly similar to that at Gournia and Lasithi (Watrous et al. 2000, 475; Watrous 1982, 13–14) and the Mesara (Watrous et al. 1993, 225). In western Crete the Protopalatial transformation of the countryside is apparent at Chania Akrotiri, which also shows a significant increase in settlement in MM I–II (Moody 2004, 253; 2000, 186; 1987; Moody, Rackham, and Rapp 1996). The Hagios Vasilios Valley shows an increase from EM II—a period thinly represented in the survey zone—but the number of MM I–II settlements is no greater than FN–EM I (Moody 2000, 187; Moody, Peatfield, and Markoulaki 2000). This kind of sudden MM I–II regional colonization or reoccupation is also apparent in eastern Crete at Ziros (Branigan 1998) and Praisos (Whitley, Prent, and Thorn 1999), which, like Hagios Vasilios, produced little evidence of Prepalatial occupation. Divergent patterns are apparent in Sphakia (Moody 2000, 189) and the Hagiofarango (Blackman and Branigan 1977, 69), where a population decrease is the result of dramatic shifts in settlement, in the latter case most likely into the Mesara. A similar reduction in rural Protopalatial settlement may have to do with nucleation to new centers at Palaikastro, Petras, and Zakro (Driessen 2001, 59; Tsipopoulou 2002; 1999a; 1999b; 1997; 1989; Driessen and MacGillivray 1989, 102; cf. Moody 2000, 187). Fifty-three MM IB–MM II sites were recovered at Kavousi, and a definite pattern of growth is detectable in each of the late Prepalatial site clusters (Figs. 11, 20): Hagios Antonios, Chordakia (Fig. 19), and Xerambela-Chondrovolakes (Fig. 19). New settlements were also founded in the highland valley of Avgo, east of Kavousi (Figs. 11, 20). The port of Tholos is resettled on both sides of the bay, and the Kavousi village deposit from the square is full of pottery from this period. There appear to be four major concentrations of settlement, each prima facie corresponding to the locations of the late Prepalatial site clusters. The

phyllite soils were preferred because of their ability to retain more plant available moisture than the limestone terra rossa, thus making them ideal for intensive rain-fed agriculture. In the upper elevations in Avgo and Xerambela, these soils are productive even in the summer months of a first year of drought (Fig. 20). A fifth cluster of sites surrounds the Bay of Tholos, and although a good spring is located in the southwest corner of the bay, the sites are oriented toward the inlet itself, and thus the concentration of activity in the area is most likely because of the naturally protected harbor (Pl. 2B). The alluvium of the bay is poorer agricultural land than the phyllites of Hagios Antonios or Chordakia, but it is cultivable by means of irrigation from the Tholos spring or, in recent times, from a myriad of wells, which dominate the fields near the shore. As in EM III–MM IA, the sites of these clusters appear to be individual houses or small house groups (1–3 buildings) of class 1–2 size (cf. Waterhouse 1983; Xanthoudides 1906). There is continuity from the late Prepalatial period into the earliest Protopalatial phase, although the real growth of settlement in Tholos Bay seems not to have occurred until well into the Protopalatial period. South of Tholos, in the neighboring Hagios Antonios Valley, Alykomouri (8) and the rock shelter (6) went out of use by MM IB, while site 5 expanded its size (Figs. 11, 20). Furthermore, a new site (11) was founded directly in the Kambos plain, immediately below the EM I–III settlement at Alykomouri. In Chordakia (Fig. 19), only the Chrysokamino Metallurgical Site seems to have been abandoned. The rest of the sites in the various areas of the fertile Chordakia region (Lakkos Ambeliou, Katsoprinos, Kephalolimnos) continued in use, and some ten new sites were founded, essentially doubling the total number from the preceding period (Fig. 19). In the north Mount Papoura catchment area (Xerambela and Chondrovolakes) both Vronda (77) and site 68 continue to be inhabited, but eleven new sites were founded, and Kavousi village was reoccupied (Fig. 19). In the neighboring Avgo Valley four new sites were established in MM IB–II (Fig. 20). Throughout the Kavousi region there is an expansion of settlement in the areas first occupied in EM III–MM IA. The Protopalatial pattern represents the maximum number of sites and the widest dispersal of settlements in the Bronze Age. The sites themselves are small, no larger than small hamlets of

THE HISTORY OF SETTLEMENT

class 2 size, and where architecture is evident, the buildings show continuity from late Prepalatial (cf. Waterhouse 1983). As discussed above (Chapter 3), because of the sampling method employed, it was sometimes difficult to distinguish site sizes for all periods. A few multiperiod sites fall into this category. Site 1 at Tholos Bay is class 4, judging from the continuous distribution of Neopalatial pottery across the site in its maximum size. In the Protopalatial period, however, there seem to have been two smaller sites located within the northern quarter of what became the LM I town. Loci 16 and 38 both produced low but consistent percentages of MM I–II pottery. Most of the Protopalatial pottery is concentrated in the far north at locus 16, while the southernmost two-thirds of the site (locus 24) produced almost no MM I–II pottery. Similarly, at site 5, the Neopalatial town at Hagios Antonios, it is the continuous scatter of LM I pottery that defines the site’s maximum size. Protopalatial pottery is restricted to the western and southwestern areas of the site (locus 10), where the largest quantity of EM III–MM IA pottery was found as well. The northern and eastern areas of the site produced only traces of MM I–II material. Thus at Tholos (1) and Hagios Antonios (5) the MM I–II pottery is visibly restricted to areas that are less than one-third the size of the maximum distribution of Neopalatial wares. Site 28 (Fig. 19), an important class 3 site in the Chordakia area (Katsoprinos), reached its maximum size in Neopalatial and Roman periods. Only on the lowest terrace of the site was EM III–MM IA and MM I–II pottery evident. Thus, here as well, the Protopalatial remains are class 1–2 in size. Other multiperiod Class 3 sites that produced Protopalatial remains are 11, 15–17, 23, 27, 28, 34, 69, and 85 (Fig. 11). The volume and condition of the MM I–II pottery on these sites suggested limited occupation; in each instance, discontinuous patches of very worn MM I–II coarse fabrics were recovered in sampling units, whereas there was a preponderance of Neopalatial, Postpalatial, or Roman pottery. There is, therefore, little evidence that Protopalatial sites in the Kavousi area actually exceed the size of class 2, the hamlet category. Most are probably single houses or household complexes (cf. Waterhouse 1983). Admittedly, the Kavousi village site in the square remains an unknown quantity. It should be emphasized that both the size of the alluvial ridge on which the village sits and the spring

71

location make ideal conditions for a class 4 site. Could a town or village have existed here in MM I–II providing an economic center for the region in a two-tiered site hierarchy? It is definitely possible. The pattern of settlement in the Protopalatial period does suggest a landscape that is integrated socially and economically; the dispersed houses and hamlets were surely interrelated and interconnected in what must have been peaceful and mutually beneficial social and economic interaction. It remains to assess the nature of the integration. The absence of a discernible site size hierarchy in the region underscores the striking character of the site cluster pattern (Haggis 2002; 1999a). At Tholos, Hagios Antonios, Chordakia, Chondrovolakes, Xerambela (Fig. 20), and Avgo, a number of farmsteads—as few as four or five in Avgo, and as many as twelve to thirteen in Chordakia—are clustered within a roughly half-kilometer radius of a concentration of arable land and usually a permanent water supply (Fig. 20). At Chordakia, Xerambela, and Avgo, where the patterns are clearest, the small size of the individual sites would have permitted the inhabitants great flexibility in locating their houses (Figs. 11, 19, 20). Buildings were placed so as to minimize direct occupation of potentially cultivable areas; this open settlement arrangement is possibly a result of both social structure and the needs of agricultural productivity in a marginal environment. The spaces between houses in a cluster could have been tilled for small gardens and used to keep household livestock. Small rises of exposed bedrock terrain were usually selected for building, permitting optimal use of the surrounding fields. The places chosen for building were frequently higher ground than the associated fields, and on the whole, dolomite and gray crystalline limestone hillocks and hillsides are preferred site locations. The terra rossa soils generated by such rocks are, in this environment, ill suited for rain-fed agriculture. Fields are harder to till, crops require more water than in phyllite soils (because of high rock content), and the slopes themselves can be extremely erosive, requiring laborintensive terracing. The phyllites, therefore, are selected and reserved for planting. This site cluster pattern is possibly related to agricultural diversity, self-sufficiency, and the variability of productivity in diverse environments. It may also have been culturally dictated. If the sites themselves represent individual family units, what

72

KAVOUSI I

might be the social structure that defines the relationship between sites and the nature of integration on a regional scale? Are the clusters manifestations of interrelated families or extended families occupying and sharing ancestral land? Was the land communally or collectively exploited—as was the central water source—and was the identity of the household and the exclusivity of the household garden maintained or reinforced through the juxtaposition of these separate buildings? It is possible that the primary subsistence and surplus needs of each cluster were satisfied through three levels of production and spatial scales of activity: household gardens, family-owned fields, and cooperative exploitation of fields beyond the immediate area of the cluster. Communal activities might include irrigation, terracing, crop-specific cultivation, and storage. And some form of social organization beyond the household would have been required to initiate, maintain, and direct such activities. The clusters then have the character of separate, autonomous, or semiautonomous communities—essentially lineage-based villages, but lacking the typical contiguous or agglomerative plans usually associated with Minoan settlements like Gournia or Pseira. There is little evidence within each cluster for a size-ranked hierarchy of settlement. In the two best preserved clusters—that is, those not subsequently obscured by extensive Neopalatial and Postpalatial building—there are interesting features that might indicate differentiation of status or function, even if a size hierarchy is absent. In Chordakia, site size differentiation in MM I–II is negligible, most sites being 0.10 ha. or smaller (Fig. 19). One site, however, is interesting in its location, architecture, and finds. Site 44 is centrally located within the cluster and situated directly on the watershed between the arable fields of Lakkos Ambeliou in the southwest and Kephalolimnos in the northeast. The site is also located directly on the route from the plain of Kambos to the port at Agriomandra (Fig. 19). The architecture is preserved in the form of an enormous foundation of limestone boulders. Dense stone debris on the terrace of the building, and use of boulders in the foundations themselves, exhibit the extensive use of stone that perhaps distinguished the building from others in the cluster. The other sites are not nearly so well preserved, and large bouldersized stones are not evident elsewhere. The finds associated with the building are also distinctive.

Material collected on the east side of the terrace included several larnax fragments from both late Prepalatial and Protopalatial periods as well as a preponderance of pithoi, jugs, and amphorae. The distinctive character of site 44 is expressed through its location, architecture, and possible burials. Could such a single house—differentiated not by size, but by position, function, and architecture—be evidence of centralized organization in the cluster? Soles has argued that evidence for social stratification is discernible in the position and relative sizes and types of tombs at Mochlos, Gournia, Malia, and elsewhere in the late Prepalatial and Protopalatial periods (Soles 1992, 1988). And the evidence for cult equipment, such as kernoi, at these cemeteries implies a community cult place where social status and perhaps political roles were lineage based and expressed through the location, architecture, and practice of the cult itself. Although cemeteries for large nucleated sites, such as Malia and Mochlos, are located some distance from the inhabited area, others such as at Pyrgos and Gournia were built contiguous with the settlements themselves. Could site 44 have been the residence of a family of high status, the community leader’s house whose ancestral cemetery and cult place were located nearby? If the essential ties between sites in a cluster were kinship based, then the identity of the community would have been linked to the landscape, the agricultural and pastoral lands exploited, and the resources shared by that cluster. The social links between the individual houses and hamlets of the cluster are no less important than those in a nucleated village with an agglomerative plan. And we need not seek a spatial hierarchy of settlement to define the social structure in the landscape. It is the notion of a territory and a consciousness of place that defined the late Prepalatial and Protopalatial community—not the physical architectural form of the village or town. Individual family identity may also have been expressed in the cluster arrangement. Site 44 stands out as a potential central focus for local cult and community focus. Perhaps beginning as an important household in the Prepalatial period, the building and its adjoining cemetery could have become, by MM I–II, ritually and socially significant. A similar, but potentially more complex pattern is evident in the Xerambela-Chondrovolakes cluster on the foot slopes of Mount Papoura above Kavousi

THE HISTORY OF SETTLEMENT

village (Fig. 19). The central site in the cluster is 68, which, like site 44 in Chordakia, is distinctive in its situation and architecture. Site 68 is centrally placed in the settlement cluster and in the north Mount Papoura catchment area, enjoying a view of the Kambos plain to the north and west. The site is also located at the intersection of two main routes in the Siteia Mountains: one that ascends from Kavousi village through Xerambela to Mount Papoura and Thriphti, and another that branches east from Chondrovolakes traveling between Azoria and the Kastro into the Avgo Valley; today an Ottoman or Venetian period kalderimi marks this frequently used path. The architecture at site 68 is most striking as it consists of Cyclopean walls standing 1.0– 2.0 m in height, and as at site 44, EM III–MM IA pottery indicates an early foundation date for this large, centrally placed building. Larnax fragments recovered from the site may be evidence of a nearby cemetery. The situation and character of site 68 establish its central importance within the cluster. Excavations in the area of Xerambela at Vronda (Fig. 19: site 77) revealed EM IIB, EM III–MM IA, and Protopalatial material (including a sealing) underlying the LM IIIC settlement of Vronda. Just east of the central LM IIIC Building A–B on the hilltop is a large Protopalatial wall, perhaps the remains of a significant central building. While the size and configuration of the settlement for late Prepalatial and Protopalatial periods remain obscured by extensive rebuilding in LM IIIC, it is possible that Vronda was similar to site 68, a competing household within the cluster. Vronda’s position permitted its inhabitants to control the mountain passage to Mounts Papoura and Thriphti, a nearby spring, and about twenty-five hectares of arable land in the immediate vicinity. It is tempting to reconstruct two overlapping and possibly competing site clusters in the north Mount Papoura catchment, one around site 68 at Chondrovolakes and the other around site 77 at Xerambela. Sites in the clusters at Hagios Antonios and Avgo are not well preserved or are much too obscured by later MM III–LM III phases to permit identification of specific architectural features belonging to late Prepalatial and Protopalatial periods (Figs. 11, 20). The cluster pattern is nevertheless apparent from the distribution of ceramic remains. On the scale of the Kavousi region, the clusters are spaced about 2.0 km from each other, suggesting

73

minimum agricultural catchment areas of about 1.0 km radius from the presumed center of each cluster (Fig. 20). The rendering of such “territories” is well in accord with Warren’s “diurnal work factor” which he introduced as a way to reconstruct agricultural zones for Minoan communities (Warren 1984). The drawing of such “zones” on the map (Fig. 20) presents an ambivalent role for Tholos, which clearly must overlap with the Hagios Antonios cluster lying to the south. Tholos is, however, the only coastal site. While it does have a good spring, the agricultural potential of the immediate surrounding area is relatively poor. Site 1 today occupies eroded dolomite bedrock terrain, and sites 2 and 3 are well within the scree line of Mount Kapsas. As a port, harbor, and fishing facility, Tholos is a special case, lacking the primary agricultural criteria for the clusters. The focal point of the cluster was surely Hagios Antonios and the hamlets lying in the northern Kambos (Figs. 11, 20). The Protopalatial site clusters probably functioned like villages. They were agriculturally selfsufficient, occupied separate agricultural zones, and directly controlled miniature territories and local water supplies. At the time of their inception in EM III, it is likely that a kinship-based social structure permitted or even encouraged the ascendancy of a dominant household that was probably the residence of the cluster’s chief or leader, someone responsible for organized cooperative labor and public cult activities. The association of larnakes with large, perhaps even elaborate, buildings in no less than two of the clusters suggests the significance of these large central structures. By MM IIB, with fifty-three sites spread over the five site clusters, such an insular social organization could simply not translate into a viable political framework for the broader region. In MM III, sites appear to have spread out gradually from each of the clusters into the Kambos, compromising both the stability and insularity of the cluster pattern (Figs. 12, 21). Southeast of Hagios Antonios there are large potsherd scatters at sites 15–19 (Fig. 21). While some MM I–II pottery is evident, the predominant material is MM III–LM I. Similarly, north and west of Kavousi village, sites 20, 21, 23, 25, 27, 50–52, and 54 are Neopalatial settlements with only traces of MM I–II activity. These new sites could represent a late Protopalatial expansion of population and the resource base into the Kambos from Xerambela,

74

KAVOUSI I

Chordakia, and Hagios Antonios. The new settlements perhaps sought to exploit the rocky and arid terra rossa soils of the plain through irrigation. In any case, extensive agriculture, or at least systematic use of the plain, would have required a complex socioeconomic structure to bring together the various and disparate settlements into a network operating on a regional scale. Unfortunately, there is no evidence in the Protopalatial period for a site hierarchy that satisfies our notion of organization in the landscape. Could it be that a single site or a single cluster was a dominant center in the region, providing the cohesion necessary for a larger regional community? While a larger-size village or town—a higher-order center—might be sought outside the limits of the survey zone (or perhaps underneath Kavousi village itself), it remains possible that organizational hierarchies might exist in the Kavousi region, but not manifest in a simple size-ranked distribution of sites. Recent studies of Protopalatial society have effectively argued for a decentralization of organizational structure on various scales—the town, the palace, and the countryside. Regionalism in archival systems, ceramic distribution and consumption, and craft production (Knappett and Schoep 2000; Day and Wilson 1998; cf. Branigan 1990; 1987; MacGillivray 1990, 432–433) emphasizes a distinctly heterarchical structure (cf. Crumley 1987) even for the palace centers themselves (Schoep 2002a; 2002b), in which power is not unilaterally centralized, but distributed, counterpoised, and potentially negotiable. More relevant for survey is that this decentralized state has been extended to the analysis of the palatial hinterland (Knappett 1999; 1997) where a segmented structure encouraged social reproduction on the regional scale (cf. Vansteenhuyse 2002; Cunningham 2001). In this framework, influence of the center on the hinterland might not have been economic or politically coercive; the evidence

strongly suggests that center-periphery interaction was varied in form, communally ritualized, and largely ideological in nature. The roots of this complex and decentralized system are perhaps to be found in the Prepalatial period (Cunningham and Driessen 2004, 109; Haggis 2002; Schoep 2002b, 21). Extending the model beyond the town and palace into the hinterland itself we might discern aspects of the structure underlying the Protopalatial settlement system at Kavousi. The undifferentiated MM I–II site clusters present a decentralized and essentially heterarchical pattern. The apparent distribution of sites might emphasize regional integration, a structure in which a multiplicity of sociopolitical units—farmhouse and hamlet clusters and villages—represent both competing and cooperative interests; links to the wider community could have been both socially (through kinship ties) as well as ritually (through corporate and communal feasting) determined. The problematic lack of hierarchy should not indicate a lack of structure—the absence of an arrangement of size-ranked settlements—nor should we have to look to an extra-regional source of centralized power, such as a palatial authority at Gournia or Ierapetra, to make the pattern comprehensible (Watrous et al. 2000, 476). Is it possible to conceive of power relationships in the Kavousi area as being flexible and negotiable, operating on both local and regional scales, and being played out between clusters, around the Mirabello Bay and across the north Isthmus region in a variety of different ways? The MM I–II settlements at Kavousi comprise a distinctly non-hierarchical and “non-dendridic” (cf. Cunningham and Driessen 2004, 108) material pattern, suggesting local agency in economic and political relationships and an essentially corporate structure (Cunningham and Driessen 2004, 108). This stable and long-lived Protopalatial sociopolitical equilibrium was evidently disrupted in the subsequent Neopalatial period.

Middle Minoan III–Late Minoan I In MM III, there are striking changes in settlement patterns (Fig. 12). A marked decrease in the total number of sites corresponds to the dissolution of the Protopalatial site clusters. Considering the

longevity, constant growth, density, insularity, and agricultural independence of the site clusters from EM III through the Protopalatial period, the subsequent changes in MM III–LM I are nothing short

THE HISTORY OF SETTLEMENT

of revolutionary. Such change signals a significant sociopolitical restructuring of the landscape. This contracted pattern is echoed across the island in the Neopalatial period (Hayden 2004a, 118–120; Cunningham and Driessen 2004, 105– 106; Driessen 2001, 56–60). In the neighboring Gournia and Vrokastro areas (Hayden 2004a, 112– 115; Watrous et al. 2000, 476), the reduction in the number of sites is highly variable with no real evidence yet for settlement nucleation causing the change. This is to say that different areas of these regions may have responded differently to the restructuring of the MM III–LM I transition. It is interesting that Watrous (et al. 2000, 476) documents some kind of demographic consolidation at Gournia, with the town itself actually shrinking in size. The drop in site numbers in Lasithi and Malia (Watrous 1982, 15; Müller 1998, 548–552; 1996) is even more dramatic than in the Mirabello Bay area, while the Mesara shows a more subtle decrease attributable to centralization of population to the larger settlements (Watrous et al. 1993, 226). Divergent patterns are found in western Crete, where there is a clear growth in settlement from Protopalatial to Neopalatial in the Hagios Vasilios Valley (Moody 2000; Moody, Peatfield, and Markoulaki 2000) and in the Chania area (Moody 2000, 186), but the reduction in settlement in the Akrotiri itself (Moody 2004, 253) indicates conditions similar to central and eastern Crete. At Ziros, Chania, and Mesara there is considerable continuity in the use of Protopalatial sites (Moody 2000, 186; Branigan 1998; Watrous et al. 1993, 226). At Kavousi, the total number of sites drops from fifty-three in MM I–II to not more than forty in LM IA. The overall decrease is not as striking as the changes apparent in each of the areas of the Protopalatial settlement clusters. At Avgo, Xerambela, and Chordakia (Fig. 12), the pattern is identical. In these areas there is abandonment of 50 percent of the sites, and significant rebuilding at others. This pattern has been confirmed by recent excavations at site 34, the Chrysokamino Habitation Site, in the Chordakia cluster (Figs. 12, 21; Betancourt and Floyd 2000–2001; Floyd 2000). In every area it seems as if numerous Protopalatial farmhouses were abandoned in favor of large nucleated sites in the Kambos, specifically in areas northwest of Kavousi village and at Hagios Antonios and Tholos (Fig. 21). This process of nucleation and population

75

centralization is not as clear in Vrokastro or Gournia (Hayden 2004a; Watrous et al. 2000), but is very likely to be the case in the Mesara (Watrous et al. 1993). In Avgo, sites 85 and 90 were abandoned, while sites 87 and 88 continued in use. The imposing Cyclopean wall at site 87 (Hayden 1997b; Boyd 1904) is possibly a Neopalatial addition, emphasizing an important characteristic of the LM I landscape: the cluster of Protopalatial houses was eventually replaced by a large single building strategically positioned to command both routes and discrete areas of arable land. While megalithic buildings located strategically along inland routes have been called watchtowers associated with Protopalatial administrative systems (Tzedakis et al. 1990; 1989; cf. MacGillivray 1997), it seems that many of these have a significant LM I phase. In Xerambela, only seven of the original thirteen Protopalatial sites (Figs. 11–12) seem to have survived the transition to Neopalatial. Only site 69 produced extensive MM III–LM IA pottery and megalithic wall foundations. In Chordakia (Fig. 21), the reduction is more striking still, as a mere seven of the original sixteen sites (Figs. 19, 21) seem to continue in use. Here only two sites—28 on the hill of Katsoprinos over looking Chordakia valley, and the Chrysokamino Habitation Site (34) on Mount Chomatas, overlooking Lakkos Ambeliou— produced extensive Neopalatial pottery and contemporary house wall remains. The pattern in each of these clusters involves abandonment of sites, population reduction, and the foundation or reestablishment of new class 1 sites in each locale. These new sites are typical Neopalatial rural houses or “megalithic farmsteads,” and their locations on hilltops above fertile alluvial valleys points to some connection with agricultural production (Betancourt and Floyd 2000–2001; Hayden 1997b; MacGillivray 1997; Tsipopoulou and Vagnetti 1995; Tsipopoulou and Papacostopoulou 1997; Nixon 1987; Wiener 1984; 1987; Warren 1984; Hood 1983; McEnroe 1982; Cadogan 1971). The repetition of such buildings across the landscape in specific areas could mean that new agricultural strategies involved systematic, specialized, and intensive exploitation of localized arable land for certain crops (Hayden 2004a, 118– 119; Bevan 2002). In two instances where house foundations were well preserved and the buildings

76

KAVOUSI I

excavated—at Panagia in Avgo (87) and the Chrysokamino Habitation Site (34)—the ground plans conform roughly to the simple “type 3” house as defined by John McEnroe (Betancourt and Floyd 2000–2001; Floyd 2000; McEnroe 1982). In Chordakia (Fig. 21), two Neopalatial houses supplant the Protopalatial cluster, and their positions are important. Site 28 is located on the easternmost point of a ridge of Mount Chomatas (Katsoprinos) overlooking the Chordakia Valley. Site 34 is on the prominent peak at the southern end of Mount Chomatas; it has a southern aspect, overlooking the alluviated basin of Lakkos Ambeliou. That is, both houses command very distinct environments. Another striking change in the ProtopalatialNeopalatial transition is the emergence of class 3 and 4 sites in the Hagios Antonios and Tholos areas (Fig. 21). In Hagios Antonios only one Protopalatial site is abandoned (7) while sites 5 and 11 grow in size. Site 11 is class 3—a hamlet—and site 5 is the size of a large village or small town (3.0 ha.). North of Hagios Antonios at Tholos Bay is site 1, a substantial Neopalatial town. Like site 5, it is no smaller than 3.0 ha. Along with these towns at Hagios Antonios and Tholos, there is a proliferation of class 3 hamlets in the plain. While the central alluvium remains thinly settled, three large hamlets developed by LM I on either side of the Platys River north of Kavousi village (Figs. 12, 21). These sites are 23 and 27, and west of Kavousi in the small plain of Petras is site 52. The most pronounced concentration of habitation is in the northern Kambos, southeast of Hagios Antonios on the east bank of the Platys River (Fig. 21). Here the small class 1 sites 18 and 19 continue in use in LM I, probably as farmhouses, but sites 15 and 17 are large hamlets of class 3 size. Together with sites 5 and 11 across the river at Hagios Antonios, the northern Kambos plain appears to have been the most densely settled area of the Kavousi region. These large class 3 and 4 settlements are located roughly equidistant from each other on both sides of the Platys River. While the Hagios Antonios-area inhabitants, like their Protopalatial ancestors, could have continued to exploit the good phyllite soils and constant water supply from the spring for irrigation, the occupants of sites 15–19 on the east side of the river faced a different environment altogether. The north Kambos region lacks accessible water, and the dense angular rock debris and scree from Mount Kapsas are combined

with cobbles and boulders from the banks of the Platys River, making the limestone soils of this area difficult to manage. Given these conditions, it is likely that these sites (15–19) used the river itself for irrigation. While check dams have not been found in the modern Platys streambed, the wide overbank deposit on either side of the present river course could have obscured such structures. Dams have been identified elsewhere at Mochlos, Pseira, Gournia, and in the Mesara (Hope Simpson 2005; Hayden 2004a, 98; Watrous et al. 2000, 476; 1994, 736; Soles 1997; Hope Simpson et al. 1995, 374–375, 396; Hope Simpson and Betancourt 1990, 322–323). Landscape modification was surely practiced in LM I in the area, as suggested by the agricultural terrace walls found at site 5 in the Hagios Antonios Valley. There, the erosive limestone slope of Mount Schinias required constant maintenance, and agricultural terracing would have been a necessity. Similar terrace walls were excavated and studied on the neighboring island of Pseira across the bay from Tholos (Betancourt and Hope Simpson 1992). The position of these new hamlets in the north Kambos plain and the growth of a villagesized settlement at Hagios Antonios suggest systematic agricultural exploitation of the plain. The large size of the settlements (class 3 and 4) and their location on either side of the Platys River, combined with the evidence of terracing, are all signs of significant changes in agricultural practices and perhaps social systems in the transition from Protopalatial to Neopalatial periods. In LM I, the settlement pattern at Kavousi is nucleated as population appears to have been drawn away from the dispersed site clusters of MM I–II into the hamlets and villages of Kambos and Tholos. At the same time, the focus of land use shifted from the phyllite soils of the clusters to the open alluvium of the Kambos plain. This extensive exploitation of the marginal terra rossa soils of Mount Schinias, Kambos, and Tholos, by means of terracing and irrigation, was probably the result of changing economic interests. More land was brought under cultivation for the larger nucleated population concentrated in the north Kambos plain and Tholos Bay. But why were the productive phyllite soils and springs of Avgo, Xerambela, and Chordakia largely abandoned for the villages in Hagios Antonios (sites 5 and 11), the north Kambos (sites 15–19), and Tholos (site 1)? Several reasons are possible:

THE HISTORY OF SETTLEMENT

1. The insular clusters could not sustain population growth beyond the carrying capacity suggested by the resources in each of the locales. 2. Agricultural specialization encouraged by the towns at Hagios Antonios and Tholos led to selective intensive use of the phyllite soils and controllable water supplies. 3. Intrainsular and off-island trade routes and (palace-based) systems of interregional and island-wide exchange encouraged a thriving port-town economy (e.g., neighboring Priniatikos Pyrgos, Gournia, Mochlos, and Pseira)—such an economy would have been a powerful draw away from rural farms. 4. The population threshold reached by the end of MM II in the Kavousi region either required or encouraged a hierarchical and highly connected relationship between sites. Could population have grown to such an extent by MM IIB that both soil and water depletion necessitated shifts in agricultural practices? Warren has argued that a high population density in the Mirabello area may have affected settlement development in LM I. Citing population growth in areas such as Avgo and Hagios Antonios, he has suggested that depletion of resources in such intensively exploited and densely inhabited areas may account for increased off-island interests in the Neopalatial period (Warren 1984). Indeed, the scarcity of arable land and the paucity of water resources, especially in eastern Crete, might give credence to Warren’s view. The effects of even minor increases in local population would have been profound, especially set against a backdrop of environmental changes that might have encouraged economic diversification and new sociopolitical configurations manifest in the development of the Minoan palatial economy (Moody 2004, 254). The environmental conditions notwithstanding, in the Neopalatial period, changes are evident in settlement patterns suggesting new social organization or economic conditions. On the land occupied earlier by the Protopalatial site clusters—and depopulated by the end of MM II—one or two prominent Neopalatial houses are found. Are these individual structures evidence of agricultural specialization or lingering landed elite? Both cases are possible.

77

Intensification of production and crop specialization in well-watered, highly productive fields could have been managed by elite families who were also dependent on the goods, services, and cult apparatus of the neighboring towns of Hagios Antonios, Tholos, Kavousi, or a higher-order center lying outside the survey zone. The buildings that possibly represent these emergent rural elites are the “megalithic farmsteads” at Avgo (87; Hayden 1997b), Chondrovolakes (69; Haggis 1996a), Chrysokamino (34; Betancourt and Floyd 2000–2001; Betancourt, Muhly, and Floyd 1998), Katsoprinos (28), as well as the “villa cluster” or town houses at Hagios Antonios (5; Warren 1984, 40–41; Soles 1991, 75–76). Furthermore, the Cyclopean buildings at Chondrovolakes (68) and Ridopoulia (59) continue in use into LM I. At the former site, a cushion seal was found. While these buildings are not “villas,” in the sense that they do not exhibit palatial architectural features or formal characteristics, they are much larger, and for the most part better constructed, than their MM I–II predecessors or counterparts. Their prominence in the Neopalatial settlement pattern may suggest that they should be classified as “country houses,” with specific agricultural functions or special economic importance (Hayden 2004a, 118–119; Floyd 2000; Floyd et al. 1995; Haggis and Mook 1993, 287– 288; Nixon 1987, 95–98; Wiener 1987, 266; 1984, 18; Watrous 1982, 14; Cadogan 1971). A contemporary parallel might be the large farmhouse at Chalinomouri in the Mochlos plain, which was presumably subordinate economically to Mochlos town in LM IB. According to Soles’s persuasive argument, this large house, overlooking the eastern end of the plain, was involved in a specialized industry of flax cultivation and weaving. A nearby source of serpentine might also have been quarried and worked at this important rural site (Soles 2003; 1997). Could flax, vines, olives, fruits, nuts, or wheat, have been special crops raised in discrete areas of the Kavousi plain and mountains, either supplying the populations of Hagios Antonios or Tholos, or providing a surplus of certain crops for export from the ports at Tholos and Agriomandra? The emergence of large class 3 settlements in the Kambos plain, and of town-sized class 4 sites at Tholos and Hagios Antonios (and possibly Kavousi), is evidence of economic changes at the beginning of the Neopalatial period. The concentration of

78

KAVOUSI I

population at the northern end of the Kambos and the growth of Tholos illustrate a new orientation to Mirabello Bay and a renewed interest in a maritime economy. The proximity of Tholos and Hagios Antonios permits the reconstruction of a town-port relationship, probably not unlike Gournia and its shore buildings and ship sheds at Elatzomouri, or on a much larger scale, Malia and its harbor near Quartier Theta. The transformation of Hagios Antonios into a nucleated settlement and the growth of Tholos are directly related to new patterns of maritime and interregional trade in the Neopalatial period. Organizational changes in the Neopalatial period are evident in the broader region of eastern Crete, where palatial features proliferate across the landscape. New palaces are constructed or rebuilt at Zakros, Petras, and Gournia—all coastal locations benefiting from, if not specializing in off-island exchange and coastal communication (Platon 2002; Soles 2002; Tsipopoulou 2002; Watrous et al. 2000). The harbor of Petras was apparently an elaborate affair located at the mouth of a river that extends inland several kilometers, connecting the palace with numerous “villas” and rural buildings such as Zou and Klimataria (Tsipopoulou 2002; 1997; Tsipopoulou and Vagnetti 1995; Tsipopoulou and Papacostopoulou 1997). Other coastal sites in East Crete exhibiting palatial architectural features are Palaikastro, Makrygialos Plakakia, and Myrtos Pyrgos (Davaras 1997; Cadogan 1977–1978). Major ports and transshipment points developed at Makrygialos, Diaskari, Atherinolakos, and Livari (Fig. 1). The wealthy towns of Mochlos and Pseira, with evidence of ashlar and gypsum embellishments, were either independent administrative and economic centers or second-order sites subordinate to the palace at Gournia (Cunningham and Driessen 2004, 108; McEnroe 2001; Soles 2002; 1991; Soles and Davaras 1996). Either way, Mochlos was clearly an important port and gateway community in Prepalatial times, and it is unlikely that this role had changed in LM I (Branigan 1991). The size of the settlement and its natural harbor are clear indications of a regional maritime center. Neighboring Pseira also reached its maximum size in LM I, thriving no doubt from coastal trade in Mirabello, East Crete, as well as the Aegean and eastern Mediterranean (Betancourt and Banou 1991). Thus the transformation of Tholos and Hagios Antonios into

class 4 sites fits well into a broader pattern of coastal nucleation in eastern Crete in LM I. It is also possible to reconstruct a three-tiered hierarchy of settlement in Neopalatial Kavousi, with a broad interdependence between class 1–3 inland sites and class 4 coastal towns. But what was the relationship between Hagios Antonios-Tholos and neighboring coastal centers outside the survey zone such as Gournia, Pseira, and Mochlos? Soles has argued that the palace at Gournia was the center of a regional Mirabello polity. In such a scenario, Mochlos with its ashlar building B-2 was a secondorder center (Soles 2004; 2002; 1991; Driessen 2001, 64; Watrous and Blitzer 1999; Watrous et al. 2000; Soles and Davaras 1996). If such a polity is a viable model for the Mirabello region, what was the role of sites such as Pseira, Priniatikos Pyrgos, and Hagios Antonios-Tholos? And why do Mochlos and Pseira exhibit subtle differences in their integration of palatial architectural features that surely indicate some differences in status or function? The various levels of the organizational structure in the region remain hazy. Our chief problem is that social, political, and economic configurations and spheres of influence are not mutually determinable constructs; nor do they manifest themselves in identical patterns in the archaeological record (Cunningham and Driessen 2004, 109). One of the most striking changes in the archaeological landscape of Kavousi in MM III–LM I is the disappearance of Mirabello fabrics—the granodiorite-tempered coarse pottery manufactured in the area of the Gournia Valley and farther west to Kalo Chorio. In the late Prepalatial and Protopalatial periods, these Mirabello imports were a dominant component of the Kavousi-area pottery assemblage. The presence of Mirabello fabrics at Kavousi-area sites in EM III–MM II suggests dynamic production and long-distance distribution of coarse utilitarian pottery, and active exchange of the vessels themselves (such as pithoi or cooking tripods and trays) or in what they might have carried (in the case of ovalmouthed amphorae, hole-mouth jars). The social insularity suggested by the Protopalatial cluster pattern at Kavousi is perhaps balanced by local exchange and dynamic interregional trade. What products did the farmhouse clusters exchange for the variety of ceramic vessels that they received from across the Isthmus? This MM I–II picture is the opposite of the Neopalatial pattern. In the latter,

THE HISTORY OF SETTLEMENT

the insularity of the clusters gives way to nucleation in the coastal zone. A nucleated town society emerged that looked outward to Mirabello Bay and the Aegean. Locally produced phyllite-quartzite-tempered pottery, long competing with the Mirabello fabrics, became the exclusive coarse-ware fabric of LM I Kavousi, displaying both greater standardization of shapes and a surprising uniformity in fabrics (Barnard 2003, 8). Is this new predominance of local phyllite-tempered fabrics a sign of economic independence or increased decentralization in a highly connected palatial economy? Both scenarios are possible (Cunningham and Driessen 2004, 108). Hagios Antonios, with its port at Tholos, may have begun to specialize in its own ceramic and agricultural production—like the neighboring community of Mochlos—either directly fulfilling demands of a palace center at Gournia, or indirectly responding to a new economic environment in the broader region, created by a palatial sociopolitical system. The study of the relationship between ceramic spheres of distribution and political and economic systems needs both more evidence and a coherent methodology. In Kavousi, the MM III–LM IA settlement pattern is one of nucleation. The settlements are no longer integrated clusters of farmhouses in arable zones; they now appear decidedly separate, placed strategically to take advantage of trade routes, ports, and perhaps the Platys River for irrigation (Fig. 21). The entirety of the central Kambos was, in LM I, given over to extensive plow agriculture. The northern coastal orientation of the largest sites is a sign that significant parts of the local population were no longer engaged in self-sufficient farming. The fertile pockets of alluvium that had accommodated the

79

Prepalatial and Protopalatial site clusters, in the Neopalatial period have only one or two large houses (Figs. 12, 21). Such buildings clearly represent second- or third-tier sites in a hierarchy, and their function no doubt involved the exploitation of highly productive soils for perhaps single crops. Finally, the linking of sites along the north-south axis through the Kavousi plain and into the north Isthmus demonstrates connections between settlements, emphasizing their role as economic units. The regionally varied relationship between human settlement and the environment, local patterns of land use, and local-level social and economic interaction appear, in LM I, to have been superceded by a regional hierarchical structure. Sites are separated from specific environments that would have permitted people to use land for small-scale farming, private gardens, and household stock rearing and herding. Concentrated arable land was surely exploited in LM I, but like the 19th century, the land was probably used for wheat, high-yield olives, fruit trees, and vines, less for barley, small mixed gardens or pulses. The results of analyses on floral and faunal remains recovered from the Chrysokamino Habitation Site at Mount Chomatas (site 34) should provide fascinating results, either substantiating or refuting the picture presented here (Floyd 2000, 68). Coupled with changes in land use come clear signs of a shifting emphasis from the hinterland to the coastal zone. The Kavousi region had become less of an insular consumer, while turning to the expanding scope of extraregional influences. The impetus for this change might have been the establishment of the LM I palaces.

Late Minoan II–Late Minoan IIIB Evidence for LM III activity was recovered from only eight habitation sites and three cemeteries (Fig. 13). The sites are also smaller in size than their LM I predecessors, and the only conclusion to be drawn from these data is that the Kavousi area experienced an extreme decrease in population during the years following LM IB. The changes in site functions before and after the Thera eruption are not easily discernable on the regional level; that is, it is very

difficult to distinguish sites that continue into LM IB, from sites that are abandoned or undergo formal changes (Knappett and Schoep 2000, 369; Watrous et al. 2000, 476; Driessen and Macdonald 1997). Recent excavation at Mochlos has however demonstrated conclusively that the LM IA Theran eruption was followed by a prosperous period of settlement expansion in LM IB (Brogan, Smith, and Soles 2003). Evidence for this post-eruption LM IB

80

KAVOUSI I

development was unfortunately lost in the coarse resolution of the survey’s ceramic study. The subsequent LM II and early IIIA phases were also not easily identifiable in the survey sample. Notwithstanding the difficulty in identifying LM II–IIIA pottery in surface survey in East Crete (Hayden 2004a, 133; Watrous et al. 1993, 227; Hayden, Moody, and Rackham 1992, 327; Hayden and Moody 1990, 51), there appears a pronounced drop in site numbers, with the best evidence for reoccupation in the Kavousi village deposit (24) and at the town site (5) of Hagios Antonios. The absence of identifiable LM II pottery should not indicate complete abandonment in this chronological phase. LM II as a ceramic phase has been defined at Knossos and Kommos (Popham 1967, 337–345; Watrous 1992, 119–125; Betancourt 1985, 149–158). In East Crete a LM II assemblage per se may be difficult to identify using north-central Cretan criteria. Even at Palaikastro—a site where Neopalatial ceramic features tend to emulate or follow Knossian styles much more than elsewhere in East Crete—LM II is assignable only to individual sherds (MacGillivray 1997). LM II features are also recognizable at Mochlos, allowing the excavators to define an advanced LM II phase at the site, while recognizing that in this area of the island it remains difficult to distinguish stratigraphically LM II from LM IB and LM IIIA:1 (Soles 2004; Brogan, Smith, and Soles 2003). In the survey at Kavousi, sherds dated unequivocally to LM II were not recognized at all. The assumption is that in this area of the island, during the chronological LM II phase, LM IB styles of pottery were still being used (MacGillivray 1997), and while recently a local LM II phase has been documented at Mochlos (Brogan, Smith, and Soles 2003), the features of this assemblage were not sufficiently understood to have been recognizable in the survey assemblages. The implications of the Mochlos study, however, are that LM II is likely to be the initial phase of reoccupation of the region, continuing into LM IIIA:1. LM IIIA–B pottery was recovered, but identification was based almost exclusively on fine-ware shapes and decorated sherds. LM III tripod cooking pots, however, are distinctive in shape and fabric, and certain coarse-ware lekanis shapes are also diagnostic, but not common. Several sites (24, 26, 34, 80, 87) with LM IIIA or LM IIIB phases were actually excavated—either archaeologically or

accidentally—thus providing valuable information on this period of occupation in the Kavousi area (Fig. 13). Accurate site sizes were on the whole difficult to determine. Sites 34 (Chrysokamino Habitation Site) and 87 are definitely individual farmhouses, as they reuse LM I foundations (Floyd 2000) (Fig. 21). While the actual disposition of the Kavousi village site 24 remains unknown, a substantial variety of LM III shapes was recovered from the dumped excavation debris; furthermore, LM III graves have been excavated on the outskirts of the village at Kamara tou Tholou (26) and Ridopoulia (60). Thus, it seems possible that reoccupation of the Kavousi village area could have been extensive. The best evidence for immediate reoccupation comes from Hagios Antonios (5) (Figs 21). Although the port of Tholos is probably abandoned after LM IB, Hagios Antonios continued in use, but on a reduced scale. Concentrations of LM IIIA–B pottery were discovered in only two discrete areas of the site—in the far southwestern corner (locus 10) and at the northeastern edge (locus 15)—suggesting a pattern of reoccupation similar to that recovered from excavations at Gournia and Mochlos (Hayden 2004a, 127; Watrous 2001, 89). In the latter cases abandoned houses were reinhabited or significantly rebuilt to accommodate a much smaller population. Despite the difficulties in identifying LM IIIA–B material in survey samples in East Crete, the apparent LM II–III contraction and partial abandonment in the Kavousi area, even if exaggerated, does seem to follow a pattern suggested by results of excavations. Although traces of LM III pottery were found at village-sized sites previously inhabited in Neopalatial times (e.g., 17, 57), the most extensive reoccupation clearly occurred in the areas of the old Protopalatial site clusters: the town at Hagios Antonios (5); two farmhouse sites at Chordakia dated to LM IIIA (34) and LM IIIA:2–B (36); in the Avgo Valley, the LM I house at Panagia (87) continued in use into LM III with a cemetery (88) located on the slope to the south. In the north Mount Papoura catchment area there is little evidence of LM IIIA–B activity with the exception of the sites already mentioned around Kavousi village (24, 26, 60), and at the northern edge of the drainage at Sta Lenika (57). In general, the preferred site locations are near springs and phyllite soils, suggesting a pattern of agricultural dependence similar to that of the late

THE HISTORY OF SETTLEMENT

Prepalatial and Protopalatial periods. LM IIIA–B sites are generally small—single houses or hamlets (34, 36, 87). Even at Hagios Antonios, where parts of the LM I town were resettled, the LM III remains are localized and discontinuous scatters suggesting a fairly small-scale reuse of the site, probably no more than two or three houses. Sites are chosen away from the coast in sheltered valleys looking inward to the Kambos plain. A significant concern of the inhabitants in Avgo (87), Kavousi village (24), and Hagios Antonios (5) was the perennial water supply from the springs at these sites. The LM IIIA–B distribution of sites is clearly a retracted form of the Neopalatial settlement pattern. In the neighboring Vrokastro area, the pattern in LM IIIA–B consists of a significant settlement reduction, similar to that at Kavousi (Hayden 2004a, 130– 133). Sites are generally small in size, and continuity in the use of LM I sites suggests settlement contraction, a characteristic of excavated sites such as Mochlos and Gournia. In the Gournia area, the drop in site numbers is even more pronounced than in Vrokastro (Watrous 2001, 89; Watrous et al. 2000, 477), suggesting that the entire Mirabello region not only suffered a major population loss following the LM IB destructions, but was also slow to develop a complex and integrated settlement structure in the Mycenaean period. Bennet (1990; 1987; 1985; cf. Driessen 2001, 64) has long observed the regional differences in LM III settlement structure, which can be seen in the survey data from across the island. The Mesara shows a growth in population with some reduction in settlement size (Watrous et al. 1993, 228; Borgna 2003, 159). Moody (2004, 255) records an overall reduction in site numbers on the Akrotiri, while the Hagios Vasilios Valley experiences a significant increase in settlement.

81

In Kavousi some farmhouses were reoccupied, while others were not. Towns and villages were reinhabited, but only on a reduced scale. In every case, the areas were chosen for their constant water supplies and arable land suitable for rain-fed agriculture. Agricultural dependence is reinforced by the lack of a site hierarchy that might help explain the nature of site interaction in the region. The evidence points to three or four very small and separate communities, perhaps not unlike the original Prepalatial site clusters, coexisting in largely self-sufficient but not isolated groups. Change comes again to the Kavousi region at the end of LM III with a remarkable movement from the lowlands of the Kambos plain to the precarious mountain sites of the Kastro (80) and Katalimata (100). This shift in settlement patterns culminates in the Early Iron Age with the foundation of several large and prosperous villagesized sites scattered throughout the Kavousi mountains. The LM IIIB–C transition involves, in many areas of Crete, an abrupt and radical reorganization of settlement (Hayden 2004a, 153–155; 1988, 2, 21; Moody 2004, 256; 1987, 309–315; Borgna 2003, 164–165; Watrous 2001; Watrous et al. 1993, 228– 229; Kanta 1980, 326; Rutter 1992, 69; Haggis and Nowicki 1993, 334). In eastern Crete the dominant settlement type was the defensible or refuge settlement (Nowicki 2001; 2000a; 1999a; Whitley 1998; Haggis 2001; 1993a), while Borgna (2003) and Wallace (2001, 85) have recently presented more nuanced views of the LM IIIC countryside, emphasizing a variety of settlement forms and economic strategies. The period from LM IIIC through PG at Kavousi is one of rapid growth and settlement expansion on a scale not seen in the area since the Protopalatial period (Figs. 14, 22).

Late Minoan IIIC–Archaic By LM IIIC, the plain of Kambos was completely abandoned (Fig. 14). The Early Iron Age settlement pattern is comprised of large habitation sites —essentially clusters of large villages—in two locations: the Avgo Valley, and Xerambela on the heights immediately south and above Kavousi village. In Avgo four sites are established on the heights of Trapeza and Melisses, overlooking the valley

floor; between them is a spring. The sites, 85 and 89–91, vary in size and state of preservation. Site 85 at Melisses is the best preserved (Fig. 14). It consists of a sherd scatter of about 0.56 ha. on a defensible bedrock ridge on the south side of the Avgo Valley. The spring, located just northeast of the site across the revma, is easily accessible. The location of the site was chosen for defense, proximity to the water

82

KAVOUSI I

supply, and the accessibility of the valley floor and terraces on the slopes above the site. Unfortunately, the slopes above and on the Trapeza Ridge are extremely eroded. Accurate site sizes were impossible to determine because of dense vegetation—a problem encountered throughout the Avgo Valley—and deep deposits of phyllite colluvium. Sheet erosion from intensive browsing on the hill slopes below Bebonas has caused a downward flow of the soil veneer, obscuring sites on the northeast slope of the Avgo Valley. The steep terrain rising to the east above the Trapeza Ridge all the way to the Bebonas watershed (92) is a mixture of phyllite and terra rossa soils, gravel, and cobble scree that has been stabilized by terrace walls for barley and vines. Shepherds, however, have encroached on the slopes above Avgo and Bebonas with the shift in settlement from the mountains to the plain since World War II. The resultant overbrowsing of these slopes has caused both sheet erosion and dense garigue ground cover, both impeding visibility on the Trapeza Ridge itself. Additionally, the abandoned hamlet of Trapeza entirely obscures the southernmost quarter of the ridge. The sites 89– 91—based on sherd densities—appear to be small class 1–2 hamlets, but it is possible that the sherd concentrations actually represent one or two class 3 sites, the erosion and vegetation creating gaps in the swath of artifacts extending from the Protopalatial site 88 in the south to the LM IIIC scatter at site 91 in the north. Like neighboring Melisses, the Trapeza Ridge is defensible, situated above the Avgo spring, and is positioned to make optimal use of the fertile valley below as well as the phyllite terraces above the sites to the east. LM III burials have been excavated both on the ridge itself and in the valley below, at the base of Panagia (86). These sites may mark the locations of LM IIIC cemeteries. A single Orientalizing pithos was revealed by a bulldozer on an agricultural terrace southwest of Melisses (89). While this vessel may have been a burial vase, no human bone was recovered. Better preserved than the sites in the Avgo area are those in Xerambela and the northern Mount Papoura drainage (Fig. 22). Vronda (77), Kastro (80), Azoria (71), and Panagia Skali (70) form a site cluster in essentially the same region occupied by the Protopalatial settlements of Xerambela and Chondrovolakes (Figs. 19, 22). As in the Avgo Valley, the sites are located in defensible positions and

each is roughly equidistant from the main spring, east of the Vronda settlement. The sizes of the settlements vary, and no doubt changed considerably from phase to phase. Vronda in LM IIIC was about 0.60 ha., Kastro in PG surely exceeded 0.80 ha., while Azoria, in LG–Archaic, may have exceeded the class 3 size ranking, becoming the main settlement in the Kavousi region in the Archaic period. Panagia Skali (70), just below the Kastro and neighbor to Azoria, is the smallest site in the group, probably not exceeding 0.25 ha. (Fig. 22). The cemeteries appear to have originally belonged to specific sites, even if the chronology of their use is not exactly parallel. Chamber tombs were excavated by Boyd somewhere at Chondrovolakes in the valley west of Panagia Skali and Azoria; in the region of the Kastro there are tholos tombs at Skouriasmenos (81), Plaï tou Kastrou (79), and Skala (78); Vronda’s cemetery, consisting of several tholos tombs, lies to the north of the LM IIIC habitation site. While the date of the construction of the Vronda tombs is uncertain, they are evidently used after the abandonment of the LM IIIC settlement; some tombs were used in PG–LG. Furthermore, the Vronda settlement itself eventually became a cemetery, the abandoned houses accommodating cist tombs and scattered inhumations and cremations. The sites in Avgo and on the north slope of Mount Papoura form two discrete site clusters, separated neatly from each other by the vertical cliff of the Makellos torrent that runs down the east side of the Kastro. Each group of sites is located within a 0.5 km radius of a perennial spring, and the sites were built on bedrock outcrops, probably to maximize the benefits of the defensible terrain while avoiding direct occupation of good arable land. The surrounding phyllite and mixed terra rossa–phyllite soils are excellent for rain-fed agriculture, but irrigation of small gardens would have been possible from the springs controlled by each cluster. Extensive terracing—probably not dissimilar to the traditional and modern practice—in such terrain would have been necessary to preserve the shallow and fragile veneer of soil. A terrace wall (83), found in the Avgo Valley, is similar in construction technique to Bronze Age walls excavated at Hagios Antonios and at Pseira, and could be evidence of Early Iron Age agroterracing practice. The sites in Avgo are smaller than those in the Xerambela area, but their inhabitable and cultivable areas are also smaller and the

THE HISTORY OF SETTLEMENT

terrain is less conducive to settlement expansion. Communication between the clusters is most direct from Azoria by following the Avgo drainage due east. At a point roughly equidistant between the clusters is the site of Pachlitzani Agriada or Makellos (82), which was a rural house temple (Pl. 4A; Fig. 22). Its use from PG until the Archaic or Classical period was confirmed by Alexiou’s excavations in 1950 (Alexiou 1956). Such a shrine may have served as a local marker of agricultural or territorial boundaries between the clusters, as well as a common meeting place along the only passable route between Avgo and the Kambos, the north Isthmus, and the Bay of Mirabello. The sites in each of the clusters are interdependent hamlets or small villages, sharing the water and land resources in topographically isolated areas (Fig. 14). Pasturage for the Avgo cluster was no doubt the slopes of Kliros, Orno, and the Bebonas watershed. The Xerambela cluster’s territory probably included the highland plain of Mount Papoura, located directly southeast of the sites. The existence of separate cemeteries in close proximity to each site, and separate tholos and built chamber tombs with multiple burials, suggest lineage-based burial practices. Individual families and extended family groups would have probably used individual tombs and tomb groups. Borgna (2003, 165–166, 171) has recently argued that the evidence from upland settlements in eastern Crete in LM IIIC suggests that emergent chiefdoms legitimized their authority by reproducing elite symbols—such as tholoi, bench shrines, special function buildings or “big houses” centralizing storage for public feasting—of an earlier lowland Mycenaean tradition (cf. L. Day and Snyder 2004; Eliopoulos 2004; Klein 2004). The correlative evidence from excavated houses at Vronda (LM IIIC) and the northwest building on the Kastro (LM IIIC–Orientalizing) illustrates clearly the growth of individual households into increasingly large agglomerative groups of buildings (Glowacki 2004; Mook 1998). These had become veritable neighborhoods, separated from each other by topography, streets, and courtyards. The specific form of social structure in these nascent LM IIIC communities is perhaps difficult to extract from architecture and other archaeological evidence—differentiation of house types and household assemblages may be subtle or invisible in the archaeological record (Day and Snyder 2004;

83

Nowicki 1999a; Whitley 1991a; Hayden 1981, 180; Jameson 1990, 108; L. Day and Glowacki 1993; Haggis 1993a, 151; Mook 1995; 1998; Gesell, Day, and Coulson 1995, 116). Looking beyond the settlement, however, allows us to see patterns in the overall dispersal and clustering of settlements. Regional social and economic integration is indicated by adjacent and overlapping spheres of influence and agropastoral landscapes (cf. Borgna 2004, 168; Wallace 2001, 85); the exploitation of potentially diverse environments suggests complex and longlasting relationships between sites and regions, including crucial linkages between highland pastures and lowland centers (Borgna 2003; Wallace 2003, 624; Cherry 1988). Such relationships, while potentially competitive and even destabilizing (Borgna 2003; Watrous 2001, 92; Wallace 2001, 85; Whitley 1991a; 1991b) would have required and probably even stimulated social interaction and eventually political integration (cf. Wallace 2001). Within each site in Early Iron Age Kavousi, family groups occupied distinct areas of the settlement and cemeteries, defining themselves in the community by means of separate tombs and household clusters. The formation of the clusters is an interesting material pattern (Haggis 2001; 1993a) that has parallels elsewhere in Crete and the Aegean (Borgna 2003, 171; Farenga 1998, 190–182; Foxhall 1995, 244–245; Donlan and Thomas 1993, 65–66; Morgan 1993, 21; Snodgrass 1990, 130; Coulson 1986, 71–78; McDonald and Rapp 1972, 142–144). But what was the relationship between sites in a cluster? Interdependence and economic and social cooperation between sites is assumed because of topographic isolation and shared water supplies, agricultural land, and pastures. Such cooperation was probably originally along kinship lines. Population growth, however, was restricted by the environmental and topographical exigencies of the region. The same conditions affected settlement development. Extrapolating from the excavated evidence of physical growth of household units on Vronda and Kastro, it may be possible to reconstruct the broad pattern of site expansion and interaction in the Early Iron Age (Mook 1998; Coulson et al. 1997; Gesell, Day, and Coulson 1995). Settlement in the region may have begun on the Kastro: founded at the end of LM IIIB or beginning of LM IIIC, and then expanded to Vronda and Azoria in LM IIIC. Vronda

84

KAVOUSI I

flourished in LM IIIC, only to be abandoned thereafter as a habitation site. Kastro on the other hand expanded in size throughout the PG and Geometric periods, but experienced gradual and systematic abandonment from LG onward. This variable pattern—regional growth by fits and starts and regional shifts in settlement—is characteristic of the early part of the Early Iron Age in Crete (Borgna 2003, 168). Azoria parallels the use of the latter sites, but grew in size and was occupied throughout the Orientalizing and Archaic periods—until sometime in the early fifth century (Haggis et al. 2004). The social links between sites would have been maintained on the level of related households, no doubt sharing or exchanging rights to ancestral land. Various scales of local identity (the family, the household, the clan, the village, and ultimately the cluster) would have been mutually reinforcing principles of social interaction. Eventually cooperative interaction and even familial bonds could have formed between the Kavousi and Avgo clusters as a matter of economic and biological necessity. The Early Iron Age clusters of Kavousi and Avgo inhabited and exploited small-scale regions—areas of land that formed not the full spatial extent of their economic catchment areas but the conceptual center of their community consciousness. The community of place is the essential notion around which the society was centered. If the site cluster pattern is a manifestation of the Early Iron Age organizational structure, then the cluster itself, not individual sites, represents the highest-order autonomous political unit in the region, as well as the central focus of the community. The organization must have been expressed on a regional level through familial and perhaps religious ties that make up the fabric of the cluster. This is not to say that a complex social or political structure did not exist, or that power hierarchies did not contribute to the organization of activities of people in a cluster. The household itself, for example, may have been an important social unit within the cluster. Large, centrally located and potentially special-function houses such as Building A-B at Vronda, House C at Katalimata, or Building M on the Kastro, with its elaborate stereobate retaining wall, could be evidence of social and economic stratification, as well as political organization (Day and Snyder 2004; Coulson et al. 1997; Haggis 1993a). Their storage facilities and adjacent courtyards could be evidence of special economic and

social-religious functions. Differentiation in tomb architecture and cemetery location could also represent ranking among families. These subtle signs of differentiation are ambiguous if important indications of complexity on the scale of the single site. But even so, the lack of any clear regional hierarchy of settlement (such as ranked size, differentiated architecture, or preferential location) emphasizes the potential sociopolitical implications of the cluster model. Settlement development and social structure cannot be examined critically on the single-site level without assessment of regional dynamics, as the proximity and shared resources of the site indicate that none would have functioned exclusive of the others. The sites form a highly integrated unit, inextricable from their surrounding topography and physical environment (cf. Borgna 2003). The reuse of the area and debris of the abandoned LM IIIC houses of the Vronda settlement as a cemetery in the Geometric period might emphasize the closeness of the relationships among sites, as well as the continuity of use of the cluster throughout the Early Iron Age and the connection of the settlements with the physical landscape. The cluster pattern itself demonstrates site interdependence, as well as lineage ties that link various areas and contribute to the territorial unity. The Early Iron Age settlement clusters at Kavousi and Avgo appear to be new foundations in LM IIIC, coinciding with the abandonment of the Kambos for the agriculturally productive and defensible terrain of the Kavousi Mountains (Fig. 14). Even a very conservative estimate of about 600–1000 people for the entire Kavousi area is perhaps a significant increase from the preceding LM IIIA–B period. The population is a minimum for the Early Iron Age (cf. Nowicki 1999a), based on an average of fifteen households per settlement (ca. five persons/household), and a total of eight settlements in the region. Since all of the settlements in the area were not used as habitation sites continuously and contemporaneously throughout the period, the number is meant to reflect the minimum number of people per household, and households per site in the period LM IIIC– PG. Evidence from excavations demonstrates clearly a gradual decrease in population on the Kastro from LG though EO and a concomitant increase at neighboring Azoria, which reaches a maximum size 700 to 600 B.C. (Fig. 15). 7th- and 6th-century remains are sporadically represented in Avgo, where

THE HISTORY OF SETTLEMENT

chance finds at sites 84 and 89 show Orientalizing and Archaic activity. A veritable synoicism or centralization encompassing both Kavousi and Avgo clusters may have ultimately resulted in the transformation of Azoria into a class 4 nucleated site—the central town or city in the region by the end of the 7th century (Fig. 15). The visible architectural remains on the hilltop at Azoria cover an area larger than 2.5 ha., but the dense vegetation on the site’s surface precludes accurate assessment of site size in various periods. The site’s total area exceeds 14 ha. The date of surface pottery collected during the survey (LM IIIC to Archaic) agrees generally with Boyd’s description of the excavated material, and this date has been confirmed in recent excavations (Haggis et al. 2004). Azoria is strategically located along the river route connecting Kavousi and Avgo clusters. While it occupies a defensible position, the site is accessible from both the plain and hinterland, requiring the massive retaining/fortification wall that surrounds the buildings on the uppermost terraces of the hilltop (Fig. 15). Azoria’s position is not only central in the overlapping spheres of the Early Iron Age site clusters; it sits squarely at the confluence of two major rivers and at the juncture of the two main passages from the Isthmus of Ierapetra into the Siteia Massif. The river from the watershed at Bebonas descends through the Avgo Valley, which is the single major route connecting the Mirabello Bay and eastern Crete. The second route follows the torrent that drains north from Hagios Niketas at the

85

watershed at Thriphti (Aori) (Fig. 2). From this watershed, another major route leads eastward to Orino and the southern side of the Siteia Mountains. Thus, Azoria’s accessibility from the plain—and coastal and Isthmus transportation and communication routes—made it a kind of gateway to the Kavousi Mountains and Siteia Massif, which was densely settled throughout the Early Iron Age (Nowicki 2000a; 1996; 1994; 1990; 1987c). It is not surprising that the rural PG-A shrine at Makellos (82), the central cult place in the region, was positioned at the foot of the Azoria hill. Azoria was a nucleated town, perhaps an asty (city center) of a nascent polis, which by the 7th century B.C. had become the physical center of the combined territories of the north Mount Papoura catchment area and the Avgo Valley (Fig. 22). Recent excavations at Azoria have confirmed the urban character of the site, identifying an agora, andreion and monumental civic building (Haggis et al. 2004). The linking of these two regions, and the emergence of a larger community, could well have been a process taking place during the LG and EO periods and finally manifesting in the nucleation of population with the abandonment of habitation sites on the Kastro, Melisses, and Trapeza. Azoria is both the spatial center of the region as well as a physical link between the coast and the hinterland, and between mountains and plain. The rural pattern by the 6th century consists of a weak two-level hierarchy— farmsteads surrounding the urban center.

Classical and Roman Periods The gradual depopulation of the Kastro after LG, and ultimately, the abandonment of several settlements in the Kavousi and Avgo clusters, was caused in part by the emergence of Azoria (71) as a nucleated center in the 7th through early 5th centuries. Both Azoria and the adjacent shrine at Makellos (82) continue into the Archaic period. This thinning out of rural settlements by the 6th century is perhaps parallel to the gradual depopulation of Lasithi, a process which Watrous has linked to the formation of a polis at Lyttos (Watrous 1982, 20–24; Hayden 2004a, 188). The location of the site of Azoria, at the

geophysical transition between Kavousi village and the Avgo Valley, and between Kambos and Mount Papoura, permitted the site to control communication between the coast and the hinterland. From Azoria, there would have been easy access to lowland fields as well as highland pastures. The Kastro was used for a time in the Orientalizing period, probably as a seasonal site, and Vronda’s cemetery continued to be used into the 7th century. By 600, however, the old Early Iron Age clusters were abandoned. The trend toward nucleation of population could have been a result of growing regional unity

86

KAVOUSI I

and increasing extraregional economic interests. The 7th and early 6th century inhabitants of Azoria were linked to a world that was decidedly wider than the mountain clusters with their local agricultural and pastoral interests. Even Azoria itself was abandoned late in the Archaic period. Whatever the cause, people seemed to have moved either south across the Isthmus to the city of Hierapytna, or west along the coast to Minoa, Oleros, Istron, or even Lato. The only other Classical site in the Isthmus region is Larisa, which is mentioned by Strabo (9, 5, 19, C440) to have been subsumed in a process of synoicism into the territory of Hierapytna (cf. Hayden 2004a, 226; 1997a, 96, 140–141; Guizzi 2001, 309–310; Watrous and Blitzer 1995, 313; Bowsky 1994, 6). If the LM IIIC–Classical site of Prophitis Elias, above the modern village of Kato Chorio at the center of the Isthmus, was ancient Larisa, as Watrous has suggested (Watrous et al. 2000, 477; Watrous and Blitzer 1995), then this would be a city both close by and sufficiently large to have absorbed the population and territory of Azoria. Population eventually shifted to urban centers outside of the Kavousi survey zone in the wake of new economic and political systems and a new hierarchy of settlement. The Sherratts have characterized the first half of the first millennium B.C. as a period of expanding state territories, where “merchant enterprise, rather than state controlled exchange, became the dominant mode of trading activity,” and where “tension between land-rooted and commercial interests became more explicit” and “manifested in a complimentary relationship between territorial empires and mercantile citystates on their boundaries—and especially on coasts—through which foreign trade was channeled” (Sherratt and Sherratt 1993, 362). Such a view of the economy of the Early Iron Age Mediterranean is a vivid backdrop for changes in settlement patterns in east Crete in the 7th and 6th centuries. The gradual dissolution of the Geometric settlement clusters (ca. 750–500 B.C.) outlined above is symptomatic of the development of new sociopolitical systems with the rise of the Cretan polis, as well as a significant tension between local clan-based agrarian interests and the town-oriented economies of new coastal communities. The development of such a coastal emphasis in the 7th and 6th centuries, accompanied by increasing off-island commercial interests, accords well with what we know of the

fully developed Hellenistic and Roman polities, whose wealth and power were based primarily on piracy, harborage fees, fishing rights, murex fishing and purple production, and the slave trade (Chaniotis 1988; Petropoulou 1985; Spyridakis 1970, 38). The epigraphic, literary, and archaeological records for east Crete are clearly dominated by references to such a maritime economy (Spyridakis 1970). Could prominent east-Cretan ports such as Lato pros Kamara, Hierapytna, Siteia, Stalis, Leuke, and Itanos, have attracted population into their urban centers as early as the Archaic period? The completeness of the abandonment of the Kavousi region is as striking as its duration, from the middle of the 5th century until sometime in the 3rd century B.C. This pattern stands in marked contrast to contemporary developments at both Vrokastro and the western Mesara. In the Mesara, there is constant growth, and a dispersed and stratified pattern from the 7th century until the Hellenistic period (Watrous et al. 1993, 230–231), while in the Vrokastro region the initial dispersed Archaic pattern gives way to increased nucleation (Hayden 2004a, 185; 1997). The Kavousi and north Isthmus areas remain sparsely populated (Watrous et al. 2000, 477–478). Two sherds of Hellenistic pottery were found in Lakkos Ambeliou, and Boyd reports a Hellenistic deposit from Hagios Antonios, but on the whole, there is no major activity to speak of until the beginning of the 1st century B.C. Recent excavations at Azoria have demonstrated the limited reuse of the southern peak of the site in the 3rd– 2nd centuries B.C., in the form of a tower or circular building which was brought to light during Boyd’s early excavations in 1900 (Boyd 1901). This installation could reflect the furthest northern limits of Hierapytna’s territory and its border conflicts with Lato to the west. While Hellenistic settlement in the north Isthmus of Ierapetra remains very thin throughout the Hellenistic period (Watrous et al. 2000, 477–478), Hierapytna’s influence on the north coast had finally manifested itself in the establishment of port facilities at Mochlos by the 2nd century B.C. (Vogeikoff 2004; 2000). By the 1st century, population seems to have returned to Kavousi village (24), Tholos (1), Hagios Antonios (5), and Chordakia (28, 40–42) (Fig. 16). Early Roman settlements are estimated to be small class 1–2 size sites, although frequently their actual

THE HISTORY OF SETTLEMENT

dimensions are obscured by Late Roman, or postRoman (Byzantine, Venetian, and Ottoman) remains. The areas chosen by Roman-period inhabitants are typically the phyllite soils of Chordakia and Hagios Antonios. These are probably farmhouses dependent on a nucleated settlement in the north Isthmus or perhaps at Kavousi village. By the 2nd century A.D. there are about nine sites total bordering the Kambos plain and exploiting the harbors of Tholos (1) and Agriomandra (39). Roman warehouses at Sta Lenika (57) and Tholos (1) indicate the dynamic commercial activity in the Kavousi and north Isthmus region (Fig. 16). Roman influence had clearly affected rural production and settlement patterns in the 1st c. A.D. (I. Sanders 1982, 30–31; Chaniotis 1988, 79–83; Harrison 1988, 150; cf. Alcock 1993, 80–85 and 93–127). In this context, it is important to recognize that the Kambos plain was not an inward-looking arable zone—merely the hinterland of Tholos, protected from the sea—but an important corridor linking Pseira Bay and the harbor at Tholos with the Isthmus route to Hierapytna (Fig. 1). In the Roman period, the Kavousi region was very likely the northern edge of Hierapytna’s territory (Spyridakis 1970, 37–39; Bosanquet 1939–1940, 69–70; Hayden, Moody, and Rackham 1992, 318, 331–336; Chaniotis 1995, 74–75; N. Papadakis 1986, 15; Bennet 1990, 200–202). As such, the port of Tholos was a gateway to the quickest land link between Cretan and Libyan seas, a veritable bridge between the Aegean and Mediterranean (Haggis 1996b; cf. Rickman 1980a, 226, 266; 1980b). At the southern edge of the Isthmus corridor was Hierapytna, which had become wealthy by securing harborage fees, fishing and purple fishing rights, and control of ports in eastern Crete. The warehouses at Tholos and Sta Lenika (cf. Rickman 1971) are probably less facilities for the storage and mobilization of local agricultural and pastoral produce for export, than temporary storage and transshipment points for Aegean and eastern Mediterranean goods falling within the sphere of Hierapytna’s influence or direct control (cf. Bowsky 1994, 7; Bennet 1990, 201; Fulford 1989, 179–188; Chaniotis 1988, 79–82; 1995, 74–75; Harrison 1988, 144–148). By the Late Roman period, the rural population in the Kavousi area reached its highest level since the Early Iron Age. This growth and ruralization of population is the opposite of the trend in the Mesara, which saw a substantial decrease in settlements

87

following a steady growth from the Classical and Hellenistic periods (Watrous et al. 1993, 231–234). In the Vrokastro area, there is continuity from Early to Late Roman, but a slight contraction of size (Hayden 2004a, 210). No fewer than five villages existed at Tholos (1), Kambos (17), Chordakia (37), Petras (53), and Sta Lenika (57) (Fig. 16). These class 3 sites are remarkably uniform in size, and they are spaced at surprisingly regular intervals of 1.5–2.0 km down the length of the entire Kambos plain. Such an equidistant distribution suggests local catchment areas of about a 1.0 km radius around each site (cf. Hayden 2004a, 210; I. Sanders 1982, 17). The pattern of ruralization of settlement, common throughout Late Roman Crete (Tsoungarakis 1990; 1988, 135), manifests itself at Kavousi as a series of nucleated villages or large hamlets of almost identical size, population, and area of adjacent agricultural land. The regular spacing of sites of similar size and surrounding land is an indicator of the potential carrying capacity of this marginal agricultural environment in this period. The Late Roman pattern could reflect the maximum population sustainable by this region. It is noteworthy, however, that between the Early Iron Age and Byzantine period, apparently little interest was paid to the fertile and well-watered highlands of Mount Papoura and the Avgo Valley. Roman-period settlements are entirely lacking in these upper elevations. While this is not evidence that such areas were not exploited, it would point to strong ties to the Kambos and to the Mirabello coast and Isthmus regions. These communities were probably self-sufficient in the exploitation of the fields immediately surrounding each village, but their location in the lowland plain emphasizes a continuing interest in the economic benefits derived from the Isthmus corridor. The site of Kephalolimnos (37) overlooks the rich phyllite soils of the Chordakia area. Olive oil-producing machinery—a trapetum mortarium and millstone (orbis)—indicates the intensive exploitation of these soils for cash crops. A nearby well provided a convenient water supply. Olive oil could be exchanged for products from neighboring villages, or more likely, from the larger-order sites in the Isthmus. The port of Tholos (1) continued to thrive well into the 6th century as did Chordakia-Kephalolimnos (37) and Agriomandra (39), testifying to the continuing importance of interregional, Aegean, and Mediterranean trade and communication. It may be possible

88

KAVOUSI I

to trace two converging routes through the plain (Fig. 16): one descending south from Tholos to site 17 on the east side of the Kambos, and then continuing straight south along the base of Mount Kapsas to Kavousi village and Sta Lenika (57); another route from Agriomandra (39) and Chordakia (37) would have turned south along the west side of the plain to Petras (53). Both routes converged in the

north Isthmus where countless similar villages dotted the passage to Roman Hierapytna. In the 4th to 6th centuries A.D., the Kavousi region had developed a village-farming settlement pattern, but the meaning of the pattern lies not in the self-sufficiency of these individual communities, but in their relationship to a broad regional and ultimately Aegean and Mediterranean economic sphere.

6

Gazetteer of Archaeological Sites

The first title of each entry contains, in boldface, the site number, followed by the site name, which is usually a local toponym. The site numbers correspond to those on the maps throughout Chapters 3 and 5 (Figs. 5–23). Below the site name are the original locus number(s) (Haggis 1992), field designation (Fig. 4), grid coordinates (British Admiralty Map, Sheet 22, Ierapetra), and the site’s size in hectares. Minimal topographical information and elevation, as well as the chronological phases suggested by the pottery are also included. For sites that consisted of single isolated installations—such as cisterns, rock-cut cists, and rock shelters—or single vessels, an area measurement in hectares is not given. In instances of sites with multiple loci, the description is broken down into locus subheadings in italics. The site description consists essentially of five parts: (1) location, physical environment, and terrain; (2) topography and architecture; (3) sampling methods and artifact summary; (4) bibliography; and (5) artifact catalog. References in the site

description to modern landmarks (such as field and terrace walls, recent buildings) and land use (method of irrigation, vegetation, and cultigens) are as they were recorded from 1989 to 1991. While some of these features of the recent physical and cultural landscape are of course ephemeral, they are left in the site catalog as part of the documentation and as a potential guide for individuals wishing to locate the sites. In the catalog for each site, the artifacts are numbered sequentially. Each artifact number (in boldface) consists of a site number prefix followed by a period and the number in the sequence. The original catalog number, indicated by the prefix “KTS” (Haggis 1992), follows immediately after the artifact number. Unless otherwise noted, all measurements are in meters. The verbal descriptions of the colors of ceramic slips, paints, and fabrics that accompany each Munsell number (and formal color name) are conventions that have been established for the Kavousi Project. Detailed descriptions of the coarseware fabric types are found in Appendices 2 and 3.

90

KAVOUSI I

1. Tholos Loci 16, 24, 38, and 39

On the west side of Tholos Bay, there are three main areas with ancient remains (Fig. 23; Pl. 9A): (1) a continuous scatter of Minoan pottery north of the modern church of Panagia (loci 16 and 38); (2) Roman buildings south of the church (locus 39); and farther south, (3) Minoan houses comprising locus 24. Minoan pottery extends continuously across the west side of Tholos Bay from the southern limits of locus 24, on the hill slope east of the Roman warehouse at locus 39, west of the church at Panagia (locus 38) and culminates at locus 16, on the point north of the church. Although increases and decreases in sherd density suggest concentrations of activity, or at least justify the separation of areas into different loci, the Bronze Age pottery is constant across the entire west side of the Tholos Bay, indicating a continuous settlement over 3 ha. in size.

old buildings” on the point north of the church of Panagia; it is likely that these walls are the Minoan buildings at locus 16. The pottery dates to the Bronze Age, but there are a few sherds of Late Roman and post-Roman (B–V) date as well. A single north to south line was collected from the northeast corner of the east building. A high proportion of coarse-ware types I (32%), VI (18%), and IX (5%) are a certain indication of LM I, MM, and EM date. Early Minoan is represented by fabric type IX; MM I–II is represented by types VI and III (12%); thin oval tripod feet and basin rims in type VI are indications of MM I–II date. Type IV is proportionally equal to type I (33%) and may be MM or LM, but is probably MM in this context. The fine ware collected is MM I–LM I. The scatter size and extant architecture suggest a settlement of a number of houses. Some 30 m east-southeast of the site is a natural roadstead, used today for fishing boats and in the Roman, Ottoman, and Modern periods, as a port facility.

LOCUS 16

LOCUS 38

Tholos, field B; 788172; 1.05 ha.

Tholos, field B, Panagia; 788168; 0.29 ha.

Some 75–100 m north of the church of Panagia on the west side of Tholos Bay, inland from shore ca. 30 m, are a dense scatter of pottery and the foundations of at least three separate buildings (Fig. 23). The scatter covers an area of 70 m east–west x 150 m north–south, beginning just north of the church and extending along the west side of the bay onto the small akrotiri that forms the western edge of the inlet. Low garigue vegetation is dense (Pl. 9A, B). Several wall foundations were distinguished, forming at least three separate structures, and covering an area of ca. 0.16 ha. Foundations of two large rectangular buildings are constructed of local dolomite blocks. The walls of the westernmost structure measure ca. 12 x 12 m, and the eastern building, less well preserved, is ca. 6 x 10 m (Pls. 9B, 10A). The blocks are square and rectangular dolomite, roughly dressed, between 0.90 and 1 m in length, and 0.50–0.60 m high and deep. Harriet Boyd (1901), in 1900, while investigating the Roman warehouse (locus 39), saw “walls of several

On the west side of Tholos Bay, behind (westsouthwest) the church of Panagia, and north-northeast of the Roman warehouse (locus 39) is a dense scatter of Bronze Age pottery, measuring ca. 45 m east–west x 65 m north–south. While the material appears to be a continuation of the settlement at locus 16, there is a discernible rise in density. The area is garigue covered, extensively browsed, and extremely eroded. The predominance of type I (63%) and thick oval tripod feet in this fabric is certain evidence of MM III–LM I activity; type VI (4%) and associated thin oval tripod feet are MM I–II. Two conical cup bases are LM IA. Fabric type III (4%) is in low quantities and, like type VI, represents MM activity. Type IV (30%) is probably MM. The church of Panagia, at the west edge of this locus, was built in the 19th century. Older walls are visible at the southern edge of the church terrace, and one wall projects in a line east from the northeast corner of the church ca. 5 m, with the

Tholos, fields B and C; 788170; 2.59–3 ha. Coastal plain and foothill; 12 m AMSL. EM I–II, MM I–II, MM III–LM I; ER; LR; B–V; Ott.–Mod.

GAZETTEER OF ARCHAEOLOGICAL SITES

same orientation as the north wall. Boyd excavated a Roman grave ca. 3 m from the northeast corner of the church, and Sanders—and after him Bolanakis and Tsoungarakis—suggested that the tomb and the wall remains underlying the church are of a Late Roman or Byzantine basilica. The grave provides confirmation of the northern limits of site 1 in the Late Roman period.

LOCUS 39 Tholos, field B; 788167; 1 ha.

Southwest of Tholos Bay and south of the church of Panagia is an area of Roman activity (Figs. 23, 24). The concentration of Roman pottery and tile is ca. 1 ha. in size, and consists of three distinct parts (Fig. 23): (1) the oblong building on the footslope of Mount Schinias southwest of the church of Panagia (“Roman Warehouse”; Figs. 23, 24); (2) a series of rectangular buildings extending ca. 100 m along the west bank of the Platys River alluvial fan; and (3) the remains of a spring house in a tide pool at the southwest corner of Tholos Bay. The narrow oblong building on the footslope of Mount Schinias (Figs. 23, 24; Pl. 10B) has a north to south orientation and measures ca. 55.70 m (north-south) and 9.60 m (east-west) . The walls are constructed of a thick concrete core consisting of a cement matrix with cobble-sized rubble or layers of fitted stones in an opus incertum fashion. The interior is lined with brick (0.04 m thick x 0.24 m long) separated by 0.05 m of cement. The exterior wall surface consists of large, regular, square dolomite blocks (ca. 0.30 m high x 0.30–0.50 m wide x 0.20 m deep) set in even courses (Pl. 10B). Sometimes the wall face has alternate courses of small, flat schist stones. The west wall, ca. 2.08 m thick, is built against an earthen terrace and upon the natural bedrock, which slopes from west to east. The east wall, some 2.15 m thick, is buttressed at no fewer than eight points along the wall face. Six buttresses are extant, and there are traces of two others (Pl. 10B). The buttresses are 1.56–1.73 m thick and spaced regularly at intervals of 3.90 m, except at the north end, where the distance from the building corner to the buttress is exactly 3 m. The south end of the building has been disturbed, and the entire southeast corner has fallen away. The northernmost buttress is the best preserved and projects from the wall face ca. 2.75 m (Fig. 24).

91

The building’s interior consists of remains of three screen walls (0.85 m wide) dividing the building into four cells measuring 5.20 m wide and ca. 12–13 m long; the length of each cell from north to south is: 11.80, 12.70, 12.70, and 13.30 m. The highest extant walls are in the north and measure between 3 m and 4 m. The floor of the building consists of a white plaster matrix containing sea pebbles and terracotta fragments. Within each cell are eight beam holes (0.25 m wide, 0.30 m high, and ca. 0.30 m deep) spaced at regular intervals (ca. 1.55 m) from north to south; the first in the north begins ca. 0.3 m from the north interior corners. In the northeast interior corner, evidence of vaulting suggests a concrete roof (as might the lack of roof tile from the immediate area of the building). The building has an eastern aspect; it is situated on a steep bedrock slope, which required buttresses and the greater thickness of the east wall. No doorway to the building is evident, but the ruined condition of the building’s south end does not preclude an entrance in this area. Associated with later use of the building are beam holes, ca. 1.63 m above the floor (ca. 0.13 m diameter), perhaps for some makeshift shelter within the building. Also of later, probably modern date are small bins (constructed of field stones and modern cement) in the northwest and southwest corners. Finally, facing the east interior wall of the second cell from the north is white plaster covering an area of ca. 5.40 x 2.10 m and containing a narrow (0.05 m) black-painted band. About 50 m east of the warehouse along the west side of the alluvial terrace (formed at the mouth of the Platys River) are more walls, also of Roman date, and appearing to form an elongated structure, ca. 100 m long and 10–15 m wide (Fig. 24: “Roman warehouses”). No certain plan of this structure could be determined; however, several north-south wall segments along the west side of the terrace and two cross walls suggest a single long structure divided at intervals of 20 m. Wall construction is of field stones and concrete. The associated artifacts are roof tiles, wall tiles, and amphora fragments. Datable fine wares include eastern Sigillata A and B, 6th-century combed ware, and Byzantine–Venetian glazed pottery. While a few Roman walls exist between the building on the river terrace and the warehouse, the pottery on the eroded slope east of the warehouse is

92

KAVOUSI I

primarily Minoan and probably associated with Bronze Age activity along this hillside between loci 24 in the south and 38 in the north. It is possible that Minoan buildings were destroyed by the construction of the Roman warehouse and the earthen terrace, which supports its west wall. Roman pottery is concentrated along the river terrace. About 50 m north-northeast of the Roman buildings at locus 39, and 122–124 degrees east-southeast of the church of Panagia, at the southwest corner of Tholos Bay, in a tide pool, there is evidence for a Roman spring house (Fig. 24). The tide pool is a roughly semicircular eroded cavity of bedrock with a 3.0–4.0 m radius, ringed with a bridge of Roman concrete (cement containing field stones and terracotta fragments). At the north edge of the pool, fresh water can be seen emanating through the sand and salt water.

LOCUS 24 Tholos, field C; 788167; 0.25 ha.

About 50 m immediately southeast of the “Roman Warehouse,” ca. 150 m south-southwest of Tholos Bay, are house foundations and a dense scatter of Bronze Age pottery covering an area ca. 50 x 50 m (Fig. 24). The densest concentration and visible architecture is located where the north-south field road, extending south from the west end of the Tholos beach, passes a taverna on the east. The site is located 150 m from the beach along this road; the road bisects the site, and house foundations are visible immediately west of the field road and taverna. Walls are extant on the east and north, forming a 2 broad terrace ca. 12–15 m . These walls are house foundations constructed of large dolomite blocks (ca. 0.70 m high, 0.60 m deep, and 1.0–1.20 m long). The scatter extends to the west up the east slope of Mount Schinias some 40–50 m from the field road. West of the shepherd’s fence are more wall remains, but no plan was discernible. The scatter extends across the hill slope east and in front of the Roman warehouse. The taverna, built in 1990, is situated at the south end of the Platys River alluvial terrace, immediately south of the Roman building and east of locus 24. In the earthen-fill foundations of the taverna’s terrace, there is much LM I pottery, suggesting that the settlement of locus 24 extended

eastward across the field road to the edge of the river. The pottery is LM I. Coarse fabric type I is predominant (88%); type IV is present in very low quantities (8%); and type VI is also present (two sherds total from two 50 m lines were collected; 2%). Two rims and one base of semiglobular cups, a stemmed cup base, and thick oval tripod feet are diagnostic for LM IB–II.

Bibliography Bolanakis 1987, 258–259; Boyd 1901, 155; 1904, 13; Haggis 1996b; Harrison 1990; 1993, 188–191; Sanders 1982, 91, 140–141; Seager 1910, 5; Tsoungarakis 1988, 306.

Artifacts LOCUS 16 (Fig. 25) 158 coarse-ware sherds. 1.1 KTS 93. Base of coarse amphora or cooking stand. D. 0.11. Coarse fabric type I. Orange fabric, thick gray core to inner surface; large purple phyllite inclusions. Rib above base. MM III. Betancourt 1990, 183; Barnard and Brogan 2003, 88, fig. 52: IB.603–608. 1.2 KTS 94. Base of coarse amphora or cooking stand. D. 0.10–0.12. Coarse fabric type I. MM III. Betancourt 1990, figs. 29:609, 64:1855; Barnard and Brogan 2003, 88, fig. 52: IB.603–608. 1.3 KTS 95. Shallow basin or tray with spout. D.? Coarse fabric type I. Whole profile. 1.4 KTS 96. Rim of jar or cooking pot. D.? Coarse fabric type VI. MM I–II. 1.5 KTS 97. Rim of dish. D. ca. 0.40. Coarse fabric type I. 1.6 KTS 98. Fragment of strainer. Th. 0.01. Mediumcoarse fabric type I. 1.7 KTS 99. Base of conical cup or tumbler. D. 0.04. Well-worn pinkish orange fabric (2.5YR 5/4–5/6 reddish brown to red). MM IB–II. Betancourt 1985, fig. 76:A–B. 1.8 KTS 100. Base of conical cup. D. 0.03. Semifine pinkish orange fabric (2.5YR 5/6 red). 1.9 KTS 101. Base of conical cup. D. 0.03. Soft, worn, orange fabric (2.5YR 5/6–6/6 red to light red). MM I–II. Betancourt 1990, fig. 44:1007 for shape (MM IB–IIA). 1.10 KTS 102. Tripod foot. Th. 0.015. Thin oval section. Coarse pinkish brown to orange fabric type IV (5YR 5/4 reddish brown). MM I–II. 1.11 KTS 103. Cooking-pot rim. D.? Angular/triangular flat rim. Coarse pinkish orange fabric type VI

GAZETTEER OF ARCHAEOLOGICAL SITES

(2.5YR 5/6 red; 5YR 5/6 yellowish red; 5YR 5/3–4/3 reddish brown). MM I–II. 1.12 KTS 104. Tripod foot. Th. 0.016. Thin oval section. Coarse fabric type VI (5YR 4/4–5/3 reddish brown). MM I–II. LOCUS 38 (Fig. 25) 84 coarse-ware sherds. 1.13 KTS 105. Head of terracotta figurine. Max. Pres. H. 0.037. W. 0.036. Soft pinkish orange fabric (2.5YR 6/8 light red). Pinched eyes, nose; impression at mouth and ears. LOCUS 39 (Fig. 25) 1.14 KTS 106. Rim of a bowl or dish. D. ca. 0.18–0.20. Pinkish orange fabric (2.5YR 6/8 light red); hard pink slip (2.5YR 6/6 light red). Phocaean Red Slip Ware. Hayes form 10. Boardman 1989, 90–93; Hayes 1983, fig. 2:8, 11; 1972, 344, fig. 71:A (6–7th c. A.D.). 1.15 KTS 107. Rim and neck of amphora or jug. D. 0.06. Hard, fine pinkish tan fabric (5YR 7/6–6/8 reddish yellow); some white and dark grits. 1.16 KTS 108. Rim of tray or dish. D. 0.32–0.34. Hard brown fabric (7.5YR 5/2–5/4 brown) with tiny white and dark grits. 1.17 KTS 109. Rim of a bowl. D. 0.19. Hard pinkish tan fabric (5YR 6/6 reddish yellow), pinkish orange slip (2.5YR 5/8 red). Scored interior/exterior; carinated exterior. Sigillata B (A.D. 100–150). Coldstream 1973b, fig. 17:J16; Hayes 1972, 380–381, fig. 82:11 (4–7th c. A.D.); Robinson 1959, G1 76, pl. 67 (1st–2nd c. A.D.). 1.18 KTS 110. Base of bowl or plate. D. 0.11. Hard, pinkish tan fabric (5YR 6/6 reddish yellow), reddish pink slip (10R 5/8 red). Ring foot. Grooves or scoring forming five concentric circles on interior; single tiny groove/rib exterior. Sigillata B. Coldstream 1973b, fig. 17:J15; Robinson 1959, 87, pl. 61:M31 (1st–2nd c. A.D.). 1.19 KTS 111. Bowl base. D. 0.04. Fine, micaceous, tan fabric (7.5YR 7/4–7/6 pink to reddish yellow), dark pinkish orange slip interior (2.5YR 5/6 red). Sigillata A. Crowfoot et al. 1957, 291, figs. 65:1–2, 80:6 (1st c. B.C.). 1.20 KTS 112. Body fragment of a bowl. Hard, fine, pinkish orange fabric and slip (2.5YR 5/6 red). Beveled rib or carination. Sigillata B. Sackett 1992, pl. 172: D1.2–4; Boardman 1989, 88–89. LOCUS 24 (Fig. 26) 119 coarse-ware sherds. 1.21 KTS 113. Base of cup or jar. D. 0.055. Buff fabric/slip (5YR 8/1–8/2 white to pinkish white). Slightly beveled flat base. Trace of black or brown paint exterior. LM IB–II. Popham 1977, 186–187, fig. 1:c. 1.22 KTS 114. Base of conical cup. D. 0.04. Pinkish orange fabric (2.5YR 6/8 light red); soft, well-worn;

93

tiny white and dark grits. Trace of black paint interior. MM II. Pyrgos II–III. For shape see Betancourt 1990, fig. 44:1007. 1.23 KTS 115. Base of conical or hemispherical cup. D. 0.04. Buff fabric (7.5YR 8/6 yellowish red), buff slip (7.5YR 8/4 pink); traces of black and red paint interior/exterior. LM I. 1.24 KTS 116. Base of conical cup. D. 0.04. Soft pinkish orange fabric (2.5YR 6/8–5YR 6/8 light red to yellowish red). LM I. 1.25 KTS 118. Base of conical or hemispherical cup. D. 0.04. Dark pinkish orange fabric with small dark grits. LM I. 1.26 KTS 119. Base of hemispherical cup. D. 0.03. Fine, buff fabric, dark brown paint interior/exterior. Base is flaring and wall thickens at middle. Thin ridge or rib above base. Possibly the same cup as rim 120. LM IA–B. Sackett and Popham 1970, fig. 13:11. 1.27 KTS 117. Rim of hemispherical cup. D. 0.09. Tan fabric (7.5YR 7/6 reddish yellow), buff slip (7.5YR 8/6 reddish yellow). Black bands on exterior. LM I. Catling, Catling, and Smyth 1979, fig. 31:226 (LM IA); MacGillivray et al. 1989, fig. 5. 1.28 KTS 120. Rim of hemispherical cup. D. 0.08. Same fabric as 1.26. LM I. Sackett and Popham 1970, fig. 13:9–11; MacGillivray et al. 1989, fig. 5. 1.29 KTS 121. Rim of cup. D. 0.06. Tan fabric (7.5YR 7/6 reddish yellow), buff slip (7.5YR 8/6 reddish yellow). 1.30 KTS 122. Rim of a rounded or bell-shaped cup. D. 0.08. Soft buff fabric; solid black paint interior. 1.31 KTS 125. Rim of cup or bowl. D.? Dark pinkish orange fabric (2.5YR 6/8 light red), pinkish orange slip (2.5YR 6/6 light red). Knob-like projections below rim exterior. 1.32 KTS 123. Body fragment of closed(?) shape. Th. 0.02. Dull buff fabric/slip; dark metallic black horizontal lines and bands. 1.33 KTS 124. Rim of tray or dish. D. 0.50(?). Coarse fabric type I. 1.34 KTS 127. Rim of tray or dish. D. ca. 0.40. Coarse fabric type I. 1.35 KTS 126. Rim of cooking pot. D. 0.20. Type A cooking pot. Coarse fabric type I. Thick gray core. LM I. Betancourt 1980. 1.36 KTS 128. Rim of jar. D. 0.16. Coarse fabric VI. MM I–II. 1.37 KTS 130. Rim of cooking pot. D.? Coarse fabric type I; burned black core/surfaces. Rim depressed to form spout. Betancourt 1980, C816. 1.38 KTS 129. Coarse cup. Rim D. ca. 0.05; Base D. 0.06; H. 0.073. Fabric type I. Handmade.

94

KAVOUSI I

2. Tholos LOCUS 23 Tholos, east shore; 793171; 0.02 ha.? Coastal beachrock, 12 m AMSL. MM I–II, MM III–LM I.

North of site 3 ca. 100 m is a very light, illdefined scatter of pottery on the eroded limestone

cliff above the east shore of Tholos Bay (Fig. 23). The material is visible on bare bedrock at the edge of the scree line and consists of coarse fabrics I and III. The area was buried in 1990 by the construction of a wide terrace west of a modern house. The site is MM–LM I.

3. Tholos LOCUS 22 Tholos, east shore; 793168; 0.2 ha. Coastal beachrock, 12 m AMSL. MM I–II, MM III–LM I.

On the east side of Tholos Bay, ca. 75 m north along the coast from the taverna (Fig. 23), is a light scatter of pottery and limestone wall foundations. The scatter extends north along two vine terraces, beginning 20 m north of the taverna and covering an area of at least 80 m north-south and 25 m east-west. The material is visible at the edge of the Kapsas scree line, where the deep stone debris has eroded into the sea or has been cleared for the recent grape arbor. The minimum size of the locus is 0.2 ha. At the north end of the vine terrace are wall blocks of

blue-gray limestone forming a short (1–2 m) linear accumulation east-west. A north-south line was collected across the site. The coarse fabrics are types I (49%), IV (39%), VI (9%), and III (3%). Diagnostic fine wares include two conical cup bases and a tumbler or straightsided cup base. A high percentage of type I is evidence of LM I; the relatively high percentages of types VI and IV are evidence of MM as is the straight-sided cup base.

Artifacts 30 coarse-ware sherds.

4. Analypse LOCUS 36 Tholos, Analypse; 786168; 0.12 ha. Coastal foothill; 71.30 m AMSL. MM I–II, MM III–LM I.

On top of Analypse Hill (Fig. 23; Pl. 9A), an eastern spur of Mount Schinias, directly west and upslope from the church of Panagia (ca. 200 m) and west-northwest of the Roman warehouse, is a flat ridge upon which are a light scatter of pottery and foundations of a single building. Local dolomite blocks form a rectangle ca. 12.50 m east-west and 11 m north-south. The site has an eastern aspect, overlooking the Tholos/Kambos plain and the Bay of Tholos. Pseira is visible to the north.

Two bisecting lines (35 m long) were collected across the site on cardinal points. Fabric types I and IV were represented exclusively; type I is dominant (60% of the total sample). Type IV pottery (ca. 27% of the total) is a single pithos scatter and is thus overrepresented. Diagnostic for LM I is a thick oval tripod foot of type I; a conical cup base (4.1) is probably MM I–II (Fig. 26). The site is either a single house or lookout for the port of Tholos. The site had a limited period of use judging from the sparseness of the sherds (2–5 sherds per 5.0 x 2.0 m unit). An abandoned mandra is located 25 m northeast of the Minoan house. The eroded dolomite slope of Mount Schinias is

GAZETTEER OF ARCHAEOLOGICAL SITES

winter grazing land today: 40–60 percent of the visible ground is exposed bedrock and 20 percent is scree. Intensive agricultural use in the Bronze Age seems unlikely, given the paucity of remains in adjacent areas and the lack of evidence for terracing.

95

slip (10YR 8/4 very pale brown). MM I–II. Betancourt 1985, fig. 49:B; Andreou 1978, fig. 11:15–17; Pelon 1970, pl. 10:7. 4.2 KTS 132. Base of jug or jar. D. 0.08. Mediumcoarse fabric type VI. MM. 4.3 KTS 133. Rim of bowl. D. 0.08–0.10. Same fabric as 4.1; pinkish tan slip (5YR 7/6 reddish yellow).

Artifacts (Fig. 26) 48 coarse-ware sherds. 4.1 KTS 131. Base of conical cup. D. 0.05. Pink to pinkish orange fabric (2.5YR 6/6–6/8 light red), buff

4.4 KTS 134. Base of pithos. D. ca. 0.24. Coarse fabric type IV.

5. Hagios Antonios Loci 10/61 and 15 Tholos/Mount Schinias, field A; 784164; 3 ha. Inland plain; 20 m AMSL. MM I–II, LM I, LM IIIA–B, R.

On the southeast side of Mount Schinias, 140 degrees and 200 m east-southeast of the church of Hagios Antonios and northeast of the Hagios Antonios revma, is a dense sherd scatter. Architectural remains are situated above the road; pottery is visible in the scarp of the road cutting. (Fig. 27; Pls. 11A, B, 12A). The site covers an area of at least 3 ha., extending over the southern footslope of Mount Schinias. The architecture lies northwest of the field road that runs north to Tholos on the west side of the Kambos plain. Road scarp deposits are at the base of the Hagios Antonios revma, on the west side of the field road, where it winds around Mount Schinias and turns east and north toward Tholos. A shepherd’s fence bisects the site on a contour west and above the road. Loci 10 and 61 are composed of a continuous scatter of pottery and are collapsed into a single locus (Fig. 27). A separate scatter of LM III pottery was recovered in locus 10. Locus 15, a road scarp deposit at the edge of 10/61, was recorded separately. Visibility is hampered by vegetation, and preservation of architecture is poor because of the loss of the colluvial veneer. Upslope and westnorthwest of the Tholos field road, the pottery remains are generally light and disparate, while in the road, at the footslope of the hill, slope wash debris and in situ material provided the bulk of the sample.

Architectural remains are visible at the base of the slope of Mount Schinias, immediately above the field road, and continue upslope ca. 100 m. The walls at the base of the hill are made of large blocks and boulders of local limestone. No plan could be discerned. Farther upslope at locus 61, ca. 100 m northwest from the edge of the plain are foundations of several buildings (Fig. 27). One building (Building A) consists of a level terrace, ca. 10 x 10 m and constructed of large limestone boulders, preserved to two courses (Pl. 11A). House foundations and ancient terrace walls are scattered 25 m westsouthwest and exactly south of this building. The foundations of another building, Building B (Pl. 11B), are 25 m east (288 degrees). This is also a built terrace 8.50 m north-south and 13.50 m eastwest, constructed of megalithic limestone blocks. On the south side of the building is a long wall (15–20 m) forming a 2.5–3 m wide terrace in front of the building. About 140 degrees and 20 m downslope to the southeast are two terrace walls running east-west with the contour of the hill (Pls. 7B, 12A). The first is constructed of limestone boulders and preserved to three courses (1.40 m) and ca. 20 m extant. Farther down the slope and slightly southeast of this wall is a second, also preserved to three courses and extant to 20 m along the contour. These walls have no returns, and support terrace fill with Bronze Age sherd material (coarse fabrics I and IV). Some Roman sherds were found downslope from these terraces. At locus 15, immediately west of the road scarp deposit and shepherd’s fence, are three rock-cut

96

KAVOUSI I

cists, possibly graves. The best preserved cist measures 0.80 m wide x 1.45 m long x 0.50–0.70 m deep. This southeast corner of Mount Schinias and the northeast edge of the Hagios Antonios Valley is extremely eroded dolomite limestone with a thin, patchy colluvial veneer. Where pockets of sediment collect, dense olive, lentisk, and spiny broom flourish. Mount Schinias today is winter pasture, as are most of the limestone hills that border the west side of the Kambos. The slope on the south of Mount Schinias (Hagios Antonios) is ca. 40 degrees; on the east, the incline is more gradual, but nonetheless steep (20–30 degrees). The Minoan house walls here required considerable terracing, and slope conservation is evident in the two terrace walls southeast of the houses. The instability of the limestone slopes would have required maintenance to hold a sufficient colluvial surface for agriculture. In locus 10/61, coarse fabric type I is 47 percent of the total sample, type VI is 25 percent, and type IV is 25 percent of the total collected. The remaining type represented is XX/XXI, which is EM III– MM IA. The even distribution of MM type VI, MM/LM III type IV, and LM I type I is good indication of continuous use of the area from MM I to LM III. Other artifacts include a terracotta loom weight, two stone weights, shell fragments, two possible terracotta figurines, and one Roman amphora handle (pseudo-Koan type). In locus 15, type I is 62 percent of the total sample, type IV is 30 percent, and type VI is 3 percent; type III is also represented by 2 sherds (5%). Thus the artifact profile is similar to 10/61; however, there is a notable lack of the typical MM cooking-pot and basin fabric VI. Diagnostic material at both loci is MM–LM III (Figs. 28–30). At locus 15, a rounded oval tripod foot is LM III. At locus 61, a type IV pithos wall with band of incised chevrons is LM III. In 1900 Boyd excavated a “well preserved piece of Cyclopean wall” at this location “on a terrace about 15 m above the foot of the hill [Hagios Antonios].” The pottery she describes is clearly LM I–III. The decorations are in red or black paint on buff (“fine light lemon yellow clay”) and include bands, dots, spirals, floral decorations, and dotted fish scale patterns; white paint is applied to the red spiral. Boyd, not finding return walls or associated features behind this “Cyclopean” wall, suggests that it is an agricultural terrace wall. In the same year, Boyd excavated a small pit at Hagios

Antonios: “Near the bottom of this hill in a cubic foot of space were found three whole lamps, pieces of eight other lamps, fragments of terracotta ‘icons’ and of jugs—all of Roman or late Greek make.” The lamps she describes as “bowls closed by a concave perforated cover with stamped ornament...a ring handle opposite the nozzle, and in two instances, scroll ornament on each side of the nozzle.” The decorations include Zeus Amon with eagle, a goddess with turreted crown and cornucopia, a cock with wreath of leaves, a bull, a sphinx, and a warrior or Amazon attended by a dog. The terracotta 2 plaques or “icons” measured 0.10 m , with molded frame and ring for hanging, and usually have a representation of a horseman. The pit seems to have been a votive deposit or burial, perhaps associated with a hero shrine. While some Roman material is apparent on the slope of Mount Schinias, there is no concentration of Hellenistic or Roman material to justify a settlement in this area.

Bibliography Boyd 1901, 130–131, 156; 1904, 22; Soles 1991, 75–76; Warren 1984, 40–41.

Artifacts LOCUS 10/61 (Figs. 28–30) 85 coarse-ware sherds. 5.1 KTS 176. Body frag. of a closed vessel. Pink fabric, buff slip, burnished exterior. Two black bands below red dots and spirals; two converging black lines above spiral and dot; traces of white paint applied to red. LM I. Popham 1967, P339. 5.2 KTS 177. Tumbler base. D. 0.05. Fine, hard, tan fabric (7.5YR 6/6 reddish yellow); flaky metallic black paint on interior, exterior and bottom; wheel-made; base slightly concave at bottom. MM IB–II. Andreou 1978, fig. 11:10–13. 5.3 KTS 178. Jug rim. D. 0.06. Pinkish tan fabric (5YR 7/4 pink), buff slip (10YR 8/4 very pale brown); rim is thickened and slightly everted. 5.4 KTS 179. Wall of conical cup. Pinkish orange fabric (2.5YR 6/6–6/8 light red), pinkish tan slip (5YR 7/6–7/8 reddish yellow); handmade. MM IA. 5.5 KTS 180. Bowl rim. D. 0.16. Fine hard light pinkish orange fabric (7.5YR 7/6 reddish yellow); red paint (2.5YR 4/8 red) interior and exterior; finger marks interior/exterior; handmade(?). MM IA–B. 5.6 KTS 181. Conical cup rim. D. 0.08. Pinkish orange fabric and slip (2.5YR 5/8 red); rim in-turned. MM III. Betancourt 1990, 40, fig. 25:478.

GAZETTEER OF ARCHAEOLOGICAL SITES

5.7 KTS 182. Rim of hemispherical or bell cup. D. 0.08. Buff fabric, slip (10YR 8/4 very pale brown); black paint exterior. MM III–LM I. Betancourt 1990, 116, figs. 34:723, 61:1800. 5.8 KTS 183. Conical cup rim. D. 0.09. Pinkish tan (5YR 6/8 reddish yellow) to pinkish orange (2.5YR 6/8 light red) fabric; tiny mica inclusions. MM III. Pelon 1970, pl. 37:6. 5.9 KTS 184. Cup base. D. 0.05. Tan fabric (7.5YR 7/6 reddish yellow), buff slip (7.5YR 8/4 pink); wheel ridging pronounced on interior. 5.10 KTS 185. Cup base. D. 0.04. Soft buff fabric/slip (10YR 8/3 very pale brown). 5.11 KTS 186. Cup base. D. 0.04. Soft pinkish tan to dark buff fabric (5YR 7/6 reddish yellow). String-cut base. 5.12 KTS 187. Cup base. D. 0.06. Dark red fabric (2.5YR 4/6 red), pinkish orange slip (2.5YR 6/8 light red); tiny mica inclusions; string-cut base. MM I–II. 5.13 KTS 188. Straight-sided cup base. D. 0.054. Fine, soft, dark buff fabric (10YR 7/4 very pale brown); straight profile, rounded at base with slight bevel; string-cut. MM IIB(?), MM III. Betancourt 1990, 40, 45, figs. 60:1707, 67:1972–1973. 5.14 KTS 189. Bowl or cup base. D. 0.07. Buff fabric (7.5YR 8/4 pink), buff slip (10YR 8/4 very pale brown), reddish pink paint interior and exterior (10R 5/8 red). MM I–II. 5.15 KTS 190. Bowl base. D. 0.06. Fine pinkish tan fabric (5YR 7/6 reddish yellow), buff slip interior/exterior (7.5YR 7/6 reddish yellow). MM III. 5.16 KTS 191. Jar rim. D. 0.16. Pink to orange fabric; coarse fabric type XXI. Rim internally thickened. EM III–MM IA. Andreou 1978, fig. 3:13. 5.17 KTS 192. Cooking-pot or jar rim. D. 0.12. Medium-coarse fabric type I. MM III–LM I. 5.18 KTS 193. Jar rim. D. 0.24. Coarse fabric type IV. MM. 5.19 KTS 194. Stone weight. L. 0.083; W. 0.056; Th. 0.033. Gray-blue limestone; naturally weathered river stone. 5.20 KTS 195. Stone weight. L. 0.077; W. 0.055; Th. 0.043. Gray-blue limestone; naturally weathered river stone. 5.21 KTS 196. Terracotta weight. D. 0.077; Th. 0.021. Coarse fabric type IV; fired hard; no finished edges; possibly a reused sherd. 5.22–5.23 KTS 197–198. Cooking trays or dishes. D. >0.30. Coarse fabric type IV. MM. 5.24 KTS 199. Cooking tray or dish. D. >0.30. Coarse fabric type VI. MM I–II. 5.25 KTS 200. Figurine frag. Max. Pres. L. 0.063; W. 0.052; Th. 0.022. Hard, well-fired, orange fabric (medium-coarse fabric type IV). 5.26 KTS 201. Figurine frag. Max. Pres. L. 0.049; W. 0.04; Th. 0.016. Coarse fabric type IV.

97

5.27 KTS 597. Pulled-rim bowl. D. 0.15. Pink-buff to orange core; tiny purple schist and mica inclusions. Reddish orange monochrome exterior and interior rim band. Burnished surfaces. LM IIIA:1. Brogan, Smith, and Soles 2003, 106, fig. 7: SM11075. 5.28 KTS 598. Kylix or bowl rim. D. 0.16. Tan clay; reddish brown monochrome exterior. Burnished. LM IIIA. Brogan, Smith, and Soles 2003, 106, fig. 7: P3273 (LM IIIA:1); Betancourt and Floyd 2000–2001, 18, fig. 9: X1700 (LM IIIA:2). 5.29 KTS 599. Bowl rim. D. 0.12. Pink-buff core; mottled orange-gray surface; tiny gray and white quartzite and schist inclusions. Black monochrome exterior and rim band interior. Burnished surfaces. LM II–IIIA. Brogan, Smith, and Soles 2003, 106, fig. 7: P1135. 5.30 KTS 600. Spouted bowl rim. D. 0.15. Soft pink core; mica inclusions; dull gray to dark brown monochrome exterior/interior. LM IIIA. 5.31 KTS 601. Bowl rim. D. 0.14. Same fabric and ware as 597. Brogan, Smith, and Soles 2003, 106, fig. 7: SM11075. 5.32 KTS 602. Conical cup. D. 0.10. Semifine grayish tan clay; microscopic dark grits. Wheel ridges prominent on interior. Worn exterior slip. LM III. 5.33 KTS 603. Conical cup. D. 0.10. Semifine; orange-red clay; tiny white and gold mica inclusions. Wheel ridges prominent on interior. LM III. 5.34 KTS 604. Cooking-pot rim. D. 0.20. Fabric type IV. LM III. 5.35 KTS 605. Cooking-pot rim. D. 0.16. Fabric type IV. LM III. 5.36 KTS 606. Basin or bowl rim. D.? Fabric type IV. LM III. 5.37 KTS 607. Tripod cooking-pot foot. Round section. Fabric type IV. LM III. 5.38 KTS 608. Lekanis rim. D. 0.27. Fabric type IV. LM IIIA–B. 5.39 KTS 609. Jar base. D. 0.033. Medium-coarse type IV. Gritty, porous clay. LM III(?) 5.40 KTS 610. Conical cup base. D. 0.035. Semifine; gritty orange-red clay. LM III. LOCUS 15 (Fig. 30) 37 coarse-ware sherds. 5.41 KTS 202. Conical cup rim. D. 0.09. Pinkish orange to brown fabric (2.5YR 5/8 red); slight wheel ridging apparent interior/exterior. Fine sand inclusions. MM III–LM I. 5.42 KTS 203. Conical cup rim. D. 0.10. Pinkish orange to red fabric (2.5YR 5/8 red), pinkish orange slip (2.5YR 5/8 red). MM III. Pelon 1970, pl. 37:6. 5.43 KTS 204. Conical cup base. D. 0.035. Pinkish orange fabric (2.5YR 5/6 red), phyllite and fine sand inclusions. MM III. Pelon 1970, pl. 37:6.

98

KAVOUSI I

5.44–5.45 KTS 205–206. Conical cup rims. D.0.08– 0.09. Pinkish orange fabric (2.5YR 5/8 red); fine sand inclusions. MM III–LM I. Pelon 1970, pl. 37. 5.46 KTS 207. Conical cup base. D. 0.04. Pink to pinkish orange fabric (2.5YR 6/6 light red); pronounced wheel ridging, ribbing or rilling possible. MM III(?). 5.47 KTS 208. Hemispherical (ogival) cup rim. D. 0.10. Pinkish orange fabric (2.5YR 5/8 red). LM IA. Pelon 1970, pls. 15:1, 35:4; Betancourt 1990, fig. 61:1800; Barnard et al. 2003, figs. 4–5. 5.48 KTS 209. Jar rim (bridge-spouted?). D.? Pink fabric (2.5YR 6/6 light red); buff slip exterior (10YR 8/4 very pale brown). MM II–III. 5.49 KTS 210. Conical cup. Rim D. 0.086; Base D. 0.038. Fine pinkish orange fabric (2.5YR 5/8 red); red

phyllite and fine sand inclusions. LM IA. Pelon 1970, pl. 37:9. 5.51 KTS 212. Cup base (hemispherical?). D. 0.032. Fine sandy tan fabric (7.5YR 6/6 reddish yellow) with red phyllite, sand, and mica inclusions; red paint exterior (10R 4/8). LM IA. 5.50, 5.52, 5.54 KTS 211, 213, 215. Jug/jar bases. D. 0.12. Pink to pinkish brown fabric (2.5YR 6/6 light red); medium-coarse type I. 5.53 KTS 214. Body (shoulder) frag. of closed shape (jar or jug). Fine dark buff sandy fabric (10YR 7/6 yellow); three arcing bands in brown paint (5YR 4/4 reddish brown) and dots (tendril scroll or foliate band). Barnard and Brogan 2003, fig. 57 (LM IB).

6. Hagios Antonios Rock Shelter LOCUS 62 Mount Schinias, field A (Hagios Antonios); 783165; 0.005 ha. Coastal foothill; 80 m AMSL. FN, EM I–II, EM III–MM IA.

The rock shelter is located on the south slope of Mount Schinias, on the west side of the Kambos, some 3.5 km northwest of the village of Kavousi (Fig. 31; Pls. 7A, 12B). It is situated on a steep eroded hill slope, ca. 80 m AMSL, 50 m east of the small church of Hagios Antonios (Fig. 6). Both the church and the rock shelter occupy the north side of the deeply incised torrent that separates Mount Schinias from the hill of Chomatas (Fig. 6). On the south side of the torrent that is across from the church, there is a modern cistern marking the location of a spring. The rock shelter is a natural overhang in the dolomite hill slope; an irregular cavity in the bedrock ca. 2.2 m deep at its deepest point, ca. 5 m wide, and 1.9–2 m high at the mouth from the level of the fill (Pl. 12B). The deepest area of the shelter is the western recess containing an irregular bedrock pit in the back (north). The shelter widens gradually from west to east, and fallen stone debris and a rise in the natural bedrock reduce the depth of the recess in the east. Immediately south of the mouth of the cave is a terrace made of soil and stone extending ca. 5–6 m north-south and 7–8 m east-west. The date of the use of the shelter has been established on the basis of Hall’s verbal description of the pottery, Betancourt’s publication of four objects

in the University of Pennsylvania Museum, and the author’s publication of four objects that appear in a photograph in the University Museum archives labeled, “Hagios Andoni, near Kavousi, cave burial, EM II, 5 Pottery vessels” (The University of Pennsylvania Museum Photographic Archives, Negative 139753 [copy of Negative 2630]). The surface finds from the cave terrace suggest the same chronological range as that of the material in the published catalogs. Coarse body fragments of cooking pots and trays were predominant. Two fabric types were the most common—fabric type II and type IX—suggesting EM IIA–B. The fine wares are EM I–MM I. The shelter was probably part of the cemetery for the neighboring settlement at Alykomouri (site 8).

Bibliography Alexiou 1951, 287; Betancourt 1983, 5–6; 1985, 49– 51; Branigan 1968, 41, 58, 73; 1988a, 13, 94–95; 1991, 99–100; Faure 1964, 36, 67, 73; Haggis 1993b; Hall 1914, 183–185; Pendlebury 1939, 61, 77; Soles 1992, 27.

Artifacts (Fig. 31) 6.1 KTS 135. Rim of bowl. D. ca. 0.14–0.20. Hard fine reddish brown fabric (medium-coarse type II). Striations indicating smoothing on interior and exterior. Two holes, piercing wall below rim. FN–EM I. Vagnetti, Christopoulou, and Tzedakis 1989, 24. 6.2 KTS 136. Body frag. of closed(?) vessel. Th. 0.003. Pinkish tan fabric (5YR 7/3 pink), pinkish tan slip (5YR 7/4 pink); fugitive vertical line (5YR 6/6 reddish

GAZETTEER OF ARCHAEOLOGICAL SITES

yellow). Dark-on-light ware. EM II (Hagios Onouphrios II). Blackman and Branigan 1982, 32. 6.3 KTS 137. Body frag. of pyxis or suspension pot. Th. 0.004. Fine gray ware (2.5Y 5/2 dark greenish gray); burnished interior/exterior. Band with diagonal incised hatching. EM I–EM IIA. Betancourt 1985, 40–41; fig. 21:c, f. 6.4 KTS 138. Base of bowl. D. 0.06. Coarse–medium-coarse fabric type II; pinkish brown fabric, burnished interior (5YR 5/2 reddish gray); red slip exterior (10R 5/6–5/8). Base is concave at bottom. FN–EM I. Sampled P.M. Day: Kavousi 93/73. Vagnetti 1996, pl. 1:1; Wilson 1985, fig. 42:FF7.

99

6.5 KTS 139. Rim of bowl. D.? Hard brown fabric (7.5Y 5/6 strong brown), tiny white grits (mediumcoarse type II). Burnished interior/exterior. Dark brownblack streaks from heavy burnishing. Brown burnished ware. EM I–II. 6.6 KTS 140. Rim of a tray or dish. D.? Coarse fabric type II; pinkish brown fabric (5YR 5/3–5/4 reddish brown); smoothed interior. 6.7 KTS 569. Obsidian blade fragment. L. 0.031; W. 0.009; Th. 0.002–0.003. Trapezoidal section; proximal end preserved. Very even break. No signs of wear or retouch.

7. Hagios Antonios Alykomouri Saddle LOCUS 8 Hagios Antonios, Alykomouri (saddle); 778163; 0.06 ha. Coastal foothill; 116.90 m AMSL. EM I–II, MM I–II, MM III–LM I, B–V.

Above and southwest of the church of Hagios Antonios, in the saddle joining the hills of Chomatas and Mount Schinias, is a light scatter of Minoan pottery (Fig. 6). The saddle runs north-south and divides the sea from the plain. Sherds are concentrated at the south end; at the north end of the saddle are modern field houses and several limestone caves, used as a winter mandra (Pl. 13). A fenced-in olive orchard and grape terraces extend west of the saddle to the sea. Recent rebuilding of the field houses and the construction of an agricultural road from Kambos to Alykomouri have disturbed much of

the material on the terrace. The pottery consists of types I, VI, and III and is evidence of MM–LM I activity in this area. Byzantine, Venetian, and Ottoman-Modern pottery is present as well. Artifacts include a threshing sledge blade (chert), probably recent; and a green-stone (metamorphosed chert) drill guide of Bronze Age date. Within the area of the caves at the saddle’s north end are sherds imbedded in thick animal dung (Pl. 13B). Visible sherds in situ are EM I–II cooking-pot wares and black burnished wares. It is likely that these limestone caves were EM burial locations like the Hagios Antonios rock shelter. Boyd saw ancient house walls at this location in 1901.

Bibliography Boyd, Avgo notebook, May 12, 1901; Haggis 1993b.

8. Alykomouri LOCUS 58 Chomatas, field B; 777163; 0.12 ha. Coastal foothill; 120 m AMSL. FN, EM I–III.

Southeast of the Alykomouri saddle, southwest of the Hagios Antonios cave, and immediately east of the northern peak of Mount Chomatas is a phyllite spur extending east from the northern end of 2 Chomatas. A terrace on the spur ca. 30–35 m

(0.12–0.13 ha.) contains a scatter of pottery and stone tools (Fig. 6). The site has been bulldozed to create a series of terraces for vines, and the architecture and artifacts have been buried and displaced. A modern field house is located on the eastern edge of the terrace (Pl. 14A). The pottery from this site consists of brown burnished ware, black burnished ware, dark-burnished ware (pattern burnished and Pyrgos Ware); Vasiliki

100

KAVOUSI I

Ware, dark-on-light painted wares (Hagios Onouphrios and Koumasa), and White-on-Dark Ware (Figs. 31, 32). The coarse fabrics collected are type IX (66%) and type II (34%). Other artifacts include door and window sockets, obsidian blade fragments and flakes, a saddle quern, several ground stone tools, and a number of marine shells. The site is the settlement belonging to the burial caves at Hagios Antonios and the Alykomouri saddle. The site has an eastern aspect, and its location on this hill above the Kambos maximizes accessibility to the arable phyllite deposits both on the slopes of Mount Chomatas and in the plain. A perennial spring, marked by a cistern, is located across the revma, immediately south of the church at Hagios Antonios.

Bibliography Boyd, Avgo notebook, May 12, 1901; Haggis 1993b.

Artifacts (Figs. 31–33) 23 coarse-ware sherds. 8.1 KTS 141. Rim of collared jar. D. 0.13. Fine hard dark buff fabric (10YR 6/4 light yellowish brown), tiny red and gray phyllite inclusions. White slip (2.5Y 8/2 white); red and dark red cross-hatching on exterior; traces of streaky dark red-brown paint interior. EM I. Sampled P.M. Day: Kavousi 93/75. Blackman and Branigan 1982, 29–32 (E22). 8.2 KTS 142. Rim of a chalice. D. 0.16. Fine hard orange (5YR 5/6 yellowish red), dark buff (10YR 6/3 pale brown) to dark gray (10YR 4/1) core. Thick black mottled slip interior/exterior: very dark gray (10YR 3/1), brown (10YR 4/3 dark brown), red (2.5YR 3/6 dark red); interior burnished. Exterior surface has dense black burnished cross-hatching extending down from three horizontal burnished lines. Dark gray burnished ware (Pyrgos Ware). EM I. Sampled P.M. Day: Kavousi 93/74. Blackman and Branigan 1982 (F31). 8.3 KTS 143. Shoulder frag. of jug. Max. Pres. H. 0.045. Soft dark buff fabric (10YR 7/4 very pale brown), buff slip (2.5Y 8/2 white); brownish black paint exterior: wide arcing stripe and two intersecting oblique stripes. Dark-on-light painted ware. EM IIA. Sampled P.M. Day: Kavousi 93/76. Warren 1972, 185: P443–444. 8.4 KTS 144. Bowl handle. Max. Pres. H. 0.102. Coarse pinkish orange (2.5YR 5/6 red) to gray (10YR 5/2 grayish brown) fabric, dark gray (10YR 4/1) core with limestone and gray phyllite inclusions. Two holes (D. ca. 0.006 m) pierce handle and two impressed circles of similar diameter in the upper corners. FN or EM I. Parallels for this bowl-handle type are found in the EM I cemetery at Hagia Photia, where Betancourt (personal

communication) calls them possible Cycladic imports. Cf. Vagnetti 1996, fig. 1:5. 8.5–8.6 KTS 145–146. Bases of shallow open bowls. D. 0.09–0.10. Coarse, hard brownish orange fabric (2.4YR 5/6 red) with brownish black surface wiped or rubbed to a smooth hard finish; brownish black core (5YR 3/2 dark reddish brown); white and gray angular limestone and dark phyllite inclusions. Slightly concave at bottom. EM II. Warren 1972, 168: P209. 8.7 KTS 147. Rim of bowl. D.? Cooking-pot fabric type IV; red fabric (2.5YR 3/6 dark red); surfaces smoothed, red (2.5YR 4/8 red). Wall thickened at rim. 8.8 KTS 148. Rim of jar. D. 0.14. Pinkish orange fabric (2.5YR 5/6 red), gray core, dense phyllite inclusions. Raised collar of spouted jar. EM IIB. Warren 1972, 202–204: P669, P681; Pendlebury, Pendlebury, and Money-Coutts 1935–1936, 154, fig. 9 (EM III– MM IA). 8.9 KTS 149. Rim of baking plate. D.? Coarse fabric type II (Myrtos fabric type 3). EM II. Warren 1972, 161: P102. 8.10 KTS 150. Base of bowl. D. 0.09. Red fabric (2.5YR 4/6 red) with a gray core (10YR 5/1); white limestone and red phyllite inclusions. Burnished and mottled exterior; thick pinkish orange (2.5YR 5/6 red) slip interior, burnished. 8.11 KTS 151. Rim of bowl. D. 0.22. Internal/external surfaces burnished; brown (10YR 3/2 very dark grayish brown), red (2.5YR 4/8 red); dark gray core (2.5YR 4/0 dark gray). 8.12 KTS 152. Rim of jar or cooking pot. D. 0.22. Coarse brown type II (7.5YR 4/4 brown to dark brown) (Myrtos fabric type 3); brown core (10YR 3/2 very dark grayish brown). Rim is internally thickened and beveled. EM III–MM IA. Sampled P.M. Day: Kavousi 93/80. 8.13 KTS 153. Rim of cooking pot. D. 0.30(?). Coarse brown fabric (10YR 3/2 very dark grayish brown); dark gray core; internal surface is red (2.5YR 4/6 red). Largish phyllite and white limestone inclusions. Light burnished interior. Rim is square and incurving. EM II. Warren 1972, 178: P339. 8.14 KTS 154. Rim of jar or cooking pot. D. 0.32(?). Coarse fabric type IV; brown to red fabric (10YR 3/2–2.5YR 4/4 very dark grayish brown to red) and dark brown (10YR 2.5/2 very dark brown) core. 8.15 KTS 155. Door socket. Socket D. 0.04–0.06; D. 0.175–0.185; Th. 0.028–0.057. Pinkish red sandstone. Flat stone, roughly circular with worn socket; ware indicated by striations. Warren 1972, 219: 61, 66. 8.16 KTS 156. Rim of baking dish. D.? Coarse fabric type II; pinkish orange fabric (2.5YR 5/6 red); Myrtos fabric type 6. EM II. Warren 1972, 162: P108. 8.17 KTS 157. Rim of a jar. D. 0.20. Red fabric (10R 5/6), dark gray core, burnished exterior, rough interior. N. Sampled P.M. Day: Kavousi 93/70.

GAZETTEER OF ARCHAEOLOGICAL SITES

8.18 KTS 158. Rim of bowl or jar. D. 0.24. Dark pinkish brown fabric (5YR 5/4 reddish brown). Large phyllite and limestone inclusions; hand-smoothed interior/exterior. 8.19 KTS 159. Rim of shallow bowl or dish. D. 0.22. Coarse fabric type II; pinkish brown to orange fabric (5YR 5/4–5/6 reddish brown to yellowish red). Smoothed interior; rough exterior. Shallow groove on top of squared rim. Same fabric as Warren 1972, 166: P168. EM II. 8.20 KTS 160. Rim of cooking dish or baking plate. D. ca. 0.30. Coarse fabric type II; pinkish brown to brown fabric (5YR 5/3–4/3 reddish brown); rough burned exterior, hand-smoothed interior. Warren 1972, 162: P107. EM II. Sampled P.M. Day: Kavousi 93/79. 8.21 KTS 161. Base of a cup or bowl. D.? Mediumcoarse pinkish orange fabric with a few large phyllite inclusions (2.5YR 5/8 red). 8.22 KTS 162. Rim of a bowl. D. 0.12. Reddish orange fabric and slip (10YR 5/6 yellowish brown).

101

8.23 KTS 163. Rim of a bowl. D.? Pink fabric (10R 6/6), tan slip (7.5YR 7/4 pink); fine dark grits. Warren 1972, 170: P252. 8.24 KTS 164. Quartzite cobble. L. 0.195; W. 0.047; Th. 0.03. Greenish gray color. One abraded face. Both ends pecked and battered. Blitzer 1995, type 10. 8.25 KTS 165. Quartzite cobble. L. 0.11; W. 0.061; Th. 0.029. Gray color. One abraded face. Both ends pecked; one end has concoidal fracture from use. Handstone. Blitzer 1995, type 7. 8.26 KTS 166. Quartzite cobble. L. 0.128; W. 0.082; Th. 0.047. Gray color. One abraded face; pecked ends. One end battered; possibly used for chopping. Blitzer 1995, type 7 (handstone) or type 10. 8.27 KTS 167. Quartzite cobble. L. 0.055; W. 0.039; Th. 0.031. Gray color. One abraded face; pecked margins. Warren 1972, 249–250: 191–192; Blitzer 1995, type 1.

9. Chomatas LOCUS 56 Chomatas, field B; 773160; 0.05 ha. Coastal foothill; 213.30 m AMSL. MM I–II, LM I.

On the edge of a limestone cliff, ca. 150 m southsouthwest of the north peak of Chomatas is a built terrace with well-preserved architectural remains and a dense scatter of pottery (Fig. 34; Pls. 1B, 2A, 6B, 14B, 15A). The site is a broad terrace ca. 30 m north-south and 20 m east-west, bordered by a substantial wall on the north, south, and east (Fig. 34; Pl. 15A). On the west, the cliff drops abruptly 200 m into the Bay of Mirabello (Pls. 1B, 2A). The blocks of the wall are large pieces of gray crys2 talline limestone, averaging 0.60 m ; some are as large as 1.00 m long. The wall forms an oval shape, encircling and supporting a number of rooms. Five rooms could be distinguished (Fig. 34). An entrance is in the north, where two upright orthostats form a doorway. Deteriorating mudbrick is apparent throughout the building. A pithos base, in situ, is exposed in the site’s northeast corner. The site commands a view of the entire Bay of Mirabello, Pseira island to the north, and the Kambos to the east. Mount Chomatas at this north end is a steep, eroded limestone slope of over 30

degrees (east), supporting only sparse garigue and some wild olive and lentisk. The precarious location and instability of the limestone slope are suited to the recent pastoral use of the area. The position itself is topographically defensible and strategic, while the circuit wall on the east side is a fortification. A north-south line was collected on the terrace. The pottery is primarily LM I (Fig. 34). Fabric type I is 92 percent of the total sample. Type IV is 5 percent, type VI is 2 percent, and type III is 0.80 percent. The coarse fabric profile is typical of LM I sites, where type I is dominant and typical MM types IV, VI, and III are in very low quantities, representing a vestige of early activity.

Artifacts (Fig. 34) 131 coarse-ware sherds. 9.1 KTS 168. Conical cup base. D. 0.04. Mediumcoarse pinkish tan fabric (5YR 6/8 reddish yellow); tiny limestone and phyllite grits. MM I–II. 9.2 KTS 169. Jug base. D. 0.05. Tan fabric, tan slip exterior (7.5YR 7/6 reddish yellow); traces of black paint exterior. 9.3–9.5 KTS 170–172. Cooking-pot rims. D.? Coarse fabric type I. Betancourt 1980, 5.

102

KAVOUSI I

9.6 KTS 173. Body frag. of open(?) shape. Th. 0.009. Orange to pinkish tan fabric, slip (5YR 5/6 yellowish red) interior/exterior; dark gray paint (10YR 3/1) exterior: solid circle at start of handle, bisecting stripes, and dot pattern (conglomerate decoration?). LM I.

9.7 KTS 174. Cooking-pot rim. D. ca. 0.30. Coarse fabric type I. Betancourt 1980, 5. 9.8 KTS 175. Pithos rim. D.? Coarse fabric type I. Rim is rounded and forms a drip overhanging body on exterior.

10. Chomatas LOCUS 55 Chomatas, field B; 773158; 0.01 ha. Coastal foothill; 190.50 m AMSL. MM III–LM I.

On the central peak of Chomatas, 500 m west of the plain is a light but homogeneous scatter of coarse fabric type I (Fig. 6). The site has an eastern

aspect but overlooks the Bay of Mirabello as well as the plain. Access to the plain and coast is difficult from this location, and the site might be a temporary shelter. This steep limestone outcrop in Chomatas was probably not exploited agriculturally in the Bronze Age.

11. Chomatas LOCUS 12 Kambos, field B; 783160; 1.5 ha. Inland plain; 32 m AMSL. MM I–II, LM I, V, Ott.

At the base of Mount Chomatas, on the western edge of the Kambos is a widespread, but thin scatter of Bronze Age pottery (Fig. 6). The size and shape of this site—determined during fieldwalking—have been largely obscured by overbank deposits from the Platys River and phyllite alluvium from Mount Chomatas. Dense vegetation and stone debris cover most of the ground surface. Several pithos fragments (Fig. 35: 11.1) were recovered, along with

pieces of mudbrick. The main LM I concentration appears to be an area of ca. 0.25 ha. in the plain directly southeast of Alykomouri. A total of 79 percent of the coarse-ware sample was type I; 3 percent was type III; 7 percent was type VI. The preponderance of material across the site is Minoan; only 7 percent of the sample was Venetian–Ottoman. One thin-section tripod foot is MM I–II.

Artifacts (Fig. 35) 29 coarse-ware sherds. 11.1 KTS 573. Pithos rim. D. >0.40. Fabric type I. MM III–LM IA. Barnard and Brogan 2003, fig. 43.

12. Chomatas LOCUS 59 Chomatas, field B; 780156; 0.14. Hard red-brown fabric, gray core, dense phyllite and sand inclusions; chaff/dung voids in interior and exterior. Interior surface burned black. EM II. 24.6 KTS 340. Bowl base. D. 0.12(?). Brown fabric, gray core; coarse fabric type IX. Light burnishing or smoothed interior and exterior surfaces. EM I–IIA. 24.7 KTS 341. Wall frag. of carinated bowl. Hard, medium-coarse type II. Pinkish brown fabric (5YR 5/4 reddish brown); looks brown to dark brown; surfaces are dark brown to black interior; reddish exterior; exterior is burnished; interior wiped or smoothed. FN. 24.8 KTS 342. Goblet rim. D. 0.10. Fine tan fabric (7.5YR 7/6 reddish yellow); interior/exterior surfaces are lustrous red (10R 4/8); exterior darkening to black at center. Vasiliki ware. EM IIA–B. Betancourt et al. 1979, 36, type VIIA, fig. 11:1–4. 24.9–24.11 KTS 343–345. Bowls with perforation around the rim. D.? Red-brown to brown medium-coarse type IV. FN–EM. 24.11 sampled, P.M. Day: Kavousi 93/68. Pendlebury, Pendlebury, and Money-Coutts 1935–1936, 34, figs. 8:124, 9:125. 24.12 KTS 346. Chalice rim. D. ca. 0.14. Fine, very hard, red to pinkish orange fabric (2.5YR 4/4–5/4 reddish brown), burnished interior and exterior; interior surface burned black. EM I. 24.13 KTS 347. Bowl rim. D. 0.18. Fine orange fabric, tiny white inclusions. 24.14 KTS 348. Jar or bowl rim. D. 0.14. Orangebrown fabric, coarse fabric type VI with mica. MM I–II. 24.15 KTS 349. Conical base of a chalice or bowl. D. 0.102. Semifine tan fabric (7.5YR 7/6 reddish yellow), pink to pinkish brown core (2.5YR 6/4 light reddish brown); phyllite, granodiorite, and quartz sand inclusions; dull brown paint exterior; worn. EM I–II.

24.16 KTS 350. Bowl rim. D. 0.20(?). Hard redbrown fabric; Coarse fabric type IX. Burnished interior, rough, wiped exterior. EM I. 24.17 KTS 351. Bowl rim. D.? Brown–reddish brown fabric, smoothed exterior surface, scored interior. EM I. Warren and Tzedakis 1974, 323; Pendlebury, Pendlebury, and Money-Coutts 1935–1936, 28. 24.18 KTS 352. Bowl rim. D.? Brown–gray to brown fabric; medium-coarse type IV fabric. Smoothed exterior; interior smoothed or lightly burnished. EM. 24.19 KTS 353. Bowl rim. D. and stance uncertain. Brown–gray to brown fabric; medium-coarse fabric type IV. Smoothed or lightly burnished surfaces; worn. EM. 24.20 KTS 354. Lekanis rim. D.? Coarse fabric type II. EM II. 24.21 KTS 355. Bowl or goblet rim. D. 0.18(?). Hard, pink fabric, blended pinkish tan to brown core (7.5YR 5/8 strong brown); surfaces worn; interior red paint (2.5YR 4/4 reddish brown); exterior is a glossy pinkish orange (2.5YR 5/8 red). Vasiliki ware. EM IIB. 24.22 KTS 356. Bowl rim. D.? Hard, micaceous, brown fabric; coarse fabric type II; burnished exterior, wiped interior. EM I–II. Warren and Tzedakis 1974, 323. 24.23 KTS 357. Bowl or jar rim. D. 0.10. Dark redbrown type II; blackened exterior; burnished exterior, smoothed interior. FN. 24.24 KTS 358. Bowl rim. D.? Hard, brown fabric; medium-coarse type II; scored or wiped interior; smooth exterior. FN. 24.25 KTS 359. Handle. Orange-red fabric, gray to black smoothed surface; fabric is a medium-coarse type IV. FN–EM I. 24.26 KTS 360. Bowl rim. D.? Dark red-brown; Fabric type IX; smoothed interior. EM I–IIA. 24.27 KTS 361. Bowl rim. D. >0.10. Coarse (type IV), orange-red, phyllite fabric. EM I. 24.28 KTS 362. Bowl base. D. 0.052. Fine, hard, orange fabric (5YR 5/6 yellowish red) with a narrow, dark gray core; some quartz sand and white limestone grits. String-cut base. MM IB–II. Pelon 1970, pl. 11:1, 2; Andreou 1978, figs. 1:12, 3:8, 12:16, 18:13, 19:24. 24.29 KTS 363. Deep bowl rim. D. 0.18. Fine, soft tan to red fabric (7.5YR 7/6 reddish yellow; 2.5YR 4/8); pinkish orange painted band exterior on rim (2.5YR 5/8 red), and streaky dark brown paint on interior. LM IIIB. Popham 1965, fig 4:12. 24.30 KTS 364. Bowl rim. D. 0.18. Soft dark buff fabric (10YR 6/4 light yellowish brown); gray phyllite and sand inclusions. Black paint exterior; black band on rim interior. LM III. 24.31 KTS 365. Cup profile. Rim D. 0.11; Base D. 0.045. Medium-coarse, hard orange fabric (5YR 5/6

GAZETTEER OF ARCHAEOLOGICAL SITES

yellowish red) with sand and red phyllite inclusions; traces of brown-black paint interior/exterior. String-cut base; wall rilled. MM III. 24.32 KTS 366. Amphora rim. D.? Semifine, soft tan fabric (7.5YR 7/4 pink); gray phyllite and sand inclusions; streaky tan paint exterior (7.5YR 7/6 reddish yellow). Finger marks on interior and exterior. MM I–III. 24.33–24.35 KTS 367–369. Carinated cup frags. Fine, dark buff fabric (10YR 7/4 very pale brown); 24.33 is tan (7.5YR 6/6 reddish yellow) with red paint (2.5YR 4/8 red) interior/exterior; 24.34 is rilled above carination; rim D. 0.09 (369). MM I–II. See 68.1 for references. 24.36 KTS 370. Cup frag. Th. 0.003. Fine, dark buff fabric (10YR 7/4 very pale brown); buff slip, dark brown spiral exterior; monochrome brown interior. LM I. 24.37 KTS 371. Stirrup jar shoulder frag. Spout D. est. 0.012. Semifine soft, orange fabric (5YR 5/8 yellowish

109

red) with red phyllites and sand inclusions; red-painted exterior (2.5YR 4/8 red). LM III. 24.38 KTS 574. Bowl rim. D. 0.20. Fine gray ware. Gray-brown core; orange-buff to brown slip; burnished interior and exterior. Granodiorite, quartz, biotite inclusions, suggesting Mirabello-area manufacture. Internally thickened rim. EM IIA. Warren 1972, 155: P24–P26; Wilson 1985, 296–297: P1–P3. 24.39 KTS 575. Bowl or chalice rim. D. 0.18. Burnished. Mottled black-brown. Fabric type II. EM I–II. 24.40 KTS 576. Bowl rim. D.? Brown core; thick, dull gray slip; burnished. Dense white quartz and calcite inclusions. FN. 24.41 KTS 614. Furnace fragment. Max. Pres. Th. 0.016. Phyllite-tempered; chaff voids; two holes preserved. Betancourt et al. 1999, 354.

25. Kambos LOCUS 27 Kamara, field A; 788143; 0.02 ha. Inland plain; 92 m AMSL. MM I–II, MM III–LM I.

At the far southeast side of the Kambos, northeast of the village, 275 m directly east of the olive oil processing plant at Kamara tou Tholou, at the base of Kapsas and the edge of the scree line, is an old olive terrace with a light sherd scatter and architecture (Fig. 6). The architectural remains consist of a single segment of wall running at an oblique angle (NE–SW) to the agro-terrace. The wall is 8.50 m in length and is constructed of large (0.80–1 m long) regular limestone blocks. The pottery is concentrated directly on top of the terrace and on the south side of the wall; however, sherds are scattered in adjacent fields. Visibility in this area is obscured by

dense grass cover and the limestone scree from the mountain front of Kapsas to the east. Coarse fabrics represented are typical of MM–LM I sites: I (35%), VI (12%), IV (47%), and III (6%). A thin oval tripod foot in type IV is MM I–II.

Artifacts (Fig. 39) 17 coarse-ware sherds. 25.1 KTS 489. Hemispherical cup base. D. 0.05. Tan fabric (7.5YR 7/6 reddish yellow), buff slip interior/exterior (10YR 8/4 very pale brown). LM I. Catling et al. 1979, fig. 31:226–227. 25.2 KTS 490. Tripod foot. Coarse fabric type IV. Red, burned exterior. Thin oval section. MM I–II. 25.3 KTS 491. Basin or lekanis base. D.? Mediumcoarse fabric type VI. MM I–II.

26. Kamara Tou Tholou LOCUS 109 Kamara (Kamara tou Tholou); 786140. Inland plain; 80 m AMSL. LM III A–B.

In 1912, Hall reported excavating “seven rectangular chamber tombs,” that contained fragments of human bone and stirrup jars. The exact location of the tombs is not certain, but the alluvial terrace, east of the bridge over the Platys River (immediately

south of the olive factory ca. 0.5 km north of Kavousi village), is a suitable location for the tombs. These “chamber tombs” might have been small stone-built tholoi of the type at Vrokastro, Vronda, and Skala.

Bibliography Hall 1914, 183.

110

KAVOUSI I

27. Chordakia LOCUS 5 Chordakia, field B (Kambos); 782142; 0.04 ha. Inland plain; 52 m AMSL. MM I–II, MM III–LM I.

In Kambos, south of the Platys River, 600–650 m northwest of Kavousi village, and ca. 50 m directly south of the north Kambos-Chordakia field road, are architecture and a dense scatter of pottery (Fig. 6). The site is located south of a field house and pig sty on the south side of the field road ca. 650 m west along the road from the village. The area is an old olive terrace upon which are 1–2 m high stone piles and field walls, which are probably composed of blocks and stones from ancient buildings. Pottery is visible among the stone debris. The architecture consists of a segment of wall constructed of large limestone blocks, and situated parallel and west of a huge stone field wall. The pottery is concentrated around this wall. Fabric types are I (65%), III (5%), and IV (30%). Diagnostics include a semiglobular cup

base, conical cup base, and a rim of a jar. One handmade cup wall fragment is MM IA–B, and a thin oval tripod foot is MM I. The predominance of type I along with the conical cup and ogival cup base suggest a LM I date.

Artifacts (Fig. 39) 20 coarse-ware sherds. 27.1 KTS 508. Hemispherical cup base. D. 0.035. Pinkish tan fabric (5YR 6/6 reddish yellow), gray core (10YR 5/2); a few large white inclusions. MM III–LM I. See 19.1 above for references. 27.2 KTS 509. Conical or straight-sided cup base. D. 0.04. Pinkish orange fabric (2.5YR 6/8 light red), buff slip (10YR 8/4 very pale brown). 27.3 KTS 510. Jar base. D. 0.14. Medium-coarse fabric type III. MM. 27.4 KTS 511. Jar rim. D.? Coarse fabric type I. 27.5 KTS 512. Oval-mouthed amphora rim. D.? Medium-coarse pinkish brown to orange fabric and slip (5YR 5/4 reddish brown–5/8).

28. Chordakia LOCUS 7 Chordakia, Katsoprinos; 776147; 0.35 ha. Inland plain; 52 m AMSL. EM III–MM IA, MM I–II, MM III–LM I, ER, LR, B, V–Ott.

On the far west edge of the Kambos and on the north edge of the field of Chordakia is an eastern spur of Mount Chomatas, called Katsoprinos (Fig. 6; Pls. 5B, 15B). Here is a dense scatter of pottery, covering 0.35–0.50 ha., as well as house walls and a lime kiln. The area has been severely disturbed by field roads and the bulldozing of new olive terraces. Like Hagios Antonios, the area is a band of phyllite, generating deep deposits of silty clay on the backslopes of this southern spur of Mount Chomatas. The site consists of four parallel bulldozed terraces (1–4) descending downslope into the plain. Sherds were collected and recorded separately on each terrace. Collection lines conformed to the terrace contours because the slope and modern features did not allow lines along cardinal points.

A wall is preserved in the west-northwest corner of the site (south edge of terrace 1), where the road along the west side of Chordakia bisects the eastwest road from Kavousi. It has only one face preserved and stands to two courses. The visible portion is preserved to 15 m in length and runs in a southwest-northeast direction. A few Minoan sherds and a MM–LM I terracotta lamp fragment were found on this terrace. The wall is taken to be Minoan. Below this wall and 15 m to the east is a kiln, probably a Byzantine or Venetian lime kiln, built into the deep deposit of phyllite clay forming terrace 2. The cutting of terrace 3 for an olive terrace has produced a section of the kiln, visible from the east. The kiln is a terracotta dome preserved to a height of 3.50 m from the ground level of terrace 3 and has a maximum external width of 3.50–3.70 m and an internal width (diameter?) of 2.50 m. The external width includes a 0.50 m thick wall of burned red clay, dark red to orange-red from the center out. Some stones are visible on the north and south, built into the outside of the red clay wall, but there is no

GAZETTEER OF ARCHAEOLOGICAL SITES

certain evidence of stone construction at the level preserved. No door or stoke hole is evident, and it is possible that a stone lined substructure underlies the present ground surface. Two segments of wall, of LR date (Pl. 15B), are 15 m east of the kiln and are built into the east face of terrace 3; just south of the walls is a deposit of MM pottery visible in a road cutting. One portion of the Roman wall forms the east wall of terrace 3. It is ca. 5–6 m long and contains an apsidal niche in the north end and a stone-built closet or shelf in the south. The wall stands ca. 4 m high from the ground level of terrace 4, but only the upper 1.50–2 m are visible; it is constructed of field stones and concrete; the eight courses visible are alternating thin schist slabs and squared limestone and schist blocks. The apsidal niche is ca. 0.75–1 m in diameter and is built of tile and thin schist pieces broken to form the curvature of the apse. The shelf in the 2 south end is ca. 30–35 m and ca. 0.20 m deep. At the south end of this wall is a second wall, running at a rough right angle to the first, and into the fill of terrace 3. The wall is of Roman construction (field stone and concrete) but does not bond with the north-south wall. The wall is mostly obscured by the terrace fill and slope wash; however, its construction, where visible, is of small field stones rather than regular courses of dolomite, and the lowest portion appears to use mud mortar rather than concrete. It is possible that the wall was constructed in two phases. The eastern-facing closet and apsidal niche indicate that the north-south wall was the west wall (interior face) of some building that once occupied the area of terrace 4. The east-west wall is probably the north wall of some building now buried by the deep phyllite terrace 3. Boyd excavated a “lime kiln, a cistern, and a hydrant, of uncertain date, probably Roman,” at Chordakia in 1900. The cistern is probably site 29 (see below). Immediately south of the east-west wall, is an exposed road scarp 3–4 m high and abutting the east-west wall. One hundred percent of the pottery of this deposit was collected and proved to be MM I–II. Collections were made of single lines on terraces 3 and 4 and 100 percent of the road scarp deposit. On the kiln terrace, 45 percent of the coarse ware is type I, 27 percent is type IV, and 27 percent is type VI. The representative fabrics suggest a MM I–LM I date. The fine ware from the

111

sample included a conical cup base, probably LM I, and a wall fragment of a closed shape with a black-painted band on buff slip. Other artifacts include a triton’s trumpet shell and two Ottomanmodern sherds. A porous limestone door jamb was found immediately north of the kiln. Terrace 4 produced mostly MM fabrics: type VI is 28 percent, type IV is 54 percent, and type XX/XXI is 3 percent; LM I fabric type I is in very low quantities (15%) relative to type VI and IV. The date indicated by the coarse types is again MM I–LM I; the terrace 4 sample also produced sixteen sherds of Late Roman–Byzantine date, including two Roman amphora handles and one dish rim. Comparison of terraces 3 and 4 suggests a concentration of LM I on the upper slopes of Katsoprinos (terraces 1, 2, and 3) and MM on terrace 4. This distribution may, however, be the result of erosion or recent human activity. The MM I–II sherds on the lower terraces might be more obtrusive because the LM I veneer has been stripped away by the bulldozer. The road scarp deposit south of the Roman building provides a purely MM II deposit, in situ, and underlying the kiln terrace; here the distribution is 56 percent VI, 16 percent is type III, and the remaining 29 percent (13 sherds) is a medium coarse type IV. No LM I type I was apparent. The fine wares illustrate a range of dates MM I–LM I.

Bibliography Boyd 1901, 156; Fotou 1993, 102; Sanders 1982, 141.

Artifacts (Figs. 40, 41) 128 coarse-ware sherds; terrace 3: 11 sherds; terrace 4: 72 sherds; road scarp: 45 sherds. 28.1 KTS 216. Conical cup rim. D. 0.11. Pinkish tan fabric (5YR 7/6 reddish yellow); reddish pink to red paint interior and exterior (10R 5/8–4/8). MM I–II. Pelon 1970, pl. 10:7. 28.2 KTS 217. Deep rounded cup rim. D. 0.11. Pinkish tan fabric (5YR 7/6 reddish yellow); dark pinkish orange paint interior/exterior (2.5YR 5/4 reddish brown). MM I–II. Pyrgos II–III. Andreou 1978, figs. 11:21, 12:8–9, 13; Dawkins 1902–1903, 305, fig. 5:b, d. 28.3 KTS 218. Conical or carinated cup rim. D. ca. 0.10. Fine, dark buff fabric (10YR 7/4 very pale brown); finger marks interior/exterior. MM I–II. Pelon 1970, pl. 35:9; Betancourt 1990, figs. 17:192, 18:194, 53:1311; Dawkins 1902–1903, 305, fig. 4:1. 28.4 KTS 219. Jar rim. D. 0.11. Light orange-pink fabric/slip (5YR 8/4 pink), black paint exterior.

112

KAVOUSI I

Collared jar, thickened interior. MM I–II. Betancourt 1990, fig. 39:803. 28.5 KTS 220. Conical cup base. D. 0.05. Pink fabric (2.5YR 6/6 light red); red paint on interior surface (2.5YR 4/8 red). MM I–II. Pelon 1970, pl. 35:9. Dawkins 1902–1903, 305, fig. 4:1. 28.6 KTS 221. Rounded cup base. D. 0.04. Pinkish tan fabric (5YR 6/6 reddish yellow); white limestone and phyllite grits. Disk base. MM. 28.7 KTS 222. Straight-sided cup. D. 0.05. Pinkish orange fabric (2.5YR 6/8 light red). MM I–II. Betancourt 1990, fig. 21:367. 28.8 KTS 223. Jug/jar base. D. 0.04. Dark red fabric (2.5YR 4/6 red); fine white inclusions. Base is slightly flaring. MM I–II. Betancourt 1990, fig. 44:989, 1005. 28.9 KTS 224. Hemispherical or carinated cup base. D. 0.045. Pinkish tan fabric (5YR 6/8 reddish yellow); black paint interior/exterior. MM II–III. 28.10 KTS 225. Tumbler. D. 0.04. Buff fabric (10YR 8/4 very pale brown). MM IB. Betancourt 1990, figs. 18:225, 21:354–355; Pelon 1970, pl. 35:9. 28.11 KTS 226. Jar rim. D. 0.17. Tan fabric (7.5YR 7/6 reddish yellow); black paint exterior (7.5YR N/0). MM I–II. Similar to Betancourt 1990, fig. 47:1106. Pyrgos II. 28.12 KTS 227. Spout of bridge-spouted jar. Pink fabric (2.5YR 6/4 light reddish brown), dark buff slip (10YR 7/4 very pale brown), washy gray paint exterior (5YR 5/1 gray). MM I–II. 28.13 KTS 228. Spout and wall of bridge-spouted jar. Pinkish tan fabric (5YR 7/6 reddish yellow); traces of black paint exterior; possibly diagonal stripes. MM. 28.14 KTS 229. Bowl base. D. 0.04. Buff fabric (7.5YR 8/4 pink), burnished, with black bands on interior and exterior (7.5YR 3/0). MM III–LM IA. 28.15 KTS 230. S-shaped or bell cup rim. D.? Fine buff fabric, burnished, with black spiral below band on exterior. LM IA. 28.16 KTS 231. Amphora neck/rim. D. ca. 0.11. Tan fabric (7.5YR 6/6 reddish yellow), orange core (5YR 5/6

yellowish red) with white grits and granodiorite inclusions. Fabric type XX/XXI. Finger marks from smoothing slip on surfaces; black band on rim interior/exterior; second band middle of the neck interior and elliptical blob on neck exterior; small elliptical cavity impressed in the fabric when wet (potter’s mark) on exterior base of neck. EM III–MM IA. 28.17 KTS 232. Lamp frag. Base D. 0.06; Max. Pres. H. 0.036. Coarse fabric type IV; dark brown slip (7.5YR 5/4 brown); dark brown-black residue interior. MM. Pelon 1970, 60, pl. 36:7. 28.18–28.20 KTS 233–235. Fragments of jugs/jars. Th. 0.004–0.006. Tan fabric (7.5 YR 7/4 pink), tan slip (7.5YR 7/4 pink), black painted stripes exterior. MM I–II. Pelon 1970, pl. 11:6. 28.21 KTS 236. Jar base. D. 0.24. Coarse fabric type IV. Cooking pot. 28.22 KTS 237. Basin rim. D. 0.38. Coarse fabric type III; pink fabric (2.5YR 6/6 light red), pinkish tan slip (5YR 7/6 reddish yellow), dense mica and granodiorite inclusions. 28.23 KTS 238. Jar or cooking-pot base. D. 0.18. Coarse fabric type VI. 28.24 KTS 239. Basin or cooking-pot rim. D.? Coarse fabric type IV. 28.25 KTS 240. Basin or cooking-pot rim. D.? Coarse fabric type VI. 28.26 KTS 241. Tripod foot. Thin oval section. Coarse fabric type VI. 28.27 KTS 242. Jar base. D. 0.12. Coarse fabric type IV. 28.28 KTS 243. Jar rim. D. ca. 0.14. Coarse fabric type VI. 28.29 KTS 244. Base of closed shape(?). D.? Coarse fabric type III. 28.30 KTS 245. Basin or cooking-pot rim. D. 0.20. Coarse fabric type II. 28.31 KTS 246. Basin rim. D. 0.32. Coarse fabric type III.

29. Chordakia LOCUS 94 Chordakia, Katsoprinos; 773148. Coastal foothill. 70 m AMSL. LR–B.

In the northwest corner of Chordakia, on the southern slope of the Katsoprinos spur, some 100 m northwest and slightly upslope from site 28, is a Byzantine cistern (Fig. 6). The cistern is subterranean

and is built into a phyllite terrace exactly above and north of the northernmost revma descending into Chordakia from Chomatas. The cistern is built of schist and concrete and lined with pink hydraulic cement. The plan is rectangular: 1.70 m wide, 2.93 m long, and ca. 1.50 m high at the apex of its vaulted roof. There is no evidence of a feeding channel, and run-off

GAZETTEER OF ARCHAEOLOGICAL SITES

from the hill slope must have fed the cistern from an opening in the roof. The vault is partially preserved. There is no associated pottery visible. This is probably the cistern excavated by Boyd in 1900.

113

Bibliography Boyd 1901, 156.

30. Chordakia LOCUS 95 Chordakia, field A; 776146. Coastal foothill; 52 m AMSL. R?

West of Site 28, ca. 25 m., in the north-northwest corner of Chordakia field, is a stone cut cist measuring 2.10 x 1.50 x 0.40 m deep (Fig. 6). The date is uncertain: however, it might be associated with the Roman walls and cistern at Katsoprinos. Boyd excavated at “Chordakia,” two rock-cut graves of similar dimensions. One tomb was empty, and the second

contained four skulls and other human bones, and pieces of at least five vases; in the bottom of one jug, there was an iron ring with a bezel, from which a stone had been removed, and a coin of the Emperor Gallienus (A.D. 253–268). The length of Boyd’s tombs (1.90 and 2.15 m) roughly agrees with the dimensions of the cist at site 30.

Bibliography Boyd 1901, 157; Fotou 1993, 102.

31. Kolonospelio or Theriospelio LOCUS 89 Chomatas, field A (Kolonospelio); 768152. Coastal foothill; 100 m AMSL. N?, EM I–III.

Kolonospelio or Theriospelio is a cave on the steeply sloping west side of Mount Chomatas. Facing the Bay of Mirabello, the cave’s entrance is ca. 200 m northeast of the promontory of Chrysokamino (Figs. 6, 18; Pl. 16A–B). It has a small, low mouth (ca. 1.15–1.20 m high x 3–4.70 m wide), which widens as one enters, but drops in height to no more than 0.80 m for a distance of ca. 9 m. A terrace in front of the cave entrance is ca. 10 m wide and 3.30 m deep. Beyond the western edge of the terrace, the slope drops off to 40–45 degrees. About 10 m past the mouth of the cave on the interior, the floor slopes down into a wide chamber with stalagmites and stalactites. In the far eastern recess of the chamber is a pool of fresh water. No artifacts were apparent (to the author) on the ground surface of the interior or around the mouth on the exterior slope. According to Mosso, Hazzidakis discovered one EM II cup and a sherd with white decoration on a dark ground; the latter Mosso calls MM III. Mosso also mentions that Hazzidakis found many other undiagnostic sherds.

Edith Hall excavated the cave in April 1910, probably after the visit of Hazzidakis. Although Hall does not name the cave, her detailed description of the site and its location, relative to the metal-working site of Chrysokamino, are certain indications that the cave is Kolonospelio. These excavations were never published, and no mention is made of the cave in any published reports. Two private letters to Hall’s family give details of the cave’s location and the excavation. First mention is during a trip by boat around the east side of the Bay of Mirabello to Pseira: “Later we crossed over again to the mainland to the place where they smelted bronze in Minoan times and where there is a big cave like those I dug in Western Crete. I am to go there again tomorrow, with Nikolaos and six men to clear this cave.” In Hall’s next letter of May 1, she describes the site as being “a steep rocky mountain side,” adding “[t]he cave had a very small opening so that we had to enter on all fours, but once inside it was fairly lofty, supported by splendid stalactite columns.” Later, after describing the difficulties of digging by lantern light, she remarks, “We got some very good pottery.” No mention is made of features and no description is given of the pottery. Also, no mention

114

KAVOUSI I

is made of human bone, or that the cave might be a burial location. It is possible that this is an EM–MM burial or cult site.

Bibliography Becker and Betancourt 1997, 109; Betancourt et al. 1999, 351; Foster 1978; Hall 1910a; 1910b; Hawes et al. 1908, 33; Mosso 1910, 219.

32. Chrysokamino (Metallurgical Site) LOCUS 88

Schachermeyr 1938, 472–473; Stos-Gale 1993, 124; Zois 1993, 340–341.

Chomatas, field A (Chrysokamino); 765152; 0.08 ha. Coastal foothill; 50 m AMSL. FN–EM II, EM III–MM IA, MM I–II?, V.

Artifacts (Fig. 42)

On the small coastal headland, west-northwest of Chordakia, on the west side of Mount Chomatas, jutting out into the Bay of Mirabello, is an isolated scatter of slag and terracotta furnace fragments covering an area ca. 0.08 ha. (Figs. 6, 18; Pls. 16A, 17A) Excavations by Betancourt in 1996–1997 have confirmed the date and function of the site that had been derived earlier from thermoluminescence samples. The site is a copper-smelting operation of FN–EM III date, consisting of an apsidal building surrounded by slag and furnace fragments. Surface collections recovered several basin fragments of EM III–MM II date. Fabric type II is Prepalatial, and type XX/XXI is EM III–MM IA in date. Fabric type III is Protopalatial.

Bibliography Betancourt 2000a; 1998; Betancourt and Floyd 2000–2001; Betancourt, Muhly, and Floyd 1998; Betancourt et al. 1999; Branigan 1968, 50–51; Hall 1910a; Mosso 1910, 289–290; Nakou 1995, 17;

32.1 KTS 580. Basin rim. D. 0.40. Fabric type II. EM III–MM IA. Momigliano 1991, 252; Momigliano and Wilson 1996, 51–52: P189–191. 32.2 KTS 581. Basin or bowl rim. D. >0.25. Fabric type II. EM III–MM IA. Momigliano 1991, 252; Momigliano and Wilson 1996, 51–52: P189–191. 32.3 KTS 582. Larnax or basin rim. Fabric type XX/XXI. EM III–MM IA. Haggis 1996c, 655: HT22– HT23. 32.4 KTS 583. Larnax or basin rim. Fabric type XX/XXI. EM III–MM IA. Haggis 1996c, 655: HT22– HT23. 32.5 KTS 584. Larnax or basin rim. Fabric type III. MM I–II. Haggis 1996c, 655: HT22–HT23. 32.6 KTS 585. Basin or larnax base. D.? Fabric type II. Dark red clay; sandy inclusions; few granodiorites visible. Rough bottom; smoothed interior and exterior. Exterior surface is smoothed or shaved at the base. EM III–MM IA. 32.7 KTS 586. Larnax or basin base. Fabric type XX/XXI. EM III–MM IA. 32.8–32.10 KTS 587–589. Terracotta furnace chimney fragments. Th. 0.012–0.014. Chaff voids. Vitrified interior surfaces. Holes range from 0.02–0.04 in diameter.

33. Chomatas LOCUS 49 Chomatas, field A; 771148; 0.30. Coarse fabric type IV. 36.8 KTS 258. Tray. D.? Whole profile. Coarse fabric type IV. 36.9 KTS 259. Wall frag. of amphoroid krater. Fine, hard, buff to tan fabric (7.5YR 8/4 pink); glossy burnished surface; tiny red and black phyllite inclusions; octopus decoration in red-brown paint exterior. LM IIIA:2–B. Kanta 1980, fig. 52:3. 36.10 KTS 260. Bowl or kalathos base. D. 0.052. Hard orange fabric (5YR 5/6 yellowish red), pinkish brown core (5YR 5/3 reddish brown); red phyllite and fine sand inclusions. LM III. 36.11 KTS 261. Miniature tripod base. D. 0.036. One foot preserved. Red fabric (2.5YR 4/6 red); hard, medium-coarse fabric type IV. 36.12 KTS 262. Kylix rim. D. 0.13. Everted rim and carinated profile. Pinkish orange fabric (2.5YR 5/6 red); fine, hard fabric with white sand and red phyllite

117

inclusions; black (dark brown) paint exterior and on band at rim interior. LM IIIB. Popham 1965, fig. 7:47. 36.13 KTS 263. Cup rim. D.? Dark buff fabric (10YR 7/4 very pale brown); traces of red paint (2.5YR 4/6 red) interior and exterior. 36.14 KTS 264. Base of a pyxis. D. 0.10. Dark buff to buff fabric and slip (10YR 7/4 very pale brown); two bands exterior above base red (10R 4/8), same red monochrome interior. The painted interior suggests an open shape. Kanta 1980, figs. 22:6, 33:6, 129:5. LM III. 36.15 KTS 265. Loom weight. H. 0.08; W. 0.068; Th. 0.025. Coarse fabric type IV. 36.16 KTS 266. Bowl. Base D. 0.053; Rim D. 0.102. Pinkish orange fabric (2.5YR 5/8 red); fine sand and red phyllite inclusions. LM IIIA–B. Sackett, Popham, and Warren 1965, fig. 14:P5. 36.17 KTS 267. Conical cup. Base D. 0.04; Rim D. 0.09. Orange fabric (5YR 5/6 yellowish red), fine white and dark grits and red phyllite inclusions. Rim is slightly in-curving. LM I. 36.18 KTS 268. Bowl. D. 0.11. Hard red fabric (2.5YR 4/6 red); sand and phyllite inclusions. LM III. 36.19 KTS 269. Conical cup. D. 0.038. Brown fabric (7.5YR 5/6 strong brown); same fabric as 36.16–18. Profile suggests LM III date. Sackett, Popham, and Warren 1965, fig. 14:P1. 36.20 KTS 270. Conical cup or bowl. D. 0.04. Red fabric (2.5YR 4/6 red); fine sand inclusions. 36.21 KTS 271. Conical cup or bowl base. D. 0.041. Red fabric (10R 4/4); gray core (10YR 6/2); fine white and dark grits. 36.22 KTS 272. Bowl or kalathos base. D. 0.05. Pinkish orange fabric (2.5YR 5/8 red); pinkish tan slip (5YR 6/6 reddish yellow); sand inclusions and some phyllite visible. Wet-smoothed exterior surface; stringcut base. 36.23 KTS 273. Conical cup or bowl base D. 0.041. Dark red fabric (10R 4/4); fine sand inclusions. 36.24 KTS 274. Bowl base. D. 0.045. Pinkish tan fabric (5YR 6/8 reddish yellow); wet-smoothed exterior; fine sand inclusions. LM III. MacGillivray et al. 1987, figs. 2–4. 36.25 KTS 275. Bowl base. D. 0.05. Tan fabric (7.5YR 6/6 reddish yellow); soft surface, core, with sand inclusions. LM III. See 36.24 for references.

37. Kephalolimnos LOCUS 2 Chordakia, field A, Kephalolimnos; 774144; 0.42 ha. Coastal foothill; 52 m AMSL. LR.

On a hillock, ca. 1 ha. in size, on the south side of Chordakia (field A), is the location Kephalolimnos. The site is a Roman scatter of pottery, roof tiles, and

118

KAVOUSI I

stone tools covering an area of 0.25–0.42 ha., on the mound-like rise above and south of Chordakia, field A (Fig. 6; Pls. 5B, 18B, 19A). The mound appears to be human deposition supported on the north, south, and east by terrace walls and possible ancient house walls. The hill and site are bordered on the west and southwest by the field road entering Chordakia from the east (from Kavousi), and extending along the southwest side of the field. The area is today planted in olives. At the center of the hill, ca. 30 m east of the Chordakia field road is an olive-crushing bed or trapetum mortarium (ca. 1.80 m external diameter; ca. 1.50 m internal diameter) of local limestone (Pl. 18B). About 13 m northwest of the crushing bed is a grinding stone (orbis) (ca. 0.97 m diameter), worn on the exterior (Pl. 19A). Nearby is a saddle quern. The pottery and tile scatter is light (10 sherds/5 x 2 m unit). The material consists of jug, pithos, bowl, and dish fragments and cobblesize ground stone tools. Immediately east of the crushing bed and down the east slope, there are ancient walls (north-south

orientation) and a filled-in well. The remains would indicate a hamlet involved in exploitation of the surrounding phyllites for olive oil production.

Artifacts (Fig. 46) 37.1 KTS 247. Bowl rim. D. 0.15. Fine, hard, pinkish orange fabric (2.5YR 5/8 red); two parallel grooves on top of rim. African red slip ware; Hayes form 93. 5–6th c. A.D. 37.2 KTS 248. Jar rim. D. 0.16. Pink fabric (2.5YR 6/6 light red); gritty. Rim everted and hooked. Coldstream and Huxley eds. 1972, fig. 52:65–66; possibly earlier (4–5th c. A.D.): Hayes 1983, 126, fig. 8:93; Sackett 1992, pl. 192:40–44 (3rd c. A.D.). 37.3 KTS 249. Bowl rim. D. ca. 0.22. Fine pink fabric (2.5YR 6/6 light red); stance uncertain. Coldstream and Huxley eds. 1972, fig. 51:27 (6–7th c. A.D.). 37.4 KTS 250. Bowl rim. D. ca. 0.20. Gray-brown core, pinkish orange slip (2.5YR 5/6 red); sandy inclusions. Thickened rim. 5–7th c. A.D. Late Roman C (form 6). Phocaean Red Slip Ware. Hayes 1972, 341, fig. 70; 408, fig. 92:2.

38. Agriomandra LOCUS 45 Agriomandra; 763144;