Marsa Matruh I: The Excavation (Prehistory Monographs) [Illustrated] 1931534004, 9781931534000

Citation preview

MARSA MATRUH I THE EXCAVATION

Oric Bates

MARSA MATRUH I The Excavation The University of Pennsylvania Museum of Archaeology and Anthropology's Excavations on Bates's Island, Marsa Matruh, Egypt

1985-1989

Donald White

with contributions by

Rita Gardner and Linda Hulin

PR EHI STORY

Mo

OGRAPH S

1

Published by THE

l

T!TUTE FOR AEGEAN P REH ISTORY ACADEM I

2002

P RESS

Design and Production The Institute for Aegean Prehistory Academic Press Printing Sun Printing Co., Philadelphia Binding Hoster Bindery, Philadelphia

Library of Congress Cataloging-in-Publication Data White, Donald. Marsa Matruh I : the excavation : the University of Pennsylvania Museum of Archaeology and Anthropology's excavations on Bates's Island, Marsa Matruh, Egypt, 1985-1989 I Donald White with contributions by Rita Gardner and Linda Hulin. p. em. -(Prehistory monographs ; 1) Includes bibliographical references and index. ISBN 1-931534-00-4 (alk. paper) 1. Marsaa Maotr~uoh (Egypt)-Antiquities. 2. Excavations (Archaeology)-Egypt-Marsaa Maotr~uoh. I. Gardner, Rita, 1955- II .. Hulin, Linda. III. Title. IV. Series. DT73 .M254 W45 2002 932-dc21 2002004915

Copyright © 2002 TH E IN ST ITUTE FOR AEGEAN PREHI STORY A CADEM IC PRESS

Philadelphia All rights reserved Printed in the United States of America

IN MEMORY OF LIDIANO BACCHlELI, CLAUDIO FRIGERO, AND JOHN LLOYD

AND

IN RESPECTFUL HOMAGE TO 0RJC BATES

Time, like an ever-rolling stream, Bears all its sons away.

TABLE OF CONTENTS LIST OF FIGURES IN THE TEXT ............ . ......................... ix LIST OF PLANS ................................................... xi LIST OF PLATES ........... ................ .... .. ................. xiii PREFACE AND ACKNOWLEDGMENTS ............................... xvii BIBLIOGRAPHICAL ABBREVIATIONS ................... ........ .... xxi MINOR ABBREVIATIONS ......................... . .. . ..... . . . .....

XXV

CHAPTER I. The Setting,

Donald White .. ... . .... ... ......... . .. .. ... ... . .......... 1 CHAPTER 2. Introduction to the Island's Excavations,

Donald White . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 CHAPTER 3. Environmental Morphology and History of the Island and Adjacent Lagoon Area,

Donald White and Rita Gardner ... . . .... ......... . . . ........ 23 CHAPTER 4. The Island's Late Bronze Age Occupation,

Donald White ......................... .. ................ 35 CHAPTER 5. Area VI: The Great Ridge, Bates's Libyan Cemetery,

Donald White and Linda Hulin .............................. 85 CHAPTER 6. The Island's Historical Period Occupation,

Donald White . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 INDEX ... . ... ... ............................................... . 121 PLANS ............................... . . ... . ................ ... .. 127 PLATES ......................... .... ............. . ... .. .. . ..... . 133

LIST OF FIGURES IN THE TEXT Figure I: I . Map of the principal ancient coastal site between A lexandria and the Libyan Plateau .......................... .. .... ........ .. .. xxvi Figure 1:2. Plan of Marsa Matruh West. ................ .. . ... .................... 2 Figure 1:3 . Fourtau 's 1904 sketch plan of Marsa Matruh ............................. .4 Figure I :4. Fourtau' sketch plan of Villa with Internal Colonnade along the south shore of Matruh 's West Lagoon . ..... . ........................ .4 Figure I :5. Gruoc's 1820 plan and elevations of Gasr ei-Fok, south of Ras ei-Hikma. . ...................................................... 6 Figure 1:6. Detai I of Barth's route map along the Marmarican Coast. .......... .. .. .. .... 8 Figure I :7. Bates 's map of Marsa Matruh and its surrounding .......................... 9 Figure 4:1. Plan of the island's Northern and Southern Clusters of Late Bronze Age walls and collapsed rubble .................... .. . ... .. . .... 38 Figure 4:2 . Plan of the Sponge-Divers House (S I 0 I) ......... . .. .. .. . ... .... .. ...... 39 Figure 4:3. Plan of G5-1V/SW Test. ........................................... .40 Figure 4:4. Cross-section of the south end of G5-JV /SW Test. ..... ... ..... .. ....... . .41 Figure 4:5. Plan of the H4-III test trench ..... ...... ........... .................. .43 Figure 4:6 . East-west section of fill across the north end of H4-III .................... .43 Figure 4:7. Plan of the Southern Cluster's and orthern Cluster's trenches and walled remains. . .... .. ................................ . ... .44-45 Figure 4:8. Plan ofH5-T/H5-fl and H5-Ill ........................ ..... ........ . . .47 Figure 4:9. East-west section, H5-I/H5-II,

orth Balk ........................ . ..... .48

Figure 4: I 0. North-south section, H5-III, West Balk. . .............................. 52 Figure 4:11. East-west section, H5-III, H5-III/SW, South Balk. . ................ . ... .. 53 Figure 4:12. East-west ect ion, H5-III/ W, South Balk. . ............................. 54 Figure 4:13. Plan of 16-1/II at the 2.1 - 2.5 level. ......... . .... .. ......... . ... .. .... 55 Figure 4: 14. Plan of 16-1/II at the 3.1 - 6.1 level. . . ................................. 56 Figure 4: 15. East-west cross-section of the northeast corner of 16-I/ll ................. .. 56

LIST OF FIGURES IN THE TEXT

Figure 4:16. Plan of the Southern Cluster. ............ . .......... ....... . . ....... 58 Figure 4:17. Plan ofF4-III, Sl07 .............................................. 60 Figure 4:18. North-south section, F4-III, East Balk. ...... . ............ ......... .. .. 61 Figure 4:19. East-west section, F4-III, North Balk. . ...... ..... ... ................. 62 Figure 4:20. North-south section, E4-IIIE, West Balk. .............................. 65 Figure 4:21. Plan ofE4-III and Room Sl02 ...................................... 67 Figure 4:22. North-south section, E4-III, West Balk. .......................... ..... 68 Figure 4:23. East-west section, E4-III, South Balk. ................................ 69 Figure 4:24. North-south section ofD4-I, West Balk... ............. ..... ........... 71 Figure 5: 1. Bates 's sketch plans and sections and contents of his tombs A. I and A.2 ........ . .. ........ .................... ............. ..... 90 Figure 5:2. Late Bronze Age stone and terracotta artifacts from Bates's Area VI Libyan Cemetery. . .... . ....... ... ........ .. ............ ....... ... 91 Figure 5:3. Plan of Area VIB, Bedouin(?) pits or hearths ............................ 95 Figure 5.4. Examples of Fabrics A and B. . .......... .. .......................... 98 Figure 5:5. Examples of Fabrics A, B, and C. .....................................99 Figure 5:6. Examples of Fabrics A, B, and C, and a modern jar..... ... ...... ......... I 0 I Figure 6: I. Plan of 18-JII/S. . . . . . . .. ......... ... .. .. .. ....................... I 09 Figure 6:2. North-south section, 18-III/S, East Balk ................................ 110 Figure 6:3. North-south section, 18-III/S, West Balk. . ........... ..... . ........... . Ill Figure 6:4. Plan ofG6-I. ................................................... 113 Figure 6:5. East-west section, 06-I, South Balk. ................................. 114 Figure 6:6. Plan of 18-I/Il .......................... .. .... . .... .. ............ 116 Figure 6:7. East-west section, J8-IIII, North Balk. . . .... .......................... 116 Figure 6:8. Bates's sketch plans of the island and its S I 01 Sponge-Divers House. . .... ... 119 Figure 6:9. Bates's plan and cross-section of the S 101 Sponge-Divers House. . ....... ... 120

X

LIST OF PLANS Plan I. Topographic plan of Bates's Island showing location of excavated walls and structures. Plan 2. Plan of Bates's Island showing location of excavated trenches. Plan 3. Map of the eastern end of the East Lagoon showing relati onship of Bates 's Island (Area T) to adjacent shores features (Areas ll, fll , and V). Plan 4. Redrawing, with additions, of 1938/ 1941 1:25 ,000 scale plan (Dept. of Surveys and Mines, Sheet no. 2) of Matruh harbor and lagoons east to RasA lam ei-RC.m, locating Areas I-IX. Area IV, Unun e l-Rakham area , and the ridge south of Umm e1-Rakham and west of Wadi el-Samad all Iie west of the mapped area.

LIST OF PLATES Plate 1 A. Town ofMatruh in 1913. B. Bedouins at Matruh. Plate 2 A. Vintage houses from o ld Matruh on the eastern side of town. B.World War II German seaplane, which crashed on Matruh 's quay. Plate 3 A. Satellite photo of the salt pans on the eastern end of the Eastern Lagoon system. B. Roman site on Coastal Ridge west of Matruh 's harbor. Plate 4 A. Group of surface sherds, including Cypriot White Slip II. B. Cratered interior of the Sponge-Divers House (S 10 I) before excavation. Plate 5 A. Detail of the interior of the Sponge-Divers House. B. Sample ordnance and shrapnel. Plate 6 A. Reconnaissance satellite photo of the coast from Marsa Matruh to RasA lam ei -RGm . B. Reconnaissance satellite photo of the First Eastern Lagoon . Plate 7 A. Distant view of the partially flooded sahel from Area V B. Closer view of the lagoon . Plate 8 A. Oblique aerial photograph of the island . B. Sherds lying on the submerged surface of the island's sandbar. Plate 9 A. Submerged line ofblocks (Sil l ). B. Area IX pumice layer. Plate 10 A. Detai I of pumice in front of the Area IX. B. Ras Alam ei-Rum. llA . Plate 11 A. HS-1 surface scatter of pumice. B. RasA lam el-Rum with marshy plain separating sea and barrier beach from easternmost lagoon . Plate 12 A. Coastal ridge running east from the west end of the First Eastern Lagoon. B. Deposit of wind-blown sand exposed by a bulldozer cut into the south face of the eastern coastal ridge opposite Bates's Island. Plate 13 A. Coastal ridge west of the main harbor. B. Bates's Island . Plate 14 A. Bates 's Island (Area V). B. View of Bates 's Island . C. Bates 's Island.

.ril'

LIST OF PLATES

Plate 15 A. General view of the Cyrenaican oasis village of Al-Giarabub. B. Layers of aeolianite on the west face of Bates's Island. Plate 16 A. Northern end of the island before excavation. B. Island 's sloping low ground across its sheltered eastern side. Plate 17 South end ofG5-IV/SW Test. Plate 18 The Sponge-Divers House (S I 0 I) with cleared surface. Plate 19 The island's Southern Cluster. Plate 20A. H5-I/Il walls, Sl18, S121, and S126. B. H5-l/II walls, Sll8, S121, and S126. Plate 21 A. H5-l, 2.1 shell deposit. B. Detail of H5-I, 2.1 shell deposit. Plate 22 A. H5-Ill B. H5-III C. H5-III D. H5-III

and the corner of the Sponge-Divers House (S I 0 I). with S130, the Arab wall (Sl28), the hearth (Sl32), and two small bins (S134a, b). with S 128, S 130, and the features S 132 and S 134 and wall (S 131) in between. with S 128, and the features S 132 and S 134.

Plate 23 A. H5-III's hearth (S 132). B. The bins (S 134 a and b). Plate 24 A. 16-l/ ll cleared to the 2.1 level. B. 16-1/ll cleared to the 3.1 - 6.1 levels. Plate 25 A. Bin (S114) in the southwest corner ofi6-1/II. B. Hearth (S I 17) in center, west of 16-1/II. C. View ofthe Sponge-Divers House (S101) and the island's Southern Cluster. Plate 26 A. The Southern Cluster. B. The Southern Cluster. Plate 27 F4-lll, showing S107. Plate 28 A. Interior of S 107 with a bin (S 138) and an oven (S 13 7). B. Modern tabun used in Baharia Oasis home. Plate 29 A. E4-III, E4-IV, and Room S I 02. B. E4-lll and Room S I 02, partially excavated. Plate 30 A. E4-Center, E4-fi/E, and E4-Il/E, West Balk with S 122a-c ramp. B. E4-II/E. Plate 31 A. E4-JII/W Balk and S 124 wall. B. D4-1/ll, S 119 workshop, before removal of West Balk. Plate 32 A. D4-l/ll, S119 workshop, before removal of West Balk. B. D4-1/ ll, S119 workshop, before removal ofWest Balk. Plate 33 A. D4-l/II , 4.1: metal-production waste products. B. Contemporary North Indian peasant's hut. Plate 34 A. Matruh's green belt from the area of Bates's Libyan Cemetery. B. The site of Cemetery A.

LIST OF PLATES

XV

Plate 35 A. Cemetery A after clearing. B. Tomb A.l. C. Tomb A.2. Plate 36 Finds from Bates's Tomb A.l. Plate 37 A. Area II. B. Area II. Plate 38 A. Areas VIIA and B. B. Surface spill of Shell-Tempered Ware B sherds, Area VIIA. Plate 39 A. Area IX. B. Area VI. C. Area VIB, Bedouin (?) hearths. Plate 40 A. Wadi Aghiba west of Ras Umm el-Rakham. B. Uncataloged sherd of Shell-Tempered Ware A. C. Uncataloged sherd of Shell-Tempered Ware B. Plate 41 A. I8-III/S. B. 18-III/S. C. 18-III/S. Plate 42 A. G6-I. B. G6-I's ?-shaped "storage containers" Sl13d and Sll3e. Plate 43 A. Underwater wall feature (S 111 ). B. Underwater slot feature (Sl40). Plate 44 A. J8-l/II. B. J8-IIII's wall Sl09a. C. J8-IIII's wall Sl09a. Plate 45 A. Interior of the Sponge-Divers House (S 101) before excavation. B. The west wall and center interior ofthe Sponge-Divers House (S101) before excavation. Plate 46 A. Interior of the Sponge-Divers House (SlOl) with G5-IV/ SW Test and H4-III partially excavated. B. The south half of S 101 's interior with G5-IV /SW Test. Plate 47 A. The northwest exterior corner and the west wall of the Sponge-Divers House (S101). B. The leveling bed (?)of sand mixed with shells lying over the west wall of the Sponge-Divers House (S101).

PREFACE AND ACKNOWLEDGMENTS The University of Pennsylvania Museum of Archaeology and Anthropology's Expedit ion to Marsa Matruh, to give it its full title, came about as an indirect consequence of the abrupt and, as matters turned out, unplanned conclusion of the museum 's work in 1981 in eastern Libya where severa l ofthe present authors had been previously engaged in archaeological research stretching over more than a decade.; Searching for something that might enable the museum to build on its Cyrenaican experiences, my attention was drawn to the pioneering activities at the outbreak of World War I of Harvard University's Oric Bates, author of the classic study of the native eastern Libyan population,;; at the site of Marsa Matruh on the northwest coast of Egypt. ;;; What was initially envisioned to be an extension of the University of Pennsylvania's earli er research on the historical Greeks along the northeast African littoral quickly turned into an effort to revive Bates's promising but large ly neglected work on the pre-Greek, Late Bronze Age occupation of the Matruh area. Our attention soon came to rest in particular on the tiny island now bearing his name at the end of Matruh's first eastern lagoon . A minuting of how the project unfolded between 1984 and 1989 is found in Chapter 2 and needs no further elaboration here. On the other hand, I am gratified to be able to express at long last my thanks to the many persons, some now deceased, for the project's conceptualization, its planning and organization in Egypt as well as back in Philadelphia, its financial support, and finally for its publication. It is fitting to first take note of the representatives of the Egyptian Antiquities Organization or what is now the Supreme Council of Antiquities in Cairo and Marsa Matruh during the years in which the project ran. This includes the individual past chairmen of the SCA, Dr. Ahmed Kadry, Dr. M.A.H. Nur el-Din, Dr. Sayed Tawfik, Dr. Muhammad Ibrahim Bakr, and Dr. Ali Hassan. I also wish to give special thanks to Dr. Kamal Fahmy, then Director of Excavations for Lower Egypt, for his personal interventions and enlightened assistance throughout the project's duration, and Dr. Zahi Hawass who first implanted in me the idea of work in Egypt in the mid 1970s. In Marsa Matruh every representative of the SCA provided contributions that turned out to be indispensable to the success of our work, but I should like to mention in particular Messrs Feisal As-Mahwy, Ezzat Osman ei-Hamhmy, Tarek Mohamed Farid, and Atif Abdel-Daym. The last two provided invaluable personal assistance during our final 1989 field season and in the study seasons that followed during the early 1990s.

xviii

PREFACE AND ACKNOWLEDGEMENTS

lt goes without saying that without the active sponsorship, collaboration, and encouragement of the American Research Center in Egypt, a project like ours would never have succeeded. I am, therefore, especially pleased to offer thanks to ARCE's US-based Executive Directors, Paul Walker and, more recently, Terence Waltz, and to the ARCE-Cairo directors, Richard Verdery, Robert Betts, Iliya Harik, and Mark Easton. And of course, as everyone knows who has ever been privileged to use ARCE's generous facilities, ARCE's Executive Secretary Madame Amira Khattab deserves our entire staff's warmest expressions of thanks. Domestically, the project owes much to the University of Pennsylvania Museum and its directors. Robert Dyson generously aided the project during the years we were active in the field, and Jeremy Sabloff has continued to help by supplying us with all the essential research resources required to bring the project to publication. Many of the Museum's staff members have extended a wide variety of important services. I think in particular of Alan Waldt, its Associate Director for Finance, Stuart Fleming, Director of MASCA, Vincent Pigott, former MASCA Senior Research Scientist and contributor to Chapter seven, the Photographic Section's Fred Schoch and Francine Sarin, and lastly Bonnie Crosfield. Carl Beetz, who received his doctorate in Mesoamerican archaeology at the University of Pennsylvania, undertook the final reworking of most of Ken Schaar's and Jim Thorn 's field drawings, while the remainder have been finished by Nicholas Griffith. Philip Betancourt, Consulting Scholar of the Museum's Mediterranean Section and INSTAP's Executive Director, and David O'Connor, the former Curator of the Museum 's Egyptian Section, were both quick to provide me, a novice to Egypt as well as the general area of Aegean archaeology, with a helpful stream of information , help, and patient encouragement. The project's field staff over three seasons was composed of Aala Shaheen, David Conwell, Eric Gieringer, Roy Green, Chris and Linda Hulin, Murray McClellan, Pamela Russell, Kenneth and Margaret Schaar, Stephanie Tyiska, James Thorn, and Arthur and Joan White. Murray McClellan ably served as the project's Assistant Director for the 1985 season, as did David Conwell in 1987. Working under circumstances that were seldom easy, and occasionally made even less so by the inexperience of the director, the entire team deserves the Museum's and my own personal thanks. In the same category I am happy to include all of the present co-authors, most of whom had to take time away from busy schedules to make special trips to Matruh in order to deal with their subject matter at firsthand. These include Don Bailey, Rita Gardner assisted by Martin Frost, Mohamed abil ei-Hadidi, Linda Hulin, Murray McClellan, Vincent Pigott, David Reese, Pamela Russell , and Jim Thorn. I am also gratified to be able to acknowledge here the many courtesies and services provided by Harvard University's Peabody Museum of Archaeology and Ethnology. lt was of course the Peabody that originally sponsored Bates's 1913/ 14 expedition to Marsa Matruh. This has resulted in their more recent administration and staff becoming the target of a constant barrage of importunities from me since 1984. I particularly wish to thank its directors, Carl Lamberg-Karlovsky, David Tilbeam and Rubie Watson, its Collection Managers Una MacDowell , Katherine Skelly, and Gloria Greis, its Assistant Registrars Germaine

PREFACE AND ACK OWLEDGEMENTS

Juneau , Genevieve Fisher, and Kerstein Audette, its Photo Archivists Martha Labell and Barbara Isaac, and its Archivists Sarah Demb and Susan Hasskell for their help over the past decade and a half. An excavation cannot survive on the contributions of ju t its participants in the field and lab. We have been especially fortunate in that the readiness of the Museum to provide eed money and research funds has been matched by many friends to whom , though their company has been sadly thinned by time, I owe an everlasting debt of gratitude. They are Man don Bates, Mr. and Mrs. Woodruff Em len, Mr. and Mrs. Willy Gorrissen, Mr. and Mrs. E. Gordon Keith , Catherine Lower, Mr. and Mrs. Robert Maxwell , Charles Saffer, and George Vaux. Of personal importance to me for reasons that transcend their repeated and substantial financial help are James and Natica Bates Satterthwaite. Tica, Oric Bates's daughter, and Jim, my long-suffering high school English teacher, have underwritten this work in ways too numerous and subtle to be adequately expressed here. 1 have reserved for nearly the last my thanks to the Institute for Aegean Prehistory, which joined with the Mu seum at the project's outset to provide it with major financial support for our work in Egypt as well as help in defraying the costs of final publication . From his own scientific articles on the subject of Bronze Age archaeology to his generous sponsorship of museums and research stretching from New York to Crete, INSTAP's founder, Malcolm Wiener, has proven to be a remarkable benefactor and true friend to Classical archaeology and to this project. In addition, Philip Betancourt has undertaken an immense amount of work in helping to prepare this study for publication in his capacity as INSTAP's Executive Director and thus, once again, leaves me greatly in his debt. Finally, I want here to express my love, gratitude and thanks to my wife Joan and my son Arthur, who after years of sharing the joys as well as the predictable vicissitudes of more than a decade of work in Libya, helped me initiate the Matruh project. Along with Tarek Mohamed and Atif Abdei -Daym, they set aside their own busy careers to bear the brunt of the actual digging during our final field season in 1989. Without their support and encouragement the Matruh project could never have taken the shape that it did . Donald White

Notes 1. Murray McClellan and .lames Thorn working at Cyrene, Donald Bailey at the Society for Libyan Studies project at Sidi Krebish , Benghazi. For the background to the conclusion of the Museum 's work at Cyrene, see D. White, " Editorial ote," Th e Extramural; Sanctuw y of Demeter and Persephone a/ Cyrene, Libya 1: Background and Introduction to the Excavations (Philadelphia 1984) before title page. ii. Eastern Libyans. iii. African Studies. For more on Bates ' work at Matruh during the winter of 1913114, see Resw:facing of Paraetonium, passim and White, Strange1; 168, n. 36, as well as elsewhere throughout this study.

xix

BIBLIOGRAPHICAL ABBREVIATIONS This study adopts throughout th e standard abbreviations used by the American Journal ofArchaeology 95 (1991) 3- 16 and Lexicon der Agyptologie IV (Wiesbaden 1982) ix- xiii, with th e former taking precedence over the latter in the case of overlapping titles. The works li sted below are supplementary.

African Studies

0. Bates, "Excavations at Marsa Matruh," Harvard Aji-ican Studies 8 ( 1927) 124-200 .

Apis

D. White, "Apis," The Archaeology ofAncient Egypt: an Ency clopedia (London 1999) 141 - 143 .

Archaic Burials

0. Bates, "Archaic Burials at Marsa Matruh," Ancient Egypt (London 1915, Part 14) 158- 165.

Architecture of Israel

A. Kempinski and R. Reich , eds., The Architecture of Ancient Israel from the Prehistoric to the Persian Periods (Jerusalem 1992).

Ball, Geographers

J. Ball , Ministry of Finance, Survey of Egypt: Egypt in the Classical Geographers (Ca iro 1942).

Bates, Pottery

0. Bates, " Semitic Traces in Marmarica," PSBA 37 (1915) 201 - 207 .

Conwell , Ostrich Eggs

D. Conwell, "On Ostrich Eggs and Libyans. Traces of a Bronze Age People fro m Bates's Island, Egypt," Expedition 29, no. 3 ( 1987) 25- 34.

Eastern Libyans

0 . Bates, Th e Eastern Libyans (London 1914 ).

Fourtau

R. Fourtau, " La cote de Ia Marmarique," BIE 8 (1914) 99- 126.

1-l abachi, Milit01y Posts

L. 1-labachi, " The Military Posts of Ramesses II on the Coastal Road and the Western Part of the Delta," BIFAO 80 ( 1980) 12- 30.

xxii

BIBLIOGRAPIIICAL ABBREVIATIONS

Hayden, Cretan Architecture

B. Hayden, The Development of Cretan Architecture from the LM IliA through Geometric Periods (Ph. D. dissertation, University of Pennsylvania 1981 ).

Hulin 1999

L.C. Hulin, "'Marmaric' Wares; Some Preliminary Remarks," SLS 30 ( 1999) 11 - 16.

Marsa Matruh

D. White, "Marsa Matruh," The Archaeology of Ancient Egypt: an Encyclopedia (London 1999) 469-473.

Provisional Evidence

D. White, "Provisional Evidence for the Seasonal Occupation of the Marsa Matruh Area by Late Bronze Age Libyans," in A. Leahy, ed., Libya and Egyptcl300- 750 B.C. (London 1990) 1- 14.

Reswfacing ofParaetonium

D. White, "The Resurfacing of Ancient Paraetonium and its Ongoing Reburial," Archaeolological

Research in Roman Egypt. JRA Archaeological Supplement 19 ( 1996) 61 - 81. Rowe, Aegypto-Cyrenean Relations

A. Rowe, New Light on Aegypto-Cyrenean Relations Suppl. ASAE 12 (Cairo 1948).

Rowe, Contributions 1

A. Rowe, "A Contribution to the Archaeology of the Western Desert: l," BRL 36 (1953) 128- 145.

Shaw, Minoan Architecture

J. Shaw, Minoan Architecture: Materials and Techniques (Rome 1971 ).

Shaw, Hearths and Ovens

M. Shaw, "Late Minoan Hearths and Ovens at Kommos, Crete," L'habitat egeen prehistorique (Paris 1990) 232- 254.

Shear, Domestic Architecture

I. Shear, Mycenaean Domestic Architecture (Ph.D. dissertation, Bryn Mawr College 1968).

Shear, Panagia Houses

I. Shear, The Panagia Houses at Mycenae (Philadelphia 1987).

White, Before the Greeks Came

D. White, "Before the Greeks Came: a Survey of the Current Arc haeo logi cal Evidence for the PreGreek Libyans," SLS 25 (1994) 31-44.

White, Coastal Survey

A.P. and D. White, "Coastal Sites of ortheast Africa: The Case against Bronze Age Ports," JARCE 33 ( 1996) I 1- 30.

BIBLIOGRAPHICAL ABBREV IATIO S

White, Stranger

xxiii

D. White, "Stranger in a Strange Land: the Untold Story of the 1909 Bates Expedition to Cyrene,"

JARCE 35 (1998) 163- 178. White , Water. Wood

D. White, "Water, Wood, Dung and Eggs; Reciprocity in Trade along the LBA Marmarican Coast," in P.P. Betancourt et al., eds., MELETEM-

ATA, Studies in Aegean Archaeology Presented to Malcolm Wiener as He Enters his 65th Year Ul (Liege 1999) 931 - 936. Wright, South Syria and Palestin e

R.G. H . Wright, Ancient Building in South Syria and Palestin e l: Handbuch der Orientalistik, Siebente Abteilung: Kunst und Archiiologie (Leiden 1985).

Wright, Cyprus

R.G. H . Wright, Ancient Building in Cyprus 1:

Handbuch der Orientalistik, Siebente Abteilung: Kunst und Archiiologie (Leiden 1992). 1987 Pottery

L. Hulin, "Marsa Matruh 1987, Preliminary Ceramic Report," JA RCE 26 ( 1989) 115- 126.

1985 Report

D. White, "1985 Excavations on Bates's Island, Marsa Matruh," JARCE 23 ( 1986) 51 - 84.

1987 Report

D. White, " 1987 Excavations on Bates 's Island, Marsa Matruh . Second Preliminary Report,"

JARCE 26 ( 1989) 87- 1 I 4.

MINOR ABBREVIATIONS a.s. l. b.s.l. BR c. ca . em. d. E el. frag. h. max. m. mm.

mod. N NE

above sea leve l below sea level Base Ring Ware century circa centimeter diameter east elevation fragme nt height maximum meter millimeter modern north northeast

not ill. not inv. NW pol. pres. rest. RL

s SE sq. SW th. uncat. veg.

w ws W.

not iII ustrated not inventoried northwest pollen preserved Restored Red Lustrous Ware south southeast square southwest thickness uncataloged vegetational west White Slip Ware width

Chapter 1

THE SETTING Donald White

The University of Pennsylvania Museum of Archaeology and Anthropology sponsored excavations at Marsa Matruh from 1985 to 1989. The excavation's main focus was a small islet in a salt-water lagoon , Bates's Island, locally once

called the Island of the Jews (Plan 1). The islet, which had been the focus of excavations by Oric Bates before World War I, provided substantial amounts of new information, particularly on the Late Bronze Age.

The Town of Marsa Matruh Since the end of World War I, the modern coastal town of Marsa Matruh that has played host to the recently completed excavations has served as the capital for Egypt 's Western Desert Province. 1 Alexandria is 290 km. to its east; the Libyan-Egyptian border at Salum lies 210 km. to the west (Fig. 1: I). This places Matruh at what is conventionally taken to be the eastern limit of the ancient Marmarica, which in antiquity lay between Cyrenaica and Egypt-proper and included the desert regions as far south as Siwa. 2 By Hellenistic times, the western limits of the Marmarica were set at Darnis (modern Derna in eastern Libya, 75 km. east of Cyrene) and included the settlements of the eastern Gebel Akhdar as far east as Antipyrgos (modem Tobruk). In post-Diocletianic times, the Mannaric coast was absorbed into Libya Inferior or Libya Sicca. Darnis was designated its early Byzantine provincial capital; eventually, that function was taken over by Paraitonion, which is today Marsa Matruh.3

The western limit of Egyptian territory evidently was fixed as early as the reign of Ramesses II at Zawiyet Umm el-Rakham, the site of A pis: ca. 20 km. west of Matruh . Following the division of Alexander's empire, Paraitonion was united administratively with Ptolemaic Egypt, and for all practical purposes the Eastern Marmarica may be thought to have begun west of Matruh. Before World War 1, the village consisted of "nothing more than a score or so of white-plastered, four-square rubble houses, mostly of one story"' (Pl. I A). By the years immediately before and after World War II, its residents were limited to a cluster of local Bedouin families (Pl. !B), a transient colony of mostly Greek and Italian sponge-divers, the odd foreign shop keeper and hotelier (mostly Greek), a few hardy summer vacationers fleeing the heat of Alexandria, and the Anglo-Egyptian officials responsible for the administration of the Western Frontiers District (Pl. 2A). 6 Today, the town serves as a popular

THE SETTING

seaside resort whose largely Egyptian population is said to swell to nearly a quarter of a million people during the su mmer months. 7 Prior to the contemporary coastal real estate boom, the only events of the present century to offset the isolation of this otherwise remote spot were offshoots of the two World Wars. As a counter-thrust against attempts by Turkish and German agents to raise the Sanusi tribesmen agai nst the Italians occupying their eastern Libyan homeland, the British launched a short campaign in late November of 19 15 that used Matruh village as an advanced staging ground for their desert operations. 8 A side effect to this minor wartime digression was the compilation of two abbreviated (but sti ll occasiona ll y usefu l) accounts of what the town and its then visible antiquities amounted to during the years immediately following the survey and excavation conducted by Oric Bates in 1913- 1914. 9

3

The years between 1940 and 1942 transformed both the town's ancient and modern topographi0 to a far greater extent than had World War I, as the by then strategically useful railhead and seaport repeatedly changed hands between the Germans and the British. Throughout, Matruh suffered intensive aerial, land, and sea bombardment (Pl. 2B) . Despite the massive accumulation of wartime accounts that take oblique note of the town, however, details of the topography seem to have been universally passed over, leaving the creation of perhaps the most atmospheric description of Matruh 's mid-century appearance to wait for the return of an ex-so ldier five years after the war's close. 11 During the 1950s and 1960s, the eastern end of the eastern lagoon system apparently was exploited for salt pans visible in the sate llite photo taken in the early 1960s (Pl. 3A).

Archaeological Investigations Before and After Oric Bates After Bates's final departure in the spring of 1914, 12 Matruh retreated into scho larly limbo until the 1930s when several articl es reported on the chance recoveries of various post- Bronze Age monum ents. 13 The only fieldwork undertaken during this period was carried out by the civi l engineer G.F. Walpole, who was ordered by the Office of the Surveyor-General to investigate an anci ent underground aqueduct located 12 km. west of the town with a view to its conversio n to modern use (Fig. I :2). 14 In more recent times, the Egyptian Antiquities Organization has excavated two later Roman period comp lexes (Pl. 3B) constructed a lo ng the ridge north of the West Lagoon in the vici nity of the shore rock features known locally as "C leopatra 's Bath." 15 Publication of the results of this promising work, carried out under the guidance of Chi efInspector Ezzat Osman e i-Hamhmy, has been pending for many years. Before Bates's arrival in 1913 , the only published archaeological study of the site, at least of which I am aware, was undertaken in 1904 by

R. Fourtau, who conducted an area reconnaissance as part of a larger survey of the coastline's antiqu ities. 16 Fourtau's sketch plan (Fig. I :3) positions ancient Paraitonion along the south shore of Matruh's West Lagoon, 17 where, according to his text, the town remains could be traced for a distance of more than 2 k::m. The lagoon's southeast corner was occup ied by a spacious cut-stone quay or pier, from which a series of stone jetties projected into the water. The pier turned north at the lagoon's southeast corner and continued at right ang les up the tongue of land separating the lagoon from the modern harbor to its immediate east. The resulting north-south lin e of pier terminated at both ends in stone towers that also are visib le on Fourtau's plan. 18 He also plotted (Fig. I :4) the northwest corner of what he termed a domestic vi ll a, facing, one gathers, what was the southern shore of the lagoon. The structure's most co nspicuous feature was a row of interi or co lumn s. While the town remains and the pier were sti II in evidence, Fourtau 's "vi ll a" and stone towers had either vanished or

4

MARSA MATRUI-I

Fig. 1:3. Fourtau 's 1904 Sketch Plan ofMarsa Matruh (ji-om BIE 8, 1914, 117Jig. 8). Redrawn by Nick Griffiths.

Fig. 1:4. Fourtau 's Sketch Plan of Villa with Internal Colonnade along the south shore of Matruh 's West Lagoon (fi-om BIE, 8, 1914, 120, fig. 9). Redrawn by Nick Griffiths.

were passed over by Bates when he arrived only nine years later. 19 East of Matruh village, Fourtau noted the presence of cut-stone stairs above the shoreline at the southeast corner of the East Lagoon, later picked up by Bates and eventually recorded by the Pennsylvania Ex pedition. 20 He also tantalizingly referred to "interessantes traces d ' habitation" along the lagoon system east of the main harbor in the direction of RasA lam el-RQm 21 but otherwise left us with no detail s. The remainder of hi s description was limited to various rockcut tombs associated with the so-called "G reat Rid ge" south of the rr.odern town and a cemetery along the "C leopatra Ridge" north of the West Lagoon .22

Early Travel Accounts The 18th and 19th century travel reports that so enrich our knowledge of Siwa oasis to the so uth and the Cyrenaica to the west are, for the most part, conspicuously silent when it comes to Marsa Matruh .23 Most European travelers seem to have struck out across the desert for Siwa from Cairo and in that way avoided the longer and potentially even more dangerous route pursued by Alexander the Great along the northwest coast before turning south. 24 Down to the outbreak of the Turkish-Italian War of 1910/ 11 , the bulk of the explorers of the Libyan Pentapolis embarked

on their overland travels from Benghazi or some other point farther west, or, alternatively, they arrived directly by ship. 25 For the most part, the Marmarica was either avoided or ignored. Those who did pass by Matruh had little to say about its Classical remains and, predictably enough, nothing about its pre-Classical phases. With one exception (Fig. I :5), they illustrated none of the monuments from its immediate vicinity; the renderings of Matruh 's harbor and adjacent lagoon sys tems incorporated into the route maps and

THE SETTI G

nautical charts before 1900 are, for the most part, too small and inaccurate to be of use. Despite the overall lack of specific information, the association of the site of Matruh with ancient Paraitonion appeared surprisingly early. 26 In the spring of 1792, W.G. Browne was the first recorded European to penetrate into the interior of this part of Africa as far as Siwa Oasis. Although Browne had abandoned the northern coast just west of Abusir and by traveling straight into the desert had bypassed Matruh by a wide margin, his general map labeled the site "Bareton," the conventional Arabic corruption of the ancient place-name. 27 Co l. William Eaton's diary account of his epic razzia launched from Alexandria against Derna in 1805 was topographically too vague to be of much use. 28 Fifteen years later, the German consul at Cairo, Genera l H. von Minutoli, led a large party of savants and exp lorers along the same coastal route. 29 Before turning in to the desert for Siwa with his topographer, Herr Gruoc, 30 Minutoli directed the group's artist to record the remains of the so-called Gasr el-Fok (Fig. 1:5), positioned more or less south of Riis el-Kenais (known today as Riis ei-Hikma or ancient Leuce Acte,l 1 ca. 50 km. east of Matruh). He then accompanied Ehrenberg and Gruoc on a side excursion to visit the site of Paraitonion, once again identified with Bareton, 32 but instead of the anticipated ancient ruins, all the group found was a ruined Arab fort, partially assembled from robbed blocks of ancient masonry. 33 The socalled Gasr Medsched was located on the expedition's route maps south of the modern harbor (characterized by Minutoli as a "ganz leidlichen Hafen") at some distance from the shore. 34 The French explorer J.-R. Pacho, so familiar to students of the Cyrenaica, visited the harbor site in 1824. 35 Although known to him as "Baretoun," his route map labeled Matruh "Berek-Marsah

5

(Paraetonium)." The Arab gasr, which appeared on Pacho 's plan in the same location as on Minutoli's, received some further description, whi le its destruction was attributed to Mohammed Ali. 36 Philip Smith's 1840 map , Ancient Africa or Liby a, Part 11,1 7 omitted any reference to the gasr and instead simply labeled the site "Paraetonium or Ammonia," the last being its occasional ancient name, as Pacho was already aware. 38 The celebrated Africanist Heinrich Barth must have brushed by Matruh in 1846 but left us with nothing other than the captions on his route map (Fig. I :6). The "K. Medsched" nomenclature was reemployed, the harbor labe led "Mirsa Labeit" ("Pt. Labeit," referring to the western horn of the modern harbor in Bates 's day), and the site Paraetonium.39 In 1847, Messrs. Laport, Forty, Longshaw, and St. John skirted within a few miles of Matruh on their way to Siwa, but according to their leader, Bayle St. John, "circumstances did not permit me to trace this coast a little farther, and visit the ruins which occur at Kassaba, and at the place which geographers called Bareton. Bareton is supposed to be identical in position with the ancient Paraetonium . . . " 40 The British traveler James Hamilton, whose 1852 narrative was filled with such useful impressions of Cyrenaica, also merely swung by Matruh while making his way to Alexandria through Siwa : "We reached the level of the sea, encamping, at the end of the ninth hour, in a meadow of rank grass, separated from the sea by a ridge of grass. A point to the west forms here a small bay, and Baretoun, the ancient Panetonium [sic], lies a little beyond this. The nam e given to the place we encamped in is Berbetat-elmudar."41 The 1883 account by the explorerbotanist G. Schweinfurth, who made a journey along the Marmaric coast west to Tobruk, contributes nothing of interest. 42

Oric Bates For present purposes, far and away the most important of the earlier generation of investigators to concern themselves with Marsa Matruh is

Oric Bates, who, at the time of his death in 1918, was Curator of African Archaeology and Ethnology at the Peabody Museum of Harvard Univer-

MARSA MATRUH

6

~........--

-

r- T

-

1

-

~

llrl·

1---

r- -

w.--,

"

.0 "1"

,..._-

.,.. , __ JI#, ... ,..,,

s

Fig. 1:5. Gruoc 1820 plan and elevations of Gasr el Fok, south of Ras el-Hikma (from H. von Minutoli, Reise zum Oase des Jupiter Ammon in der libyschen Wi.iste, Berlin 1825, pl. 4), courtesy of the British Library.

THE SETTING

sity. 43 Best remembered for his classic study of the eastern Libyans (The Eastern Libyans, London 1914), Bates deals with Matruh in four separate articles, of which the most ambitious is a long but partially incomplete report edited for publication by Natica Bates and published nine years after his death as part of the eighth volume of the Harvard African Studies series, which he had founded and edited. 44 The report is based on survey and excavations that Bates carried out at Matruh with the assistance of the Egyptologist W.J. Harding-King between late December 1913 and the end of February 1914. The idea to investigate the Matruh area for what it might reveal about the ancient Libyans appears to have come to Bates three years earlier when he used the coastal village as the staging area for launching the eastern Marmarican portion of his field reconnaissance that formed the basis of much of

The Eastern Libyans. 45 Before his death, Bates completed the Harvard African Studies report's first three chapters (the "Introduction" "Graves" and "Rock-Cut ' ' Tombs"). Chapter 4 ("The Byzantine Site and the Island of the Jew") was completed posthumously by his widow, Natica Bates, from the expedition's journal. A fifth, concluding chapter ("Archaeological Finds in the Neighborhood of Marsa Matruh") was based on the unedited field notes gathered together by Harding-King in 1913 before Bates's arrival at Matruh. While the chapter in which Bates described the "Island of the Jew" (Chapter 4) is father to the present study, it is Bates's opening chapter which provides the classic description of the surrounding area's physical configuration (Fig. 1:7), the layout of Matruh 's harbor and surrounding lagoons, its natural resources, and its climate. 46 Although topographical arguments for linking Matruh with the Hellenistic-Roman town of Paraitonion first emerge in travel accounts as early as the time of Minutoli and Pacho, if not before, they find their most authoritative presentation here, 47 along with a thorough survey of the etymology of the various other place-names attached to the harbor site both during and after antiquity. The town's history is examined thoroughly from the time of its traditional foundation

7

by Alexander 8 through its brief moment of international notoriety when occupied by Marc Antony and perhaps even Cleopatra, 49 and down to its early Byzantine phase when Paraitonion shone as the chief city of the eparchy of Lower Egypt, the seat of the dux limitis Libyci, and home to a bishop. Setting aside the prehistoric remains associated with the Eastern Lagoon's "Island of the Jew" and the Libyan cemetery on the Great Ridge, Bates's and Harding-King's most important topographical discoveries were two sets of town remains flanking the West Lagoon, 50 the West Lagoon's "large masonry quay," 51 the cross-wall sealing off the western end of the coastal ridgetop settlement attributed to the Justinianic period/ 2 a third set of town remains spreading southward behind the southeast corner of the first East Lagoon, 53 a Byzantine house site on the shore opposite the modern harbor's mouth, 54 and the nearby, so-called Byzantine Bath. 55 In addition to these walled monuments, during their residency Bates and Harding-King were able to explore and record no less than ten cemeteries of mainly Roman-Byzantine date associated with either the Great Ridge south of the modern city or with the regions to the west and southeast of the town. 56 In addition to the separate cemeteries, they discovered and cleared three large rock-cut chamber burials. The first and second, T I and T2 , were found ca. 500 yards inland from the southeast corner of the modern harbor where they presumably lie still buried today beneath the modern town. 57 A third tomb, T3 , was found on a low mound ca. 1000 yards west of the first two. 58 Eight additional independent chamber tombs (T4- Tll) were then recorded in somewhat more abbreviated fashion than the first three. 59 All the tombs were associated with the Great Ridge area to the south and southwest of the modern town , and some may have represented tombs more cursorily noted by Fourtau. 6° Finally, in the preface to his Chapter 3, Bates promises to discuss "a number of rock-cut tombs, called Site 100, of the classical period, some of which are rather imposing because of their size," but he says nothing further about them in the extant text. 6 1 According to his site map (Fig. I :7), Site I 00 was located on

THE SETTING

9

10

MARSA MATRUH

the "Cleopatra Ridge" area within the walled limits of coastal Paraitonion-proper, which once again had been earmarked by Fourtau as the site of an important cemetery but today reveals no obvious traces of burials so far as I am aware. Apart from the odd monument south of the

modern harbor as well as farther afield away from the ancient town, this group of sites concludes what Bates was able to say about Matruh with the exception of its Libyan cemetery and so-called "Island of the Jew," treated below.

Chapter 1 Notes Through Egvpt in War-Tim e (London 1918) 111 - 130. W.

I.

Marsa Matnth, 469-473.

2.

Hence the oasis's occasional name Marmaricus

Massey, The Desert Campaigns (London 1918) 132- 153. A.

Hammon or "bath of Marmarica." For additional references

De Casson, Mareotis (London 1935) 196- 199. Morton (n.

to the region's topography, see 1985 Report, as well as

6) 180. E.E. Pritchard-Evans, The Sanusi of Cyrenaica

Resurfacing of Paraetonium, 65 , and White, Coastal Sur-

(Oxford 19 73) 125- 130. Resurfacing of Parae/onium,

l 'e_l ',

11 - 30. 3.

Marsa, Mersa, and even occasionally Mirsa are all

67- 68. 9.

In addition to the interesting narrative by Martin

variant transliterativc spe llings of the obsolete Arab word

Briggs who was stationed in Matruh (sec n. 8), there exist

meaning " bay" or "harbor" that are routinely attached to the

various unpublished notes and drawings assembled by the

site. See Aji"ican Studies, 128. For Paraitonion 's identi fica-

chaplain to the British forces, the Rev. L. Goodenough,

lion with Matruh, see Afi"ican Studies, 128- 136. For Paraito-

which eventua lly came into the hands ofthc late R.G. Good-

nion 's post-Diocletianic history, see Aji·ican Studies, 135;

child. See "A Byzanti ne Chape l at Marsa Matruh (Paraito-

RE 14.2 ( 1930) 1882; D. Roqucs, Sym'!sios de Cyrene et Ia

nion)," JARCE 28 ( 1991) 20 I 211. See also Reswjc1cing of

Cyrenai"que du Bas-Empire (Paris 1987) I 09- 1 12 , 326- 329,

Paraetonium, 68, fig. 6.

331, 337, 340; Reswfacing ofParaetonium, 65- 66, n. I S. 4.

10.

While it is difficult to separate what was destroyed

RE 1.2 ( 1984) 2807. Rowe , Aegypto-C_rrenean

in the war (when ten percent of the town was totally obliter-

Relations, 77, fig. 5. Rowe, Contrihutions I, 139. Ball, Geo-

ated and twenty percent of its buildings damaged) from what

graphers, 76 . Habachi, Milital) ' Posts, 13- 30. Coastal Sites,

was pulled apart and/or buried beneath modern con-

20, nn. 43-44. Apis, 141 - 143.

struction , the harbor's edge was badly shelled, which may

5.

"With two or three exceptions these houses are the

have been what eventually obliterated Bates 's so-called

'stores' or 'canteens' of the Greek traders who all along this

Byzantine House ("Site 400 A"), whereas the nearby

coast exploit the Bedawin." African Studies, 127.

"Byzantine Bath" ("Site 400," now republi shed by Good-

G.E. Simpson, The Heart of Libya, The Si11'a Oasis,

ch ild as a church) probably already had been covered by the

Its Peoples, Customs and Sport (London 1925) 42- 53. B.

enlargement of the old Governor's residence before World

6.

Khun de Prorok , Mysterious Sahara (Chicago 1929)

War II. See Goodchild (n. 9) 201 - 211. See also Ajhcan

248- 251. H. Morton, Through Lands of the Bible (London

Studies, 177ff., map I. The ruins along the southern shore of

I st ed. 1938; 6th. ed. 1948) 182- 188, whose reference I owe

Matruh's West Lagoon , which Bates identified as "Contra-

to J. Thorn. W. MacArthur, Auto Nomad in Barbarv (London

Paraitonion," may well have s uffered considerable war

1950) 325- 326, 338- 343. Reswfacing of Paraetonium,

damage but were eventually totally destroyed after the war

passim.

by the construction of concrete docks for deep-water shi p-

7.

This rapid growth is part of the larger phenomenon

ping. See African Studies, 136, 189, map I , "S ite 31."

of the explosive development of coastal real estate since the

Bates's candidate for Paraitonion-proper is the cluster of

mid 1980s that is especially prevalent between Alexandria's

remains along the sandy coastal ridge separating the West

western suburbs and el-Aiamein but at the present time is

Lagoon from the sea (his "S ite I 00") known locally as the

increasingly concentrated in the territory to either side of

"Cleopatra Ridge" and already the target of some excava-

Matruh. See Resurfacing of Paraetonium, 61 ff.

tion under the direction of Ezzat Osman el-Hamhmy, for

8.

The fighting was over by March, 1916. M. Briggs,

which see below. As far as I can tell , relatively little wartime

II

THE SETT I G

damage took place in thi s zone, althoug h Bates 's " Ju sti ni an-

interior contains a man-made chamber conn ected to the sea

ic Wall" (hi s "S ite 102'') at the town 's west end is certainl y

by a carefully cut tunnel. See Restufttcing of Parae/onium,

more difficult to trace today than it was in 1914 and may

73, nn. 53- 54 .

have been bombed. See A.fi"ican Studies, 135- 136, 195 and

Resurfacing of Paraetoniwn, 73- 74, fig . I I. By way of contrast, the exposed sandstone coasta l ridge north of the First

16. ll/11 ,

Fourtau, 98ff. See also Reswfacing of Parae/oni-

67, figs. 4 , 5. White, Coastal Survey, 11 - 12, n. 6.

17.

In other word s, where Bates was to position hi s so-

East Lagoon, especially in the area around the recently dis-

cal led Contra- Paraitonion, for which see n. 50, as well as

covered Area V church, was shelled heavily. See 1987

Reswfttcing of Parae/onium , 67, 73.

Report, 88- 93, fig s. 3- 5, and Reswfacing of Paraetonium, 79, fig . 15. The same is true of the East Lagoon island that

18.

It is just as well that the pl an leaves no doubt as to

which lagoon Fourtau intended us to understand stood in

forms the main subject of thi s monograph as we ll as our

front of hi s city because the written text (p. I 17) refe rs to it

Area Ill remains on the slope a bove the southeast corner of

as th e "East" rather than the " West."

the lagoo n, ca lled by Bates " Site 54" and which in hi s day

19.

See African Studies, 13 1, 136, 189, nos. 30, 3 1 for

co nstituted a more extensive set of ruin s than today. See

what was visible during the w inter of 1913/ 14. See a lso

African Studies, 191. I I. For th e wartime acco unt s, see Resurfacing of'

c learly has undergone continuous morphological change

Bates, PorteiJ; 20 I. The shape of the Matruh lagoon system

Parae/onium , 74, n. 59. Our post- World War II so urce is R.

since th e beg innin g of thi s ce ntury. Fourtau him se lf

Maugham, Joumey to Siwa (Lo ndon 1950) 28ff., pi s. opp.

desc ribed th e ground se parating the West Lagoon from the

pp. 18, 38--4 1. It is Maug ham who provides the percentages

main harbor as " deux banes de sa ble rei iant un !lot aux deux bandcs littorales" (ibid., p. 11 7) and placed th e northernmost

of wart ime damage cited; see n. I 0 . 12.

For more on Bates's earlier activities at Matruh and

of his two towers on an island- li ke bul ge on his modest

e lsewhere in the Western Desert, see White, Stranget;

sketch plan . No channel is indicated for the passage of sea-

165- 168.

wate r into the lagoon. Bates's more e laborate plan (F ig . I :7)

13.

shows th e ground between the two bodies of wa ter as a

E. Breccia, "U na statuetta del buon pastore da

Marsa Matruh ," BSRA A 26 ( 193 1) 247-257. E. Breccia,

sa ndy isthmus; about 300 feet south of the lagoon's north

Ann. du Musee Greco-Romain 193 1- 32, 2 ( 1933) 24 (a G reek inscribed altar). 0 . Gueraud, "S ignature d'un Platri-

shore, it is cut through by a narrow, meandering water-

er," BSRAA 30 ( 1936) 31 - 33 (a c istern, found 7 km. to the

or ruin s of a pier. A I :50 scale British Admiralty chart of

channel.

othing is said abo ut either Fourtau's stone towers

so utheast of Matruh, inscribed in 6 b.c . with the name of

Marsa Matruh , no. 3567, initially drawn in 1917, a mere

a Marsa

three yea rs after Bates's departure, depicts the land forma-

Matrouh," Ann. du Musee Greco-Romain 1935- 1939, 3

tion se parating lagoo n from harbor as a near-i sland sur-

Isaia s the plasterer). A. Ad ri ani , " Deco uvertes

( 1940) 159- 161 (rock-cut tomb s and assoc iated sto ne por-

ro unded by water to the north, south, and west with traces of

trait scu lptures discovered at Hakfet Abdel Razek Krim

an anc ient pier visible at its so uth west co rn er.

so utheast of Matruh). In additi on, Bates's " Site 400," identified as a Byzantine bath , is described in the 1920s Baedeker :~

20.

Fourtau, 11 9. African Studies, 185, 19 1, field notes

entry 54 . 1985 Report, 56, fig s. 4, 5.

Egypt, ed. by I. Masser (Norw ich 1929) 3 I, as a double-

2 1.

Fourtau, 1 19.

apsida l Ch ri stian bas ilica . Resw fac ing of Paraetoniwn,

22.

68- 7 1; Marsa Matnth , 469--473 .

23.

Resutfacing of Parae/onium , 67. Much of the followin g discussion has already

14.

Walpole, 1- 30. Reswfacing of Parae/onium, 7 1. D.

White, " Introd uctory Observations for a Comparative Study of Urban Raised Water Delivery Systems: Cyren aican A po lIonia and Marmarican Paraito ni o n," forthcoming. In addition, see Chapter 3, for a di sc uss ion of the implications of Wa lpole's aqueduct for estab li shing sea level during th e Rom an period. 15.

Illustrated in R. Maugham (n. II) pl. opp. p. 36.

appeared as part of the article Resurfacing of Parae/onium, 66ff. 24.

Ea rly trave l to Siwa is discussed by A. Fakhry, Th e

Oases oj'Egypt 1: Siwa Oasis (Cairo 1973) 98- 112. 25 .

R. Goodchi ld, "A Hole in th e Heavens ," Select

Papers of th e Late Richard Goodch ild (eel. by J. Reynold s, London 1976) 273- 297. White, Stranger, 163- 164. In a bi za rre switch from norm al practices, in 1762 or sho rtly

The " bath" is a rock ledge, washed by the waves; its outer

thereafter, James Bruce traveled overland a ll the way from

face and top have been modified by quarrying while the

Algiers to Ptolema is on the coast of Cyre naica roughly ha lf-

MARSA MATRUH

12

way between Benghazi and Cyrcnc where rumors of tribal warfare, plague, and civil disturbances a t Derna pers uaded him to take to sea in a "small Greek junk ." The craft sank in a storm off Benghazi where Bruce was robbed and beaten by loca l inhabitants who had found him lying exhausted on the shore. From here he escaped in another boat to Greece

31.

Ball, Geographers, 67. White, Coastal Survey,

22 23, fig. 14. 32.

Minutoli (n. 29) 65. llcmprich and Ehrenberg (n .

26) 85, n. I. 33.

Minutoli (n. 29) 64 65: " Diesc Burg ubrigens von

nicht geringem Umfang, und in den Grundmauern derselbcn

from where he eventually resumed his celebrated voyage.

crblickte ich regelmassig bchaucnc Quadcrn, we lche fruher

See J. Bruce, Travels to Discover the Source of the Nile I

offenbar griechischen oder romischen Bauten angehort

(2nd. ed., Edinbu rgh 1804) 44-52.

haben mussten." Hemprich and Ehrenberg (n. 30) 84.

26.

See Ajhcan Studies, 128 for its possible appearance

in corrupted form as early as the Middle Ages and Renaissance.

34.

See Hemprich and Ehrenberg (n. 30) foldout map,

where the caption reads: " Kasr Medsched (Paraetonium)." 35.

J.-R. Pacho, Relation d'un voyage dans Ia Mar-

W.G. Browne, Travels in Afhcct, Egypt, and Syria,

marique, Ia Cyrenai'que et les oasis d'Auclj'elah el de

.fi'Oin the Year 1792 to 1798 (London 1799). A. Fakhry (n.

Maradeh (republished Marseille 1979) map opposite p.

27.

24) 98- 104. For cl-Baratun, here simply Bareton, sec

xxxii. G.R.H. Wright, "Cyrcnc: Other Men 's Memories,"

Afi'ican Studies, 128.

SLS 30 ( 1999) I 00- 102.

28 .

C. Prentiss, T!te Life oft!te Late Gen . William Eato11

(Brookfield, Conn . 1813) 304:

36.

Pacho (n . 35) 29 31.

37.

Issued for T!te Sociell'.fiJr the Di/]itsion of U1-ejitl

Knowledge in London. March 13th. Marched twenty five miles. These three days we have passed low sand vallics [sic] and rocky, desert plains. Few vestiges of ruins except a castle fifty

38.

Strabo 17.799. Pacho (n. 35) 30. H. Kecs, "Paraito-

nion," RE 18,3 ( 1949) 1184. 39.

H. Barth , Wanderlungen durch die Kiistenliinder

miles from Arab's tower [Abusir], which has all the

des Millelsmeeres ( Bcrl in 1849). The paucity of information

appearance of Grecian architecture .. .. At this camping

may be explained by the fact that Barth was attacked by rob-

ground, on the seaboard, are numerous wells of great

bers and all his records lost. See Goodchild, "A Hole in the

depth cut through solid rock, and producing water of

Heavens," Libyan Studies, Select Papers of the Late R.G.

good quality. A little to the westward is a handsome bay

Goodchild (ed. J. Reynolds, London 1976) 278 .

about twenty miles deep and fifteen broad [Abu Hashaifa

40.

B. St. John , Adventures in the Libyan Desert and

March 14th. Marched twenty six miles over a barren

th e Oasis ofJupiter Antnwn (London 1849) 76- 77 . 41. James Hamilton, Wanderings in North Afhca (Lon-

rocky plain. Passed some vestiges of ancient fortifica-

don 1856) 309- 310. Wright (n. 35) 104 105.

Bay between Riis Alam cl-RCim and Ras el-Kanayis'J].

tions [Gasr Medschcd?]: raised and encamped upon the dividing ridge between Egypt and Tripoli [above SalCim atop the Libyan plateau for which see White, Coastal Survey, 15- 18].

42.

G. Schwcinfurth, "La cote de Ia Marmariquc,"

Compte.\· Rendus de Societe geographique ( 1883) 484-486. 43.

Bates died on October 8, 1918 from pneumonia

contracted while stationed in the army at Fort Zac hary Taylor, Louisville, Kentucky. He was 35 years o ld at the time.

In October of 1820, accompanied by Drs. Ehren-

For an account of his li fc , sec A. Coolidge, "Oric Bates,"

berg, Hemrich, and Scholz, the painter Liman, the topogra-

29 .

HAS II ( 1918) before preface, as well as the rema rk s of hi s

pher Gruoc, and the draftsman Boldrini. H. von Minutoli,

widow, Natica Bates, in Afhcan Studies, 177, together with

Reise :::um Oase des Jupiter Ammon in der libyschen Wiisle

the short death notice in the same vol um e by Dows Dun-

(Berlin 1825)6 1- 69. 30.

The rest of the exped ition continued along the coast

ham. A third important source isM. DeW. Howe, Memoirs

of the Harvard Dead

i11

the War Against Germany 4 (Cam-

as far west as Derna at the western limit of the Marmarica.

bridge 1924) 547- 566. For more discussion of Bates's

W. Hemprich and C. Ehrenberg, Natwgesc!tichtliche Reisen

career as we ll as hi s work at Matruh, see White, Strange1;

durch Nord-Afhka und Wesi-Asien in den Jahren 1820 his

passim; and Reswjacing of Parae/onium, 6 1, 68, 7 1- 74.

1825, 1: Reisen in i{gypten, Lihyen, Nuhien und Dongola

Wright (n. 35) I 07 I 08. J. Uh lcnbrock, "Cyrcnc Papers: the

(Berlin 1828).

Second Report. The Oric Bates Expedition of 1909," SLS 30

THE SETTING

13

( 1999) 77- 97, fig. I. To judge from a perso nal letter written

tion seems strained in retro spect; both sites were surely part

by Bates to his father from Marsa Matruh on Janua ry 23,

of the same town. These points arc further discussed in

19 14, at least part of the costs of the expedition to Matruh

ResUJfacing of Parae/onium . 71 - 74.

were defrayed by a fellowship from Harvard Univers ity. 44.

5 1.

African Studies, 126, 13 1, 136, 189 . Bates, Pol/e1y,

His "Excavations at Marsa Matruh" (referred to

20 I . Because Strabo 17.799 described Paraitonion as a "c ity

throughout this study as African Studies). The three addi -

with a large harbor, about 40 stades in extent (3.9 nautical

tional articles are " Semitic Traces in the Marmarica," PSBA

miles long)," Bates argued that in ant iqui ty the West Lagoon

3 7 ( 1915) 201 - 207 (here Bates, Po11e1y) "Archaic Burials at

had to have been lin ked to what is today the main harbor by

Marsa Matruh," Ancient Egypt IV (Lo ndon 19 15) 158- 165,

a narrow channel and that their modern near-separation is a

and "Ethno log ica l Notes from Marsa Matruh," JRAS (Oct.

recent phenomenon , caused by "sand-bars of recent forma-

19 15) 736- 739. While not referring spec ifi cally to Matruh ,

tion." As already stated, he does not see m to be aware of

a fourth art icle, "Nomad Burial s in Marmarica," Man 13

Fourtau's north-so uth lin e of pier, set a long the east end of

( 19 13) 159- 162, conta ins cognate material.

the West Lagoon and topped at either end with towers, but

45.

Aji·ican Studies, 173 a lludes to the ea rli er visit. See

also White, S11·ange1; 168, n. 37. 46.

In particular African Studies, 126- 127. Two obser-

vatio ns bear repeating:

the pier presumably stopped short of the connecting water channel and therefore does not inva lidate the theory ca lli ng for a larger harbor. 52 .

Aji·ican Studies, 136, 195, and site map no. I 02. By

the summer of 1990, the remains of this wall , which mu st The port of Matruh is therefore geographically signifi-

have been conspic uo us in Bates's day, was just barely visi-

cant, being situated nearly halfway between the fertile

ble above the drifting sand. Its north end run s to the edge of

region of Cyrenaica on the one hand and the Ni le Delta

the steep, coastal bluff overlook in g the sea west of"C ieopa-

on the other; and its importance is further enhanced by

tra 's Bath" (seen. 15). The settled, intramural ridge area east

the fact that it is, with the exception of Marsa Tobruk, the

of the wa ll is littered with Roman-period surface shcrds as

on ly good haven between the gu lf of Bombah in the

we ll as traces of buildings, in c luding remains of an apsida l-

west, and Alexandria in the east. . . Although dependent

ended structu re. I co ul d detect no traces of the cross-wa ll in

for moisture on somewhat uncertain rainfalls and, per-

a deep construction trench laid to the north of the modern

haps, on seepage from the desert highlands to the south-

road, and it is no longer appa rent where the wall met the

wards, this plain, carefu lly cultivated, could- and in

north sho re of the West Lagoon. Some excavation wou ld be

Roman times apparently did- support a large popula-

required to clarify its construction and date. See also Resur-

tion. The rich brown earth is extraordinari ly deep, and the

facing of Parae/onium, fig. I I . The question of whether the

agricu ltural possibilities of this district are not sur-

remainder of the town was fortified during its Hellenistic-

passed elsewhere in the Marmaric littoral zone.

Roman phase is discussed, ibid., n. 49. 53 .

Sec a lso Eastern Libyans, 6- 8. 47.

Add Ball, Geographers and H. Kces, "Paraito-

nion," RE 18,3 ( 1949) 1182- 1184.

Aji·ican Studies, 136: "a la rge Roman site on theSE

of the first east lagoon." See ibid. , 185, 191 no. 54, pl. 56, fig . 16, where the remain s are associated with three stone sta irs lead in g down to the water and with various other fea-

48.

See C hapter 5, n. 6 a nd Vol. II , C hapte r 15.

tures. One set of sta irs, still visible in the 1980s and perhaps

49.

Vol. II , C hapter 15 , nn. 99 and I 00.

those orig ina lly picked up by Fourtau (n . 16) was recorded

50.

See African Studies, 189 no. 3 1, 195 no. I 0 I . The

by the Pennsylvania exped iti on: D. White, 1985 Report, 56,

quay or pier and the town remains a lo ng the lagoon's south-

fig . 5. Bates believed that the East Lagoon, like the West

ern shore had been discovered previously by Fourtau, whose

one, must have originally joined with the main harbor, a nd

work Bates either did not know or chose to ignore. Fo urtau

in this he may well have been right. See Bates, Po11e1y, 20 I .

identified the southern set of rema ins as Paraitonion-proper.

African Studies, 136, preferred to ca ll it Contra-Paraitonion a nd reserved the title of Paraitonion for the wa ll-enclosed

54.

Site 400 A. See African Studies, 180- 18 1, 183- 184,

pis. 52, 54- 56, 65. 55.

Site 400, located about I 00 yards to the southwest

ruins occupying the northern coastal ridge separating the

of the Site 400 A house. Afi'ican Studies, 177ff., pis. 50- 53,

lagoon from the sea, i.e. the "Cleopatra Ridge." The distinc-

60- 62, 64. R. Goodchi ld (n . II ) identifies it as a church .

14

MARSA MATRUH

African Studies, 141 - 154. The Pennsylvani a ex pe-

ever else was found in Bates's other post-Bronze Age buri-

dition ha s not ve rifi ed how many of these have survived th e

als, is no longer known, but they see m to have remained in

expansion of the modern town.

Egypt after his final departure.

56.

57.

African Studies, 156- 168, pis . 17-20, 33-40, 49,

58.

African Studies, 168- 173, pis. 21, 23- 25 , 41-43 ,

map, Site 24. T I, which was redrawn and again described by

map, Site C.T3 was also unrobbed at the time of its entry

the Rev. Goodenough (see Resurfacing of Parae/onium, n.

and contained human skeletal remain s as well as artifacts

32, fig. 6) was unrifled at the time of its initial discovery and

assigned by their discoverer to the early Ptolemaic period.

yielded a significant range of Hellenistic-Roman artifacts.

59.

African Studies, 173- 176, pis. 22, 44-47.

T2 had been robbed in recent times and produced only a few

60.

See Fourtau and n. 16.

dislocated goods, again of the Ptolemaic-Roman period. The

61.

Aji-ican Studies, 136.

whereabouts of the contents of both tomb s, along with what-

Chapter 2

INTRODUCTION TO THE ISLAND'S EXCAVATIONS Donald White

When I first visited the island at Marsa Matruh with my wife Joan in the summer of 1984, we co ll ected a sample of Cypriot White Slip and Roman period surface sherds. 1 Most of the former (Pl. 4A) turned up in the vicinity of what wil l be henceforth referred to as the Sponge-Divers House (SlOl) (Pl. 4B) ; the later sherds were picked up either on the island 's surface or from shallow water over the sa ndbar off its northeast

corner. When ana lyzed together with the 23 mostly Cypriot sherds that Oric Bates sent from the island to the Peabody Museum back in 1914/ the ceramic material appeared sufficiently promising for the University of Pennsylvania Museum to send a full excavation team to Matruh in 1985. The resulting 1985 season combined area survey with excavation and lasted two months. 3

Summary of the University of Pennsylvania Project The 1985 survey activities were limited to mapping the island and its adjacent lagoon shoreline,4 combined with a preliminary effort to rediscover Bates's Libyan "Cemetery A" (Area VI) atop the Great Ridge southeast of the modern town by means of his area map (Plan 4, Fig. I :7). 5 Three areas adjacent to the shoreline around the eastern end of the East Lagoon were selected for investigation in addition to the island (Plan 3): Area II, a low sandstone outcropping directly east of Bates 's Island; 6 Area III, the site of ancient stairs and a rock-cut chamber at the southeast corner of the lagoon; 7 and Area V, the site of a ruined church and tombs on the south slope of the coastal ridge. 8 The coasta I ridge west of Matruh 's harbor

and the territory around Ras Umm el-Rakham , the site of ancient Apis ca. 20 km. west ofMatruh, were explored with special attention directed toward the Ramassid fort at Zawiyet Umm elRakham.9 The bulk of 1985's work was concentrated on excavating the island. A second season of excavation followed an involuntary year's delay and ran for roughly a month during the summer of 1987. 10Architectural plans were executed of Area II's rock-cut features 11 and of the Area V church's surface remains.12The sebka system was reconnoitered as far east as RasA lam el-Rtim and as far west as the coastal territory of Ras Umm el-Rakham in an effort to recover evidence for the presence of the

16

MARSA MATRUH

Late Bronze Age Libyan population. ll A second attempt was made to relocate Bates's Libyan "Cemetery A" in Area VI atop the Great Ridge; although this attempt was again unsuccessful, it led to the retrieval of Libyan Shell-Tempered Ware sherds on the ridge surface west of the supposed location of Bates's discoveries. 14 The remainder of the season was devoted to the excavation of Bates's Island. A final season of excavation and study took place in the summer of 1989. 15 A major objective of that season was to survey the shoreline areas to the east and west of the modern town, with particular attention to the geology of the lagoon basins distributed along the sebka between Matruh's modern harbor and Ras Alam el-RC!m to the east. 16 At the same time that Gardner and Frost conducted their geological survey, other expedition members collected samples of surface pottery over the same area, in particular from the coastal ridge area and its southern slope below the Area V church remains (Plan 3). 17 Soil samples from the island were turned over to Cairo University's N.H. elHadidi, who used his brief stay with the expedition

to collect samples from soil deposits in the vicinity of the island for further evidence of the lagoon's previous floral environment. And lastly, some limited excavation was carried out on the Late Bronze Age walled remains in the island's H5-IIl grid square. The project's faunal remains subsequently were tabulated by Reese in 1991. 18 The Pennsylvania Expedition's final return in 1996 to survey potential Late Bronze Age ports west ofRas Gibeisa involved no activity at Marsa Matruh. 19 It should be said that the 1985, 1986, and 1989 seasons were insufficient to bring about the total clearance of the island, and while its complete excavation was never envisioned as a necessarily appropriate objective, a number of practical obstacles prevented the Pennsylvania Expedition from exposing as much as we would have liked. The net effect is that the present study is based on a useful samp le of what the island has to offer archaeologically but leaves some questions still unanswered, hopefully for future investigators to resolve.

The Archaeological Sequence on Bates's Island The Pennsylvania Expedition's work has made it clear that the island initially had a consistent and prolonged occupation 20 from perhaps the 15th to the 13th century B.C., with its floruit in the 14th century. Based on what has been gleaned from its remains, the Bronze Age occupation was followed by roughly five centuries in which no detectable use of the island took place. Some reuse, but traceable only in the ceramic record, seems to have occurred between the late 8th and later 6th centuries B.C. This is then followed by a more or less continuous occupation, manifested by both walls and artifacts, which appears to have extended from the late 6th century B.C. to the first half of the 5th century A.D., after which the archaeological record again breaks off. By the late 17th century, life on the island picked up again with the construction of a house for sponge-divers (S 10 I) built from reused ancient

materials. It continued to be used in this marginal capacity into the 19th century when the house was abandoned, which is how Bates found it in 1913. Since then, the island has been used seasonally by fishermen and hunters whose shotgun shells and rubble bird-blinds litter the surface. It seems unlikely that it was not used at least for fishing during much of the time of its supposed abandonment in the past. The effects of World War II were drastic. The island's surface was left peppered with shrapnel and rounds from smallarms weapons, with the center blown out of the Sponge-Divers House (Pl. SA, B). At least one foreign national, probably a German soldier, met his end on the island some time before 1943. 21 To judge from the evidence (mainly of overlapping wall remains), the periodization of the island breaks down into six stages, punctuated by various sub-phases.

INTRODUCTION TO THE ISLAND 'S EXCAVATIONS

I. Late Bronze Age (3 sub-phases?) 2. Late Archaic 3. Late Hellenistic/Early Roman Imperial (2 sub-phases?) 4. Later Roman Imperial (2 or more subphases?) 5. Arab (2 sub-phases?) 6. Modern ( 1914- present) The apparent lack of walled remains attributable with certainty to the Classical through Hellenistic years is troubling and may be considered a consequence of insufficient excavation. Clearly, the site's artifactual record, combined with subtle differences in constructional technique observable in the various separate structures, indicates the existence of a considerably more complex and diverse occupational history in keeping with the island's long occupation.

17

Range of Occupational Phases Located Off the Island The Pennsylvania Expedition made no attempt to obtain a complete archaeological record of the neighboring Matruh area in addition to the island, but in the course of its various activities, the following phases of the history of the surrounding area came into focus (Plans 1-4):

Late Bronze Age/Pre-Greek: Areas II, VI, VII, VIII, IX, Umm el-Rakham Area Hellenistic: Area III Roman-Byzantine: Areas ll, IV, V; ridge south of Umm ei-Rakham and west of Wadi ei-Samad.

Archaeological Material from the Matruh Area at Harvard As mentioned above, 22 Harvard University's Peabody Museum of Archaeology and Ethnology possesses a small collection of pottery sent back from the island by Oric Bates following his 1913 to 1919 seasons of survey and excavation. Most of the collection is Late Bronze Age Cypriot pottery. We received the Peabody's permission to study the pottery in 1984, and it is incorporated

in Russell's chapter on the island's Aegean and Cypriot pottery. 23 In 1991, the Peabody Museum staff was able to relocate the various shells, terracotta vessels, and stone implements discovered by Bates in his "Cemetery A" southeast of the modern town (Area VI) / 4 and they were also made available for laboratory testing and eventual presentation in the present publication.

Previous Pennsylvania Expedition Publications Relatively little publication of the results of the University Museum's work at Matruh has preceded the present study. The two JARCE seasonal reports for 1985 and 1987 already have been mentioned on a number of occasions above. 25 The latest of these publications was accompanied by a detailed but preliminary report by L. Hulin on the 1987 pottery finds. 26 The 1985 season provided sufficient information to attempt a tentative outline of the possible implications of a Late Bronze Age Libyan occupation of the main-

land area immediately surrounding the island for the 1986 SOAS-sponsored London conference on "Libya and Egypt prior to 750 B.C." 27 The same subject was further pursued shortly thereafter in a more popular vein. 28 The 1993 colloquium on Cyrenaican archaeology at Newnham College, Cambridge produced a further study of the Bronze Age Libyan presence in both the Marmarica and the Cyrenaica that touches on the evidence from Matruh. 29 A general review of the evidence for Greek, Roman, and Byzantine

18

MARSA MATRUH

Matruh was produced for the British Museum sponsored conference on Greco-Roman Egypt that took place later in the same year. 10 Mention of the site was made as part of a general study of coastal sites along the eastern Marmarican coast that appeared in 1996. 31 A short article on the reciprocal movement of goods to and from the Late Bronze Age island appeared as part of a Festschrift honoring Malcolm Wiener. 32

Finally, abstracts have appeared of the field reports on the 1985- 1987 seasons presented at annual meetings of the Archaeological Institute of America and the American Research Center in Egypt, 33 and copies of the end-of-season reports submitted to the SCA at the conclusion of the 1985 and 1987 seasons have also appeared in print. H Summaries of all seasons, including the 1989 season, are published in Orientalia. 35

Project Design The voyage along the coast of this sea is exceedingly long, and any landing is especially difficult; for from Paraetonium in Libya as .far as lope (i.e., Joppa) in Coele-Syria, a voyage along the coast of some five thousand stades, there is not to befound a safe harbor exceptfor Pharos. (Diod. 1.31 .3, Loeb trans.) Marsa Matruh/ Paraitonion initially was picked for study and excavation because of the obvious centrality of its geographic position at the edge of the eastern Marmarica; in other words, it is set nearly equidistant between the Libyan Pentapolis to its west and Alexandrian Egypt to the east (Fig. 1: 1). Moreover, the historically important oasis of Siwa lies to its south-southwest, at the end of another stretch of ca. 190 miles. 36 For all intents and purposes, the coastal zone west of Matruh up to the edge of the Libyan plateau, which undergoes a precipitous rise at the border town of Salum/ 7 was largely an archaeological unknown in Bates's daylx and stil l remains much the same to this day. The more important protected roadsteads and minor ports between Ras Gibeisa just west of eiAiamein and the Libyan border surveyed for the University of Pennsylvania in the summer of 1996 failed to generate any compelling evidence for the existence of harbor facilities anywhere along this coastline prior to the historical period, apart from the one clear exception of Bates's Island. 39 This, in effect, bears out the essential accuracy of Diodorus's observation that Matruh provides the only significant natural harbor between Tobruk (eastern Cyrenaica) and Alexandria, a stretch of ca. 425 miles of other-

wise bleakly monotonous sandy shoreline. For the moment, one cannot do better than quote Bates's own words: Marsa Matruh, finally, lies almost under the eastern extremity of Crete, distant only 250 miles to the north by west; it is opposite the mouth of the Aegean, and but 370 miles south-easterly from Cyprus. When one considers all its factors, one would be prepared to find that, at a very early period in the history of the Mediterranean, Matruh had served as a center both for the Libyan neighbors of the Egypt and for the Sea Peoples. 40 At its outset, the project's purpose principally was hinged to examining Paraitonion's role as a trade and cultural link between Naucratis and the Cyrenaican Pentapolis before the arrival of Alexander, or at least to verifying whether or not such I inkage (whether trans-desert, by coastal route, or by sea) existed before the town's traditional later 4th century B.C. foundation ." 1 As the fieldwork unfolded between 1985 and 1989, the expedition undertook to investigate and record all phases of the Matruh area's archaeology that came to its attention in the course of its work

INTRODUCTION TO THE ISLAND ' S EXCAVATIONS

both on as well as off the island, including the late Roman period and beyond. Our interest in these scattered later aspects of the area's history is based on a conviction that the archaeological zone comprising the modern town and its environs carries a considerable potential importance.42 On the other hand, with the exception of the various haphazard investigations and discoveries that took place before World War TT 43 and some promising excavation of remains of Roman date along the town's coastal ridge north of the West Lagoon undertaken by the SCA but never published, the bulk of Matruh 's later archaeo logy still remains to be discovered and made known. That sa id, it is plain that the principal interest of the present project lies in Matruh's Bronze Age development. The main purpose of this publi cation is, therefore, to lay forth the results of what can be said about the site long before the permanent sett le ment of Greeks on the Cyrenaican plateau and in the Nile Delta. It should be viewed as a large ly but not exclusively Bronze Age study.

Site Survey Plans and Recording Conventions Whatever "excavation" occurred in connection with work undertaken in the Area II watercollection facility, Area ITT rock-cut stairs, Area V church remains, and the Area VIB hearth basically was limited to clearing away surface soi l or sand in preparation for drawing and photography. Consequently, the project's formal excavation activities were for all intents and purposes limited to Bates's Island, and what follows is a description of the methods and conventions adopted for recording the work on the island. Island Survey Grid: A 10 m. square grid was established over the island 's surface , oriented to true north (Plan 2). Individual survey points made of iron reinforcing rod were cemented into the ground or exposed bedrock but probably will not survive long into the future because of rust and the determined efforts of visitors to vandalize the site. Reference to individual grid squares on the site

19

and individual plans is made by citing their upper right-hand (i.e., northeast) coordinates. Each grid square is then quartered, with Roman numerals I- IV assigned to each quarter beginning with its upper left-hand (i.e., northwest) corner. Structure Numbers: Separate architectural or built features are assigned three digit structure ("S") numbers (Plan 1). The first digit refers to the "Area" (which in the case of the island is "1 "); the second and third digits indicate the sequence in which each individual feature was identified, starting with 0 I. Thus, for example, the seventh architectural feature recorded during the first season of work was S I 07, a Late Bronze Age room that came to light in the southeast quadrant of grid square F4-lll. Island and Lagoon Survey Plans: Two I :200 plans have been prepared of the island! 4 One plan (P lan I) records the natural rock features and contour elevations. Unbroken contour lines occur at meter intervals and refer to elevations above the 1985 sea level. The island's contour plan also plots al l its walled and otherwise manmade architectural features, accompanied, space permitting, by their relevant three-digit "S" number. The second site plan (Plan 2) plots the layout of the 10m. grid and the position of the expedition's trenches designated by their grid coordinates and relevant quadrant number. It also repeats the island's exposed rock features but not the contoured lines. Finally, a I: I 000 scale plan (Plan 3) was prepared of the eastern end of the Eastern Lagoon in order to illustrate the relationship of the island (Area I) to the surrounding Area II, III, and V features as well as to define the outline of its sandbar that at one time seems to have connected the island to the adjacent eastern shore in the vici nity of Area II. The broken contour lines indicate below 1985 sea- leve l depths in intervals of half-meters. Elevation Readings: All plan and section elevations are taken from sea level recorded during the summer of 1985. Recording of Stratigraphy: The island's stratigraphy is recorded by deposits, for present purposes defined as any observable accumulation of

20

MARSA MATRUII

soil, fill (including anthropogenic deposits of refuse, bones, shell, etc.) or (particularly germane in the present instance) sand, distinct from what surrounds it. Each deposit is referred to by two numbers, e.g., "Deposit 2.3". The first number indicates the deposit's actual layer, that is to say, its position relative to what lies above or beneath it. Thus, the nomenclature of any deposit stratified above "2.3" must begin with a "I," while any deposit stratified below it must begin with a "3." The second number refers to the sequence in which deposits were excavated in layers that are otherwise stratigraphically equal. Thus, "2.3" is the third deposit excavated in the layer directly below the ground's surface, which for present purposes is defined as "1." Each trench has its separate sequence of deposit numbers; a 2.3 deposit found in one trench does not necessarily have anything to do with a 2.3 deposit found somewhere else. Where, however, an otherwise homogeneous deposit is divided by either a feature (e.g., a wall) or a point arbitrarily selected by the excavator, its subdivision is indicated by a slash, e.g. 2.3/2.4. Trench Nomenclature: Patterned after the designation of grid squares described above, trenches are designated by both the grid coordinates and the quarter section of the square in which they are located. Thus, "Trench F4-III" is found in the lower right or southeast quarter of the F4 grid square. When trenches extend into more than one quarter, their position is indicated with a slash: e.g., "Trench D4-IIII" is located in the northern half of the D4 grid square. Test probes that are appreciably smaller than an individual quarter-square generally are referred to with the addition of a short, descriptive title, e.g., "G5-IY, southwest test" or "D4-I/ll, north balk." Area Designations: As is apparent already from the above narrative, Marsa Matruh and its surrounding environs have been divided into various "areas" for purposes of reference. The major divisions are as follows: Area !. Area II.

Bates 's Island (Plan I) The rock outcropping on the shore area east of the island (Plan 3)

Area !fl.

The stone stairs and rock-cut chamber along the shore south of the island (Plan I) Area IV The coastal ridge north of the West Lagoon (Plan 4), otherwise known as the "Cleopatra Ridge"45 A rea V The coastal ridge north of the island associated with early church remains (Plan 3) Area VI. The "Great Ridge" southeast of the modern town where Bates's "Cemetery A" was located (Plan 4 )46 Area VII. The ridge formation between Area ll and the Second Eastern Lagoon Area Vfff. The hill between the Second and the Third Eastern Lagoon (Plan 4) Area IX The ridge area directly above the northeast corner of the Fourth Eastern Lagoon (Plan 4) Some of the above areas have been further divided into "A," "B," etc.; these subdivisions will be defined where appropriate. Areas are referred to by Roman numerals as part of object inventory numbers. For example, 851-P-5 refers to a piece of pottery found in 1985 on the island and 87IY-C-l indicates a coin found on the "Cleopatra Ridge." Area also is designated by the initial Arabic numeral when part of an "S" number designates an architectural feature. Thus, "S lOl " is the so-called "S ponge-Divers House" on the island. Although in theory "S5 0 I" should be used to designate the church remains on the Area V ridge, in practice "S" numbers are used only in the case of the island, and they always commence with a "1." Catalog System: A dual record-keeping system, consisting of daily deposit tally sheets on the one hand and a traditional catalog on the other hand, is used to record the project's finds. The catalog system, which utilized typed 5-by-7 inch cards now backed by a computer database, lists objects organized by three-part inventory designations as follows:

INTRODU CTION TO THE ISLAND ' S EXCAVATIONS

Year combined with Area Number Artifact Type designated by a letter Sequence Number within artifact categories designated by the fo llowing abbreviations:

AS

G M MO

pottery stone object terracotta object

so TC

Thus, "891 -P- 1" refers to the first piece of pottery cataloged from the island during the 1989 season. " 871-M-5" is the fifth metal object from the island cataloged in 1987. "85 IV-C-1 " is the first coin cataloged in 1985 from the "Cleopatra Beach" area west of the main harbor.

architectural stone com glass metal miscellaneous object

c

p

21

Chapter 2 Notes I.

By D. and J. White in the summer of 1984.

obj ects draftsman and architect; Linda Hulin, pottery ana-

2.

1985 Report, 52, n. 7. Pamela Russell has included

lyst and ca taloger; Stephanie Tyiska, conservator and cook;

the sherd s, invento ri ed by the Pea bod y Museum as

Eric Gieringer, Roy Green , and Dr. Aa la Shaheen, fie ld

Accession no. 46-4-40, cat. no. 5917, in her study of the

superv isors. Dr. Carl Beetz was responsibl e for putting th e

island's Aegea n and Cypriot wa res (Vol. II, C hapter 7)

architectural field drawin gs in their final form. D. White

where they appear as 7.79- 7.98 .

se rved as project director and photographer.

3.

The staff consisted of the followin g: Dr. Murray

II .

1987 Report, 88, fig. 2. See Vol. II , Chapter 6.

McC lellan, ass istant field director; Dr. Kenneth Schaar and

12.

1987 Report. figs. 3- 5. See a lso Vol. II .

Dr. Marga ret Schaar, surveyors; James Thorn, field architect

13.

1987 Report, 93- 95. 1987 Pol/ery, 11 5- 11 6. See

and draftsm an; Joa n White, conservator; Dr. Pamela Russell ,

Chapter 3.

pottery consultant and cataloger; and David Conwell , area

14.

1987 Report, 93. 1987 Pollery, I 16. See Chapter 3.

supervisor. Dr. Carl Beetz was responsible for putting the

15 .

As with the 1987 season, the year's delay was

architectural field drawings in th eir fin al fom1 after their

ca used by factors beyond o ur contro l and did not reflect a

return to Philadelphia. The present writer was project director

des ire on the U ni versity of Penn sylva ni a Museum 's part to

and photographer.

spread out the fieldwork ; both seasons a lso were shorter

4.

Ca rried out by K. and M. Schaar. 1985 Report, figs.

2, 8, and 9. 5.

1985 Report, 53 . Provisional Evidence, 8ff. See

than originally planned. The 1989 staff consisted of the following persons : Linda Hulin, ass ista nt directo r and Bronze Age ceramics consultant; Chri stopher Hulin , technician

a lso Vo l. II.

and cook; Dr. Rita Gardner and Dr. Martin Frost, geo lo-

6.

Its surface features were mapped, photog raphed,

gists; Dr. Donald Bailey, post-Bronze Age pottery consult-

and record ed but not otherwise excavated. See 1985

ant; Dr. David Reese, fa un al spec ial ist; James Thorn ,

Report, 54-55, fig . 3.

objects drafts man and architect; Joa n and Arthur White,

Already recorded by Fourtau and Bates. See Chapter

excavators; and D. White, proj ect director and photogra-

I. Again, not excavated but instead photographed, drawn ,

pher. The ex pedition was joined for a day by Dr. N. Hab il

and described: 1985 Report, 56, figs. 4-5. See also Vo l. II.

el-Hadidi , palaeobotanist. Dr. Carl Beetz once again com-

Drawn and photographed. 1985 Report, 56 . See

pleted the finished architectural drawings in Philadelphi a.

7.

8.

a lso Vol. II.

9.

1985 Report, 51, n. 2. Provisional Evidence, 6.

For a summ ary of the res ults of thi s season, see Or 59 ( 1990) 336- 337.

See C hapter 3. Apis, 141 - 143. SeeS. Snape, " Ram ses ll 's

16.

Forgotten Frontier," Egyptian Archaeology I I ( 1997)

17.

See a lso Vol. II.

23- 24.

18.

See Vol. II , Chapter I 0.

The staff consisted of the followi ng indi vidual s:

19.

White, Coastal Survey, and below.

Dr. David Co nwell , ass istant field director; James Thorn,

20.

As opposed to va ri o us forms of peri od ic and occa-

I 0.

See C hapter 3.

22

MARSA MATRUH

sional use, which our recovery techniques probably were

Settlement on Bates's Is land, Marsa Matru h, ARC£ Annual

not sufficiently sensitive to record.

Meeting, Program and Abstracts (Phi ladelphia 1989)

21. n. 59.

Cf. 1985 Report, 63. Reswfacing of Parae/onium ,

54- 55.

Seen. I .

34. 8- 12. 35.

22. 23.

I wish to thank Professor Lamberg-Karlovsky,

NARC£ 130 (1985) 3- 12; NARC£ 139 ( 1987) Or 55 (1986) 237; Or 56 (1987) 293-294; Or 57

individuals: Ms. Una MacDonald, Ass istant Co ll ections

( 1988) 308; Or 58 ( 1989) 337- 338; Or 59 ( 1990) 336- 337. 36. " Marsa Matruh is thus situ ated some 175 miles west of Alexandria by land, or 150 miles in a direct lin e by

Manager; and Ms. Germaine Juneau , Assistant Registrar,

sea: while from es-Se llum [sic] the land distance is rough-

then director of the Peabody, for his permission to restudy Bates 's sherd s. Thanks are a lso due to the following oth er

for

arranging

their temporary

transference

to

the

Univers ity of Pennsylvania Museum.

24.

See Vol. II , Chapter I 0. These pieces have been

reported prematurely as lost (see Provisional Evidence, 9. 1987 Report, 93. Conwell, Ostrich Eggs, 32). l wish to

thank the Peabody 's staff: its then director, Prof. David Tilbeam; its Col lections Manager, Ms. Katherine Skelly; its Assistant

Registrar, Ms.

Genevieve

ly 150 miles. The port is thus geographically significant, being situated nearly halfway between the ferti le region of

Fisher;

its

Archaeologica l Col lections Manager, Ms. Gloria Greis; and Mrs. A.B. Isaac, its Photographic Archivist, for relo-

Cyrenaica on the one hand, and the Ni le Delta on the other." African Studies, 126.

3 7.

White, Coastal Survey, 15- 18.

38.

Bates himself was unaware of Fourtau's 1904 sur-

vey of this area, for which see White, Coastal Survey.

39.

White, Coastal Survey.

40.

African Studies, 126. This point is discussed fur-

ther in White, Coastal Survey.

cating this material as well as for their assistance in mak-

41.

See Vol. II , Chapter 6.

ing it available for publication.

42.

See also my survey of the area's post-Bronze Age

25. 26. 27. 28.

1985 Report, 51 - 84 and 1987 Report. 87- 11 4.

I.e. , 1987 Polle1y, 115- 126. Provisional Evidence, 1- 14. Conwell, Ostrich Eggs, 25- 34. D. O'Connor,

"Egyptians a nd Libyans in the New Kingdom ," Expedition

29, no. 3 ( 1987) 35-37. See also "An Archaeologist Finds a Safe Harbor in Egypt," The Pennsylvania Ga::::e/le 84, no.

5 (1986) 17- 18. 29. White, Before the Greeks Came. 30. Resurfacing of Para e/onium, which includes a summary of the pottery evidence by D. Bailey.

31. 32. 33 .

developments, Resu1j"acing of Parae/onium, passim.

43. 44.

See Chapter I. Both represent redrawings on the original 1985

:I 00 sca le survey of the island undertaken by K. Schaar.

45 . For more on the name a nd extent of thi s area, see Resurfacing of Parae/onium, n. 54. See also R. Maugham, Journey to Siwa (London 1950) pi s. opp. pp. 18 and 36. 46. The 1985 Report, 56, incorrectly located Area VI south of the main road leading from Marsa Matruh east to el-Alamein and Alexandria. That road, in fact, li es south of what Bates ca ll ed the Great Ridge, which act ual ly li es

White, Coastal Survey, 11 - 30, passim .

parallel to and south of the minor coastal road that con-

White, Wate1; Wood.

nects Matruh with the Riis Alam el-Rum area to its east.

D. White, "Excavations of Bates 's Is land, A Late

Neither road appears on the l :25 ,000 scale area map as it

Bronze Age Egyptia n Trading Station," AJA 90 ( 1986)

205- 206 . D. White, "Excavations of a Late Bronze Age

is rep rod uced he re.

Chapter 3

ENVIRONMENTAL MORPHOLOGY AND HISTORY OF THE ISLAND AND ADJACENT LAGOON AREA Donald White and Rita Gardner No attempt was made by the Pennsylvania Expedition to exp lore Matruh 's West Lagoon (Figs. I :2, I :4, I :7) for the simple reason that the eastern half of its southern shore, where at one time Fourtau and Bates were able to observe remains of a Classical urban settlement as well as an ancient pier, had been transfonned before our arrival into a mod-

ern concrete deep-water docking facility. The rest of the shore area, farther to its west, was posted offlimits by the Egyptian military authorities. 1 Instead, the expedition's attention was concentrated primarily on the five separate lagoons that connect Matruh's modem harbor with the Ras Alam el-RCun headland ca. 8 km. to the east (Plan 4; Pl. 6A, B).

Fluctuations in Water-Level from the Archaeological Perspective Donald White

Fluctuations in Recent Times A cursory comparison between the area maps published by Fourtau and Bates (Figs. l :3 and l :7) w ith the versio n emp loyed for the present study (P lan 1) 2 will indicate that real differences exist between the various plans avai lable for judg ing the size and outline of the indiv idu al lagoons . These differe nces may reflect not merely fluctuating standards of cartographic accuracy. Sixty years separate the present day from the 193 8/ I 94 1 survey that forms the basis of our own area map. By the late 1930s the Third and Fourth Lagoons were joined together by a thin water-

channe l, whereas today they are independent and appear to be smaller, at least during the dry summer months. This change could result simp ly from the time of year in w hi ch they were drawn or some other natural variable; during the summer of 1987 when lagoon leve ls were at their lowest for the year, the co mbinatio n of moon tides and wi nd drove a thin sheet of water over the sand flats separating the First Lagoon from the Second Lagoon, in effect transforming them for a day or two into a uni fied lake (Pl. 7 A, B). Moreover, we k11ow that Bates's Island, which is sited at the eastern end of Matruh 's First Eastern Lagoon, is today separated from the shore by a

24

MARSA MATRUH

depth of 1.30 m. or more of water throughout most of the year. And yet, living observers have reported seeing it connected to the adjacent eastern shore by an exposed, dry sandbar (Pl. SA). The rise in level seems to be attributable to the fact that a paved causeway, under which seawater now flows from Matruh's inner harbor into the East Lagoon, was installed in 1946 together with a water-channel and sluice-gate. By way of contrast, in Bates's day 80 years ago, ca. 350m. of dry land, still visible on the 1938/ 1941 survey plan, separated the lagoon from the inner harbor; in 1913/1914 the island's sandbar was covered by ankle-deep water in the summer and waist-high water in the winter. 3 The impression that one is left with is that sizeable rises and falls in water-level have occurred throughout the entire lagoon system over the past century. Apart from man-made terrain modifications such as the causeway, not to mention the effects of World War II shelling and aerial bombardment, change is brought about routinely by the seasonal fluctuations in the coastal weather, sea storms, and tidal action. What is calculated less easily is how much the annual runoff of winter rain over the coastal plain has led to silting up the water-channels connecting the various lagoons, and to what extent this can be held responsible for modifying the shape and size of the lagoons over the short run. In any case, the impact of erosion cannot be discounted in terms of longer intervals oftime:

Greco-Roman Era Fluctuations Fourtau 's 5 and Bates's6 observations of docking facilities along the southern and eastern shorelines of the West Lagoon at the beginning of the century suggest that the water-level at Matruh must be today more or less what it was in Roman times. This conclusion is further sustained by Ball's observations concerning the floor level of the aqueduct (Fig. 1:2) surveyed by Walpole during the 1930s west of Matruh: This subterranean aqueduct, whose floor is only a few centimeters above present

Mediterranean-level , was evidently excavated for the purpose of obtaining supplies of potable water for a Greco-Roman settlement that existed in its vicinity, by tapping the thin sheet of drainage-water that in this locality flows underground from the inland hills to reach the sea; and after having been cleared of the sand which had partially choked it in the course of the intervening centuries, it now forms an important source of drinking-water for the modern town of Matruh. Had the sea-level at the time when the aqueduct was excavated been some two or three meters lower, relatively to the land of the locality, than it is at the present day, it would seem impossible that the underground sheet of fresh water, which overlies salt water and does not exceed a meter or so in thickness, could have reached to a level sufficiently high for it to be tapped by the aqueduct. 7 More directly relevant to the present discussio n is the fact that if the ancient sea-level had been any higher, it would have contaminated the aqueduct's channel with salt water seeping through the sandy coastal ridge into the coastal plain. On the other hand, scatters of submerged sherds on the sa ndbar extending off the northeast tip of Bates's Island, as well as from three other off-shore sites linked to the island,8 strongly suggest that the post-Ptolemaic period water-level in the East Lagoon stood at a lower level than it does today, and that the island was connected to the land during at least some of the Imperial period. We have seen how variations in weather patterns, time of year, and localized silting have created similar short-lived "dry" conditions both during and since Bates's day. In addition, there is archaeological evidence to bolster the conclusion that, as far as Matruh's First East Lagoon is concerned, the water-level in the Greek period stood more or less where it is today. During the Roman period, drop in waterlevel may have persisted for a fairly extended period of time despite the absence of supporting sedimentary evidence. 9 Apart from the underwater scatters of sherds (Pl. 8B), which could owe

ENVIRONME TAL MORPHOLOGY AND HISTORY

their presence on the lagoon floor to chance dumping either from boats or from once being on dry land, the basis for our assuming a lower water-level for antiquity rests on the presence on the island of a wall whose footings lie under present-day sea- level and three additiona l submerged wall features recorded in the deeper waters to either side of the island's sandbar. The land wall, S 11 2, which could date as early as the 6th century B.C., constitutes the lowest set of built remains found on the island (Plan 4). 10 It came to light in the 18 square due west of the sandbar. During the course of its excavation, despite being set back from the water's edge nearly 8 m., S 112 's lowest footings were inundated w ith lagoon water each morning before the su n had a chance to dry them out. In winter, when the water-level ri ses, they must be immersed fully, suggesting that at one time the island was alm ost certainly slight ly larger and its surrounding water-level lower than at present. A short distance south of S 112, a line of east-west wal l (S I 04) was noted running nearly to the beach. If S I 04 's line is continued east on paper, it forms a right angle with the underwater extension of a north-south Iine of ashlars (S 139) which came to li ght in the summer of 1989 at a depth of ca. 1.00 m. below sea-leve l in grid square G 10. If the two walls were once part of a single dry-land structure, the island's Greco-Roman shoreline would have extended at least seven meters beyond its present beach line. Nine meters east of the island 's northeast tip, a short length of north-south running ashlars (S Ill ) was recorded at a depth of 1.30 meters below sea-leve l in the summer of 1985 (Pl. 9A). Consisting of only four stones, the wall was surrounded partially by a thick scatter of sherds on the lagoon's floor. Finally, a sma ll rectangular slot (S140), 0.49 by 0.27 by -0.04 m., was observed (summer 1985) cut into the sandstone she lf and subm erged in roughly a meter of water 2.0 m. off the northern tip ofthe island (grid square L7). Too little is known to specu late whether this feature was part of a larger dry-land extension of the island. In theory, all four elements, Si ll , Sl31, Sl39, and S 140, could have belonged to piers or jetties

25

run out into the lagoon to se rvice the docking of small boats. The build-up of sherds surrounding S 11 I and those in the water off the north and east sides of the island (grid squares L8 and 06/ 07 respectively), however, seem to point instead to their belonging at one time to a dry-land extension of the island. ln later antiquity and for short intervals in the 20th century, the island was linked to the Area 11 shoreline by a land-bridge (Pl. 8A). The archaeological data suggest that the initial radiocarbon dates reached for Gardner's seq uence of fluctuations in lagoon water-levels (see below) need some adjustment.

Bronze Age Period Fluctuations The combined archaeometrica l/archaeologica l evidence for the water- leve l of the preArchaic Late Bronze Age lagoon system suggests the following regional model. 11 A ca. 0.15 m. deep bed of rounded, sea-borne pumice (P is. 9B, 1OA) was surveyed in 1989 at ca . 0.50/0.75 m. a.s.l., a short distance inland from the so utheast shore of the most eastern of Matruh 's five lagoons, located less than I km. from the Ras A lam ei-Rum headland (Pis. 6A, B, I OB). According to the results of Gardner's analys is, 12 the samples originated from Thera, which, ifwe are dealing with the eruptions of the approxi mate mid-2nd millennium B.C., 13 would indicate that the water-level throughout the lagoon system shortly before the Late Bronze Age occupation of Bates's Island might have stood only 0.50/0.75 m. above its present-day levels. More pieces of pumice were excavated from island deposits associated with everything from contemporary surface scatters to sea led Late Bronze Age levels. For examp le, we found the ground surface in H4-ll off the northwest corner of the Sponge-Divers' House (S I 0 I) and west of the H5-J trench libera lly sprinkled with pumice (Pl. I I A). Some of the pebble-sized stones may have floated onto the lower slopes of the island during Gardner's final high-water-level period when the lagoon water may have stood ca. 2.5 m. above its present level. 14 On the other hand, there

26

MARSA MATRUH

is some real possibility that a significant percentage of the island's pumice was introduced by its occupants for a variety of purposes rather than havi ng been deposited by waves. It would be, of course, useful to know precisely when the pumice bed northwest of the Fifth Lagoon was establi shed. The samples themselves have not been dated by laboratory methods. Even though Gardner has ruled out the effects of a tsunami or giant sea-wave, the deposit can be attributed only to natural causes, which means that the individual water-borne pebbles had to be floated in. Despite the fact that a half-meter rise in water would have been theoretically enough to join all five lagoons into a single inland lake, 15 it seems implausible that such a large concentration of pumice would have drifted east from the mouth of Matruh 's modern harbor without being deposited somewhere first along its 5 km. length. This all combines to suggest that the lagoon system was open to the sea at its opposite, Ras Alam e i-RCm1, end during the Late Bronze Age period (Pl. 11 B). 16 No archaeologica lly significant deposit of Late Bronze Age material was excavated on the island lower than a level of 2.25 m. a.s.l., 17 and it has been something of an article of faith with its investigators that the Late Bronze Age waterlevel stood significantly higher in the lagoon than it does today, perhaps by as much as nearly 3 m. 1x This position now must be modified to accommodate the evidence of the pumice bed at ca. 0.50/0.75 m. a.s.l. Unless the latter owes its deposition to volcanic activity on Santorini at some period other than its later I 7th (or less

plausibly mid-16th) century B.c. eruption, 19 it would seem to confirm the fact that Bates 's Island was smaller in the Late Bronze Age than it is today, but neither as smal l nor as well-protected as the archaeological evidence from the island previously had seemed to indicate. A rise of0.75 m., the maximum amount admissible according to the evidence of the pumice bed, would put the lagoon 's water-level across the sand bar connecting the island to the shore at just over a man's head. This would have afforded the Late Bronze Age occupants a real measure of protection from the local pastoral Berber-Libyan population who cannot be presumed to have been readily attracted to swimming or boating. At the same time, a 0.75 m. rise would have had the effect of flooding a number of today 's dry-land hilltop sites situated down the length of the then sing le lagoon basin that connected the modem harbor mouth with an outlet to the sea by RasA lam el-Rum (Pis. 6A, B, 1 I B). These newly created "islets" would have been surrounded by only waist-high water, which afforded minimal protection against intrusion. This may be why no traces of a Late Bronze Age occupation by either Egyptians or foreigners simi lar to that on Bates's Island were found on the high ground sites associated with the lagoon system.20 On the other hand, a series of surface scatters of poorly dated Marmaric Ware sherds recovered along the lagoon system could be argued to indicate the presence of Libyan pastoralists in the vicinity of Bates's Island perhaps as early as the time of its Late Bronze Age occupation. 21

Marsa Matruh Environmental History Rita Gardner

Environmental Setting of Marsa Matruh Area The Matruh area has an unusual geomorphological setting compared with other parts of the western Mediterranean coast of Egypt. 22 It is characterized by a number of linked lagoonal

basins, only one of which (the Matruh lagoon) has direct access to the sea at the present time. There are seven lagoonal basins in all (Pl. 6A, B), a lthough the five easternmost basins are largely silted up and have saline playas in their bases. The two western lagoons are used for shipping and military purposes, and it is likely

ENYIRO MENTAL MORPHOLOGY A D HISTORY

that they have been dredged and deepened artificially over the past 50 years or so. The present entrance to the lagoonal systems from the Mediterranean through the Matruh lagoon , however, appears to be natural , because a welldeveloped marine-cut shore platform is found close to sea-level around the entrance. The existence of the lagoons is a direct result of earlier coastal development, probably in the late Pleistocene, when three major ridges of aeolian (dune) sand accumulated in the area and were then cemented by calcium carbonate into aeolianite. The first ridge, like the others trending east-west, was blown up against, and partially over, the Pliocene rocks which previously formed the coastline. It is found today as a wellcemented aeolianite with a calcrete capping. The outer two ridges , which are roughly parallel to the inner one, are also aeo lian in origin, but they are less well-cemented. Thus, they are probably significantly younger in age; late Pleistocene is most likely given the age of many comparable aeolianites around th e Mediterranean. The ridges were formed at a time of lower sea-level relative to land, as they pass beneath the sea today. The lagoons lie between the outer two aeolian ridges , occupying the old interdune corridor. There is no evidence that this corridor was flooded before the Holocene, at which point the lagoons came into being. The outer ridge (Pl. 12A), which separates the lagoons from the sea, has been eroded and breached in only one place, namely at Matruh. The only other possible entrance for the sea is at the eastern end (i.e., by Ra s Alam el-RCun) of the whole lagoon system where, owing to a change in the direction of the coastline, the two aeolianite ridges are open to the sea. At present, this access is closed-off by a barrier beach with patches of blown sand behind it (Pl. liB) . The outer ridge also is covered partially in places on its southern side by modern blown sands (Pl. 12B). The area in which aeolian activity is most active now, however, is to be found on the coast to the west of Matruh. Thus, the lagoon system is a highly protected environment; it is protected from the waves and the wind by the outer ridge ofaeolianite, which reaches heights in excess of

27

34m. above sea-level (Pl. 12A). While the main aeolianite ridges trend east to west, it is their detailed topography and that of the interdune hollows that control the size and shape of the lagoon basins. The ridges were not formed perfectly straight (Pl. 13A). The interdune hollow was not uniform , and subsequent erosion also will have modified the topography. Small dunes' masses appear to have occupied part of the interdune area, poss ibly linking the two ridges in places; these locations are found today as upstanding areas, islands arising from the lagoon floors. Bates's Island is one such example, being found toward the eastern end of the Matruh lagoon, ri sing to a height of over 6 m . above the level of the lagoon (Pl. 13B).

Holocene Stratigraphy The Holocene stratigraphy was investigated in the area between Matruh and Ras Alam el-Rum at the eastern end of the lagoon system. Most emphasis was placed on the stratigraphy around the Matruh lagoon , especially on its sheltered northeastern and eastern sides where the sedimentary record was best preserved and which was, fortuitously, the area closest to Bates's Island. An examination of the many exposures created by gullying and anthropogenic activity revea led a consistent stratigraphy for the Matruh lagoon . This stratigraphy a lso was supported by a much smaller number of investigations in the other lagoon systems farther east. Before considering the stratigraphy in detail , it is worth mentioning the problems and advantages characteristic of this type of environment. Most importantly, the lagoons are an enclosed and sheltered environment and have , therefore, suffered little from substa ntial erosion or from the introduction of large volumes of sediments from outside the immediately surrounding area. Thus, the lagoons are characterized by the accumulation of sandy sediments that were derived largely from local sources including the aeolianite. Furthermore, in such an environment, a good record of sedimentary hi story often is preserved. This enclosed nature also mean s, however, that

28

MARSA MATRUH

there can be a problem of reworking of the sediments; that is, older materials can be re-eroded locally and incorporated into younger sediments. In this instance, a radiocarbon date on an uncontaminated shell sample will represent a maximum age for the enclosing sediment (i.e., the sediment contains reworked shells that are older than the enclosing sediment). The sediments can be divided into four types: (a) Dark gray sandy clays, which occupy the

deeper levels examined in terms of stratigraphy. These sediments were observed in pits dug below present lagoon level to the east and southeast of Bates's Island. They are typical of a lower energy lagoon environment in which the finer particles tend to be concentrated. The sediments contain abundant cockle shells (Cerastoderma edule glaucum) in life position. (b) Sandy beds containing shell fragments and

whole shells, but where the whole shells are not in life position and clearly have been transported and accumulated in shallow water near the margins ~fthe lagoon. The sands are usually near-horizontally bedded, although occasionally crossbedding can also be seen. These sediments were found around the margins of the present lagoon, extending from about present lagoon level to a height in excess of 2.5 m. above present level. All of the shell species identified by David Reese, with the exception of the land snail and the freshwater gastropod, were found within these sediments. Those most commonly encountered were Cerastoderma and Cerithium. (c) Rounded, water-worn particles of pumice interbedded with the shelly sand beds. The pumice typically was found in laterally continuous layers at or just above (to 0.5 m.) present lagoon level around Matruh lagoon, and possibly at elevations of up to 0.75 m. above present lagoon level in lagoons farther to the east (Pl. 9B). The pumice fragments occur in one or two beds of up to 15 em. thickness; if there are two beds, they are close together. Sizes of fragments were found to vary between 0.5 em. to, rarely, over 5.0 em.; most were in the range of 12 em. diameter. The pumice beds contained little in the way of sand or other particles. (d) Sandy, poorly to well-bedded hills/ope deposits (colluvium) that do not contain marine or

lagoonal shell species, but which can contain small land snails. These sands have derived from erosion and redeposition downslope of particles released by weathering of the aeolianite. Aeolianite fragments also can be found within the sands which tend to accumulate at the change of slope where the aeolianite ridges merge into the lagoon. Pottery fragments and charcoal accumulations from fires are common in these deposits, probably because these materials formed the past surfaces of the hill slopes close to the lagoon. Gullies incise through these deposits in several places. All of these deposits were observed in the eastern portion of Matruh lagoon, and all except the gray clays were observed in the other lagoons farther east. The stratigraphy of these deposits around the Matruh lagoon is as follows: Lagoon Deposits

Height (a.s./.)

shelly sands pw11ice bed shelly sands gray clays

to 2.5 m. c.O.Sm. c. 0.0 m. > -0.3 m.

Hills/ope Activity Colluviwn m Gullying Colluviwn II Colluvium II Colluviw11 II Clays Gullying Colluviw11l

Late Holocene Chronology Volcanic eruptions are common in the Holocene history of the Mediterranean; not all of them are known, and relatively few are well documented. The type of eruption needed to supply pumice in large quantities is a volcaniclastic eruption; Thera and the volcanoes of southern Italy are of this type. Thera has, of course, been studied far more than any of the others. Sea-borne pumice has been documented from within raised beach sediments in several areas of the eastern Mediterranean, including Israel, Cyprus, Crete, and the northern Aegean islands. ".1 The sea-borne pumice is not the same, however, at all locations. It falls into two main groups mineralogically and chemically: Theran and rhyodacite pumices. The groups are differentiated mainly on the presence (rhyodacite type) or absence (Theran

E VIRO MENTAL MORPHOLOGY AND HISTORY

type) of biotite and on the ratios of silica, aluminum, and potassium oxides. The two types are thought to be almost coeval, as they are found together in some locations; those of the rhyodacite type appear to be more widespread than those of Theran type. Unfortunately, the origin of the rhyodacite ones is as yet unknown. Analyses of samples from Egypt were undertaken to try to determine whether they matched either of the Theran or rhyodacite types, and hence to determine, if possible, their origin and approximate age. Samples from four locations (Bates's Island, northeastern Matruh lagoon, and two from the lagoons farther east) were analyzed chemically using the ICP (inductively coupled plasma spectrophotometer) and mineralogically in thin section. This examination revealed that they are more similar to the Theran pumices chemically than to the rhyodacite type; moreover, they do not contain biotite. Thus, it would seem that the pumice is likely to be Theran in origin, and therefore from the 17th century B.C., given the recent revision in dates for this event. 24 This conclusion implies that the sea-level was close to or just above that of the present level at about this time. It is difficult to be very precise regarding height, partly because of the problems of defining modern mean sea-level in a fluctuating lagoon system. The Theran eruption would have been accompanied by a tsunami (giant sea wave), which may have pushed the pumice up to a level above that of normal sea-level. This method of deposition for the pumice deposited here, however, is not very likely for three reasons: first, the tsunami would have reached the coastline before the wind-blown pumice traveling across the sea; secondly, the effect of any tsunami would be far less within the lagoon than on the open coast; and thirdly the pumice forms a clear layer devoid of other sediment that would have been washed up in a maJor wave.

Radiocarbon Dating Radiocarbon dating was undertaken on nine shell samples and two charcoal sa mples in order to date the sea-level hi story of the Matruh lagoon

29

and the lagoon to the east and, by implication, the whole lagoon system. The shells were cleaned carefully, and subsamples from two samples were checked for contamination both using X-ray diffraction analysis of powdered shell (to check for contamination of aragonite with calcite) and in thin section. Both methods revealed that there was no contamination of the shells checked. Nevertheless, there is still some risk of contamination because it is impossible to check exactly the same shells that are dated, and because it is impossible to check the sa me shells that were submitted to the laboratory for radiocarbon dating and destroyed in the process of analysis. Given the results of the checks, it is unlikely, however, that the shells dated had suffered from contamination with more modern carbonate. Moreover, in shells of this young age, the effect of a small amount of contamination is relatively in sign ificant. If contamination had occurred, then the dates would be underestimates of true age. If the sediments had been affected by reworking and the incorporation of older shells, then the radiocarbon ages of the shells would overestimate the age of the sediments. The charcoal samples in this environment are unlikely to have suffered any contamination at all. The resulting dates are internally consistent and consistent with the stratigraphic position from which the samples were extracted. The results are as follows, with the analysis supplied by Beta Analytic, Inc., Coral Gables, Florida. 25 The years before present (B.P.) are subtracted from A.D. 1950. This dating is consistent with the island's occupants having transported water-worn pumice up to the top of the island (pumice found in the Late Bronze Age occupation leve ls). In other words, the eruption predates the occupation. Sea-level was at or just above present level at the time of the Theran eruption; water-borne pumice was carried by currents and waves into the Matruh lagoon. The pumice accumulated at the shoreline and has remained there within the sheltered lagoon environment, more or less undisturbed. The date of ca. 1165- 965 B.C. suggests that sea-level was sti ll close to present leve l after the deposition of the pumice. This date is

30

MARSA MATRUH

supported further by the date of ca. 873- 773 B.C. for cockle shells in the gray clay. It would appear that there was little deposition taking place at the margins of the lagoon over the period from the deposition of the pumice until the late 2nd century B.c. to early 1st century A.D., when sea-level showed the first signs of rising. (This sequence is supported by the stratigraphy at site F in the lagoon to the east.) From the late 2nd century B. c ./early Jst century A.D. to apparently the early 9th/first half of the 1Oth century A.D. at least, there was a rise in sealevel relative to the land . It was initially fairly sudden, a rise to 1.5 m. having taken place by the 1st/mid-2nd century A.D. It is possible that it occurred even more suddenly as the sedimentation would naturally lag behind the rise in level. There is then a gap in the dated record until the 9th/early 1Oth century A.D. when the level seems to be approximately 2.5 m. above present. The dates suggest that this high level persisted until approximately mid 14th/mid 15th century A.D., and since then the sea-level relative to the land has fallen back to present level. It is not possible to say from the field evidence available when this drop in relative sea-level occurred, but it would appear to have happened quite suddenly, as there is little in the way of a regressive sequence of sediments preserved. Over the same time interval, a sequence of sedimentation and erosion was taking place on the aeolianite hill slopes adjoining the lagoon. This sedimentation is best preserved on the northern side of the Matruh lagoon. Deposition of colluvial sands commenced prior to the Theran eruption recorded by the pumice. This earliest visible colluvium (I) was gullied prior to being covered by the second colluvium (II), which in its base contains pockets of windblown sand in which small rounded fragments of pumice (less than 2 mm. in diameter) are present. It is suggested that these fragments were removed by wind erosion from the pumice exposed at the strandline and then incorporated into the colluvium. In areas away from the rising sea, Colluvium II continued to form as the sealevel rose. Where the sea cut in against the aeolianite hill slopes, the colluvium was removed.

Thus, along much of the northern edge of the Matruh lagoon, colluvium only is observed again once the sea-level started to fall. It often is found both underlying and overlying the lagoonal sediments in areas close to the lagoon . Colluvium II was gullied quite intensely prior to the deposition of Colluvium Ill, which, like the upper parts of Colluvium II, contains abundant pottery. Colluvium III differs, however, in showing signs of substantial disturbance by human activity.

Archaeological Chronology There are two aspects to the archaeological chronology: the archaeological materials observed within the sections studied for sedimentological purposes and the independent archaeological analysis based upon studies of Bates 's Island.

Archaeological Materials Within Sections Studied Pottery was observed in a number of sections in situ: Roman pottery was observed widely within the upper part of Colluvium II and in Colluvium Ill, which is a disturbed reworking of Colluvium II in large part. This is consistent with the dates obtained from radiocarbon dating. Hellenistic pottery, identified by Donald Bailey, was observed at a height of 1.3 m. a.s.l. in section C, from which dates of A.D. 15- 153 and 51 - 168 were obtained for shells at a similar height. It is unlikely that reworking of coastal material would give an age of both shells and pottery that was concordant. Pottery always was absent from Colluvium I and from the horizons with pumice in situ.

Archaeological Studies Roman walls with associated painted wall plaster are in the northeast corner of Bates's Island, the highest being at J.3 m. a.s.l. Beneath these walls are two sets of earlier walls which go back to the Greek Late Archaic period and whose footings are

ENVIRONMENTAL MORPHOLOGY AND HISTORY

YEARSB.P. c 1

31

UNCALffiRA TED

CALmRATED

2.5 m a.s.l. (N of Bates's Island, shells)

950 B.P. +/-70

A.D. 1000+/-70

A.D. 1344-1446

2

2.4 m a.s.l. (E of Bates' s Island, shells)

1530 B.P. +/-50

A.D. 420 +/-50

A.D. 802-934

3

between 1.0-1.5 m a.s.l. (in lagoon to east, shells)

2240 B.P. +/-50

290 B.C. +/-50

A.D. 51-168

4

between 1.0-1.5 m a.s.l. (in lagoon to east, shells)

2260 B.P. +/-60

310 B.C. +/-60

A.D. 15-153

5

between 1.0-1.5 m a.s.l. (in lagoon to east, shells)

2280 B.P. +/-50

330 B.C. +/-50

A.D. 12-124

6

overlying pumice (0.5-{).7 m a.s.l., shells)

2370 B.P. +/-60

420 B.C. +/-60

114 B.C.-A.D. 34

7

gray clay, -0.2 m a.s.l. (not around lagoon edge, shells)

3000 B.P. +/-60

1050 B.C. +/-60

873-773 B.C.

8

sands at 0.1 m a.s.l. (not around lagoon edge, shells)

3210 B.P. +/-70

1260 B.C. +/-70

1165-965 B.C.

9

top of colluvium 1 (charcoal)

3180 B.P. +/-120

1230 B.C. +/-120

1615-1310 B.C.

10

base of colluvium 1 (charcoal)

3930 B.P. +/-90

1980 B.C. +/-90

2566-2290 B.C.

11

on Bates's Island (shells)

3550 B.P. +/-60

1600 B.C. +/-60

1543-1408 B.C.

Table 3:1. Sample Radiocarbon Dates. just above present sea-leve l. 26 Roman sherds from the island and sandbar date to the period 100 B.C.- A.D. 600. 27 Two sets of walls, currently underwater off the island 's north tip, look to be Roman in age. 28 Fourtau 's (1903) observations suggested that the water-level in the ancient Roman period in the west lagoon was sim il ar to the level existing in 1903. This conclusion was based on the presence of Roman dock structures. 29 A Roman aqueduct exists at Marsa Matruh; its floor is at a level close to present sea-level (Fig. 1:2).30 A higher sea-level (by more than about 0.9 m.) would have rendered the aqueduct inoperable, because it would have tapped saline water, and a

lower sea-level wou ld have left it unable to tap any water. 3 1 The implication from all of these factors is that the sea-level was no higher than today during part of the Roman period, and it has been suggested above that it even may have been lower, thus joining the eastern end of the island to the mainland.

Reconciling the Chronologies There is little doubt in my mind, given the stratigraphy and dates, that sea-level during the Late Bronze Age was at, or possibly just above (max. 0.5 m .), present level ("present" equaling

32

MARSA MATRUII

the time we undertook the fieldwork, at which time the sluice was open). During the Late Bronze Age, the lagoon system was open at both ends, at Matruh and at the eastern end. This situation is shown by the presence of a pumice layer at a depth in the sediments that have silted up the eastern lagoon. The pumice is overlain by shelly sands of lagoonal origin. The depth of water in the eastern end of the lagoon is unknown in the Late Bronze Age, a lthough the aeolianite is fairly close to the surface, suggesting that it was quite shallow. There may, however, have been natural, deeper channels that would be very difficult to trace without undertaking a resistivity survey. There is also little doubt that sea-level started to rise in the late 2nd century B.c./early first century A.D. Thus, the walls built close to sea-level on Bates's Island and dating to ca. 500 B.C. are not inconsistent with this story. The real problem comes with the Roman period, when so much archaeological evidence suggests a relatively low sea-level, and yet the evidence presented here suggests that a high sealevel persisted throughout this period. It seems to me that there are two possible explanations. The first is that there was a negative oscillation in the period between the 1st/early 2nd century A.D. and the 9th/early lOth century A.D. The field evidence, however, offers no support for such a hypothesis. There is no evidence of gullying of the deposits at this time, of any unconformity in the depositional sequence, or of an intervening colluvial layer.

The second possible explanation is that the dates for the highest sea-level are contaminated by younger shells being incorporated into the deposits. Thus, the sea-level high could have peaked earlier, say around 50 B.C., and then fallen to about present level. This situation would leave the Roman period free for the various constructional activities on Bates's Island and elsewhere at a height comparable to modern sealevel. This latter hypothesis is preferable because, of all the materials analyzed, these are the ones most likely to be contaminated.'" This view may be supported by the presence of later Hellenistic and Roman pottery within gullies in the colluvium." Some of the gullies are eroded down close to modern lagoon level. This erosion could suggest a lower sea-level at around that time, but it need not, because the gullies could have formed at any time after 50 B.C., and the pottery could have been washed into the gu llies during a storm. lfwe accept that the highest shell deposits may be misleading in terms of age, then there is conclusive evidence for a rise in sea-level to between 1.5 and 2.0 m. above present level, at which time the notch on Bates's Island may have been eroded by the sea. (Alternatively, the notch could relate to an earlier sea-level of this approximate height as seen in notches cut in the coastal cliffs at RasA lam el-Rum). This rise in sea-level started in the 1st century B.c./early 1st century A.D. and probably only lasted for a period of between 200 and 400 years, if we take into account the archaeological evidence.

Chapter 3 Notes I.

The east end of the West Lagoon, the location of

is based on the I :25,000 scale map, revised by the Egyptian

Fourtau's pier and towers, lies outside the military zone but

Dept. of Survey and Mines in 1941 from the 1938 survey by

also has been subjected to heavy building. See Resu1j"acing

Khalcf Mursi (not 1932 as erroneously stated, 1987 Report,

of Parae/onium, 67, 71 - 74 as well as Chapter 2.

89, fig. I), Matruh Area sheet no. 2. This is the same as the

2.

Fourtau's drawing (Fig. I :3) is a rough thumbnail

German staff map, Matruch. Sonderausgabe VJJ sheet no. 2,

sketch, but Bates's drawing (Fig. I :7) is based on the

1941. For further discussion of the differences between

Egyptian Survey Department maps "District of Marsa

Fourtau's and Bates's plans, see Chapter l, n. 19.

Matruh" and "District of Ras Allcm cl RCtm ," the British Admiralty chart no. 3567, and personal observation. Plan I

3.

Afi'ican Studies, 185 186. Bates seems to have had

an antipathy to getting wet since he writes of being carried

ENVIRONME TAL MORPHOLOGY A D HISTORY

over to the island by one of his laborers as well as by camel. Understandably, working in wet clothes on a cold, windy day in January had little appeal. 4. See Ajhcan Studies, 127 for winter downpours. E. Semple, The Geography of the Mediterranean Region (New York 1931) 85- 93 and W. Fitzgerald, Africa, a Social, Economic, and Political Geography of its Major Regions (7th. ed., London 1950) 61 - 63 provide conventional statements on the normal aridity of the region. Bates's theory that the West Lagoon and the First East Lagoon originally were integrated with the main harbor assumes that their formation into separate parts was caused by silting. See Chapter 2, nn. 43 and 45. 5. See n. I. Because of the very sketchy character of Fourtau's plan, it is difficult to be certain whether his quai ran along the water's edge or was set back. The likeliest interpretation is that he meant the reader to understand that the quai continued the outer contour line of his lagoon basin unless his contour line was tracing the upper limit of the lagoon's sandy beach. His text fails to clarify the problem. 6. Chapter I. 7. J. Ball, "Recent Land-Subsidence in the Delta," Contributions to the Geography of" Egypt, Survey of" Egypt

(Cairo 1952) 67. For Walpole, sec Chapter I, n. 14. 8. The scatter of sherds over the sandbar where it attaches to the island is matched by sherds trailing into the water off the beach of Area II , a low outcropping of rock directly opposite the island. This si tuation suggests that the island and Area II were connected at one time by a neck of dry land. On one of the few calm days experienced in 1985, an additiona l submerged cluster of sherds was observed off the southeast shore of the island (in grid squares D6/ D7) at a depth of -1.00 m. A second cluster of submerged shcrds was found at roughly similar depth off the north tip of the island (grid square LS); a third was located at -1.30 m. in conjunction with the underwater wall, SIll, off the northeast corner of the island (grid square K I0/ L I0), to be discussed in Volume II. See also 1985 Report, 60, 62 , fig. I 0. Because of their severe deterioration through long-term exposure to seawater, none of the shcrds can be identified with certai nty, but the majority appear to be late. 9. See pp. 30- 32. I 0. See Chapter 4 for a full description of S 112. I I. For a summary of the standard arguments for a general rise in Mediterranean water since the Late Bronze Age, see A. Knapp, "The Thera Frescoes and the Question of Aegean Contact with Libya during the Late Bronze Age," Journal of Mediterranean Anthropolog_1· and Archaeology

33

1/2 ( 1981) 263- 267. 12. See p. 29 . 13. That pumice from Thera shou ld end up on the coast of the eastern Marmarica shou ld cause no surprise, because it already has been documented for Dor/Caesarea on the coast of Israel. See V Francaviglia, "Sea-Borne Pumice Deposits of Archaeological Interest on Eastern Mediterranean Beaches," Thera and the Aegean World Ill, 3, Proceedings of th e Third International Conference 1989 (London 1990) 19- 23. For a list of finds of pumice in the Aegean, see PP Betancourt, "Chronological Conclusions: the Relation between Pseira and the Eruption of the Volcano of Thera," in P.P Betancourt and C. Davaras, eds., Pseira II. Building AC (The "Shrine") and Other Buildings in Area A (Phi ladelphia 1998) 118. For arguments in favor of the earlier chronology for Thera's Minoan period eruption, see S.W. Manning, The Absolute Chronolog1· of the Aegean Early Bron::.e Age (Sheffield 1995), P. Kuniholm, "Overview of the Evidence of the Date of the Eruption of Thera," Th era and th e Aegean World Ill , 2, Proceedings of the Third International Conference 1989 (London 1990) 13- 18, and

P.P. Betancourt, "High Chronology or Low Chronology: the Archaeological Evidence," ibid. , 19- 23 . For a date in 1645 B.C. based on the presence of high su lfur content and vo lcanic ash in the icc strata of Greenland, see C.U. Hammer et al., "The Minoan Eruption of Santorini in Greece dated to 1645 BC?" Nature 328 ( 1987) 517- 519 and P. Schmid et al.,

"Separation and Analysis of Thcran Volcanic Glass by INAA, XRF and EPMA," Mikrochimica Acta 133 (2000) 143- 149. For the later chronology, which places the eruption ofThera in the middle third of the 16th century B.C., see P.M . Warren and V Hankey, Aegean Bronze Age Chronology (Bristol 1989), to which add the more recent bibliography and new information in M.H. Wiener, "The White Slip I of Tell ci-Dab'a and Thera : Critical Challenge for the Aegean Long Chronology," in V Karageorghis, cd., Th e White Slip Ware of Late Bron::.e Age Cyprus (Vienna 200 I) 195- 202. Perhaps the most unusual recent argument for the low chronology is the one based on the Pharaoh Ahmose's Tempest Stele of ca. 1530. See E. Davis, "A Storm in Egypt during the Reign of Ahmose," Thera and the Aegean World Ill, 3, Proceedings of" the Third International Conj"erence 1989 (London 1990) 232- 235. Also K. Polinger Foster, "Ahmose and the Eruption ofThera," NARC£ 171 ( 1996) I, 9- 10. To complicate matters further, scientists speak of twelve major explosive cycles of volcanic activity occurring in the eastern Mediterranean over the past 200,000 years centering on Thera, Milos, Kos, Nisyros, and Yali. See J. Keller,

MARSA MATRUH

34

Th. Rehren, and E. Stadbauer, "Explosive Volcanism in the

16.

Seep. 27.

Hellenic Arc: a Summary and Review," Thera and the

17.

See pp. 36-37.

Aegean World III, 2, Proceedings of the Third International

18.

Provisional Evidence, 4- 6. 1987 Report, 93- 95, I 03 .

Conference 1989 (London 1990) 13- 26. Thera's post-

19.

Seen. 14.

20.

The absence of imported Bronze Age pottery report-

Minoan period has been marked by nine episodes of " intracalderic activity:" 197 B.C., A.D. 46-47?, 726, 1570- 1573,

ed in the I 987 Report, 94 was confirmed by further survey

1701 - 1711, 1866-1870, 1925- 1928, 1939- 1941, and 1950.

of the lagoon system carried out in 1989. Sec Vol. II, Chapter 9.

See M. Fytikas, N. Kolios, and G. Vougioukalakis, "Post-

21.

Minoan Volcanic Activity of the Santorini Volcano. Volcanic

22.

But see White, Coastal Survey, 14, n. 31.

Hazard and Risk, Forecasting Possibilities," ibid. , 185- 186.

23.

See n. 13, passim.

See Chapter 5, pp. 93-94.

See also R. Strothers and M. Rampino, "Volcanic Eruptions

24.

See n. 13 for bibliography on the subject.

in the Mediterranean before A.D. 630 from Written and

25.

For the calibration of C-14 dates used here, see M.

Archaeological Sources," Journal of Geophysical Research

Stuiver, P.J. Reimer, and T.F. Braziunas, "High-precision

88, no. B8 (1983) 6358, 6366-6367. M. Fytikas, N. Kolios ,

Radiocarbon Age Calibration for Terrestial and Marine

and G. Vougioukalakis, " Post-Minoan Volcanic activity of

Samples" Radiocarbon 40 (1998) 1127- 1151 and M.

the

Stuiver et al. , "INTCAL98 Radiocarbon Age Calibration,

Santorini

Volcano.

Volcanic

Hazard

and

Risk,

Forecasting Possibilities," ibid., 184- 187.

24,000- 0 cal BP," Radiocarbon 40 (1998) 1041 - 1083. For

Just when this took place is not clear. It has been

the calibration of the marine samples, see M. Stuiver and T.

radiocarbon-dated to the 5th- I Oth century A.D., which could

F. Braziunas, "Modeling Atmospheric 14C Influences and

14.

correspond to the eruption of A.D. 726 reported above by

14C Ages of Marine Samples to I 0,000 BC," Radiocarbon

Fytikas et al. (seen. 13) 185. Gardner's dates, however, arc

35 ( 1993) 137- 189. M. Stuiver ct al., "High Precision

contradicted by the archaeological evidence which seems to

Radiocarbon Age Calibration for Terrestial and Marine

demonstrate quite conclusively that the entire island as

Samples," Radiocarbon 40 ( 1998) I 127- 1 151.

presently constituted remained occupied as late as the

26.

See Vol. II, Chapter 6.

5th- 6th centuries A.D. and, therefore, cannot have been inun-

27.

For the sandbar, see n. 8. The island 's pottery and

dated even partially during the later Roman period.

other Roman period artifacts are discussed in Vol. II,

According to Gardner, the dates arrived at by her tests may

Chapter 12.

have been skewed by contamination from younger carbon-

28.

See Vol. II, Chapter 6.

ate sand particles. See p. 32 , and n. 32.

29.

Fourtau, 117- 118. See Chapter I. Also Reswfacing

15.

The 1987 Report, 94, makes the point that according

of Parae/onium, 67.

to the area survey map used as the basis for Plan 4, the third

30.

and fourth lagoons were connected by a narrow stream of

3 I.

See Chapter I. Also Reswfacing of Parae/onium.

water as recently as the 1930s although the two were

32.

The reason for this situation is that the shells were

See n. 7 and Chapter I, n. 13.

mapped as separate features in Bates's day. A causeway

very small, and although a rigorous effort was made to

built in 1965 to connect the town with the eastern horn of

remove all carbonate sands from within them (by sawing

Matruh 's modern harbor (the so-called Rommel Beach area)

them in half, etc.), it is possible that some younger carbon-

has separated its inner harbor (see 1985 Report, 56- 57)

ate was included in the dating. More importantly, these two

from the First Eastern Lagoon. It is not clear from the

samples giving the dates of 1500 and I 000 B.P. (ca. A.D.

records currently available whether more water flowed over

802-93 4 and 1344- 1446) were extracted from very close to

the sand flats from the harbor into the lagoon before or after

the land surface, on which surface wash would occur during

the construction of the causeway. The flow of water from

heavy rainstorms, and onto which small (younger) shells

the harbor to the lagoon is controlled by a channel running

could be thrown up during heavy storms. These circum-

under the modern road and sluice-gate. If more water

stances were only true for these two samples; all the others

flowed into the lagoon system prior to 1965, the sand flats

were taken from depth within excavated sections. The

between the First Lagoon and the Second Lagoon would

uppermost surface of a beach is very difficult to sample

have been covered with a thin sheet of water, but Bates's

unless it has been covered subsequently by slope materials

and Fourtau's maps, here Figures I :3 and I :7, would seem

and hence fully buried. Therefore, I suspect that these sam-

to call such a conclusion into doubt.

ples may have been contaminated by younger shells.

Chapter 4

LATE BRONZE AGE OCCUPATION Donald White

Appearance of the Early Island Settlement The subject of this study lies in chest-high water ca. 170 m. off the eastern end of Matruh's first salt-water lagoon, east of the modern harbor (Pl. 14A- C). The lagoon is mostly land-locked, since, whatever its configuration might have been during the Late Bronze Age period, it today possesses only one entrance to the sea, which is out through the mouth of the main harbor. From perhaps the 15th to the 13th century B. C., 1 the island supported a small population of resident foreigners during at least part of the year. The most intensive occupation occurred in the 14th century B. C. Activities were centered in a cluster of modest communal room units along the northeast-southwest ridge of the island. Sharing narrow party walls and small open courts or perhaps light wells, none of the rooms seem likely to have risen above a single story. Because of their bad state of preservation / it is impossible to be certain what divided one unit from another and whether they belonged to a single structure or a series of attached but· separate shelters or walled workspaces. When seen originally silhouetted against the sky from the nearby shore, their collective outline might well have resembled contemporary rubble buildings (Pl. 15A) or mudbrick attached-housing on the mainland 3 or, indeed, any cluster of fishing shacks on a remote shoreline around the Mediterranean. To judge from at least what is left, the island settlement

possessed little in the way of architectural formality and order. 4 On the other hand, because whatever originally occupied the space eventually was reused for the modern Sponge-Divers House (S I 0 I), which was severely damaged during World War II (Pl. SA), we will never be entirely sure of its original aspect.

The Island's Current Appearance Today, Bates's Island measures approximately 135 m. northeast by southwest and 55 m. northwest by southeast (Plans I and 2). Its maximum elevation is 6.10 m. a.s.l. Like the coastal ridge to its north and Area II to its east, the island is built up from layers of aeolianite (calcium-carbonate cemented dune sandY that presents a wind-eroded rocky scarp across its western face, largely swept clean of sand (Pl. 15B). By way of contrast, much of the island's axial ridge, its eastern beachfront, and the sloping low ground occupying its northern sector are covered with layers of loose sand, which in places can reach depths of more than 2m. over the aeolianite core which will be referred to hereafter as "bedrock" (Pl. 16A, B). 6 With the exception of certain limited argillaceous deposits in the island's lower northeast corner (Trench 18-TII/S), which may have been introduced from the mainland, the successive

36

MARSA MATRUH

levels are made up of granular sands; each is stained with various intrusive impurities such as carbonized ash and decaying organic material rather than consisting of soil or humus. 7 Apart from a normal growth of scrub evergreens and dune succulents covering its sandy surface in patches between rock outcroppings, the island is today mostly bare of vegetation. x It shows no signs of ever having been cultivated. It has no evidence for springs or wells, and whatever freshwater was available would have had to have been collected in cisterns that still remain to be discovered, or it would have been carried over from the mainland. Before we started excavation, the island's only visible wall remains consisted of the demolished shell of the Sponge-Divers House (Pl. SA)'> and tumbled-down duck blinds built by local hunters from the fallen rubble stones lying about the island's surface. 10

Size of the Inhabited Island in Antiquity To judge from the spatial distribution of its archaeological remains, the island's Late Bronze Age occupation was limited to what is today the upper 3.6 m. of its dry-land mass. With one apparent exception, no Late Bronze Age sherds were recovered from the island's surface below the 3m. a.s.l. contour-line (Plan 1), and a meterdeep probe down to sea-level at its northeast corner in Trench 18-111/S produced no relevant Late Bronze Age material. 11 Similarly negative results were recorded in the nearby J8-IIII trench whose sloping surface lies between 1.0 and 2.0 m. a.s.l. ~ Trench G6-I, east of the Sponge-Divers House (SlOl), contained a small numberoftransplanted Late Bronze Age sherds mixed with Hellenistic/Roman material at 2.25 m. a.s.l. 13 Because of the mixed nature of these deposits, it is difficult not to conclude that their contents tell us nothing about the shape and size of the Late Bronze Age island. Moving farther up the island slope, at an elevation of ca. 2.50 m. a.s.l., Trench 16-1/II brought forth a minuscule sampling of Late Bronze Age sherds in what may have been part of 1

an uncontaminated early level immediately beneath a tangle of Hellenistic period cooking and storage installations. 14 On the other hand, the sherds were not associated with any walls, and they could owe their presence at this level to later backfilling activities in the island's northwest quadrant. In Trench HS-Ill immediately northeast of the Sponge-Divers House, 15 two Late Bronze Age storage bins (S 134a, b) were excavated at ca. 2.80 m. a.s.l. Collectively, the pair mark the lowest Late Bronze Age built features brought to light in the course of our excavations; all of Bates 's discoveries came from higher levels associated with the Sponge-Divers House. 16 If the absence of Late Bronze Age occupation below 2.50 m. a.s.l. (the level of the 16-l/ll sherds) may be taken as an indication of a rise in the level of the surrounding water, 1- the abovewater Late Bronze Age island would have been less than half its present size, and it would have appeared from the shore as little more than a dot of land adrift on the lagoon's surface. If one includes the possibly displaced pottery from the northeast corner of Trench 16- IIII, Late Bronze Age artifacts were spread over a space of ca. 56 m. from north to south. Excavated Late Bronze Age walls, as opposed to sherds and other transportable artifacts, extended ca. 43 m. northsouth, starting with the south wall of the S I 19 workshop in D4-l/ll and running to the point where the S 118 wall disappears into the north balk ofHS-I/11. On the other hand, the total (documented) east-west distribution of Late Bronze Age occupation is considerably less, being limited to the space cleared along the island's ridge in Trench HS-1111 / III , the Sponge-Divers House ( S 10 1), and the various trenches south of S 10 I. Since many of the Late Bronze Age finds from S I 01 were clearly dispersed from their original depositions by the construction of the SpongeDivers House, the actual east-west spread of excavated Late Bronze Age occupation theoretically is restricted to a narrow band averaging no more than 8 m. in width. If the walled settlement originally extended farther in any direction, it would have been to the north in the vicinity of Trench 16-1/II and east across the island's saddle where the ground is

37

LATE BRO ZE AGE OCCUPATIO

protected from the prevailing northwest wind 18 by the lee of the aeolianite ridge (Pl. 16A). The rocky, wind-etched western face seems to have been written off as uninhabitable during all phases of the island's use down to the present day (Pl. 15B). Development of the ridge toward the south was inhibited by a sharp drop-off in ground south of the S 1 19 workshop in Trench 04-I/11. Consequently, while the minimum Late Bronze Age occupation of the island could have amounted to less than 350 sq. m., it probably was more like ca. 1,100 sq. m. 19 The above estimates are flawed to the degree that they are not based on the total excavation of the Late Bronze Age island's hypothetical dryland mass but instead on educated guesses based on sample probes. In terms of what may have been missed as a consequence of this fact, it is worth noting that the expedition was unable to turn up conclusive proof for anything resembling an outer perimeter defense or barrier wall. ' 0 Moreover, we found no traces of Late Bronze Age docking facilities that in theory could have occupied the area of the island's now dry-land protected saddle.

Area of the Sponge-Divers House (SlOl) in the Late Bronze Age Figure 4: 1 indicates how the northeast corner of the post-antique Sponge-Divers House (S I 01) lies athwart the line of the Late Bronze Age ridge settlement, cutting off its northern extension in Trenches H5-I/II/III (referred to henceforth as the island's Northern Cluster) from its principal wall remains in F4-III and farther south (the Southern Cluster). When Bates opened up the Sponge-Divers House in 1914,11 his investigations led him to believe that its outer walls were built on top of ancient fill, but that the fill from its interior had been re-deposited upside down during its construction. More specifically, its lowest deposits contained Ottoman period artifacts, and the highest stratum was sprinkled through with Late Bronze Age Cypriot sherds. From what he relates, it may be understood that Bates's procedure was to excavate to

bedrock in the house's northern two-thirds, a part of the structure which he interpreted, probably correctly," as a separate chamber from the southern third. 13 The earth was thoroughly disturbed down to bedrock, and about 80 em. from the bottom was found an iron cannon-ball of about 6 lbs. weight, another of the same calibre, of stone, and a gun-flint. Below this level were found shards of modern Arabic wares, dating from the later seventeenth or early eighteenth centuries. As the higher levels contained numerous shards of Roman red ware of the second century A.D. , no further proof of the confusion of the debris was needed. More was, however, afforded by the discovery, in the upper part of the filling of the chamber, of a number of shards belonging to the "second Cypriote period" ( 1500- 1200 B.C.). He publishes 6 Cypriot sherds from this context. 24 Bates's digging operation then moved to outside the house, "a few metres west" (i.e., somewhere in H4 or H5 but west of H5-I/IIIIII), where he again dug a small probe to bedrock. This excavation uncovered the following, more orthodox sequence: 15

Stratum A

0- 50 em. Sand; shards ofmodern Arabic ware.

Stratum B

50-120 em. Reddish earth, containing fragments of Roman pottery (second century A.D.) and two shards of Ptolemaic hard buff ware, semi-lustrous black slip.

Stratum C

120- 185 em. Sand and loess.

Stratum D

185- 215 em. Loess; ashes, nw11erous shards of Cypriote fabric; two shards of other fabric.

Stratum E

215- 260 em. Sand and grit. (260 em. bedrock).

MARSA MATRUH

38

Northern Cluster

r----,----t i 1

I I

I

I

l---,

I ______ j

:I

~

I

~

L- ~ -- -!

S101 (Sponge-Divers House)

N

r---..,

r - - - -~

I I

I

I

I

I

I

I

I I I

1

I

L____ J

I

,--1

Southern Cluster

I

I

0

2

1, .. 1

I

.~

I

I

~---Jl I

·- - - - -

Fig. 4:1. Plan of island s Northern and Southern Clusters.

4m

LATE BRO ZE AGE OCCUPATION

Bates published a detailed description of 18 sherds from Stratum D: nos. vii- xxi are Cypriot White Slip, while xxii- xxiii are said to be "cognate of that class;" xxiv is attributed to Minoan Crete. 26 Since the ground's surface in H4 and H5 is at least 5.0 m. a.s.l., the bottom of Bates 's "St. D"

39

is higher than the leve l of the Late Bronze Age sherds excavated in 16-I/II and, therefore, does nothing to modify the previously set forth views on the size of the Late Bronze Age island.27 Because Bates implied that he already had cleared S 101 's northern two-thirds to bedrock,

N

-..._ I

-

I

I I

--_,

I

-

I

_, ----------·----------------------J H4-lll

+

S101

0 Q 5 -IV/SW Test

:------ ----1/n8co I

I

b'

--0

:

1

'

I

2

4m

Fig. 4:2. Plan of the Sponge-Divers House (SJOJ).

_jo

40

MARSA MATRUH

the Pennsylvania expedition avoided re-excavating the bulk of the house, and decided to concentrate instead on cleaning and drawing its exposed walls (Fig. 4:2). The fact that we found SlOl 's interior badly cratered by a World War II bomb or exploded artillery shell that had scattered rusting shrapnel across its surface gave a further reason for not excavating.28

Bates's Dump When we began to explore Bates's dump in the H3-III and H4-IV sectors by the northwest corner of S 10 I, it quickly became apparent that it was made up of a mixture of fills taken from the house interior as well as the test trench Bates had sunk outside the house 's northwest corner. In places the dump was nearly 1.5 m. deep. 29 Aithough much of it was simply sterile sand, sifting led to the recovery of 740 sherds, along with marine invertebrates, fish bones, and bird bones, several bags of bronze fragments , a terracotta loomweight (Vol. 11, 13.78), a Late Bronze Age crucible fragment (Vol. II, 9.28), and a flint. In addition, it produced a number of scraps of iron (probably World War II shrapnel) and modern bottle glass mixed with Roman blown glass. Most of the sherds came from the upper 0.05 m. of loose surface sand, although the damp, beigecolored sand that made up the rest of the dump also contained a sparse scattering of sherds. In terms of chronology, the dump's pottery spanned the full range of the island's occupation, including the earliest phase (Vol. II, 7.41, 7.43, 7.60). Wall cleaning led to recovering a substantial number of artifacts dating from the 19th century back to the Late Bronze Age and, also, some faunal materiaJ.3° While obviously stripped of any stratigraphical value, the dump finds have meaning simply by virtue of their association with the house. In other words, when considered along with the early stray finds from the surface fill inside the house and associated with its four walls, the dump's artifacts help to substantiate the fact that this particular area of the island played a major role during the Late Bronze Age period . The house's walls, moreover, tell us some-

thing about the island 's earlier occupation because they appear to be made up in large measure of reused stones. Apart from the early sherds from this area, probably the most significant class of Late Bronze Age objects are nine crucible fragments (Vol. II, 9.27, 9.29- 9.36). In addition to wall and surface cleaning, two small test trenches were excavated in S 10 I 's interior. Both trenches confirmed Bates's view that in its present configuration nothing predates the 17th century A.D., despite the fact that probably most, if not all , of S 101 's masonry had been removed from ancient buildings, almost certainly including elements of the Late Bronze Age settlement. More importantly, the tests corroborated the fact that the zone overridden by the house had been originally occupied during the Late Bronze Age and had then been subsequently disturbed by later building activity. We nowhere encountered a repetition of Bates's bizarre, but perfectly believable, upside-down stratigraphy.

GS-IV/SW Test (Fig. 4:3; Pl. 17) A 2 by 2 m. test (Fig. 4:3) was cleared to bedrock in the northwest corner of G5-IV, using the

r·-·-·-·-c-·-·-C3:if·-·-r-·-·-·- ·-·-·-·-; i i

i

\

. I

...........,

\

r-1~

·

f

1

'----.J

__j

i

/

4.3 _/

.

i . !

Undug

below 3 .1

1

'~ ~............

i i i

- 1.58

I I

.

!

- · -·-·- · - · - · - · - · --~

C) 4.1

N

Wall"A"= (]

S133

l Ji.~1~6-·-·-·-·-·-·-·-4.2

- 1~8

l·-·-·-·-·-·-·-· 0

Fig 4:3. Plan ofG5-IVISW Test.

_i 2m

LATE BRONZ E AG E OCCU PATION

41

2.1. Compacted deposit of gray sand 32 mixed with small stones that buried the lower 0.20 m. of the wall as well as running under its footings. Circa 0.20- 0.50 m. thick. Permeated with roots of still living succulents. Veg. taxa and pol. (Vol. II, 11.24, 11.25, 11.53). 3. 1. Circa 0.40 m. thick layer of light-colored sand with no living roots and largely free of stone; contained some chips and pebbles. Tunneled into by small animals. Not associated with any architectural features .

3.1 e

·= 0

V 4.2

o

"'o

oo, w E4-lll, South Balk

Fig. 4:23 . East-west section, E4-Jlf, South Balk.

2.2.

Small number of uncat. late sherds.

2.7.

Animal bones. Ostrich eggshell (Vol. II,

9.96). Late Bronze Age imported sherds (Vol. II , 7.1 ; 8.37, 8.97). Canaanite lamp (Vol. II, 8.4). I faience vessel frag. (Vol. II, 9.85). Bronze barbless arrow-head (Vol. II, 9.1); 2 bronze nodules. 2 pieces of iron slag. Stone pounder (Vol. II, 9.63). Stone bead (Vol. II, 9.75). Fish and animal bones. Marine invertebrates. Ostrich eggshell (Vol. II, 9.95). 2.3.

2.4. Unexploded World War II artillery round . Late Bronze Age imported pottery (Vol. II, 7.22 , 7.45, 7.55; 9.59) and Egyptian pottery (Vol. II, 8.47, 8.48). 5 scraps of bronze. Chert core or hammerstone flake (Vol. II, 9.59). Bronze Age stone weight (Vol. II, 9.68). Anima l bones. The artillery shell has penetrated into what appears to be otherwise a purely Late Bronze Age level.

3 .I. Sterile, apart from single unidentifiable sherd. 3.2.

nodules . Marine invertebrates. Fish, bird, and animal bones. Ostrich eggshell (Vol. 11, 9.97). A Late Bronze Age deposit. 3.3. Imported Late Bronze Age and Egyptian pottery; Egyptian sherds (Vol. II , 8.54, 8.55). 4 scraps of bronze. 4 pieces of chert. Pumice (Vol. II, 13.59). Hearth ash. Marine invertebrates. Fish and animal bones. Ostrich eggshell (Vol. II, 9.98). A Late Bronze Age deposit. 4.1 .

Late Bronze Age imported pottery (Vol. II, 7.44, 7.53). Egyptian pottery (Vol. II, 8.49, 8.50). Late Bronze Age lamp (Vol. II, 8.5). Chert. Fish and animal bones. Marine invertebrates.

Late Bronze Age imported pottery (Vol. II ,

7.76, 7.46). Bronze pin (Vol. Jl, 9.9); 14 bronze

Sterile.

2.5 .

2.6. Egyptian pottery (Vol. II, 8.51- 8.53). 3 nodules of bronze; undatable flat triangular sheet of bronze. 3 frag s. of iron. Fish and animal bones. Marine invertebrates.

Late Bronze Age imported pottery (Vol. II, 7.73). Numerous chunks of plaster. Bronze pin; 12 bronze nodules. Glass bead (Vol. II, 9.84). Marine invertebrates . Fish and animal bones. Fresh-water shells. Land snails. A Late Bronze Age deposit. 4.2.

4 .3.

Sterile.

70

MARSA MATRUH

Late Bronze Age Architectural Remains Associated with E4-III S102 (Fig. 4:21; Pl. 29A, B). Apart from the two small pits linked with the 2. 7 and 4.2 deposits , the trench's only architectural features are the north, east, and south walls of a small Late Bronze Age chamber (S 102). The latter's most significant stratigraphical fills are the 2.5 deposit linked with its east wall (S 102c) and the 3.3 occupation build up over its interior (4.3) floor. The 3.2 deposit outside the south wall (S 102d) seems to be composed of similar occupation debris. All three deposits held exclusively Late Bronze Age material. Ashes and bones in the 3.3 build up suggest that the room was used more for eating63 than for storage or industrial purposes. The small pit (deposit 4.2) that lay a short distance to its south contained lumps of white plaster that, on analogy with room S I 07 and wall Sll8, 64 might originally have been used to seal the inner face of S 102 's walls, although no traces of plaster were found actually sticking to the stones. Sl02 measures internally roughly 1.80 m. north-to-south. When its west wall, S I 02a, eventually was cleared from the E4-III/W Balk, the room's east-west internal width turned out to be approximately 1.90 m. Externally, Sl02 represents a very approximate 3 by 3 m. square. Its north wall, S I 02b, measures ca. 0.60 m. across and stands to a height of only two or three courses of flat rubble. S I 02c on the east also measures 0.60 m. across, but is somewhat better preserved to a height of approximately 0.40 m. The south wall, S I 02d, is the most flimsily preserved except at its external southeast corner where it spreads into a meter-wide spur made up mostly of small rubble chips that may have continued into the unexcavated space to its south. Its narrowing-down to two small parallel slabs just off the interior southeast corner creates an impression on the evidential plan of a threshold. This is, however, misleading. When seen in its actual excavated state, the corner displays no conclusive evidence for the existence of a door.65 Instead, as in the case with Sl07 (the site's only other Late Bronze Age chamber with a fully pre-

served set of outer walls), 66 no clues survive for how the room was entered. This situation constitutes something of a mystery because the aeo1ianite pavement (4.3) across its interior, which was covered by ca. 0. I 0 m. of occupation debris (3.3), suggests that more survives of its walls than simply their foundations, and that on entering one stepped down into a slightly lowered interior. No evidence for either mudbrick or upright timbers and/or saplings to carry up the walls was recovered. As is so often the case with other island structures, the archaeological record for S I 02 's superstructure has been largely nullified by time.

E4-III/W Balk (Pl. 31A) The principal contributions of this 0.80 by 4 .00 m. test was to expose the remains ofSl02's west wall (S 102a), along with a short length of possibly Archaic Greek period wall (S 124, Pl. 31A). In addition, a squatter wall of recent date (S125, 1.0 meter long) was found in 1.1 fill in the northwest corner.

E4-IIIIW Balk Deposit Sequence I. I. Loose, off-white sand, ca. 0.15 m. thick, whose surface before excavation was covered with randomly scattered stones. Buries wall S 125; penetrated by tops of the four plug-shaped blocks making up wall S 124. 2.1. A more compact, reddish layer of sand, nearly entirely free of stones in its southern half. Northern half covers part of the S I 02a wall as well as supporting footings of wall S 124. Circa 0.30 m. thick. 3.1. Reddish yellow, crumbling aeolianite pavement, mixed with white flecks, underlying S I 02a. It represents the extension W of the 4.3 deposit in E4-III. Cleared of its surface accumulation but otherwise not excavated.

LATE BRONZE AGE OCCUPATIO

71

········· .... •

-1.2 ".

, •.

•

,

'.

. I ' , , •. • •.

-~~·-.

·-· , ·;·.:.------.--.•.>·•···< .

·::: .. __ _

.· ... . ')di~?~.~

-::;-.··· ·

..·._.. ·: .. ... .

~:_ ._:,.,,.-:_•."''.-:.::;:;:.·:-:.

1

0

2m

I

I

I

----------------------~~N

04-I,West Balk

Fig. 4:24. North-south section of D4-1, West Balk.

Contents of E4-IIIIW Balk Deposits 1.1. Miscellaneous Late Bronze Age and uncat. late sherds. 3 bronze frags. Animal bones. 2.1. Three intrusive uncat. late sherds. Standard range of Late Bronze Age imported sherds. 8 bronze frags. Animal and bird bones. 3.1. Two WS sherds. Bronze nail or small pointed blade (Vol. II, 9.21 ). Animal bone.

Late Bronze Age Architectural Remains Associated with E4-IIIIW Balk S124 (Pl. 31A). This 1.15 m . length of wall is made up of four squarely proportioned blocks running north-south immediately west of the S 102a's northwest corner. Averaging ca. 0.40 m. wide, 0.50 m. deep, and 0.20 m. high, their flattened, western faces and rounded edges create a plug-like appearance unlike any other set of walls encountered on the island. Since it is bedded on 0.15 m. of 2.1 fill , which could be early, S124 may be a Late Bronze Age construction . As such, it could represent an extension at a slightly different angle of the S122 "Ramp" . On the other hand, it does not align with the west wall ofS102

and, moreover, gave the impress ion, at the time of its excavation, of belonging to a distinctly separate phase of island occupation. It has been earlier suggested 67 that its masonry style perhaps argues for an Archaic date (more congruent with 2.1 's single Red-Figure sherd) , in which case its blocks would have had to have been set down into ca. 0. 15 m. of earlier fill. S102a. The west wall of this small enclosure was revealed by the balk's clearance, with few surprises. Its width has been depleted by later building activity; some of its blocks may have been swept into the little pile of rubble found at the south end of wall S 124. The Late Bronze Age character of the overall room was further confirmed by the presence of Late Bronze Age sherds in the 3.1 fill underlying S1 02a.

E4-IV The westernmost extension of the E4 trench series, this 2.0 by 4 .0 m. test was intended to expose either the western perimeter of the S 102 chamber or the existence of another Late Bronze Age construction. As it turned out, it brought to light ne ither. The position of the western wall of S I 02 was subsequently confirmed with the opening-up of the E4-JTI/W Balk, but E4-IV turned

72

MARSA MATRUH

out to be empty of Late Bronze Age built features. Its most conspicuous find turned out to be skeletal rather than architectural, belonging to the hastily interred bodily remains of a (German?) soldier evidently killed by she ll fire during World War II. The badly disarticulated parts of a human skull, ribs, and long bones were excavated from the surface level of the southern half of the trench. The loose 1.2 sand failed to preserve even the outline of a scooped-out burial pit, and no scraps of either the dress or battle insignia turned up with the victim's jumbled bones; his identity is based on information supplied by the workmen that a German was rumored to have been buried on the island. If a rubble marker or cairn was ever mounted over the remains, it was probably pulled apart by local fishermen or hunters. A short line of small rubble stones appeared in 2.1 fill at the northeast corner of the trench. Since these stones run more or less parallel to the enigmatic pi ug-shaped stones making up wall S 124 (Pl. 31 A), they may relate to the latter, particularly if S 124 does, in fact, represent a change of direction of the S122 ramp. On the other hand, given the prevalence of random scatters of stones in all levels of the island, their survival in this pattern could be coincidentai. 6R

E4-IV Deposit Sequence 1.1. Very fine wind-blown white sand across northern half of trench; scattering of surface stones. Circa 0.25/0.28 m. thick. 1.2. Contin uation of the same sand across southern half of trench and surrounding human skeleton. Marked by patches of surface burning (left over from modern camp fires?). Circa 0.10 m. thick. 2. 1. Damp, compact reddish sand, ca. 0.10- 0.20 m. thick. 3.1. Unexcavated aeo li anite ridge pavement, a.s.l. elevation 5.55 m.

Contents of E4-IV Deposits I. I. Small number of miscellaneous sherds. 3 nodules of bronze. Animal bones. 1.2. Small number of miscellaneous sherds, including Late Bronze Age Cypriot. Iron shrapnel. 2.1. Small number of miscellaneous Late Bronze Age sherds and a 500-450 B.c. RedFigure sherd (Vol. ll, 12.14), wh ich may or may not be intrusive. Small frag. of bronze; intrusive piece of iron shrapnel. Animal bones. Although this unit might represent a disturbed Late Bronze Age deposit, any conclusive evidence is lacking. 3 .I .

Sterile.

D4 Grid Square (Figs. 4:16, 4:24; Pis. 31B, 32A, B) Grid Square D4 brought to light the most southerly and in some ways the most provocative set of architectural remains in the series of Late Bronze Age structures making up the island's Southern Cluster. Most of its S 11 9 workshop came to light in D4-l /ll and the extension, D4-I/Il N Balk; elements of the west wall, S 119c, were found in the adjacent D4-l Balk. Bates reports that the local residents of his day told the story of how either a single Jew or two Jewish men had been buried long ago on the island; according to the latter version, the buried men, Ishak and Hugah, were go ldsmiths. 69 He describes how he found "a large ruinous sto ne circle (ca. 3 m. in diameter)" sti ll marking the site of the burial(s). 70 While sketchy to the point of being nearly useless, Bates's map of the island does indicate that the stone circ le or cairn, as he terms the burial elsewhere, probably lay in the 04 grid square. 71 Bates himself was never able to investigate the "tomb," a lthough it was certainly his wish to do so. When it was eventua lly opened in 1987, what came to light was not a burial but a small industrial workshop area, S 119. When the Pennsylvania expedition tested the 04 grid square, all that was left was a sma ll

LATE BRONZE AGE OCCUPATION

mound-like outcropping of bedrock and sand. Its uncleared surface was studded with stones, particularly across its southern half (presumably representing the scattered remnants of Bates's cairn), and it also contained Late Bronze Age sherds. Our 3.20 by 2.20 m. test trench initially was laid down in two parts, the "N Balk" referring to a 1.0 m. wide cut between the E4-lll square to the north and the D4-IIII trench-proper to the south. For purposes of the following description, D4-l/II and 04-1/II N Balk are treated together since they share the same deposit sequence and are not differentiated in the area's north-south section .

D4-I/II and D4-JIJI N Balk Deposit Sequence 1.1 . Loose, light brown sand, mixed with a thick scatter of rubble stones, 0.10/0. 15 m. thick. Occupies eastern three-quarters of the trench. 1.2. Loose, light brown sand with pockets of gray/reddish yellow soil across western quarter of trench. Differentiated from I. I by its somewhat heavier scatter of stones marking the tops of the S 119c wall blocks. 0. I 0/0.18 m. thick. 2.1. Gray/reddish yellow loose sand containing scattered stones, ca. 0.25 m. thick. Basically a wind-borne layer, like 1.1 , but colored by vegetational activity. Filled in around the four walls of Sll9. 2.2. Dark gray ash layer, containing lumps of charcoal and white plaster-like flecks. Apparently the remnants of a broken-apart hearth or furnace (originally plaster-lined?) in the NE corner of Room Sl19. 0.07- 0.10 m. thick.

a.s.l. elevation of 5.35 m. Circa 0.05- 0.10 m. thick. 3.2. Another concentration of gray/black ash, mixed with bones, pebbles, sand, and stones, located beneath the 2.2 hearth-furnace layer. This unit appears to represent the earliest phase of hearth-furnace activity in this room. It rests on a scattering of stones set on S 119's 4.1 floor. 0.05 m. thick. 4.1. Moist reddish yellow sand, mixed with white flecks from the crumbling natural aeolianire pavement. S 1 19's primary floor level at an a.s.l. elevation of 5.30 m. Circa 0.05 m. thick.

Contents of D4-I/II and D4-I/II N Balk Deposits 1.1. Large range of Late Bronze Age imported Canaanite and Egyptian pottery (Vol. II, 8.21, 8.22 , 8.149). Uncat. late sherds. 3 frags. of plaster. 3 frags. of worked stone. Marine invertebrates. Animal and fish bones. 1.2. Broad range of Late Bronze Age sherds; uncat. late juglet fragment. 2 glass beads. 3 small bits of bronze. Charcoal frags. Animal bones. 2.1. Late Bronze Age Egyptian and Cypriot sherds, including pot mark (Vol. II, 8.16, 8.84). 5 frit and glass beads (Vol. Il, 9. 78, 9.80, 9.81 ). 2 pieces of iron, 1 bronze large frag., 1 bronze frag. Plaster. Charcoal. Fish bones. 2.2. Late Bronze Age imported and Egyptian sherds that date S 119 's furnace to the island's Bronze Age occupation. Animal bones. 2.3.

2.3. Shapeless concentration of white plaster bits accumulated against the eastern edge of the 2.2 "hearth" deposit. 0.04 m. thick. 3 .1. Compacted reddish yellow sand, filled with white flecks , across most of trench and serving as a secondary floor level to Room S I 19 at an

73

Plaster. Otherwise no datable material.

3.1. Large range of Late Bronze Age imported sherds (Vol. II, 8.23, 8.85-8.87, 8.150- 8.152) and Egyptian pottery (Vol. 11 , 8.23) that date Sll9's secondary floor to island's Bronze Age occupation. Blue frit (Vol. II, 9.84). Long bronze needle (Vol. II, 9.23). Large lump of bronze, 12

74

MARSA MATRUH

small bronze frags. Pierced stone fishing-net or loomweight (Vol. II, 9. 71 ). Ash. Marine invertebrates. Animal and fish bones. Ostrich eggshell (Vol. II, 9.86, 9.87). 3.2. Late Bronze Age imported and Egyptian sherds; Late Bronze Age lamp (Vol. II, 8.1). 4 bronze frags. Animal and fish bones. Ostrich eggshell (Vol. II, 9.88). Primary phase of S 119's hearth-furnace is, thus, clearly Late Bronze Age. 4 . 1. Late Bronze Age imported sherds (Vol. 11 , 8.88, 8.89); single uncat. intrusive late sherd. Bronze fishhook (Vol. II, 9.24). Large concentration oflumps of metallic waste material including at least one lump of hematite (Pl. 33A) (Vol. II , 9.50), left over from some type of metal production activity. Marine invertebrates. Fish and animal bones. Fresh-water shells. Land snails. A Late Bronze Age level.

Late Bronze Age Architectural Remains Associated with D4-I/II and D4-I/II N Balk S119 Workshop (Pis. 31 B, 32A, B). Small even by the standards of the cramped room spaces elsewhere on the island, the external measurements of this feature are ca. 2.20 m. eastwest by 2.60 m. north-south. Internally, it measures a meager ca. 1.75 m. square. Where measurable, its wall width does not seem to exceed 0.45 m. The best preserved elements are its S 119c (west) and Sll9b (south) walls, but all four sides have been pulled apart too badly to say more than that they were laid-up with random-sized fieldstones and have left no traces of their original superstructure to indicate whether they continued up in stone, some variety of wattle and daub , or, in the case of its windward-facing west side, in the form of a wind-screen of some kind of perishable material. n The plaster bits found in the 2.2 and 2.3 deposits are better associated with the 2.2 hearth-furnace than its outer walls. The crumbling 4 . 1 aeolianite pavement taken to be Sll9's primary floor is slightly lower than its foundations, producing, once again, a sunken

floor. Indeed, given its reduced size, what is called a room may have been little more than a sheltered, open-air work area. Once again, it is unclear how S 119 was entered, although it see ms likely that some kind of entrance was in its north wall to communicate with the nearby S I 02 room. Despite all its limitations, S 119 contributes four elements to our understanding of the island 's early hi story. First, its 3.1 and 4.1 "floor" levels, which are associated with the 2.2 and 3.2 furnaces located in the center and northeast corner of the room , clearly belong to separate occupation phases. This stratigraphy is, once again, a sign that the island underwent some kind of extended Late Bronze Age occupation even though its exact chronological limits remain poorly understood. Secondly, the biggest concentration of bronzeworking waste particles discovered anywhere on the island was scattered over S 119's 4.1 "floor." Despite the fact that no crucible fragments turned up anywhere in D4, their frequent recovery in Late Bronze Age levels close by strongly suggests that this space was used for pouring molten bronze to cast simple trade items such as the 0.18 m. long "sail" needle found in 3.1 fill (Vol. II , 9.23) and the bronze fishhook in 4.1 (Vol. II , 9.24). The two 2.2 and 3.2 burned deposits have lost their original configurations, which hinders any close interpretation of their function. As already suggested, the bits of plaster mixed with the burned matter in 2.2 may have something to do with an otherwise lost fire chamber. Of course, both deposits simply could have been used as open hearths for heating and cooking, as they were elsewhere, since some animal bones were present in the 4.1 , 3.1, 2.1, and 2.2 fills. On the other hand, because of the production dross scattered throughout 4 .1, the greater likelihood is that they mark the location of metal-melting furnaces set on the ground, 71 rather than furnace bowls or pits set into the ground, 74 of which we have no trace. In either case, they would have been fueled with wood, charcoal, or perhaps even animal dung, with oxygen supplied by the strong prevailing winds, possibly supplemented by some system of bellows or blowpipes. 75 The

LATE BRONZE AG E OCCU PATION

smoke would have been vented to the sky through an open roof. Thirdly, Egyptian wares , particularly open bowls, dominate the pottery found in the upper Bronze Age levels of S 119 (i.e. , 2. 1, 2.2, 3.1, 3.2), while according to Linda Hulin, they are absent from its lower 4.1 level. Lastly, 60% of the island 's admittedly slender but entire stock of either glass or faience Late Bronze Age beads was found inside Sl19. Could this sih1ation, together with the presence of the Egyptian bowls, indicate that S 119 and the area immediately around it were used for processing

75

more trade items than just metal tools for local exchange?

D4-J/ W (Fig. 4: 16) A small rectangular test was excavated directly west of S119c, mainly in order to clarify the appearance of the western face of that wall. Although this excavation brought to light parts of what seems to have been a modern duck hunters' blind (S127) , it produced no Late Bronze Age material of any particular significance and so requires no further description here.

Summary As has been made plain throughout the above narrative, the Late Bronze Age constructional remains on Bates's Island are, for the most part, exiguous. In addition to the three largely complete rooms or enclosures (S I 02, S I 07, S 119) making up the Southern Cluster (Pis. 27, 28A, 31B, 32A, B), we have the negative imprint of what must have been two additional similar spaces (S 120, S 136, Fig. 4: 16). The existence of an exterior paved area or court (S 125, Fig. 4: 17) seems relatively certain. On the other hand, the so-called " Ramp" (S 122, Fig. 4: 16; Pl. 30A) has been pulled apart so badly by later builders that we could, instead, be dealing with a section of an otherwise lost outer banier or defensive wall, although this seems unlikely. Building evidence in the Northern Cluster is even more impoverished, consisting mainly of short, disconnected lines of walls (S 118, S 121 , S 130, and S 131 ). Only the obtusely angled S 126a and b walls resemble what seems to be a section of an actual room or enclosure (Fig. 4:8; Pis. 20A, B, 22A-C).

Southern and Northern Cluster Architectures Walls As preserved, wall s normally consist of flat, undressed fieldstones of various sizes that were

presumably carried onto the island and then laid up in rough courses. 76 In the case of the S 118 and S 121 walls (Fig. 4:8; Pl. 20A, B), the builders first set up inner and outer facings of medium-sized rubble stones, laid in courses in the case of S 118, and then used fist-sized stones to pack the core as well as to chink the outer facings. No cut-stone or hammer-dressed blocks appear to be emp loyed before the historical period. Wall widths are kept consistently under a meter, unless the ca. 2.0 m. wide S 122 feahrre (Pl. 30A), in fact, functioned as a barrier wall rather than a ramp, which, as said before, seems unlikely. 77 They range from 0.40 (S 121) and 0.45 m. (S 119, Fig. 4:24; P1s. 31B, 32A, B) to 0.60 m. (S 102, Fig. 4:21 , Pis. 29A, B, Sl18 and 126a and b, Fig. 4:8, Pl. 20A, B) and 0.80 m. (S130, Fig. 4:5, Pl. 22B, C). The relatively narrow widths of some and consistently low heights of all the extant rubble remains could mean that they were either foundations or, at best, low socles intended to carry superstructures built in something other than stone. In that vein, no direct evidence survives to indicate how the island 's Late Bronze Age walls continued up to roof level. The use of fieldstones and rubble for foundations and socles is nearly universal throughout the eastern Mediterranean during the Late Bronze Age. At the same time, complete walls of rubble, normally bonded with soil, mud, or clay as well as reinforced with wood, also occur

76

MARSA MATRUH

regionall/ 8 and may have been what were used on the island. Mudbrick superstructures are an unlikely alternative, if for no other reason than we found no deposits of collapsed mudbrick in the course of excavation. The generally sandy nature of the island's surface and what is found throughout the lagoon shore area probably explains why the technique was avoided. Walls made from puddled mud occur from a very early time onward elsewhere,79 and reinforced sun-dried brick walls seem to be virtually ubiquitous. 80 Walls of pise, which involves pressing mud or clay onto a wooden framework that is subsequently removed, occur at Mycenae 81 and in the Levant82 but do not seem to enjoy much of a vogue elsewhere. Again, the absence of any site evidence along with a paucity of mud would seem to rule out the use of pise on the island, although the argillaceous sand bottoms of the nearby lagoons would have sufficed for simple daubing. Wattleand-daub is normally associated with early round buildings, 83 but marsh reeds or saplings coated with lagoon mud could have worked for walls with rubble socles 0.45 m. wide and less. 84 It is at least theoretically possible that the island's wider walls were built with fieldstones up to ceiling level, and that the narrow (partition?) walls were erected in wattle-and-daub. Finally, if we are right in assuming that the strong prevailing northwesterly winds supplied air to the bronze-melting furnaces situated in at least some of the Late Bronze Age room units extended along the island's exposed ridge, the possibility should be kept in mind that their windward-facing "walls" may have been erected as movable screens that could be taken down or moved aside to let in the wind. 85 In such a case, the screens could have been put together from skins or cloths stretched over wood or reed frames. For obvious reasons, no direct evidence survives for such an arrangement.

Sand Slurry A muddy and sandy slurry was used to bind the rubble cores ofthe Sll8 and Sl21 walls. The S I 07 walls exploited a similar mix to consolidate

their rubble courses. The remainder of the island's Late Bronze Age walls appear to have been laid up dry. Wet clay and mud are used as mortar to bind rubble walls on Crete. 86 Clay mortar combined with rubble masonry has been reported at Mycenae.87 On Cyprus, mud mortar is employed with both rubble and mudbrick construction; 88 in the Levant clay and mud are both used to bind rubble. 89

Wall Plaster A coarse plaster was used to coat the inner wall surfaces of the S I07 room and the inner face of wall S 118. Whether it was applied to the walls the Sl02 room and the curved line of wall Sl21 is more problematic. The surviving plaster fragments are mostly white, but chunks tinted a pale green were found inside S I 07. None of the fragments preserve any obvious imprint to help establish the types of surfaces to which they were applied. Perhaps its absence suggests the use of mudbrick walls , but even mudbricks could be expected to have left some kind of mark on the stucco surface. Painted and plain lime plaster is used extensively on Crete to revet interior walls,90 and the same is true for Mycenaean housing. 91 Although its use on walls occurs in the Levant as early as the Neolithic period, it seems to be less well documented there in the Late Bronze Age. 92 According to Wright, the evidence for prehistoric Cyprus is still not fully understood, but the technique again appears to be very early. 93

Roofs No direct evidence was generated for how the rooms might have been covered, but a reed or marsh grass thatching carried on a framework of samplings would have been adequate for such narrow spans. There seems to be no reason to suggest that the island builders resorted to a timber and mud plaster technique despite the near universality of that combination down to the present (Pl. 33B). 94 On the other hand, wall plaster, whether interior or exterior, needs some form of protection from rain. The fact that some of the rooms were

LATE BRONZE AGE OCCUPATIO

plastered and, therefore, somehow roofed, implies that at least a limited occupation could have been kept up on the island during the rainy winter months. The fact remains, however, that the island was probably occupied most intensively during late spring, summer, and fall. Thatched roofs have been argued for early housing on Cyprus 95 and the Levant. 96 On Crete and on the mainland, the preferred method appears to be beaten soil or clay laid on reeds, which are in turn set over wooden rafters .'n Some kind of simplified version of the latter is at least theoretically possible for the island. Room Sll9, the metal workshop area believed to have been used for casting operations, was perhaps left open to the sky, and the same was presumably true for the S 136 enclosure.

Windows and Doors Not surprisingly in light of the state of the walls, no evidence exists for windows. Perhaps the small interiors received enough illumination through their doors to eliminate the need. Less understandable is the lack of any traces of doors. 98 The trampled sand floors inside the S I 02 and S 107 rooms and the S I 19 workshop were lower than what we take to be the bottom of their respective surviving outer walls. This situation makes the absence of thresholds in all three units puzzling. Presumably, someone entering had to step over some kind of raised threshold element, now missing, down into the interior. A wooden board may have been enough to keep the interiors dry. In the houses at Mycenae, the thresholds were made of wood at least some of the time, and they were often raised above floor level. 99 Wooden window frames were fairly universal , especially on Crete; 100 wood thresholds , however, do not occur elsewhere with much frequency.

Floors and Pavements The island's best-preserved rooms used the crumbling surfaces of the aeolianite ridge for their primary occupation or "floor" levels. In the case of S 107, an ash layer from its small interior oven (S 13 7) appears to have been trampled deliberately

77

into the natural aeolianite to become a secondary " floor" level. The S I 19 workshop had two trampled-sand occupation levels. Similar occupation levels were recorded in connection with the S 118 and S 121 walls. On the other hand, none of the island 's room or enclosure interiors seem to have been provided with paved or otherwise man-made floors, unless the S 135 walk surface in the 16 grid square, which was a peculiar amalgam of clay, white plaster, and disintegrated bones, 101 somehow represents a survival from the Late Bronze Age period. Evidence does suggest that the S125 courtyard and its extension into S 120 may have been paved with stone. If, however, S 120 actually was a separate room, we are dealing here with a paved interior. 102 The spaces off to either side of the stump of wall S 130 (in H5-lll/SW) also seem to have been paved, but the preservation is too poor to tell much. The same is unfortunately true ofthe S 122 "ramp" whose paved surface now has been almost entirely robbed away. The sunken floor feature, which has been inferred for S I 02, S I 07, and S 119, is paralleled by the early round houses at Jericho 103 but does not seem to occur with any frequency elsewhere throughout the eastern Mediterranean. Similar to a root cellar, its purpose could have been to insulate foodstuffs susceptible to spoilage from high temperatures, particularly if dampened loose sand was heaped around the lower levels of the outer walls. 104

Hearths, Ovens, and Furnaces The stone hearth s that have been linked with a number of the island's walls and rooms are a standard feature of Late Bronze Age occupation. 105 These features include the following: the two pulled-apart hearths in the center and northwest corner of H5-II near wall S 118; 106 the circular hearih built of stones (S 132) in the H5-III grid square (Pis. 228- D, 23A); hearth traces in possibly Late Bronze Age 4.1 fill in the 16 grid square; the pulled-apart hearth (S 117) in 3.2 fill occupying the center of 16-l/II 's western half (Pl. 258); and E4-Ill, 3.3 hearth debris in the interior of Room S 102. Room S I 07 has an enclosed oven (S 137) with a terracotta cooking compartment

78

MARSA MATRUH

similar to a traditional beduin tabun or bread oven 107 set against its west wall (Pl. 28A, B). Melting operations seem to have been carried out in the S 119 workshop during two occupation phases, without, however, leaving sufficient traces of permanent furnaces to be assigned "S" numbers. Similar operations also may have taken place in some of the remaining Late Bronze Age units in rooms on the ridge top . 108 That production was confined to some variety of small aboveground furnace with the forced air supplied by either some kind of bellows system 109 or blowpipe(s)1 10 might seem a given, but the northwesterly winds buffeting the island throughout much of the year 111 are strong enough to raise the temperature of a charcoal-fueled fire to the ca. 85011083 °C range required for melting bronze 112 without the use of bellows or blowpipes. The lack of precise evidence for the shapes of the missing furnaces and their oxygen delivery systems is offset by the D4-I/II, 2.2 and 3.2 ash deposits and the relatively plentiful scraps of bronze production debris scattered over S 119's 4.1 floor.

Bins Two small bin-like compartments (S 134a and b), constructed from thin, flat stone slabs set upright in the sand, were excavated in the H5-III grid square north of the S I 0 I Sponge-Divers House (Pl. 23B), and a third bin (S 138) was found inside room S107 (Pl. 28A). The prevalence of what are elsewhere called "cupboards" or "enclosures" in other Bronze Age sites is discussed by Shaw. 113 While the bottom of S 13 8 may have been lined with flat chips, S 134a and b were left unlined . There is no evidence to show if and how a ll three were covered. Despite the fact that their specific purpose remains unclear, all three bins were found close to hearths or ovens and may be presumed to have had something to do with food preparation or storage. The interior of S 138 contained pieces of a Cypriot Plain White Ware bowl. It is also noteworthy that no less than nine similar stone-slab compartments were excavated in Hellenistic-Roman levels in grid square G6 just a short distance east of the Sponge-Divers House.

Periodization ofWalls A structural analysis of the H5-1/II walls indicates that the S 126a and b unit evidently was built first, and that it was followed by the erection of the S 121 wall , which in turn was followed by the Sll8 wall. By just exactly what intervals of time these events were separated appears impossible to spell out in detail. The relatively large number of fragments of pithoi, Canaanite storage vessels, and Cypriot vessels in coarse fabric reported from this sector of the island, including H5-III, confirms that the island's Bronze Age community made provisions to maintain some form of extended residence. Likewise, the deposit stratigraphy from seven areas exhibits sequential build-ups of Bronze Age levels: H5-ll, 4.4/ 5.4, 6.1 H5-111 , 3.1, 4.1 H5-Ill/SW, 3.1 , 4.1 16-I/ll, 4.1, 5.1155 F4-III (Room SI07), 3.1, 3.2, 3.3, 4.1 , 4 .2, 4.4, 5.1 E4-III/W Balk, 2.1, 3.1 D4-IIII (S119 workshop), 2.1, 2.2, 3.1, 3.2, 4.1 In the case of the last area, the fill from inside the workshop appears to be associated with two separate floors and two distinct stages of furnace development.

Room Size None of the Northern Cluster's isolated walls appear to belong to a sing le room, unless S 118 and S 121 are somehow part of a common feature. As already stated, it is unlikely that the two belong together, 115 but if they did, the linear distance separating them is 2.4 111. The island's only comp lete or semi-comp lete room units occur in the Southern C luster. Their interna l measurements are as follows. SI07

2.5 X 1.90 m. 1.50 X(?) 111.

LAT E BRONZE AG E OCC UPATION

Sl36 117 SI02 Sll9

1.60 x 2.30 (?) m. 1.80 x 1.85 m. 1.75 X 1.75 m.

These quarters are certainly diminutive by normal standards for domestic housing of any period, 118 providing in the case of three out of the five units barely enough space for an adult person to lie down. If there were any compelling reason to restore a second story, perhaps the room s could be explained as basement or cellar storage units that supported more spacious rooms above them , but no such evidence exists. Perhaps, the easiest way to explain what is otherwise a seemingly strange situation is to look to the ca. 12 m. square area of the SpongeDivers House. As pointed out elsewhere, 119 bomb damage coupled with post-Bronze Age building have obliterated what, to judge from the relative abundance of pottery discoveries made by Bates and the results of our own reinvestigation of his dump , 120 may have been the central architectural feature of the island. It is certain that some of the house 's outer walls were built from reused ancient material that includes a triglyph block and possibly a voussoir. 12 1 It is further I ikely that a significant portion of the rubble making up the walls ultimately stems from Late Bronze Age building activity. This being so, we may have lost the island's main Bronze Age structure, and we are left instead with only the remnants of outlying workroom(s) and storage areas.

Contiguous Unit Construction Given the disturbed nature of what survives (Figs. 2:2, 4:7, 4: 16), it is impossible to determine whether we are dealing with a single, badly eroded building complex strung along the island's ridge or a series of separately maintained but contiguous rooms and open-air courts. The Late Bronze Age walls in the Northern Cluster run into

79

the interior northeast quadrant of the S l 0 I Sponge-Divers House but cannot be shown to reemerge to its south to join up with the Southern Cluster. The F4-I and F5-I/W test squares south of SlOl are completely sterile, except in F5's southwest extension where some scattered rubble trails off in the direction of SlOl. Unfortunately for present purposes, the space between the two test squares was not investigated, but it seems to hold little promise of ever providing much in the way of a " missing link" between the Southern Cluster, the S I 0 I area, and the Northern Cluster. With this being the case, the original layout of structures along the island's ridge may have provided a separate area for storage and work to the south (the Southern Cluster) and something of a more administrative and residential quarter to the north (Sponge-Divers House and the Northern Cluster). Although the question of individual ownership, touched upon above, is probably moot since the entire complex may well have been under some form of communal management, contiguous individual housing is a common enough feature of Bronze Age architecture throughout the region. 122

Chronology For a full discussion of the island's Late Bronze Age occupancy, the reader should refer to the relevant sections in Linda Hulin's and Pamela Russell's discussions of the various Late Bronze Age pottery wares (Vol. H, Chapters 7 and 8) as well as the final chapter (Vol. ll, Chapter 15). The standard dating currently available for the imported pottery in fine fabrics on the one hand and the various local as well as imported coarse vessels on the other hand do not appear to be cosynchronous despite the occurrence of both types in the same archaeological deposits. What they do reflect is that the floruit for the island's occupation occurred in the 14th century B. C. , with possible extensions into the J5th and 13th centuries B.C.

MARSA MATRUH

80

Chapter 4 Notes I.

See, p. 79 and Vol. II, Chapter 15.

the vicinity. The rapidity with which the wind-driven

2.

Because the walls have been so badly mauled over

sand will eat into the soft rocks is astonishing.

time, thickness cannot be used by itself to distinguish interior partitions from party walls. Seep. 57. 3.

A. Fakhry, The Oases of' Egypt 2: Bahriyah and

Farafra Oases (Cairo 1983) 17, fig. 2.

Our experience over three summers' work on Bates 's Island did little to mitigate hi s harsh judgment. See 1985 Report, 59, n. 22. For an assessment of the Etesians as a determinant

4.

But see under Southern Cluster, pp. 57- 59.

in the location of roadsteads and harbors along the entire

5.

See Chapter 3.

eastern Marmarica, see White, Coastal Survey, I I If., and

6.

For more on the effect of rapidly forming aeolianite

passim.

on ancient sites along the Marmarican coast, see White,

Coastal Survey, 25- 30, figs. 16- 18.

19.

This assumes that the total north-south spread was

on the order of 56 m., and that the width of the island set-

7.

See Vol. II , Chapter II.

tlement above 2.8 m. a.s.l. might have averaged out to

8.

Seep . 27.

something around 20 m.

9.

1985 Report, 63- 67. See alsop. 119.

20.

If, in fact, the Late Bronze Age ridge settlement

The exact nature of what gets shot when has not

ever possessed such a feature, which, given its modest size,

been established but the underlying purpose of these con-

hapha za rd appearance, and seasonal character, is perhaps

I 0.

structions is clear. See p. 41. I I.

Two Late Bronze Age sherds were reported from its

2.1 /2.4 layer out of a total of I ,539 sherds from the same deposit, which makes the former statistically meaningless. I

unlikely. But see pp. 65 66 for the S 122 "wall," which, if it were not a ramp, must have been a segment of an otherwise missing perimeter wall. 21.

African Studies, 186 187. Bates, Pottery, 202- 206,

am inclined to view them as intrusive or modern "strays."

pis. 20- 21, figs. 2- 22. The attribution of the house to

See Vol. II, Chapter I 0.

sponge-divers is not Bates 's idea, but ours. For their activi-

12.

Three Late Bronze Age Cypriot sherds were report-

ties both before and after Bates 's arrival until at least the

ed out of a total of 584 shcrds, including one from the

1950s, see 1985 Report, 66, n. 35. See also G. Simpson, The

trench 's surface and two buried in clearly Roman contexts.

Heart of Libya, the Siwa Oasis, its Peoples, Customs and

13.

Seepp. 112- 115 .

Sport (London 1925) 49- 53. H. Morton, Through Lands of

14.

For the contents of the 16-1 / 11, 4.1 level, see p. 57.

the Bible (London I st ed. 1938; 6th ed. 1948) 185- 188. W.

15.

See pp. 52, 54 for the identification and archaeo-

logical history of this feature. 16.

See pp. 37- 39 for the lowest levels uncovered by

Bates containing Late Bronze Age sherds. 17.

See pp. 25- 26.

18.

The so-called "Etesian" winds. Herodotus, 2.29.

MacArthur, Auto Nomad in Barbwy (London 1950) 343. Sec Vol. II , Chapter I I, for the house site's later occupations and building phases. 22.

Sec pp.l19- 120 for possible traces of a cross-wall,

partitioning the house into two unequal parts. 23.

Apart from tracing S I 0 I 's outer walls, our own

Diodorus, 1.39.6. Pliny, Nat. 18.335. After spending the

investigation of the house was limited. It included the

winter of 1913/ 14 in residence in Marsa Matruh, Bates

removal of surface debris and the excavation of a single deep-

(A(i'ican Studies. 127- 128.) described their effects thusly:

level probe across its south wall after it became apparent that either World War II bombing or shelling had cratered the inte-

Because of the flatness of the surrounding country

rior to the point where no traces survived of Bates's excava-

there is nothing to lessen the violence of the winds,

tion throughout its northern two-thirds, and that the supposed

which are frequent and severe . We had, when in the

east-west division wall was no longer present. Judging from

field scarcely a day which was calm, and no three con-

what was left of the badly disturbed northern half, it is diffi-

secutive ones without experiencing at least one of cold

cult to believe that Bates actually cleared as much as his pub-

blustering weather. The summer was little better.

lished records (whose accompanying sketch plans contain

The frequent gales have caused much sand-blasting

egregious inaccuracies) would suggest, and it seems more

and wind erosion of the soft limestone formations of

likely that he, too, limited his digging to a single test.

LATE BRONZE AGE OCCU PAT ION

24.

Bates, Pottety, 203, pis. 20-22, figs. 2- 7a, nos.

1--6 . Today, this pottt:ry is in Harvard Univers ity's Peabody Museum and has been cataloged by P. Russell , Vol. II , C hapter 7, 7.79- 7.98 . See also 1985 Report, 52, n. 7.

4 1.

81

These represent the continuation of the chunk of

115-1 4 .2 deposit excavated cast of S I 18. 42.

A point not fully understood at the time of the pre-

liminary report. Compare 1987 Report, 107. 43.

This has to be said because the coll apse and later

25 .

Aji-ican Studies, 186. Bates, Polle1y, 203- 207 .

26.

Bates, PollefJI, 204-205, pis. 23- 25, figs. 8- 24.

redistribution of the island's Late Bronze Age wa ll s can cre-

27.

See pp . 36- 37.

ate problems in fixing with complete assura nce their origi-

28.

Much of the island 's surface is sti ll covered with

nal o utlin es. Obviously, the main difficulty here lay in sepa-

shrapne l and spent ordnance. The expedition used an unex-

rating S 121 from S 126a and b at the point where all three

ploded tank Hi gh Exp los ive round fou nd in sha ll ow waters

converge, an observation a lready raised in the 1987 Report.

east of the island for a doorstop until commo n sense pre-

44.

While most of this area fa ll s in H5-lll , i.e. , the grid

va il ed. We cou nted ou rse lves fo rtun ate in not encounteri ng

square's so uth east quarter, the western end of H5-111 /W2

mines.

sli ghtly overlaps H5-IV

29.

That it was a dump and not undisturbed fill was

45.

Bedrock in H5- lll is nearly ha lf a meter lower than

made clear by the fact that wind has swept the rocky edge of

the aeo li anite pavement to its immediate north, mea surin g

the western face of the island's ridge bare of sand every-

3.38 m. a.s.l. across its eastern edge but then dipping to 2.84

where e lse except in this one spot. Our digging made it clear

m. in the southeastern corner of H5- lll.

that the dump was tipped in from the direction ofS I 0 I, with

46.

S ll8 , Sl21 , and Sl26 in H5-l and II .

most of the sand deposited at the foot of the ridge's shou l-

47.

Eventually removed in the course of the 1989 sea-

der just above the beach Iine.

son. For more o n S 128, seep. 52 and Fig. 4:12.

30.

See Vol. II , Chapter I 0.

3 I.

See n. I 0.

northeast corner to a depth of 0.30 m. where steri le fill was

32 .

As sha ll become more evident as this report pro-

enco untered. Elsewhere, the trench was c leared to the top of

48.

gresses, va ri ations in the color of the island 's various deposits, wh ich in the majority of cases were made up largely of sa nd, depend partly on the presence of roots (and hence orga ni c activity) and partly o n moisture. The deeper the fill, the wetter and, therefore, darker its color is the rule on ly until the su n had dried out the lower layers.

A 1.5 x 2.0 m. sq uare test was dug into 16- 1/ ll 's

its 3. I leve l. 49.

See p. 48, and n. 39.

50 .

For its vegetat iona l cover and pollen types , sec Vol.

II , 11.1 - 11.3, 11.42- 11.44. 51.

Sherds from this deposit were mixed by error with

the Late Bronze Age 4. I finds excavated beneath it.

The test was positioned to correspond with the spot

52.

See n. 5 1.

indicated by Bates with an "x" on his sketch plan: "U nder

53 .

The other was the S I 02 room located in the E4 grid

33.

the corn er, x, was observab le a distinct stratification," etc. Compa re Aji-ican Studies,

186. Also Bates, Pol/ery,

sq uare to the sou th. 54 .

T here is, for examp le, a 20th ce ntury beduin tabun

203- 207.

ex hibited in the loca l eth nology museum at Marsa Matruh

34.

See n. 33.

that is remarkably close in shape and design. See also Pl. 28B.

35.

No drawn sections of hi s work on the island accom-

36.

World War II damage to the ho use's interior g reatly

contributed to the difficult in sorting o ut what had taken place. 37.

55. 56.

be Late Bronze Age in date. But seep . 57.

39 .

African

Studies,

Sec pp. 55- 56 for two post-Late Bronze Age leve l

bins, S ll4 and SII5. 57.

If this were the case, the S 131 posthole could also

38.

See pp. 54-55 for the Late Bronze Age S 134a and

b bin(s).

pany Bates 's descriptions.

Its 2. I deposit did produce a possibly 17th century

European cook ing pot (Vo l. II, 14.11 ) and a cylindri cal vessel that is e ither Islam ic o r mode rn (Vol. II , 14.17). It a lso contained some early sherds (Vo l. II , 7.9, 7.57, 8.71 , 8.111 ,

134- 135.

Reswfa cing

of'

Parae/onium, 65, n. 14 .

8. 112).

58.

To be certain of this, howeve r, further testing would

For example, what Pausanias ca ll s .1\iSou cc;ypoi or

have to be conducted in the space between F4- l and F5-l/W,

field-stones. Pausanius, 2.25.7 . R. Martin , Manuel d'archi-

whic h, for reasons beyond the immediate control of the

tecture grecque (Paris 1965) 372.

exped iti on, could not be carried out.

40.

MARSA MATRUH

82

59.

See pp. 59, 75 for this feature.

lowed the trade of gold-smithing. The latter legend,

60.

See p. 59.

Iike the name of the island itself, is very curious in the

61.

See pp. 71 - 72.

face of the facts presently to be set forth, [i.e., the pres-

62.

Incorrectly referred to in 1985 Report, 80, as the

ence on the island of Late Bronze Age Cypriot pottery]

"3.1 deposit." 63.

But probably not sleeping because of its extremely

restricted size. See p. 79.

though the reader should bear in mind that it is hardly probable that we are here confronted with a case of a folk-memory running back for such a considerable

64.

See p. 62.

period as three thousand years.

65.

Seep. 77.

70.

Afi-iccm Studies, 187.

66.

See p. 62.

71.

Afi·ican Studies, I 86, feature "B" on map. Bates,

67.

1987 Report, 110. In terms of masonry style, SI24's

plug-shape blocks resemble certain of the Archaic walls of the

"Ethnographic Notes from Marsa Matruh," JRAS (Oct. 1915)737:

Extramural Sanctuary of Demeter and Persephone at Cyrene. Sec D. White, The Extramural Sanctuar.\' of Demeter and

I learnt from another source that Sidi 'Awam was sup-

Persephone at c_,.,.ene, Libya. Final Reports V: The Site :5

posed to have come to Matruh in a boat "from the

Architecture, its First Six Hundred Years of Development

west"; that he was ever held, even in his lifetime, to

(Philadelphia 1993)7,9, II, 18,29,41, 179,n.28.

have been a waly; and that he died a natural death. The

68.

For this reason, no feature number has been

assigned to them. 69.

As recounted in Afi·ican Studies, I 87, the legend of

the island ran as follows:

informant from whom I had this version accounted for the name of the Gezireh-t-cl- Yahudy by saying that in old times there lived on the islet two Jews, who were goldsmiths. He added that these Jews were called ' Ishak and Hugah (?) .H e could give no account of their

Once upon a time there was a fisherman here who

death. Some of our local workmen of the expedition

owned a little falucca in which he used to go out fish-

pointed out a ruinous cairn on the islet as the grave of

ing. He was very lucky in his catches, which was natu-

the Jew (or Jews), from whom the place was named.

ral, for he was very good and pious. ow there lived in those days a Jew, and when he saw

72.

For the possible usc of wind-screens that could be

dropped to allow the wind to supply forced air to the metal-

how many fish Sidi 'Awam used to get, he built himself

melting furnaces housed insideS 119 as well as perhaps some

a fa Iucca and went out fishing. But the Jew being a Jew,

of the other room units, sec p. 76 and Vol. II, Chapter 9.

and a hideously wicked man, caught nothing. At this

73.

All that is required for the melting of ingots is a

the Jew grew very jealous of Sidi 'Awam and he final-

ring of stones, a pile of hot charcoal, and, in addition to the

ly murdered him. But the Arabs killed the Jew and

wind, a clay tuyere connected to a bellows of which little can

buried him on the littl e island in the first lagoon, which

be expected to surv ive. See R.F. Tylccote, A HistOIJ' of

to this day is called Gezireh-t-e l- Yahudi. And the good

Metallwgy (2nd. ed., London 1992) 21, 25, 38, fig. 7. Of the

'Awam became a saint.

sort of furnace depicted in the much analyzed metal-workers relief from the Old Kingdom Tomb of Mereruka,

Bates recounts a second version of the story in 1987 Po/fer\·,

Saqqara. Egypt, see C.J. Davey, "Tell edh-Dhiba 'a and the

20 Iff.:

Near Eastern Metalworking Tradition ," in R. Maddin, ed.,

The Beginning of the Use of Metals and Alloys (Cambridge, In the eastern Lagoon is a small island, known among the Au lad 'Aiy Arabs of the vicinity as the Gezireh-t-elYahudy. This name- "The Isle of the Jcw"- is locally

Mass. 1988) 63- 64, fig. 6.3. 74.

For bowl-furnaces, a clay-lined hole in the ground,

see R.J. Forbes, Studies in Ancient Technology 8 (Le idcn

explained in two ways. According to one story, the

1964) 123. For pit furnaces sec R.J. Forbes, Studies in

island served as the burial place of a wicked Jew, the

Ancient Technology 9 (Lciden 1972) 33, 149. For the recon-

slayer of the patron saint of Matruh and its neighbor-

struction of a Tell edh-Dhiba 'a melting furnace set in a pit

hood, Sheykh Sidi 'Awam. Another and quite different

see Davey (seen. 73) fig. 6.4.

explanation is to the effect that, long ago, there lived on the isle two Jews, named ' Ishak and Hugah (?),who fol-

75.

See White, Wate1; Wood, 932- 933 . Fuel and oxygen

supply are further discussed in Vol. II , Chapter 9. For the

LATE BRO NZE AGE OCCUPATION

need for forced air to melt copper, see R.J. Forbes (seen. 74) 33. Also see Tylecote (seen. 73) 38 and seen. 109. 76.

Wright, South Syria and Palestin e, 340 a nd

80.

83

Wright, Cyprus, 377- 378 and 418-419. On C rete,

the use of sun-dried brick was mainly relegated to the upper stories of hou ses : Shaw, Minoan Architecture, 194; Hayden,

397-400 and Wright, Cyprus, 361 and 408-409, di stin-

Cretan Architecture, 122 . Shear, Domestic Architecture,

guishes between more or less rounded fieldstones that result

432 , nn . 745, 749- 759. Shear, Panagia Houses, 8, n. 9.

from the outwash of wadis, etc., and "fl at angular stones

Reich (see n. 78) 5-7 . Ben-Dov (see n. 78) I 04. Wright,

with plane surfaces" that occur in se mi-dry region s where

South Syria and Palestine, 349ff.

" the extremes of the diurnal temperature range acting variously on the diverse mineral constituents of certain rocks breaks up the surface into a network of cracks and causes it to spall off in a fl at layer." The flat but jagged fieldstones

8 1.

Shear, Domestic Architecture, 433 , nn. 762 and

763. 82.

Wright, South Syria and Palestine, 360- 361.

83.

W. Dinsmoor, Th e Architecture of Ancient Greece

that lie scattered over the coastal pla in sloping back to the

(3rd. ed., rev. , New York 1975) 4- 5. Wright, South Syria and

"Great Ridge" south of Matruh would seem to conform to

Palestine, 23, 360, 364, 413 , 41 7. Wright, Cyprus, 385. 84.

the latter. 77.

See Wright, South Syria and Palestine, 392- 394,

esp. 394 for the use of outer and inner masonry skin s packed

D. Conwell suggests an interestin g parall e l in the

area of Pampa de las Llamas-Moxeke, Peru , wh ich dates ca . 1800 B.C.:

with rubble at Megiddo and Shechem. See also Architecture

of israel, 136- 142. For the ca. 2.0 m. wide city wall assoc i-

Most of the population- most likely farmers, laborers

ated with St. Vat Teii-Abu-Hawam , see L. Gershuni, ZDPV

and artisans- lived in small, crowded irregular clusters

97 (1981) 36-44 and QDAP IV (1935) 11 - 13, pl. II. 1.

of room s with mud-coated cane or wood walls . These

Weinstein , " Was Te ll-Abu-Hawam a 19th. Century [sic]

dwellings left few rema ins except for low, sto ne-wall

Egyptian Naval Base?" BASOR 238 ( 1980) 83- 90. The site,

footings.

set on an island within a coastal lagoo n, has been interpreted as a trading emporium, which provides a number of inter-

S. and T. Pozo rski , "Ea rl y Andean C iti es," Scientific

esting parallels to Bates 's Island. Te l Nami, set on a water-

American (June 1994) 48.

girt peninsula 5 km. so uth of Athlit and another site of

85.

Seep. 74 and Vol. II, Chapter 9, n. 28.

potentially great interest, received an impressive seaside

86.

Shaw, Minoan Architecture, 187. For rubbl e set in a

rampart in the 13th century B.C. Whether it se rved as a bar-

mud packing at Gournia, see Hayden, Cretan Architecture, 24.

rier against the water or human attack remains to be deter-

87.

Shear, Domestic Architecture, 433, n. 764; 435-436.

mined . See A. Raban , C.M.S News 16 (Haifa 1989) . For

88.

Wright, Cyprus, 381 , 409.

Cyprus, see Wright, Cyprus, 406.

89 .

Wri g ht, South Syria and Palestine, 75 , 349 ,

For purposes of the following di sc uss ion , " region"

359- 360. At Tel Nami, Artzy mentions the use of clay

means the eastern Mediterranean and does not include Italy.

78 .

cement produced by mixin g swa mp mud with crushed pot-

For the use of rubble on Crete, see Hayden, Cretan

tery and limestone in order to deal with the " problem s of sea

Architecture, 122 and Shaw, Minoan Architecture, 77- 83.

water and sand." See C. M.S. News (Haifa 1988).

Mainl and houses: Shear, Domestic Architecture, 431, 435ff. ;

90 .

Shaw, Minoan Architecture, 207- 218 .

Shear, Panagia Houses, 7- 8; and G. Hiesal, Spii.thelladische

91.

Shear, Domestic Architecture, 443-444. Shear,

Hausarchitektur (Mainz 1977) 6. Palestine : R. Reich, " Building

Materials

and

Architectural

Elements,"

Architecture of Israel, 1- 3. M. Ben-Dov, " Middle and Late

Panagia Houses, 11 - 12 . 92.

Wright, South Syria and Palestine, 422-423. Reich

(seen. 78) 9.

Bronze Age Dwellings," ibid., 99-1 04; Wright, South Syria

93.

Wright, Cyprus, 390- 391.

and Palestine, 359- 360 and 397ff. Cyprus: Wright, Cyprus,

94.

Wright, South Syria and Palestine, 456.

360- 362 ; 408-410.

95.

Wright, Cyprus, 493.

96.

Wright, South Syria and Palestine, 456.

97 .

Shaw, Minoan Architecture, 221 - 222. Shear,

79.

Wright, South Syria and Palestine, 349 and 360.

Wright, Cyprus, 377 , 418. The construction of the older hou ses in Siwa appear externally to be solid mud, but the buildings are, in fact, made of mud-coated rubble, reinforced with palm logs.

Domestic Architecture, 45 1. Shear, Panagia Houses, I I. 98.

For an interestin g discussion of the advan tages of

placing doors by the corners of rectangul ar room s built of

MARSA MATRUH

84

rubble, see Wright, Cyprus, 474-475. 99.

Shear, Domestic Architecture, 36 n. 58, 442.

I 07.

Shaw, Hearths and Ovens, 238, 254 . See also G .

Gesell, L. Day, and W. Coulson (see n. I 05) 96, pl. 29c, for

I 00.

Shaw, Minoan Architecture, 174ff.

a Geometric period oven which "may have had a domed

I 0 I.

For adulterated lime flooring on Cyprus and in the

pise superstructure." I thank Leslie Day for this reference.

Levant, see Wright, South Syria and Palestine, 437 and Wright, Cyprus, 468-470. Lime plaster mixed with clay is known from mainland Greece. See Shear, Domestic

I 08.

For an attempt to distinguish the Northern from the

Southern Cluster of rooms by possible function, see p. 79. 109.

For pot bellows, which , if used on the island would,

Architecture, 444, n. 859. For lime mixed with crushed ter-

one would have thought, left some trace of their distinctive

racotta on Crete, see Shaw, Minoan Architecture, 216.

clay bowls, see C.J. Davey, "Some Ancient Near Eastern Pot

102.

In the Levant interior house rooms paved with

Bellows," Levant I I ( 1979) I 0 1- 1 I I. See also, C.J. Davey

stone have been taken to indicate that animals were domi-

(see n. 73) 63- 68. For traditional skin or bag bellows, see

ciled under the same roof. Wright, South Syria and

R.J. Forbes, Studies in Ancien/ Technologv8 (Leiden 1964)

Palestine, 441-442. Presumably no animal over the size of

111 - 112. Tylecote (seen. 73) 38.

a sheep, goat, or possibly pig was kept live on Bates's

II 0.

G.A. Wainwright, " Rekhmire's Metal Workers,"

Island. Perhaps the single Bos metapodial shaft fragment

Man 44 ( 1944) 95- 96. R.J. Forbes, Studies in Ancient Tech-

reported by Reese from the 1.2 surface of room S I 19 had

nology 8 (Leiden 1964) Ill. Davey (seen . 73) 64, fig. 6.3.

been slaughtered on the mainland. See Vol. II, Chapter I 0.

Forbes, Studies in Ancient Technology 9 (Leiden 1972) 34.

I 03.

Wright, South Syria and Palestine, 435.

I 04.

The suggestion is Philip Betancourt's . The effects

of freezing temperatures, which a sunken floor could also help to offset, are unlikely to be a factor here. The S I 07

Ill.

Seen. 18. For wind used by itself to smelt iron in

Sri Lanka and copper in the Andes, see Vol. II, Chapter 9, nn. 29 and 38. I 12.

The precise temperature depends on the nature of

room contained a floor-level storage bin, S 138, probably

the bronze alloy; copper melts at I 084° C. For more on the

used for food, next to its oven, S 137.

island 's metal working activities, see Vol. II, Chapter 9.

As well as of many other prehistoric sites. M.

I 13.

Shaw, Hearths and Ovens, 238, n. II.

Shaw, Hearths and Ovens, 232- 254 . Add G. Gesell, L. Day,

I 14.

For the problematic Late Bronze Age nature of the

I 05.

and W. Coulson, "Excavations at Kavousi , Crete, 1989 and 1990," Hesperia 64 ( 1995) 70, pl. 17 :a, whose reference I owe to Leslie Day. I follow Shaw's definition of"hearth" as

16-1/ 11 sequence, see pp. 36, 57. I 15.

See pp. 50- 51.

I 16.

It is not clear whether this space belongs to a sepa-

a constructed floor fixture used with fire, regardless of

rate room or it was part of an external court, perhaps con-

whether or not its purpose was for cooking or heating, and

nected to its north with S 125. See p. 63.

"oven" refers to a partially or almost completely closable space heated with embers rather than flames. SeeM. Shaw,

Hearths and Ovens. passim for specific constructional vari-

I 17.

It is again unclear ifS 136 is an actual room. Seep. 66.

118.

See, for example, the house plans assembled by M.

Ben-Dov (see n. 78) 99ff. in which most Middle and Late

ations in hearth design. "Furnace" refers here to enclosed

Bronze Age Canaanite house rooms, as opposed to what

heating devices used to melt bron ze for casting but not for

pretty clearly seem to have been storage spaces, measure 3

smelting. That cooking and heating might have been carried

or more meters in at least one direction . The same appears

out by means of portable metal braz iers of the kind dis-

to be true of the Panagia Houses at Mycenae, and where

cussed by Shaw, ibid. 251, is of course a possibility, but no

smaller units occur roughly analogous in size to the island's

traces of portable heaters, whether metal or terracotta, were

room units, their function seems to be for storage, i.e.,

recovered from the island. I 06.

Not given "S" numbers because of their pulled-

apart condition, these hearths were associated with the H5-

House II , Rooms 16- 18, and 21. See Shear, Panagia

Houses, 36-43. The same conclusion is substantiated by the various plan s illustrated by G. Hiesal (seen. 78) passim.

ll, 3.2 and 3.3 deposits; neither are, therefore, securely Late

119.

See pp. 40, 118. See p. 40.

Bronze Age in period. According to Shaw, Hearths and

120.

Ovens, 235, hearths used for cooking normally were set

121.

See p. I 19.

against a wall or in a corner, whereas hearths positioned in

122.

E.g., Shear, Panagia Houses, plan at volume's end.

a room's center were so placed to ensure an even distribu-

Hayden , Cretan Architecture, 112- 113, 135. Wright,

tion of heat and perhaps even light.

Cyprus, 227ff.

Chapter 5

AREA VI: THE "GREAT RIDGE," BATES'S LIBYAN CEMETERY, AND SITES WITH SHELL-TEMPERED POTTERY Linda Hulin and Donald White

Much of what fol lows summarizes observations that have already appeared elsewhere. They are presented here to provide background to a final consideration of two unusual tomb groups made up of a small number of terracotta and stone vesse ls, random sherds, and a few shells. The finds were taken to Harvard University's Peabody Museum by their discoverer, Oric Bates. Bates 's own description and analysis of what he had found appeared in the posthumous monograph on his 1913/14 work at Matruh 1 and are repeated below in sli ghtly modified form to facilitate their accessibi li ty. Three years before

his death , Bates returned again to his discoveries in a short article to which Flinders-Petrie appended a g loss corroborating the ethnic (i.e., Libyan) origin and the provisional earliness of the tomb 's contents.2 Short notices of the University of Pennsylvania Expedition's attempts to relocate the cemetery area in the mid-1980s appeared in the campaign reports for the first two seasons in an article by D. Conwe ll / intercalated with somewhat more extended descriptions tied to the general subject of the Late Bronze Age Libyans. 4 As far as we are aware, nothing further of consequence has been written on the subject. 5

Bates's Libyan Cemetery Location Bates locates his "Cemetery A" as follows: About two miles east of Matruh projects from the northern face of the Great Ridge a small rounded spur, which is conspicuous as having on its summit a modern grave encirc led- as in common in these parts- with a

wa ll of loose stones. Here, under circumstances already described, was found a group of five graves. 6 The "circumstances already described" refers to a 1913 artic le 7 discussing nomadic burials that dated from the then-present back to Roman times in other parts of the Marmarica; except by inference, it sheds no further light on Cemetery A.

86

MARSA MATRUH

What Bates habitually referred to as the Great Ridge 8 is a long, elevated formation of gray limestone that starts to take shape about 250 m. south of Matruh's main harbor. According to Gardner, the ridge formed in the late Pleistocene when aeolian (dune) sand was blown up against, and partially over, the Pliocene rocks previously forming the coastline; the blown sand eventually was transformed by calcium carbonate into a ridge of well-cemented aeolianite with a calcrete capping. 9 It extends parallel to the coast for approximately 8 km . east of Matruh to where the eastern lagoon series peters out just west of Ras Alam el-RCtm. The flat land, hemmed in by the ridge to the south and by Matruh 's harbor and the First East Lagoon to the north, consists of the same deep brown soil that makes up the remainder of the coastal plain west of Matruh. The area draws water for crops from the winter rains as well as from cisterns or from wells spotted through the plain (Fig. 2:4). The latter tap the aquifers fed by the Libyan Plateau, which begins its own gradual rise from the desert floor ca. 6 km. to the south. In Greek and Roman times, the combination of relatively deep soil and adequate water (Pl. 34A) made possible an unusually intensive level of agricultural activity, which was "not surpassed elsewhere in the Marmaric littoral zone." 10 And elsewhere : Although the vegetation in Marmarica depends almost wholly on the rainfall, the richness of the soil and the size of the fertile area in the vicinity of Matruh must have attracted and sustained a population at a very early period. The evidence of agricultural activity in Roman times (threshing floors, vats, cisterns, olive-presses, etc.) are numerous; and in the late third, or early fourth, century of our era, Paraetonium was very naturally the centre at which government grain-tithes were paid in. 11 The existence of this agricultural land represents one of the key factors that has been argued to have set the Matruh area apart from all of the other roadsteads and anchorages exploited in antiquity between Catabathmus Maior (modern

Sa!Ctm) at the Libyan border and el-Alamein. 12 The Pennsylvania Expedition first attempted to pinpoint the location of the cemetery in 1985. Bates's area map (Fig. I :7) positions it along the northern rim of the ridge , ca. 1700 m. southeast of Bates' Island. 13 To reach the high ground in question, it was necessary to drive along the secondary road connecting the eastern outskirts of Matruh with Ras Alam el-RCtm for slightly over a mile and then to drive across the privately owned farmland that occupies the rising ground leading to the ridge's rim. Unfortunately, once this objective was achieved, the ridge slope and rim area indicated by the map preserved no sign of the low circle of loose stones surrounding what Bates took to be a modern Bedouin burial site, which in turn provided the ancient cemetery with its most easily recognizable feature (Pis. 34B, 35A). Nor was it any longer readily clear what Bates had meant by the "small rounded spur" projecting from the ridge's north face , whose summit he said marked the site of the tomb senes. By the later 1980s, the broken calcrete pavement of the ridgetop was still free of modern building. Its surface was covered patchily with a thin layer of gravelly soil and broken stones. By mid-summer, the grass cover visible in Bates's photo 14 mostly had withered and blown away, leaving only the odd scrub evergreen bush to mark the edges of the shallow sinkholes that naturally indent much of the stony surface. Partially filled with drifting sand, the sinkholes tend to capture enough water for their bottom surfaces to remain grassed-over into early summer. Depending on their configuration, they can look deceptively like the sorts of elliptical, man-made cist graves described by Bates. Repeated attempts were made in both 1985 and 1987 to investigate the most promising of the grave-like sinkholes in hope of either re-discovering traces of Bates's original burials or, even better, to unearth fresh, unexcavated burials. We were disappointed on both scores. The only sherds to turn up were either standard Greco-Roman coarse pottery found lying on the rocky surface, for the most part outside of the sinkholes or, far less commonly, the odd, presumably "Libyan," shell-

87

AR EA VI

tempered sherd, a few of which were actually found inside an otherwise sterile sinkhole. What was possibly the most promising area for investigation, since it came closest to Bates 's description of a projecting spur, was being used in 1987 for contemporary Bedouin burials, which put a stop to any further investigation. By the same year, much of the remainder of the ridge slope and summit had been staked out for future house construction, and presumably by now it is covered with modern dwellings. Left holding what amounts to a will-o-the-wisp, all there is to go on today are Bates's own description , photographs, and drawings.

Cemetery Description Bates reports finding five tombs (numbered A.l - A.5) clustered on the east face of the spur. Apart from the single wall-enclosed contemporary Bedouin burial and some small stone cairns weathered down to ground level on its west side, 15 the spur preserved no additional remains. The a.s.l. elevation of the five tombs was estimated to be ca. 30 m. The burials were roughly elliptical in plan, measuring ca. 1.50 m. east-west and ca. 1.10 m. north-south. Their average depth of 0.30 to 0.40 m. seems to have been obtained by simply evacuating a few centimeters of the soft bedrock underlying the stony soil. Surface markers of any kind, as well as cap-stones, were conspicuously missing. Tombs A.3- A.5 had been cleared out completely by the time of their discovery, either through water and wind erosion or by plundering. On the other hand, A.l and A.2 (Pis. 35B, C), separated from one another by an interval of ca. 0.30 m., contained skeletal remains and burial goods. A.l (Pl. 35B) contained the badly disarticulated remains of a body lying on its left side, head facing east. The body was in a partially contracted position, which left the right femur at an angle of 90° to the spine; the left leg was flexed less acutely. The left arm was straight, its hand thrust between the knees. The body in A .2 was more knocked about (Pl. 35C). It lay on the right side, head again facing

east. The left elbow was bent, with the left hand placed somewhere near the face. Only a fragment of the lower jaw and six teeth were left from the head. Given the harshness of the local climate, it is not surprising that no traces could be observed of either clothing or cords to rope the bodies in their contracted positions. Of the manner of their interment, Bates had this to say: The bodies, although both so oriented as to have the heads east, lay on different sides and in different degrees of contraction, thus showing a careless departure from a presumably rigid primitive canon . 16 After ruminating on the various objects buried alongside the bodies , Bates concluded that graves A.l and A.2 , whose contents "hint at a whole primitive culture, hitherto quite unknown, and as rich , presumably, as that of Predynastic Egypt itself,'' 17 should date between 2000 and 1500 B. C. [n his view their ethnic origin must be Libyan. 1H In his short addendum , Flinders-Petrie concurred with Bates's Libyan association and tentatively dated the stone vessels to the 12th Dynasty on the basis of parallels with Egyptian objects as well as with a small collection of similar but unprovenienced stone vessels purchased by him in Egypt and housed in the University of London collection. 19

Contents of Graves A. 1 and A. 2 Both tombs contained a similar assortment of grave offerings, consisting of sherd material , a small number of complete stone and terracotta vessels, and mollusk shells (Figs. 5: I and 5:2). Several sherds said to belong to a single terracotta pot were also found in surface fill immediately outside A.l, and they may be assumed to have been washed out or otherwise displaced from its interior. No metal artifacts were recovered, but if such ever existed, the lack of effective protective soil cover and harsh weather conditions could have caused their complete disappearance. Bates shipped the portable artifacts to Harvard's Peabody Museum some time before 1918;20 it is

Fig 5.-J. Bates S sketch plans and sections a} his tombs A. I and A .2, labeled '.'figs. 1-4," and can tents of tomb A.!, labeled 'jigs. 4-9," and A.2, labeled 'jigs. II, 12, a and b" (from Af'ican Studies, pi 26)

89

AREA VI

--~=-- ~

---

a

d ,,,~\\\\\\1\\

'\ \\ -,~ \\~

\ \\\\

b

e

f

c

0 g

1-

10

5

----1

-

:::j

Fig 5:2. Late Bronze Age stone and terracotta artifacts from Bates :S· Area VI Libyan Cemetery. Redrawn by J Thornji-om African Studies, pis. 26 and 27.

90

MARSA MATRUH

not known what happened to the skeletons. The assemblage was still intact in 1990, except for the loose sherds reportedly found on the ground outside of A.l. 21 Arrangements were made for thermoluminescence tests to be carried out on samples from two of the clay vases.

5.7 lridina shell A.l /R.4. Found just under the basalt vase 5.1 .

Not ill.

5.8 Helix Nucula Shell A.l /3. Intrusive; found by skeleton's left hand.

Not ill.

TOMBA.l

TOMBA.2

5.1

5.9 Terracotta Jar Fig. 5:2e; Pl. 36 A.2/R.l =Peabody B-128. H. 11.5, d. at spring ofneck 7.6, max. d. 9.2. Handmade. Fairly hard, uniform gray-black ware; black inside. Partly smoothed on exterior and decorated with rows of faintly incised, short, and nearly vertical strokes. Sooty when found.

Basalt Vase Fig. 5:2a; Pl. 36 A.l / R.l = Peabody 8-131. H. 8.7, max. lower d. 11.75, min. upper d. 5.2. Incomplete outward flaring lip. Tapered open shape, bored cylindrically with horizontal score marks left over from winged drill on interior.

5.2 Basalt Jar Fig. 5:2b; Pl. 36 A. Ill = Peabody 8-132. H. 9.0, d. 8.3. Incomplete outward flaring lip. Bore expanded to follow approximate profile of outer wall. Short hairlike striations left over from polishing and traces of pecking in two places mark the surface of the exterior.

5.3 Sherd Fig. 5:2c A.l /R. W. 6.0, h. 4.1, max. th. 0.5. One of several loose sherds no longer preserved. Handmade. Sandy, black fabric, pebble-smoothed inside. Traces of greenish black slip on exterior. Incised zigzag pattern between two parallel lines.

5.10

Terracotta Jar Fig. 5:2f; Pl. 36 A.2/ R.2 = Peabody 8-129. H. 8.6, d. of mouth 5.0, max d. 7.6. Handmade. Red ware with unsmoothed exterior. Inside of neck partly smoothed. Painted red. Opposing ears or lugs at shoulder.

5.11

Stone Palette Fig. 5:2g; Pl. 36 A.2/R.3 = Peabody 8-133. H. 4.6, max. d. 5.6. Small stone palette or mortar. Spheroidal piece of purplish conglomerate with greenish white inclusions. Nearly circular in outline, with a 0 .5 em. deep recession in the top. Shows evidence of polishing as well as random surface pecking.

5.4 Terracotta Jar Fig. 5:2d; Pl. 36 A.l / R.2. = Peabody 8-130. H. I 1.6, upper d. across mouth 8.2, max. d. I 0. Handmade. Soft, fairly coarse, buff fabric, with minute white inclusions. Faint reddish tinges in two places on exterior; sections of irregular black core caused by irregular firing. Inside of neck and exterior pebble-smoothed.

5.12 lridina Shell A.2/ l. Found in front of skeleton's chin.

Not ill.

5.13 I rid ina shell A.2/2. Found in front of the skeleton's chin.

Not ill.

5.5 lridina Shell Not ill. A.l /2. Thin pearly shell of the fluviatile Unionidae/ 2 found above the right femur. Said to be ofNilotic origin.

Chronology

5.6 lridina Shell A.l / R.3. Found just under the basalt vase 5.1.

Not ill.

As already indicated, Bates dated his A. I and A.2 assemblage to ca. 2000- 1500 B.C., or the Middle Kingdom to early New Kingdom. He evidently arrived at this conclusion by comparisons in technique with Old Kingdom Egyptian stone and clay vessels and by analogy with

AREA VI

Nubian mortars, but he died before citing whatever specific examples he had in mind.23 Apart from Flinders-Petrie's remarks already noted/ 4 this appears to be where the matter rested until 1992 when the Luminescence Dating Laboratory at the University of Durham, U.K., tested two samples from the Peabody collection (5.4 and 5.9). The first sample (5.9) produced a thermoluminescent date of between I 000 and 3000 B. C., and the second (5.4) a date of between the beginning of the present era and 2000 B. C. Although the results supply only very broad dates, they do not

91

contradict the idea of a local presence of Bronze Age Libyans that already was suspected from the I ithic tools and ostrich eggshells found on Bates's Island 25 as well as the scatters of Marmaric Wares found elsewhere locally on the Area VIB ridge and along the eastern lagoon system. 26 It, moreover, suggests that any future evaluations of the Late Bronze Age Libyans should take into account Bates's assemblages of stone, terra cotta, and shell artifacts in the Peabody Museum and perhaps the stone vases in the University College of London collection. 27

Sites with Shell-Tempered Pottery Reconnaissance conducted in 1987 along the lagoon system east to Ras Alam el-RQm (Plan 3) led to the discovery of surface scatters of small sherds of a distinctive hand-made pottery which was originally termed "Marmaric Shell Tempered Ware" by Linda l--lulin.28 Since it appears to be of local (i.e., of Marmarican) origin, our initial hope was that the pottery could, in theory, be used for tracking the presence of the island occupants' Late Bronze Age indigenous Libyan neighbors, or, if not them, their historical-period descendants. After the initial recognition of this pottery on high ground along the lagoons east of Bates's Island, additional scatters were observed on the so-called Great Ridge to the southeast of the modern town as well as in very limited quantities in the area of Wadi Aghiba south of Ras Umm el-Rakham, 25 km. west of Matruh. Finally, a few sherds were recovered on Bates 's Island itself, along with a single sherd from the Egyptian Antiquities Organization's late Roman excavation dump near "Cleopatra's Bath." 29

to either side of the ridge to the edge of the second eastern lagoon. Moon tide conditions and coastal storms are enough to flood the flats even during the summer months, despite the presence of a causeway (Pis. 6A, B, 7A) / 0 and it is clear that a rise of 0.75 m. in the average water-level of the lagoon system during the Late Bronze Age would have transformed the Area II hillock into a yearround island. 31 Where the north end of the ridge falls away to meet the sand flats, its surface is littered with fallen scree. At the edge of the scree line at the ridge's northeast corner (Area liB), a small number of Shell-Tempered Ware B sherds were found in loose sand in a way that suggests that they had been washed down the slope by the winter rains from their original location atop the north end of the Area II ridge. As displaced objects, however, they tell us nothing about their context other than the fact that the hill situated closest to Bates's Island seems to have been occupied in some capacity by the region's local population.32 As to period, they again reveal nothing.

Area liB

Area VIlA and VIIB

Area II is a low ridge outcropping of weathered sandstone (Pis. 3 7 A, B) that parallels the lagoon shoreline east of Bates's Island. It measures ca . 210 m. north-south and 95 m. east-west, and its maximum elevation is 7 m. a.s.l. Sand flats spread

Slightly more informative results were obtained from the large ridge formation that rises between Area II and the western shore of the second East Lagoon. Area VIlA refers to the south slope of the sandy saddle between the ridge 's two peaks; VIIB

92

MARSA MATRUH

is the east slope of the northern peak of the same ridge formation facing the second East Lagoon (Pl. 38A).ln both loci small surface spills of ShellTempered B sherds (Pl. 40C) were deployed on the ridge's sandy surface (Pl. 38B) amid patches of scrub maqui free of any association with wall masonry or any other potentially significant leftover signs of human activity. On the other hand, both scatters seemed to mark some form of occupation, and unlike the Area liB sherds, they did not appear to have been washed down the slopes to their final resting places. The Area VIJ ridge sites and the Area Vlll hill (described below) would have been surrounded by a sheet of water, which would have been easily waded across if the lagoon system's water-level were up by ca. 0.50/0.75 m. when the sherds were deposited.

Area VIII The western slope of the hill rising off the sand flats between the second and third eastern lagoons (Area VIII) constitutes a similar, fourth surface scatter of Shell-Tempered sherds. In all four instances, the sherd concentrations lay within visual contact with Bates 's Island. We do not know how the people who left the sherds used Area 11; in the case of Areas VIlA, VliB, and VIII , they deliberately seem to have sheltered their observation posts or temporary resting places just below the exposed, rocky crests of ridges which were close enough to keep Bates's Island under observation. All four hilltop sites are uncomfot1ably restricted in size, and they are, at the same time, lacking in direct access to the wells/ 1 cultivatable soil, and pasture land that were available a short distance inland from the lagoon basin. 3" In addition, it is hard to imagine desert pastoralists under any circumstances selecting what must have been at least partially water-locked sites for their principal coastal campsites. If the sherds were indeed early, i.e., contemporary with the island's Late Bronze Age occupation, we could, in theory, be dealing with the meager remnants of Late Bronze Age lookout stations used by the native element for the main purpose of keeping the island's foreign occupants under surveillance. The evidence for an early

date for the ware, however, has proven to be elusive at best. Based on recent discoveries made by the University of Liverpool team in the area of Zawiyet Umm el-Rakham, it seems best to regard, for now at least, the bulk of our samples as belonging to the Hellenistic Greek and Roman periods rather than the Late Bronze Age. 35

Area IX One final cluster of lagoon basin sherds (Pl. 40B) turned up on a low sandy saddle south of the coastal ridge to the south-southeast of the last lagoon, directly west of Ras Alam ei-Rum (Pl. 39B). Disassociated as always from any traceable man-made features , they can be assumed here to have washed down from a nearby spur of the coastal ridge whose surface was stripped bare by wind and water erosion. In any case, Bates's Island could not be observed directly from this site.

Area VIB This site (Pl. 39A) was discovered atop the Great Ridge to the southeast of the modern town , ca. 140 m. west of what we have taken to be the site of Bates's Libyan Cemetery. 36 The northern edge of the windswept ridge surface was marked by a small shallow oval depression (0.30 m. northsouth, 0.80 m. east-west) crusted over with rainwashed sand (Fig. 5:3). Random piles of rubble stones surround the depression , but it is difficult to say whether they were moved to their present location for some purpose or owe their presence simply to the accidents of time, because the entire ridge surface is marked by little heaps and scatters of stones. A relatively large concentration of ShellTempered C sherds were embedded in the depression's ca. 0.05 m. thick sand crust. Surface cleaning led to the discovery of three nearby man-made features. The first was a line of three stones forming a short, ca. 0.45 m. long, wall that ran eastwest across the notihern end of the depression. Directly south of the wall was a shallow pit measuring 0.40 m. east-west and ca. 0.12 m. northsouth. A second shallow pit (0.20 m. east-west and

AREA VI

- -

-

-

-

- --

-

-

- - --, I I I I I

N

93

ceramics must have ranged freely over the ridge at one time. On the other hand, the finds were never sufficiently abundant to suggest the existence of a campsite. Finally, the first East Lagoon and Bates's Island are visible from much of the northern edge of the Great Ridge.

Ras Umm el-Rakham Area

I

South Pit I

L------------

0

0.5

I _

_j

I.Om

Fig. 5:3. Plan ofArea VIB, Bedouin(?) pits or hearths.

0.25 m. north-south) was 0.45 rn. southwest of the first pit. A large Shell-Tempered C sherd lined the southern rim of the second pit. Other sherds were scattered in the 2.2 and 2.3 deposits filling both pits. Traces of burning indicate that both pits had served as hearths, which makes it likely that the short wall was used as a windbreaker. The antiquity of this installation was, however, immediately called into question when bits of cloth and modern iron nails appeared in the north pit and broken bottle-glass fragments 37 turned up in the south pit. In other words, both pits must be "modern" (of probable Bedouin origin), and, unless the hearthmakers had dug through a pre-existent surface scatter of old sherds and in the process mixed the modern nails and broken bottles with the old sherds, Shell-Tempered Ware C ought to be a relatively recent, local handmade fabric (Pl. 39C). Elsewhere along the northern edge of the Great Ridge, in the general vicinity of Area VIA where we had anticipated rediscovering Bates's Libyan Cemetery, we found the odd Shell-Tempered Ware A and B sherd, indicating that the users of these

The wadi systems to the south and west of Zawiyet Umm el-Rakham, the Pharaonic Apis, ca. 25 km. west ofMatruh, along with the area around the familiar Ramassid fortress (Fig. I : I ) 38 were searched for traces of surface sherds. No imported Late Bronze Age sherds of any kind were retrieved by the Pennsylvania expedition from the coastal plain immediately outside the actual fort, whose excavation dumps also failed to produce either Late Bronze Age imports or Shell-Tempered Ware. The absence of the former is perhaps surprising, but in the case of the local wares, the result is in keeping with what has been argued for the fortress's anti-Libyan defensive role. 39 The extensive Wadi Aghiba system running into the sea immediately west of Umm el-Rakham's headland (Pl. 40A) was formed out of a large natural bowl or sinkho le. Along its southern rim , set about 0.5 km. inland, were small random scatters of sherds of Shell -Tempered Ware A and B (Pl. 408, C). Two sherds were also discovered close to a long stretch of curving wall of indeterminate, but perhaps ancient date, located on the surface of the flat crest of the inland ridge directly west of Wadi el-Simad east of Ras Umm el-Rakham. About all that can be made of these essentia lly contextually empty discoveries is that, along with the Area IX sherds, they corroborate the passage of pastoral Libyans along the coastal area for over a distance of at least 30 km. in the area of Matruh.

Bates's Island Lastly, four Shell-Tempered B sherds were found on Bates's Island itself, but in surface 1.1 contexts that do nothing to clarify how and when they made their way to the island.-'0

94

MARSA MATRUH

Shell-Tempered Ware Linda Hulin The 1987 survey of the Marsa Matruh area yielded a class of pottery which, for want of any close parallels, I have called Shell-Tempered Ware. 41 The ware is divided into three classes dependent upon variations in fabric type (A- C). Despite this being a final report, the information presented here must be regarded as preliminary. Shell-Tempered Wares are defined by the overwhelming presence of crushed marine shell in their fabric and are thus indisputably local; a buff sandy deposit saturated with such matter occurs throughout the area and was observed by expedition members 42 at Wadi Aghiba (ca. 25 km . to the west of Marsa Matruh) close to clay beds. The ware is sandy and divided into three classes, mainly on the basis of fabric color. Indeed, the boundary between Fabrics A and B may be more apparent than real if, as is possible, the difference is due to variable firing conditions. Shell-Tempered A is a hard, reddish yellow fabric (7.5YR 7/6); the more common ShellTempered B is also hard, but a light brownish gray (I OYR 6/2); Fabric C is a I ight gray or brown ; the core crumbles easily, and the shell has been ground finely, resembling fibers in asbestos sheeting.

of Fabric B. Catalog number 5.54, a sadly unattached tab handle, is unique to the collection. Fabrics A and B derive from hanJmade, apparently round-bellied jars. The rim is plain (5.39), or squared off (5.20), the neck flares (5.61), sometimes extremely (5.41), and the base is flat (5.33). Decoration is incised and, according to White, it is not unlike what occurs on much earlier wares reported farther south and east as far as the Cyrenaica_4 3 It is largely confined to the neck and upper body; occasionally, some motifs do extend onto the lower body (5.36). Number 5.34 is unusual in having a visible bulge under the rim. Four decorative schemes occur: horizontal lines below the rim , either singly (5.23) or in pairs (5.35); short vertical strokes around the neck between horizontal lines (5.48); elongated crosshatching around the upper body (5.20, 5.46), occasionally meeting in points at the neck (5.61); a mixture of the above (5.30). Number 5.68 from Area IX, and 5.72, form a distinct repertoire of their own. Number 5.68 has an undifferentiated holemouth rim. Fabric C is represented by a handmade roundbellied jar with flaring neck; the rim is plain (5.37).

Shape and Coloration Distribution of Shell-Tempered Wares Shell-Tempered Fabrics A and B were found in surface concentrations of small fragments only (see below); consequently, no whole or profile shapes were forthcoming . The shapes offered here are reconstructed from Fabric B, but there were no observable differences between the rim sherds of A and B; the result must be treated with caution as an educated guess, although catalog number 5.8 appears to have been more sharply curved at the shoulder than most examples. Fabric C only derives from an excavated context; the sherds were larger, and reconstruction of shape was more possible; however, the shape does not seem to be markedly different from that

Fabrics A, B, and C were found: In surface scatters along the top of the Great Ridge (Area VIA) and VfB, 2.1 (Fig. 5:3). East of Bates's Island on low-lying hills rising off the lagoon basins in Areas liB, VI fA and B (Pl. 38A), Vlll and IX (Pl. 39A). Area I (Bates's Island): E4-IIIE 1. 1; F4-fll 1.1; H5-l 1.1 ; 16-I/II 1.1. Umm el-Rakham, Wadi ei-Simad, and Siwa!Aghurmi lie outside the immediate study area, and

AREA VI

the sherds from these areas are merely chance finds on the part of expedition members. Excluding the Great Ridge (Areas VIA and VTB) and, obviously, Area I, the factor unifying the finds in the Marsa Matruh area is their relations hip to Bates 's Island. All were situated on the lee side of hillocks or slopes that were visible from the island.

Date Shell-Tempered Fabric C was found in Area VIB in the context of a hearth in association with fragments of cloth , rusted metal , and a square g lass bottle. 44 The g lass bottle is typical of those dating to the last quarter of the 19th century (5.72, from Siwa, is also modern , i.e. , 20th century, and is included for purposes of comparison only). Fabrics A and B are older, dating at least to the late Roman period, because they were frequently, but not invariably, found in assoc iation with such sherds during surface survey. Unfortunately, they did not appear in any excavated context, Bronze Age or Roman , in Areas I or V 45 An examination of the distribution of the number of ware types from recent limited survey work to the west of Zawiyet Umm el-Rakham tends to support the contention that the fabrics are later (i.e. , GrecoRoman) rather than earlier."6 Shell-Tempered fabrics appear but rarely in publications of North African sites of the Hellenistic or Roman period, and they are always characterized as local. A copy in Fabric A of a standard Roman s hape, the unguentarium, is now in the Royal Ontario Museum;47 it is dated to the I st century A. D. and listed somewhat uncertainly under Fayoum Ware or related . Fabric B corresponds to local fabric 1 and 2 at Benghazi and occurs throughout the Hellenistic and Roman levels. It also occurred in the Archaic levels (Deposits I- III) at Tocra in the neighboring Libyan Pentapolis region .' 8 More recent work in the vicinity of Umm el-Rakham and south of Marsa Matruh indicates a date so mewh e re around the 7th century A. D., although that data has yet to be fully analyzed. 49

95

The examples of shell-tempered pottery are organized according to provenience. 5.14 Jar Fig. 5:4 Not inv. Area fiB, Surface. Pres. h. 4.3; pres. w. 2. 1; wall th. 0.7; d. of rim ca. 13.5.

Rim fragment from Fabric B jar. Very unevenly fired, fabric ranging from sandy to gray. Handmade. Inci sed decoration: hori zontal ladder pattern li ghtly inci sed just above shoulder, extending part way across the fragment. 1987 Pol/e1y, 11 7, fig. I b. 5.15 Jar Not inv. Area liB, Surface. Pres. h. 2.3 ; pres. w. 2.8; wall th . 0.7. Plain everted rim from Fabric A jar.

Not ill.

5.16 Jar Not inv. Area ll B, Surface. Pres. h. 2.2; pres. w. 1.8; wall th. 0.6. Rim fragment from Fabric A jar.

Not ill.

Not ill. 5.17 Open Vessel Not inv. Area liB, Surface. Pres. h. 2.1 ; pres. w. 1.7; wall th. 0.5. Body fragment from Fabric A vessel. Black slip on interior. 5.18 Closed Vessel Not ill. Not inv. Area liB, Surface. Pres. h. 2.0; pres. w. 0.7 ; wall th. 0.7. eck/shoulder fragment from Fabric B closed vesse l. Inci sed decoration: two left diagonal parallel lines. Not ill. 5.19 Jar Not inv. Area IlB, Surface. Pres. h. 3.5 ; pres. w. 1.8; wa ll th. 0.7 ; d. of rim ca. 13 .5.

Body fragment from Fabric B jar. Incised decoration: three parallel vertical lines descending from single hori zontal line. End of a diagonal line not connected to above scheme. Lightly incised. Same as 5.18? 5.20 Jar Not ill. Not inv. Area liB, Surface. Pres. h. 1.7; pres. w. 3.6; wall th . 0.6. Body fragment from Fabric B jar. Very red core. Inci sed decoration : traces of one ve rtical and one diagonal line (probably joined).

MARSA MATRUH

96

r·w.>· J ' ·... J r

5.21

5.14

5.23

. .

5.30

.

.

.

5

l

..· . ., . ...::.:; .. ·.... : -·::...· .. '

.

5.34

5.33 0 F

~ .·.

10 =j

Fig. 5:4. Shell- Tempered pottery.

5.21 Jar Fig. 5:4 Not inv. Area liB, Surface. Pres. h. 3.3; pres. w. 3.4; wall th. 0.6. Shoulder fragment from Fabric A jar. Slightly more angular at shoulder than most examples.

5.26 Jar Not ill. Not inv. Area VI, Surface. Pres. h. 3.4; pres. w. 3.6; wall th . 0.6; d. of rim ca. 28. Flaring plain rim fragment from Fabric B jar.

5.22 Open Vessel Not ill. Not inv. Area VI , Surface. Pres. h. 3.2; pres. w. 1.8; wall th. 0.5; d. of rim ca. 24. Flaring plain rim fragment from Fabric C open vessel. Unevenly fired.

5.27 Jar Not ill. Not inv. Area VI, Surface. Pres. h. 5.4; pres. w. 4.8; wall th. 0.5. Body fragment from Fabric A open vessel. Incised decoration: traces of two vertical parallel lines with end of a horizontal line below.

5.23 Jar Fig. 5:4 Not inv. Area VI, Surface. Pres. h. 3.4; pres. w. 3.6; wall th. 0.6; d. of rim ca. 14.5. Flaring plain rim from Fabric A jar. Unevenly fired. Incised decoration: horizontal line broken by vertical line below rim.

5.28 Jar Not ill. Not inv. Area VI, Surface. Pres. h. 2.5; pres. w. 1.8; wall th. 0.7. Body fragment from Fabric B jar. Inci sed decoration: two converging diagonal lines, one being made up of short horizontal strokes.

5.24 Jar Not ill. Not inv. Area VI, Surface. Pres. h. 2.3; pres. w. 2.4; wall th. 0.5; d. of rim ca. 2.7. Plain, slightly flaring rim/neck fragment from Fabric A jar. Incised decoration: horizontal line with trace of diagonal Iine running down from it at junction of rim and body. 5.25 Jar Not ill. Not inv. Area VI, Surface. Pres. h. 2.7; pres. w. 2.1; wall th. 0.7; d. ofrim ca. 21. Plain rim from Fabric A open vessel. Incised decoration: uneven horizontal line.

5.29 Jar Not ill. Not inv. Area VI, Surface. Pres. h. 2.3; pres. w. 2.9; wall th. 0.6. Body fragment from Fabric B jar. Incised decoration: three vertical flaring lines. 5.30 Jar Fig. 5:4 Not inv. Area VI, Surface. Pres. h. 1.5 ; pres. w. 2.1; wall th. 0.6. Body fragment from Fabric B jar. Incised decoration: hori zontal row of short vertical lines with two diagonal lines below. 1987 Pollery, 117, fig. lg.

97

AREA VI

p . · · . ~·i( 5.35

5.36

5.38

5.37

5.39

(. -.. -----------r----------------) -..~

\

l

I 1

~' - : .: 5.40

I I

~.:~;:/~t,}~ 5.46

:- . ~ --------- ---- ---- ------,---- ------------- ---- --;

:

\~, ....... \

1:

'

'

,;A;' ,

,.~"

I

:

5.48 10

0

f-Fig. 5:5. Shell-Tempered pottery.

-

=j

, I

98

MARSA MATRUH

5.31 Jar Not ill. Not inv. Area VI, Surface. Pres. h. 2.0; pres. w. 1.6; wall th. 0.6. Body fragment from Fabric B jar. Incised decoration: traces of two parallel lines. 5.32

Open Vessel Not ill. Not inv. Area VI, Surface. Pres. h. 2.7 ; pres. w. 1.6; wall th. 0.6. Neck and body fragment from Fabric B open vessel. Incised decoration: two parallel horizontal lines. Below, diagonal lines of short vertical strokes . 5.33 Jar Fig . 5:4 Not inv. Area Vl, Surface. D. of base 7.0. Base fragment from Fabric B jar. Flat base with flaring wall. 1987 Pottery, 116, fig . I c. 5.34 Jar Fig. 5:4 Not inv. Area VIB, 2.1. Pres. h. 4 . 1; pres. w. 3.9; wall th. 0.5 ; d. of rim ca . 18. Rim fragment from Fabric B jar. Plastic decoration: line just below rim. 5.35 Jar Fig. 5:5 Not inv. Area VIB, 2.1. Pres. h. 4.8 ; pres. w. 4 .2; wall th. 0.5. Rim , neck and shoulder fragment from Fabric A jar. Incised decoration: two horizontal lines a long shoulder. 5.36 Jar Fig. 5:5 Not inv. Area VJB, 2.1. D. of rim 16. Rim and upper body fragment from Fabric C jar. Plain flaring rim , rounded body. Incised decoration: parallel zigzag lines, one with a sing le hori zo ntal line joining two apexes to make a triangle. 1987 Pottery, 118, fig. 2b. 5.37

Jar Fig. 5:5 Not inv. Area VIB, 2.1. Pres. h. 3.5; pres. w. 2 .8; wall th. 0 .9. Incised body fragment from large Fabric B jar. 1987 Pottery, 117, fig. I d.

5.38 Jar Fig. 5:5 Not inv. Area llB, Surface. Plain, slightly rounded, flared nm fragment from Fabric A jar. Incised decoration below rim: row of diagonal lines, now broken, but which would have extended down body. 1987 Pottery, 117, fig. Ia.

5.39 Jar Fig. 5:5 Not inv. Area liB, Surface. Pres. h. 2.5; pres. w. 3.5; wall th. 0.9; d. ofrim ca. 17.0. Flared rim fragment from Fabric A jar. Traces of incised decoration left diagonal line at junction of neck. 5.40 Jar Fig. 5:5 Not inv. Area IIB, Surface. Pres. h. 3; pres. w. 3.6; wall th. 0.6; d. of rim ca. 14. Plain flared rim fabric from Fabric A jar. Fabric is a little coarser, with a streaky reddish tinge and occasiona l larger red grits. Remains of black slip on interior (exterior too eroded to judge). 5.41

Jar Not ill. Not inv. Area VII, Surface. Pres. h. 4.0; pres. w. 3.8; wall th. 0.8 ; d. of rim ca. 21. Plain flaring rim fragment from Fabric A jar. Inc ised decoration at junction of neck and body: two flaring lines.

5.42 Jar Not ill. Not inv. Area Vll, Surface. Pres. h. 4.8 ; pres. w. 0.3; wall th . 1.2; d. ofrim ca. 21. Plain flared rim fragment from Fabric A jar. Incised decoration at junction of rim and body : flaring lines. 5.43

Jar Not ill. Not inv. Area VII, Surface. Pres. h. 5.5; pres. w. 3.8; wall th. 0.8. Neck fragment from Fabric A jar. Traces of black slip on interior; exterior very eroded. Incised decoration at junction of rim and body: flaring lines.

5.44

Jar Not ill. Not inv. Area VII, Surface. Pres. h. 4.8; pres. w. 3.3; wall th. 0.8. Neck fragment from Fabric A jar. Incised decOJ·ation: one deep horizontal line between single rows of short vertica l lines; vertica l lines do not extend all the way ac ross the fragment.

5.45 Jar Not ill. Not inv. Area VTI, Surface. Pres. h. 2.0 ; pres. w. 2.6; wall th. 0.9. Neck and body fragment from Fabric A jar. Incised decoration: parallel z igzag lines. 5.46

Jar Fig. 5:5 Not inv. Area VII, Surface. Pres. h. 1.4; pres. w. 2.6; wall th. 0. 7. Neck and body fragment from Fabric A jar. Incised

99

A REA VI

"-,:~~.: - - -

---------------__1_____ ----------------

iiJ:';X' i ,,,,,

(-~~ ~-

0

2m

-------'-""W G6- l, South Balk

Fig 6:5. E- W section, G6-l, South Balle

(fallen from nearby S101 's E wall?); light brown in color and mixed with a few pebbles. 2.1. Circa 0.30 m. Continuation of the same sand fill as 1.1, only slightly darker in color, burying a random spill of rubble-stones. Veg. taxa and pol. (Vol. II , 11.24, 11.55) . 3.1. Circa 0.50 m. thick deposit of slightly moist yellow-orange sand and pebbles, covered by the 2.1. topple of stones. The layer buries as well as supports 8 of the trench 's n-shaped bins. Veg. taxa (Vol. II, 11.25). 3.2. Isolated fire-pit filled with a dark, slightly greasy sand fill that contains ash , in NE corner of trench (not associated with a bin). Sherds and stones show signs of burning. Veg. taxa and pol. (Vol. ll, 11.26, 11.56). 4.1. Earthy, grayish brown sand mixed with many small stones and pebbles; deposit dug to only ca. 0.10 m. depth before trench was closed. Supports a single bin (S 113h). Veg. taxa and pol. (Vol. II, 11.27, 11.57) .

Contents of G6-I's Deposits 1. 1. Miscellaneous wares, including small number of Late Bronze Age and uncat. Roman corrugated. '11 Iron. Lead weight (Vol. II, 13.44). 9 fragments of brick. 3 pieces of modern glass. Uncat. chert, poss ibly a core. Bird and animal bones. 2.1. Miscellaneous sherds, including Late Bronze Age WS sherd and uncat. Roman corrugated.21 Post-antique sherds, including 2-handled glazed dish. 3 pieces of bronze slag. Basalt pendant (Vol. II, 13.64). Numerous marine invertebrates. Many bones, including bird, animal, and fish bones. Post-antique deposit. 3.1. Late Bronze Age sherds, including WS , BR, and RL. Numerous miscellaneous postBronze Age sherds, including uncat. black glaze sherd and numerous uncat. Roman corrugated. 22 Fragmentary bronze sheet (Vol. II, 13.25), chisel? (Vol. If, 13.21), twisted handle (Vol. II, 13.23), hook (Vol. II, 13.22); 2 bronze nodules. Iron point (Vol. II, 13.33) and nail (Vol. II, 13.31 ). Marine invertebrates. Land snai Is. Bird,

THE ISLAND 'S HISTORICAL PERIOD OCCUPATION

fish, and animal bones including dog. Imperial Roman deposit. 3.2. I rim and 4 body sherds, unidentifiable. 7 shells. Bird and animal bones. Undatable. 4.1. Small number of Late Bronze Age sherds, including 3 WS. 3 uncat. glazed Hellenistic sherds. Uncat. Roman ridged rim sherd, Roman corrugated. 23 Stone core. Bronze slag. Marine invertebrates. Bird, fish, and animal bones. Imperial Roman layer.

n-Shaped Enclosures Given Bates 's Island's history of metal-working already discussed in connection with its Late Bronze Age phase/ 4 one might have expected S I 13a- h to have served a related purpose except for their lack of any traces of burning. It, furthermore, seems highly unlikely that the island was used for processing metals as late as the Roman period. Shaw has listed enclosures of similar size and design on various Minoan sites where they seem to have functioned as storage bins. 25 shaped storage bins have been reported at the LM IIIC settlement of the Vronda ridge above the village of Kavousi on eastern Crete, where they are built of large slabs set on their sides against the walls of rooms .26 A similar bin type, dated to the second half of the 8th century B.C., has been found at the site of Kastro. 27 In the earlier contexts , however, bins never appear in multiple clusters set out unsupported in the middle of rooms. By contrast, the island's G6-l examples seem to have been placed in the open. As sparse as the evidence is, I am more inclined to believe that they worked as outdoor cages for cooping up small edible animals (rabbits?) or fowl rather than storage jars bases or supports, but even this explanation has its problems. 28

n-

Underwater Wall S139 In H9, south of the isl and's sandbar, lies a short stretch of three underwater ashlars (wall

11 3

S 139). Nothing was recovered to indicate date, but when continued on paper its line forms a right angle with S 104. This perhaps opens the possibility that the S I 05 room, wall S I 04, and wall S 139 were all part of a late development of the island at a time when the lagoon's water stood at a lower level than it does now.

Underwater Wall S111 (Pis. 9A and 43A) A similar 2.5 m. long underwater line of rough ashlars was found at ca. 1.30 m. below sea level ca. I 0 m. due east of the northeast tip of the island in grid square K I 0. During an uncharacteristically calm day in 1985, sections of at least four blocks were traceable on the soft sandy bottom of the lagoon. No evidence for their period was recovered, but their shapes are compatible with a Greco-Roman date. It is also not possible to determine whether they belong to a pier or docking facility or were in stead part of the dryland extension of the island. However, as previously argued, the presence of sherds on the lagoon floor around the wall would perhaps seem to point to S I II being a land feature.

Underwater Slot S140 (Pl. 43B) A third underwater feature was also visible on calm days ca. 2.0 m. northwest of the island's northern tip. It is a rectangular slot cut into the submerged natural sandstone shelf that marks the outer perimeter of the island-proper from the lagoon bottom. The slot measures 0.49 by 0.27 by 0.04 m. There are no clues as to its purpose.

Underwater Sherds Finally, Donald Bailey has reported eleven sherds recovered from shallow water off the eastern tip of the island: two transport amphorae (Vol. II , 12.39, 12.46), an Augustan period sigillata bowl (Vol. II, 12.54), and eight coarse ware cooking pots (Vol. II, 12.61, 12.64-12.70) . None

MARSA MATRUH

114

J8 -l/ll

o

I

:

.~ ()

~

:

!

,CJ:

[JV!l

I I I

n

QIJo·D

I I

()

L______________________ J

0

I

2

4m

Fig. 6:6. Plan of J8-IIII.

of these have direct relevance for the island's usage other than to suggest that the water level stood lower in the Greek and Roman period than it does now, and that the sandbar, in whose proximity the sherds were collected, was above water.

.1.1 ;2.2'' 32 ' undug

0

J8-IIII (Figs. 6:6, 6:7; Pis. 44A-C) This 4.0 by 4.0 m. test was opened on the lower slope of the island's northern tip just before the sand surface gives way to rocky beach. A 2.0 by 1.25 m. extension was added to its northeast corner. Before excavation, two parallel lines of rough, unpainted wall plaster stood above the ground's surface. These turned out to be the vertical surfaces of a narrow plastered ashlar wall, S I 09a, that runs from the northern balk ca. 1.75 m. in a south-southeasterly direction before petering out. The only other features of the trench are the moth-eaten, patchy remains of floors , S I 09b and c, to either side of S I 09a. After excavating its 1.1 surface, the southern three-quarters of J8-J / II turned out to be sterile, and all subsequent testing was restricted to the trench's northern quarter in order to expose S I 09a's foundations. The following deposit

,,,,. '''·'"''"''

2

3m

W""--'--------

Fig. 6:7. E- W section, J8-I/II, north Balk.

sequence was recorded from a 1.0 m. wide, 0.40 m. deep sondage across the trench's northern end and its extension.

J8-IIJI Deposit Sequence (Figs. 6:6, 6: 7) 1.1. Circa 0. 10 m. of densely packed sand mixed with occasional small stones; unusually clear of sherds and other artifacts on its surface. 2.1. A thin (ca. 0.08 m. thick) accumulation of densely packed orange-tan sand W of wall Sl09a.

THE ISLA D' S HISTORI CAL PERIOD OCCUPATIO;..!

2.2. Continuation of fill similar to 2.1, E of Sl 09a. 2.3. A small patch of hard-packed floor adhering to the western face of S 109a, but not appearing in the Fig. 6:7 section. Less than 0.05 m. thick, it consists of a mixture of sand, lumps of hard plaster, and small stones. Labeled S 109b in the trench plan, Fig. 6:6. 2.4. A thin but very hard floor surviving in patches E of S 109a, but it does not appear in the Fig. 6:7 section. The floor consists of a mixture of sand, lumps of hard plaster, and small stones. Labeled Sl09c in the trench plan, Fig. 6:6. 3.1. Circa 0.12/0.15 m. thick orange-tan fill , similar in texture and color to 2.1, packed against the large ashlar set at roughly right-angles to the W face of S109a. 3.2. A dark gray sand fill, ca. 0.40 m. deep, against the E face of S I 09a's foundation and lowest course. Represents an occupation level.

Contents of J8-I/II's Deposits 1.1. One Late Bronze Age WS , two late 6th- 5th c. B.C. (Vol. 11 , 12.11, 12.91), and two Hellenistic (Vol. II, 12.26, 12.29) sherds; ca. 320 unidentified sherds. Marble revetment fragment (Vol. II, 13.44). 2 uncat. pieces of bronze. 2 uncat. iron nails and iron lump. 3 fragments of bottle glass. Numerous pieces of plaster. Animal bones. 2.1.

Circa 20 unidentified sherds. Plaster lump .

2.4.

115

Sterile.

3.1. Stray Late Bronze Age Cypriot sherd. Ptolemaic or earlier jug (Vol. II, 12.27). Numerous unidentified sherds. 3.2. Unidentified sherds, including black painted sherd. Fish and animal bones.

Architectural Remains in J8-I/II The only architecture is wall S 109a and its two associated floor levels, S I 09a and S J 09b (otherwise identified as deposits 2.3 and 2.4) that lie at a height of ca. 1.65 m. a.s.l. Wall SJ09, which measures only 0.28 m. in width, consists of three badly weathered, but otherwise relatively welltrimmed, rectangular blocks that rest on a projecting foundation footing. Their height is ca. 0.40 m. ; the foundation (Fig. 6:6) is ca. 0.20 m. high. The extreme thinness of S I 09a, coupled with the plastering coating both of its faces, indicates that it is an interior wall dividing two rooms. lts date seems to be 4th century B.c. or Ptolemaic, depending on the date of the coastal jug (Vol. II , 12.27) in 3.1 as well as the Classical period sherds in 2.3 (Vol. !I, 12.33, 12.47). Plastered walls, occasionally painted, seems to be a hallmark of construction in this sector of the island to judge from all the plaster fragments found here and in association with Room S J 05 in I8-III/S.29 In addition, a small probe conducted by R. Gardner for geological data in 1989 brought to light more decoratively painted plaster fragments (Vol. II, 13.83-13.89) north of the neck of the island 's sandbar in grid square 19. In I9, the latest pottery is Roman Imperial , dating to the 2nd century A.D . (Vol. II, 12.55, 12.56, 12.82) .

2.2. Attic cup, ca. 500 B.C. (Vol. 11, 12.12); Attic salt-cellar (Vol. II, 12.19). Ca. J 30 unidentified sherds. Bronze fragment. Bird, fish, and animal bones.

Miscellaneous Later Walls and Other Occupational Features

2.3. 5th c. B.C. coastal jar stopper (Vol. 11 , 12.33) and 5th, early 4th c. B.C. coastal amphora or jug (Vol. IT, 12.47) . This dates the floor attached to the western face of wall S 109a.

16-1/11. While clear of actual walled remains, the trench , which was excavated ca. 14.5 m. northeast of the northeastern corner of the Sponge-Divers House (SlOl) at ca. 2.75- 3.50 m. a.s.l., brought to

MARSA MATRUH

116

Iight two stone hearths (S I 16, S 1 17) and two stone storage bins (Sll4, Sll5) associated with its 2.2 and 2.4 Hellenistic levels. 30 The trench also contained, in what is probably a Hellenistic level, 3.1, a thin period pavement or floor (S 135) made up of clay mixed with white plaster, black ash, and disintegrated osseous material. From his 1989 survey of the island's surface

pottery, D. Bailey has reported a significant number of I st to 7th century A .D. sherds (Vol. II, 12.81, 12. 83, 12.89, 12.93) from the sheltered, lower (i.e., below the 2.0 m . contour line to sea level) eastern side of the island. These clearly indicate the on-going use of the island in the Roman Imperial and Byzantine period.

The Ottoman Sponge-Divers Occupation Period HS-III/W2 S 129 (Plan I) is a rough pavement of rubble stones scattered in front of the northeastern corner of the Sponge-Divers House (S I 0 I); associated with the 2.1 deposit recorded above, 1 1 it may represent an Arab period feature but cannot be dated with any certainty.

Sponge-Divers House (SlOt) (Figs. 6:8, 6:9; Pis. 45, 46, 47 A) The largest set of walled remains, S I 0 I, formed the focus of Oric Bates 's work on the island when it was still called the " Island of the Jew" or Gezira-t-el- Yahudy. ~ As already pointed out/3 its association with foreigners was not his idea but rather ours, based on information supplied by the former Chief Inspector for Matruh, Ess. Feisal As-Mahwy, that the island had served in the past as a base of operations for Greek sponge-divers. 34 Since the house's construction dates back a respectable number of years to the late 17th century to judge from its pottery' 5 and "Turkish" pipe (Vol. II, 13.81), it is not clear that it was erected by sponge-divers or even used by them before or after Bates 's arrival. Bates himself reports nothing to indicate that he disrupted its occupation, and the Pennsylvania expedition found no traces of recent, i.e. , post-World War I, occupation apart from the odd fragment of broken glass. Even assuming that sponge-divers made occasional use of its sheltering walls, their precise ethnic identity is in doubt, since Mr. As1

Mawy spoke of Greeks, but its interior produced a bronze religious medal bearing the inscription "SALVATOR MUND" (Vol. II , 13.1); perhaps it housed Italian or Maltese fishermen from time to time as well as Greeks. 36 In any case, "SpongeDivers House" is simply a convenient tag, and its significance should not be pressed too far. Anomalies abound between what Bates could report and draw in 1913/191437 and what was left for the Pennsylvania expedition to observe in the mid-1980s, but they have scarcely more than a vague antiquarian interest today. According to Bates, the building was a rectangle with two interior rooms and a centrally located door with exterior steps at its north end. While admittedly our understanding has been severely impaired by what looks to have been the explosion from a World War II bomb or shell that inconveniently blew out the center of the building (Pl. 5A), 1H S I 0 I today more closely resembles a lop-sided square; its north end preserves no trace of either steps or a door, and there exists little indication of a second room. The less sketchily executed of Bates's two published house plans (Fig. 6:8) plots an overall north-south length of ca. 13.5 m. ,39 which is not too far from its actual length of 12.9 m. The explosion probably took out Bates's interior divisional wall which created a ca. 4.5 m. long room across the structure's southern end as well as the door to the north, but it does not explain why he drew his structure as a rectangle measuring only 6.0 m. east to west.~ Bates reports that the larger room to the north was floored at its door sill level, and that there was a low-studded cellar set on the bedrock beneath its floor. What exactly constituted 0

117

THE ISLAND ' S HISTORICAL PERIOD OCCUPATION

-~

'\.;