The Roman Port and Fishery of Cosa. A Center of Ancient Trade 0691035814, 9780691609980, 9780691628837

Citation preview

THE ROMAN PORT AND FISHERY OF COSA

CONTRIBUTORS J. Lawrence Angel, Isaac Barshad, JeUe de Boer, Lionel Casson, Howard Comfort, Domenico Cozzupoli, Charles R. Crumly, David Frederick Grose, Brian Hesse, John David Lewis, Richard E. Linington, Luc Long, William Old, Martha I. Oleson, Raffaello Trigila, Brian Viani, and Paula Wapnish ARCHITECTURAL DRAWINGS

J. F. Warren and Elaine K. Gazda, Elizabeth F. Solomon, John Stubbs, and David van Zanten PHOTOGRAPHY

Anna Marguerite McCann, Barbara Bini, and Julian Whittlesey

The Roman Port and Fishery of Cosa A Center of Ancient Trade BY A N N A M A R G U E R I T E M C C A N N A N D J O A N N E B O U R G E O I S , ELAINE K. G A Z D A , J O H N PETER O L E S O N , AND ELIZABETH LYDING WILL

Princeton University Press, Princeton, New Jersey

Princeton Legacy Library edition 2017 Paperback ISBN: 978-0-691-60998-0 Hardcover ISBN: 978-0-691-62883-7

Copyright © 1987 by Princeton University Press Published by Princeton University Press, 41 William Street, Princeton, New Jersey 08540 In the United Kingdom: Princeton University Press, Guildford, Surrey All Rights Reserved Library of Congress Cataloging in Publication Data will be found on the last printed page of this book ISBN 0-691-03581-4 Publication of this book has been aided by grants from the American Academy in Rome; the Horace H. Rackham School of Graduate Studies, the University of Michigan; and the National Endowment for the Humanities This book has been composed in Linotron Bembo Designed by Jan Lilly Clothbound editions of Princeton University Press books are printed on acid-free paper, and binding materials are chosen for strength and durability Printed in the United States of America by Princeton University Press Princeton, New Jersey

To the ancient sailors and fishermen of the Portus Cosanus in memoriam

Contents

List of Illustrations

xi

List of Tables

xix

Frequently Cited References and Abbreviations

xxi

Preface

xxxi

Introduction: A Perspective Anna Marguerite McCann

3

PART I. THE HISTORY, GEOGRAPHY, AND ARCHITECTURAL REMAINS

Chapter I. The History and Topography Anna Marguerite McCann

15

Chapter II. Geography and Geology Joanne Bourgeois

44

Chapter III. Modern Scholarship and the Land and Underwater Excavations Anna Marguerite McCann

58

Appendix: The Tools and Techniques of the Underwater Excavations John David Lewis

72

Chapter IV. The Port and Fishery: Description of the Extant Remains and Sequence of Construction Elaine K. Gazda

74

Chapter V. The Spring House Complex John Peter Oleson

98

Chapter VI. The Lagoon Temple Anna Marguerite McCann

129

Chapter VII. Reconstruction and Function: Port, Fishery, and Villa Elaine K. Gazda and Anna Marguerite McCann

137

Chapter VIII. Ancient Shipping in the Portus Cosanus Lionel Casson

160

Appendix 1. The Grand Congloue Site: A Reassessment LMi Long

164

Appendix 2. The Ship Remains in the Portus Cosanus John David Lewis

166

PART II. EVIDENCE FOR TRADE

Chapter IX. The Roman Amphoras Elizabeth Lyding Will

171

viii

CONTENTS PART III.

THE CATALOGUES

Chapter X. Fine Wares

223

Roman Thin Walled Pottery Anna Marguerite McCann

223

Black Glaze Pottery Anna Marguerite McCann

233

Red-Gloss Pottery Howard Comfort and Anna Marguerite McCann

235

Chapter XI. Coarse Wares

243

Utilitarian Ware Martha I. Oleson and John Peter Oleson

243

Miscellaneous Ceramics Anna Marguerite McCann and Elizabeth Lyding Will

265

Chapter XH. Glass and Minor Objects Glass David Frederick Grose

266 266

Lamps Anna Marguerite McCann

270

Coins Anna Marguerite McCann

272>

Chapter XIII. Varia John Peter Oleson PART IV.

274

THE SCIENTIFIC REPORTS

Chapter XIV. Archaeological Prospecting in the Ancient Lagoon Richard E. Linington Chapter XV. Analyses of Cores and Sediment Samples from the Ancient Lagoon Isaac Barshad and Brian Viani Chapter XVI. Mineralogical and Petrographic Reports

285 297 309

The Provenance of Some Amphora and Tile Samples Domenico Cozzupoli and Raffaello Trigila

309

A Comparison of Some Amphora and Sand Samples Jelle de Boer

312

The Provenance of Some Tuff Samples Raffaello Trigila

313

Chapter XVII. Human and Faunal Remains

315

Human Bones J. Lawrence Angel

315

Faunal Remains Brian Hesse and Paula Wapnish

315

Land Tortoise Shell Remains Charles R. Crumly

317

Marine Snail Shell Remains William Old

318

CONTENTS PART V.

ix

CHRONOLOGY AND CONCLUSIONS

Chapter XVIII. The Chronology of the Port and Fishery Anna Marguerite McCann

321

Chapter XIX. Conclusions Anna Marguerite McCann

335

Index

343

Illustrations

List of Illustrations

T E X T FIGURES

Text Fig. 1-1. Reconstruction of the Portus Cosanus and its fishing lagoon, late second and first centuries B.C. Text Fig. II—1. Index map for cross-sections illustrated in Text Figs. II-2 to II-4. Text Fig. II-2. Generalized west-east cross-sections A-A'. Text Fig. II-3. Generalized west-east cross-section B-B'. Text Fig. II-4. Generalized north-south crosssection C - C Text Fig. H-5. Section Sl, Tagliata Canal, looking north. Text Fig. VI-I. Plan of lagoon temple, phase 1 and phase 2. Text Fig. VI-2. Elevation of lagoon temple, phase 1 and phase 2. Text Fig. VII-I. Plan of Roman fishery and saltery atjavea, Spain. Text Fig. VII-2. Plan of fish tank and concrete protective piers in sea off the Roman villa at Santa Liberata. Text Fig. VII-3. Plan of trapping area (1905) in channel between lagoon and sea at Nassa fishery, Orbetello. Text Fig. VII-4. Schematic plan of fish barrages in a terminal canal. Text Fig. IX-I. Distribution of verified Sestius stamps as of 1981. Text Fig. XIV-I. Key to symbols used in the drillhole profiles from the Cosa lagoon area, Text Figs. XIV-2 to XIV-8. Text Fig. XIV-2. Profile of drill holes 1 to 9, with corresponding magnetic anomaly. Text Fig. XIV-3. Profile of drill holes 10 to 16 and 37 to 42, with corresponding magnetic anomaly. Text Fig. XIV-4. Profile of drill holes 28 to 31, with corresponding magnetic anomaly. Text Fig. XIV-5. Profile of drill holes 17 to 24, 37, and 43, with corresponding magnetic anomaly. Text Fig. XIV-6. Profiles of drill holes 25 to 27 and 32 to 36. Text Fig. XIV-7. Profiles of drill holes 23, 26, and 44 to 47. Text Fig. XIV-8. Profiles of drill holes 48 to 50, which form an east-west profile about 58 m south of the Spring House. Text Fig. XV-I. Cross-section showing relative position, elevation from sea level, and

14 52 52 52

thickness of samples within cores from the Cosa lagoon area. Text Fig. XVI-I. Ti, Cr, Zr triangle. Text Fig. XIX-I. Plan of the ancient harbors of Piraeus (Athens): Munichia, Zea, Kantharos. Text Fig. XIX-2. Plan of ancient port of Alexandria. Text Fig. XIX-3. Plan of ancient port of Antium (Anzio). Text Fig. XIX-4. Schematic restoration of Roman harbor area, Kenchreai, Greece.

299 311 336 336 339 339

52 M A P S (following p . 354)

56 130 130 143 152 154 154 175 288 288 289

Map 1. Map of Italy, locating Cosa. Map 2. Map of the Ager Cosanus showing ancient topography and limits of ancient colony. Map 3. Plan of fortified hill town of Cosa with the Portus Cosanus and adjacent lagoon of Cosa. Map 4. The Portus Cosanus, plan of ancient remains with dates of excavation. Map 5. Lagoon of Cosa, plan of excavated remains with dates of excavation. Map 6. The Portus Cosanus, plan of ancient remains. See Fig. IV-I for key to locations of detailed plans and sections. Map 7. Lagoon of Cosa, plan of excavated remains. Map 8. Lagoon of Cosa, detail of Map 7. Map 9. Lagoon of Cosa with contour diagram of the magnetic and drilling survey results of the Lerici Foundation, Rome. Map 10. Lagoon of Cosa, key for locations of detailed maps of individual areas and trenches. Map 11. Plan, lagoon area IK, with Walls P and X. Map 12. Plan, lagoon area YZ. Map 13. Plan, lagoon, north end: water channel with Walls U, V, and X (Trench IKE). Map 14. Plan of Spring House complex in lagoon of Cosa.

289 C O L O R FIGURES (following Maps)

290 294 294 295

Color Fig. 1. Harbor of Cosa with ancient concrete Piers 1, 2, and 3. Tagliata channel and ancient limestone quarry cut into cliffs in background. Color Fig. 2. Excavation in north end of silted lagoon of Cosa, 1972, showing typical terra rossa, and black, organic-rich lagoonal muds with limestone, polygonal, embankment Wall P. Color Fig. 3. Detail of Sestius stamp, SES with trident, Type 4a wine amphora, Cat. A60.

XIl

LIST OF ILLUSTRATIONS

Color Fig. 4. Fragment of neck of garum(?) amphora of Type 5, from the Agora Excavations, Athens (P6867). Color Fig. 5. Thin walled goblet, Cat. TWlO. Color Fig. 6. Decorated carinated bowl, Late Italian sigillata, Cat. RG12.

FIGURES (following C o l o r Figures) Fig. 1-1. Promontory of Cosa/Ansedonia with silted ancient lagoon in foreground, looking southwest. Fig. 1-2. Aerial view of Cosa, lagoons of Orbetello, and Port'Ercole, May 1944. Fig. 1-3. Aerial view of promontory of Cosa with fortified town, port, and silted lagoon cut by modern drainage canal, looking northeast. Fig. 1-4. Aerial view of limestone promontory of Cosa with Spacco della Regina and Tagliata channels. Fig. 1-5. Aerial view of western portion of silted Cosa lagoon, looking north, August 1972. Fig. 1-6. Detail of polygonal fortification walls of Orbetello. Fig. 1-7. Detail of northeast gate of Cosa fortifications. Fig. 1-8. Detail of Tabula Peutingeriana, ca. A.D. 500. Fig. 1-9. Monte Argentario with Port'Ercole, looking northwest with section of the Tombolo di Feniglia. Fig. 1-10. Temple D, arx of Cosa. Fig. I—11. Drawing of frescoes of thirteenth century, Abbey of the Three Fountains, Rome. Fig. 1-12. Detail of fresco as seen today, Abbey of the Three Fountains, Rome. Middle zone: far left, Charlemagne on horseback, ships with warriors attacking castle of Ansedonia. Fig. 1-13. Half-litra bronze coin of Cosa, 273 B.C., obverse. Fig. 1-14. Reverse of Fig. 1-13. Fig. 1-15. Drawing of Fig. 1-14. Fig. II—1. Geologic map of vicinity of the Portus Cosanus, generalized from the Carta Geologica d'ltalia Foglio 135. Fig. II-2. Oblique aerial photograph of Portus Cosanus, looking north, with overlay, interpreting the underlying sediments and bedrock. Fig. II-3. Detail of map of Etruria by A. Danti (15801583), showing Ansedonia, the lagoon of Cosa and Orbetello, with inlet near western end of lagoons of Cosa. Fig. H-4. Detail of "Maritima Senesium" map by A. Danti (1580-1583), showing Cosa, Succosa, lagoons of Cosa and Orbetello, with inlet at eastern end of lagoon of Cosa. Fig. II-5. Detail of map of ancient Etruria by P. Oliver, 1624. Fig. II-6. Detail of map of Tuscany by G. lnghirami, Carta geometrica della Toscana (Florence, 1830), showing Cosa and vicinity. Fig. II-7. Aerial photograph of area of the Portus Cosanus, May 1944, looking north. Fig. II-8. Silted lagoon of Cosa from hill town wall, looking east, before excavation, 1969. Fig. II-9. Aerial view of lagoon and port of Cosa during excavations of 1972, looking south. Fig. 11-10. Map of lagoon of Cosa with key for locations

of individual trench sections, designated by arrows with figure numbers in chapter two. Fig. H-Il. Section, Trench IL, north wall. Fig. 11-12. Section, Trench IKE, looking north. Fig. 11-13. Theoretical section, western end of Trench IA, looking north. Fig. 11-14- Section, Trench U l , looking north. Fig. 11-15. Section, Trench U5, looking north. Fig. 11-16. Sections, area YZ, looking north: (a) western portion, (b) central portion, (c) Trench 23. Fig. 11-17. Section, Trench IB, north end. Fig. 11-18. Section, Trench IB, south end, reconstruction. Fig. 11-19. Section, north wall of Spring House Trench IC. Fig. HI-I. Section, Beach Probe No. VI, looking northnorthwest. Fig. ΙΠ-2. Nineteenth-century stone marker from beach of Cosa delimiting fishing boundaries between the sea and the lagoon. Fig. IH-3. View of the harbor of Cosa, looking west with concrete piers, Tagliata, quarry, and entrance to Spacco della Regina in northwest corner of quarry. Fig. III-4. Detail of limestone cliffs on western end of harbor of Cosa with open, seaward, southern section of Tagliata. Fig. III-5. Tagliata tunnel and northern channel, looking south. Fig. III-6. Modern drainage canal leading from the quarry north to the lagoon, following course of ancient Tagliata Canal. Fig. III-7. Cave in Spacco della Regina with sky-light cut by man. Fig. ΙΠ-8. Man-made sky-light in northeastern end of Spacco della Regina. Fig. III-9. Plan of Tagliata and Spacco della Regina by R. Del Rosso, 1905. Fig. IH-10. Plan and elevation of the Tagliata by R. Del Rosso, 1905. Fig. IH-Il. Marble statuette of Ceres from Tagliata. Fig. 111-12. Concrete Piers 2, 3, and 4 in harbor of Cosa. Fig. IH-13. Pier 1, western side, showing impressions for planking of ancient wooden cofferdam revealed in excavations of 1968. Fig. Ill—14. Eastern tip of main breakwater harbor side, showing exposed boulders after airlifting. Fig. 111-15. Section, Breakwater Extension D, underwater Trench D l , west side, looking east. Fig. 111-16. Underwater Trench D2, east side of Breakwater Extension D. Fig. IH-17. Detail of Trench D2, with Roman pottery seen under the breakwater rocks; nozzle of airlift in foreground. Fig. Ill—18. Section, Breakwater Extension D, underwater Trench D2, east side, looking west. Fig. 111-19. Divers moving joined steel cylinders into place with inflated bags. Fig. 111-20. Divers working airlift within steel cylinder. Fig. Ill—21. Cylinder sinking into harbor floor as sand removed by airlift inside. Fig. 111-22. Section, underwater cylinder Trench C l . Fig. IH-23. Section Trench N in harbor area, southeast corner. Fig. IH-24. Tagliata Canal leading north into lagoon,

LIST OF ILLUSTRATIONS with ancient concrete wall below modern concrete wall, resting on bedrock. Fig. 111-25. View of lagoon of Cosa during 1968 excavations, looking southwest. Fig. ΙΠ-26. Elevation of Trench IH with Wall P and Renaissance cobblestone road to east, looking west. Fig. 111-27. Wall P exposed in area of Trench IH, before excavation. Fig. 111-28. Trench IH with Wall P resting upon bedrock with retaining wall to west. Fig. 111-29. Trench IH, Renaissance cobblestone road to east of Wall P. Fig. IH-30. Cistern with rubble Walls a and b of Spring House, before excavation in 1968. Fig. Ill—31. Trench IC of Spring House excavations of 1968, with southern wing of concrete platform surrounded by wooden formwork; pier 2 on top of platform; section of Wall P, lower right. Fig. 111-32. Detail of Spring House platform, eastern end of southern wing with wooden formwork, 1968. Fig. 111-33. Section, Trench Pl, looking west with Wall P. Fig. 111-34. Section, Trench P3 and Trench ID with Wall P, looking west. Fig. 111-35. Trench ID, plan. Fig. IH-36. Trench ID with Wall P, elevation, looking south. Fig. 111-37. Section, western face of Trench IF with Wall P. Fig. 111-38. Trench IF with Wall P, looking west. Layer of white packed lime above Wall P. Fig. 111-39. Section, southern end of Wall P in Trench IG. Fig. ΠΙ-40. Trench IG with southern end of Wall P joining northern tip of bedrock of limestone cliffs of harbor. Fig. Ill—41. Spring House excavations, 1969, with modern wooden cofferdam used for excavation. Fig. 111-42. Cesium magnetometer study of lagoon in 1969 by Elizabeth Ralph, area of Trench IC; pier 2 of Spring House platform in lower left. Fig. HI-43. Excavation of northern end of Wall P with back-hoe in 1972, Trench IK-IL. Fig. 111-44. Trench U2 with embankment of amphoras of Type 4b to east of Wall U. Fig. HI-45. Piers of Aqueduct, nos. 3-8, looking east. Fig. HI-46. Pier Y, abutting Wall P on bedrock foundation under which a spring surges. Fig. 111-47. Area YZ 1 looking north with rubble Wall W of northern fish tank. Fig. 111-48. Foundation of walls of structure YZ, looking north with terracotta drain in foreground and bases of rubble piers. Fig. 111-49. Excavations in lagoon, 1972, looking west toward Spring House and cistern. Fig. 111-50. Wall S and top of pier 9, looking north. Fig. Ill—51. Divers linking cylinders together for airlifting inside. Fig. 111-52. Water jet prober in use by J. D. Lewis in sandy harbor floor of Cosa. Fig. HI-53. J. D. Lewis working airlift, west end of breakwater extension A in harbor of Cosa. Fig. IV-I. Portus Cosanus, key for location of detailed plan of Tagliata South, designated by dotted

XlU

rectangle and sections, designated by arrows. Fig. IV-2. Aerial view of the Portus Cosanus and site of ancient lagoon, looking north, 1969. Fig. IV-3. Aerial view of the Portus Cosanus with Tagliata, breakwater and its extensions, Piers 1-5, beach walls, and Torre della Tagliata (upper center). Fig. IV-4. Aerial view of main breakwater with Piers 4 and 5, looking north. Fig. IV-5. Aerial view of eastern tip of main breakwater with Pier 5 and extensions A-F, looking north. Fig. IV-6. Breakwater extension A, underwater. Fig. IV-7. Breakwater extension B, underwater. Fig. IV-8. Breakwater extension F, underwater. Fig. IV-9. Breakwater extension D, with diver moving rocks by means of inflated plastic bag, underwater. Fig. IV-IO. Breakwater extension E, underwater. Fig. IV-Il. Pier 2, western face. Fig. IV-12. Drawing of center section of Pier 2, looking west. Fig. IV-13. Pier 1.5, between Piers 1 and 2, at water level. Piers 3 and 4 in background. Fig. IV-14. Underwater view of northwest corner of Pier 4, showing heavy erosion at base, above rocks of breakwater. Fig. IV-15. Pier 5, underwater. Fig. IV-16. Wall M, looking north. Fig. IV-17. Drawing, elevation of Wall M, looking north, showing impressions of formwork. Fig. IV-18. Aerial view of beach of Cosa, looking northeast. Fig. IV-19. Detail of polygonal limestone beach Wall A, east face. Fig. IV-20. View of harbor of Cosa in 1968, with limestone cliffs in background and rubble villa Walls C, D, and E in foreground, looking west. Fig. IV-21. Beach Probe V with rubble villa Wall O with square pilaster; limestone polygonal Wall A in lower section. Fig. IV-22. Detail of western face of rubble villa Wall D with impressions for formwork. Fig. IV-23. Cistern of villa, west of Torre della Tagliata. Fig. IV-24. Villa walls below La Strega restaurant, east of Torre della Tagliata. Fig. IV-25. Villa walls north and west of La Strega restaurant (the "room"). Fig. IV-26. Aerial view of rubble walls on beach, about 1 km east of the port of Cosa, looking northwest. Fig. IV-27. Aerial view of underwater portions of walls in Fig. IV-26, about 1 km east of port of Cosa, looking northwest. Fig. IV-28. Natural outlet through sandbar at Lago di Burano, looking southwest. Fig. IV-29. Detail of polygonal wall at outlet of Lago di Burano, looking southwest. Fig. IV-30. Section elevation C of Tagliata, looking west. Fig. IV-31. Plan of Tagliata South with Tagliata Piccola. Fig. IV-32. Seaward mouth of Tagliata South, section A (looking northwest) and section B (looking northeast). Fig. IV-33. Aerial view of Tagliata South, looking northwest. Fig. IV-34. Tagliata South, looking northeast. Fig. IV-35. Tagliata South, looking southwest, with

LIST OF ILLUSTRATIONS mouth of Tagliata Piccola in right foreground and remains of concrete vaulting in left center. Fig. IV-36. Tagliata South, interior concrete domical vault in outlet tunnel. Fig. IV-37. Tagliata South, slots in east wall lined with tiles. Fig. IV-38. Tagliata South, cuttings for sluice gate along western wall. Fig. IV-39. Tunnel of Tagliata Piccola, looking south. Fig. IV-40. Tunnel of Tagliata Piccola, with remains of concrete vault. Fig. IV-41. Rectangular air shafts above seaward mouth of Tagliata Piccola. Fig. IV-42. Open section of Tagliata North with tunnel, looking south. Fig. IV-43. Tagliata Canal, western cliff face with square cuttings for sluice gates. Fig. IV-44. Northeastern exit of Piccolo Spacco into Tagliata North. Fig. IV-45. Detail of cut ledges of northeastern mouth of Piccolo Spacco. Fig. IV-46. Northeastern entrance of Spacco della Regina from quarry, marked by modern concrete pillar in lower center. Fig. IV-47. Spacco della Regina, showing scarping along fracture as it runs southwest. Fig. IV-48. Southern seaward mouths of Spacco della Regina, looking west. Fig. IV-49. Northeast-southwest channel of southern mouth of Spacco della Regina, looking west. Fig. IV-50. Spacco della Regina, seaward mouth, section D, looking west. Fig. IV-51. Spacco della Regina, seaward mouth, section E, northeast-southwest channel, looking south. Fig. 1V-52. Underwater view of northeast-southwest channel of seaward mouth of Spacco della Regina, looking west. Fig. IV-53. Spacco della Regina, seaward mouth, section F, northeast entrance of northeast-southwest channel, looking southwest showing man-cut ledge. Fig. IV-54. Detail of cuttings for polygonal blocks at northern end of quarry in western wall of Tagliata Canal. Fig. IV-55. Polygonal limestone embankment Wall PW where it joins rubble Wall d of Spring House. Rg. IV-56. Wall P in Trench IJ. Fig. IV-57. Wall P, north of Spring House. Fig. IV-58. Wall P, north of Spring House, detail of section of rubblework repair with bedrock foundation. Fig. IV-59. Wall P, north of Pier Z, detail of a repair with decorative use of amphora toes and fragments set into concrete. Fig. IV-60. Rubble column of water conduit leading from Spring House to cistern showing decorative use of amphora toes and tiles set in concrete. Fig. IV-61. Wall P, south of Spring House where it joins buttress of Wall a of Spring House. Fig. IV-62. Wall P, north of Spring House where it joins buttress of Wall b of Spring House. Fig. IV-63. Wall P with concrete foundation, north of Pier Z in area YZ, with repairs in lower course. Fig. IV-64. WaU P with concrete foundation ledge in Trench IK.

Fig. IV-65. Wall P with concrete foundation ledge, abutting foundation of Wall X at northern end of lagoon. Fig. IV-66. Northern rubble embankment Wall X, looking north. Fig. IV-67. Wall X, joining rubble Wall U with water channel, looking west. Fig. IV-68. Detail of Wall X, showing upper section at point where it ends. Fig. IV-69. Amphoras of Type 4b in Trench U l , east of Wall U. Fig. IV-70. Wall U with water channel entering under rubble Wall V of northern fish tank. Fig. IV-71. Wall W intersecting Wall V at meter stick to form northwest corner of northern fish tank. Fig. IV-72. Pier 3 of Aqueduct, with Spring House platform and piers 1 and 2 in background, looking north. Fig. IV-73. Western face of pier 6 of Aqueduct with Wall L of southern fish tank on right. Fig. IV-74. Pier 9 on eastern side of modern drainage canal. Fig. IV-75. Rubble Wall N of southern fish tank in Trench IB, looking north. Fig. IV-76. Amphora stamp of Type 4a (Cat. A99) with anchor, immured in concrete Wall L of southern fish tank. Fig. IV-77. Wall PE, ancient polygonal limestone embankment wall, which now forms eastern side of modern drainage canal; rubblework repairs in center section. Fig. IV-78. Wall PE, detail of polygonal section with Torre della Tagliata in background. Fig. IV-79. Northern rubble end of Wall PE with ashlar corner blocks of rampway in foreground. Fig. IV-80. Pier Z of structure YZ, abutting Wall P, southern face. Fig. IV-81. Channel in southeast corner of structure YZ with cover tile and wooden formwork preserved around foundation wall. Fig. V-I. Spring House complex, excavations of 1972 on western bank of ancient lagoon of Cosa, looking west. Fig. V-2. Spring House complex with piers of Aqueduct in foreground and cistern in background, 1972, looking west. Fig. V-3. Spring House complex on western embankment of ancient lagoon of Cosa with polygonal retaining Wall P running south, looking southwest with modern hotel, Il Pescatore, in background. Fig. V-4. Spring House concrete platform, bisected by drain with piers 1 and 2 resting on top. In background, water collecting basin flanked by buttresses of Walls a and b and polygonal Wall P. Fig. V-5. Spring House platform with preserved wooden formwork and piers 1 and 2, looking south. Fig. V-6. Southern wing of Spring House platform with wooden formwork and pier 2. Fig. V-7. Southern wing of Spring House platform with formwork on southern side. Fig. V-8. Wood in situ, northeast corner of room 2, detail of buckets and wooden pieces. Fig. V-9. Wood in situ, northeast corner of room 2,

LIST OF ILLUSTRATIONS detail of bucket in Fig. V-8, Cat. Wl. Fig. V-IO. Southwest corner of Spring House platform, bonded with polygonal Wall P. Fig. V-Il. Detail of wooden planking around lower step in front of Wall P and abutting the south edge of the Spring House platform. Fig. V-12. Northern wing of Spring House platform with formwork and polygonal Wall P with concrete foundation north of buttress of Wall b. Fig. V-13. Spring House basin showing drain in western Wall c and opus quasi-reticulatum. Fig. V-14. Detail of lower portion of Wall c of Spring House basin of opus quasi-reticulatum with walled-up opening just above middle tile drains. Fig. V-15. Detail of formwork in northeast corner of Spring House basin. Fig. V-16. East face of Spring House basin with immured post and formwork below. Fig. V-17. Spring House basin, northwest corner, detail of opus quasi-reticulatum. Fig. V-18. Spring House basin, northwest corner, junction of Walls c and b, upper section. Fig. V-19. Drawing and section of three mortared voussoirs from Spring House excavations. Fig. V-20. Spring House, southern face of Wall b, with buttress and section of Wall c. Fig. V-21. Spring House, Room 2, looking south to Wall a. Fig. V-22. Spring House, Room 2, northwest corner, eastern face of Crosswall 1. Fig. V-23. Spring House, Room 2, detail of bedrock in southwest corner. Fig. V-24. Spring House, Room 2, south Wall a. Fig. V-25. Spring House, Crosswall 1 and Crosswall la, looking southwest with Room 1 in background with door, steps, and projecting threshold block in south wall. Fig. V-26. Spring House, Crosswall 2, looking north. Fig. V-27. Northwest corner of Spring House. Fig. V-28. Fallen rubble wall south of Wall a of Spring House with western face of trench. Fig. V-29. Spring House, inner face of south wall of Room 1 with walled-up door, plastered walls, and broken threshold block in wall. Fig. V-30. Spring House, north door in Room 1 with Pier fin background, looking north. Fig. V-31. Spring House, southwest corner with preserved plaster on exterior of Wall a. Fig. V-32. Paving with terracotta pan tiles forming gutter south of Wall a of Spring House. Fig. V-33. West wall of trench south of Wall a at southwest corner of Room 2 of Spring House, showing drainage channel. Fig. V-34. Piers 3 through 6 of Aqueduct, looking west, with Walls d and e of Spring House water reservoir, stage 1, in upper center. Fig. V-35. Spring House, Walls d and e of water reservoir, stage 1. Fig. V-36. Rubble bases of Piers f through k of water conduit of Spring House, looking east. Fig. V-37. Pier g of water conduit and north face of Wall b of Spring House, looking west through trench to cistern. Fig. V-38. Cistern, east face with buttresses.

χ

Fig. V-39. Southeast buttress with southeast corner of cistern. Fig. V-40. South wall of cistern with southern buttress and well in southeast corner. Fig. V-41. Interior of cistern, northeast corner with floor of cocciopesto concrete. Fig. V-42. Spring House complex, key for locations of section drawings. Fig. V-43. Spring House complex, excavations: Longitudinal Section A, archaeological remains. 1 Fig. V-44. Spring House: longitudinal Section A , stage 1. 1 Fig. V-45. Spring House: transverse Section B , stage 1. Fig. V-46. Spring House: axonametric projection of floor levels, stage 1. Fig. V-47. Spring House: axonametric projection of framing, stage 1. Fig. V-48. Spring House complex: axonametric reconstruction, stage 1. Fig. V-49. Spring House: longitudinal Section A2, stage 2. Fig. V-50. Spring House: transverse Section B 2 , stage 2. Fig. V-51. Spring House: axonametric projection of floor levels, stage 2. Fig. V-52. Spring House: axonametric projection of framing, stage 2. Fig. V-53. Spring House: axonametric reconstruction, stage 2. Fig. V-54. Nineteenth-century Egyptian gear-driven potgarlands. Fig. V-55. Nineteenth-century gear-driven pot-garland, diagram. Fig. V-56. Recent gear-driven bucket-chain installation at Formia, Italy (elevated shaft variety). Fig. V-57. Recent gear-driven bucket-chain installation at Formia, Italy (elevated shaft variety). Fig. V-58. Reproduction of bucket from Spring House, stage 2. Fig. V-59. Reproduction of bucket from Spring House, stage 2. Detail. Fig. V-60. Reproduction of bucket from Spring House, stage 2. Alternate arrangement with transverse rod. Fig. V-61. Nineteenth-century Egyptian gear-driven wheel with compartmented rim (short-shaft design). Fig. V-62. Twentieth-century Balearic gear-driven bucket-chain, diagram (short-shaft design). Fig. V-63. Philo's bucket-chain, Pneumatica 65. Carra de Vaux's reconstruction. Fig. V-64. Philo's bucket-chain. Beck's reconstruction. Fig. V-65. Vitruvius' bucket-chain (10.4.4). Reconstruction (mode of dumping conjectural). Fig. V-66. SSqiya pot from Medinet Habu, late Roman. Fig. V-67. Baths near the forum, acropolis of Cosa. View of shaft and elevated cistern. Fig. V-68. Reconstruction of installation for gear-driven bucket-chain at Hermoupolis West, Egypt. First to second century A. D. Fig. V-69. Plan of installation for gear-driven potgarland, "Der groBe Doppelbad," Abu Mena, Egypt. Fifth to sixth century A. D. Fig. V-70. Reconstruction of gear-driven pot-garland at "Der grofie Doppelbad," Abu Mena. Fig. V-71. Reconstruction of tread-wheel-driven potgarland at "Der grofie Doppelbad," Abu Mena.

XVl

LIST OF ILLUSTRATIONS

Fig. V-72. Plan of installation for gear-driven potgarland at "Der Palast," Abu Mena, Egypt. Fifth to sixth century A. D. Fig. V-73. Drawing of fresco representing a gear-driven compartmented wheel, Alexandria, Egypt. Fig. V-74. Plans of tread-wheel-driven bucket-chain installations at Pompeii: Stabian Baths, House of the Queen of England, Forum Baths, ca. 60 B.C. Fig. V-75. Reconstruction of the tread-wheel-driven potgarlands in the Stabian Baths and the House of the Queen of England. Fig. V-76. Reconstruction of bucket-chain wheel in Stabian Baths. Fig. VI-I. View of the ancient lagoon area of Cosa, now silted pasture land, looking northeast toward Le Rocce Hotel, the site of the lagoon temple. Fig. VI-2. Terracotta warrior from pediment of the lagoon temple, side view. Fig. VI-3. Revetment plaque from original decoration of lagoon temple with looped palmette design. Fig. VI-4. Revetment plaque from original decoration of Temple D, arx of Cosa. Figs. VI-5 to VI-Il. Cats. LTl to LT7, architectural terracottas from lagoon temple. Fig. VII-I. Modern port of Sesimbra, Portugal. Fig. VII-2. Breakwater at Sesimbra, Portugal, with fishing boats tied to boulders. Fig. VII-3. Terracotta model from Vulci of a lighthouse. Fig. VII-4. Modern sluice gate at Nassa fishery, Orbetello lagoon. Fig. VII-5. Javea, Spain. Southern channel connecting Roman fish tank to sea, cut through rock of the Punta de l'Arenal. Fig. VII-6. Modern fishery at Lago di Burano. Fig. VII-7. Modern metal fish barrages at Nassa fishery, in channel leading from Orbetello lagoon to the sea. Fig. VII-8. Basilica, forum of Cosa, pier of Neronian odeon. Fig. VII-9. Schematic plan, Phase 1, the Portus Cosanus and lagoon, later third and second centuries B.C. Fig. VII-10. Schematic plan, Phase 2, the Portus Cosanus and lagoon, late second and first centuries B.C. Fig. VII-Il. North-south schematic section, Phase 2, the Portus Cosanus and lagoon, late second and first centuries B.C. Fig. VII-12. East-west schematic section, Phase 2, the Portus Cosanus and lagoon, late second and first centuries B.C. Fig. VII-13. Schematic Plan, Phase 3, the Portus Cosanus and lagoon, Imperial period, later first to third centuries A. D. Fig. VIlI-I. Small freighter and rowboat. Detail of relief on a sarcophagus from Ostia. Fig. VIII-2. Small freighter. Detail from the same relief as Fig. VIII-I. Fig. VIII-3. Large freighters, stevedore loading amphora. Mosaic in the Foro delle Corporazioni, Ostia, ca. A. D. 200. Fig. VIII-4. Two cargo vessels, one entering the port of Ostia, the other tied up at quay for unloading. Relief, Torlonia Museum, Rome, ca. A.D. 200. Fig. VIII-5. Merchant galley under oars and sail. Wall

painting from a villa near Sirmione. First century A.D.

Fig. VIII-6. Tugboat, with towline leading from the stern upward. Relief from a tomb in the Isola Sacra between Ostia and Portus. Third century A.D. Fig. VIII-7. Fishing boats, detail from mosaic of Neptune and Amphitrite from Constantine (Algeria), now in the Louvre. Third century A. D. Fig. VIII-8. Fishing boat with fisherman harpooning fish. Mosaic from Sousse, House of the Triumph of Dionysus. Third century A. D. Fig. VIII-9. Fishing boat with fishermen using net. Mosaic from Thuburbo Majus, Tunisia. Third century A. D. Fig. VIII-IO. Fishing boat with fisherman using baskets for catching eels. Mosaic from Sousse, House of the Triumph of Dionysus. Third century A. D. Fig. VIII-Il. Fishing boat found in the mud of what was once the harbor of Claudius at Ostia. Roman Imperial period. Fig. VIII-12. Rowboats with ram-like prow. Clay lamp from Libya. First century A.D. Fig. VIII-13. Two coastal craft at the entrance of the port of Ostia, loaded with amphoras. Relief in the Catacomb of Praetextatus, Rome. Third century A.D.

Fig. VIII-14. Aerial view of ship's hull, looking toward prow, underwater, Cosa harbor. Fig. VIII-15. View of ship's hull, looking toward prow, underwater, Cosa harbor. Fig. VIII-16. Prow of ship, underwater, Cosa harbor. Fig. VIII-17. Frames from ship's hull, underwater, Cosa harbor. Fig. IX-I. Type la 1 amphora, Antiquarium, Orbetello. First quarter of third century B.C. Fig. IX-2. Detail of Fig. IX-I. Graffito, MPK, on lower neck. Fig. IX-3. Type Id amphora (western). Said to be from Port' Ercole. First half of second century B.C. Fig. IX-4. Type 4a amphora from Cosa hill excavations (C65.94), stamped on lip with trident device. Late second or early first century B.C. Figs. IX-5, IX-6. Cats. Al,a A2. Type la amphoras. Figs. IX-7 to IX-9. Cats. A3-A5. Type Ic amphoras. Figs. IX-10 to IX-51. Cats. A6-A48. Type Id amphoras. Figs. IX-52 to IX-277. Cats. A50-A187. Type 4a amphoras. Figs. IX-278 to IX-371. Cats. A188-A245. Type 4b amphoras. Figs. IX-372 to IX-393. Cats. A247-A257. Type 5 amphoras. Fig. IX-394. Cat. A258. Type 10 amphora. Figs. IX-395, IX-396. Cat. A259. Type 11c amphora. Figs. IX-397, IX-398. Cat. A260. Type 12a amphora. Figs. IX-399 to IX-404. Cats. A261-A265. Type 12b amphoras. Figs. IX-405 to IX-407. Cats. A267, A268. Type 12c amphoras. Fig. IX-408. Cat. A269. Type 13 amphora. Fig. IX-409. Cat. A270. Type 14 amphora. Fig. IX-410. Cat. A271. Type 16 amphora. Figs. IX-411 to IX-423. Cats. A272-A281. Type 18a amphoras.

LIST OF ILLUSTRATIONS Figs. IX-424 to IX-427. Cats. A282-A285. Type 18b amphoras. Figs. IX-428 to IX-431. Cats. A286-A288. Type 20 amphoras. Figs. IX-432 to IX-435. Cats. A289, A290. Type 21a amphoras. Figs. IX-436 to IX-439. Cats. A291-A293. Type 21b amphoras. Fig. IX-440. Cat. A294. Type 21c amphora. Figs. IX-441, IX-442. Cat. A295. Type 24a amphora. Figs. IX-443 to IX-453. Cats. A298-A304. Unclassified amphoras. Figs. X-I to X-58. Cats. TW1-TW36. Roman thin walled pottery. Figs. X-59 to X-70. Cats. BG1-BG6. Black glaze pottery. Figs. X-71 to X-103. Cats. RG1-RG16. Red-gloss pottery. Figs. XI-I to XI-90. Cats. K1-K90, Utilitarian ware, Kitchen ware. Figs. XI-91 to XI-127. Cats. C1-C37. Utilitarian ware, Coarse ware. Figs. XI-128 to XI-165. Cats. D1-D38. Utilitarian ware, Domestic ware. Figs. XI-166 to XI-174. Cats. MC1-MC9. Miscellaneous ceramics. Figs. XII-I to XII-20. Cats. G1-G13. Roman glass.

XVIl

Figs. XII-21 to XII-32. Cats. L1-L9. Roman lamps. Figs. XII-33 to XII-35. Cats. C 0 1 - C 0 2 . Roman coins. Figs. XIII-I to XIII-33. Cats. W1-W16. Wood from the Spring House machinery. Figs. XIII-34 to XIII-64. Cats. M1-M28. Metal from the Spring House area. Fig. XIII-65. Cat. AD3. Architectural molding. Fig. XVI-I. Scanning-electron microscope picture of a thermally treated sample, no. 3 from Cava della Doganella clay quarry, magnification X 1400. Fig. XVI-2. Tile sample ΔΓ from Cosa lagoon, magnification X 3000, showing textural affinities due to the presence of the phyllosilicates (mainly 2Mmica). Fig. XVII-I. Cat. T l , land tortoise shell, Testudo hermanni, from Spring House, level below concrete platform. Fig. XVII-2. Cat. Sl, marine snail shell, Tonna galea galea (Linne), from Spring House, level of concrete platform. Fig. XIX-I. Aerial view of port of Sidon. Fig. XIX-2. Aerial view of the ancient ports of Ostia (Portus) looking north, showing modern coastline over 3 km to the northwest. Fig. XIX-3. Pompeian painting of harbor scene with beacon light on rocky projection, from Stabiae.

List of Tables

Table IX-I. Portus Cosanus Roman amphora totals. Table XIV-I. Magnetic susceptibilities of Cosa rock and soil samples. Table XV-I. Description of cores taken from lagoon area at Cosa. Table XV-2. Description of 1972 samples taken from excavated trenches in lagoon at Cosa. Table XV-3. Chemical and mineralogical composition of key cores and trench samples in lagoon at Cosa. Table XV-4. Particle-size distribution of samples from key cores in lagoon at Cosa. Table XVI-I. Identification and localities of clay and archaeological samples.

217 287 302 306

306 308 309

Table XVI-2. Mineralogical composition of clayey samples. Table XVI-3. Mineralogical composition of thermally treated clayey samples and archaeological findings. Table XVI-4. XRF analyses from Ti, Cr, Zr, and Sr on thermally treated clayey samples and archaeological findings. Table XVII-I. Distribution of faunal remains from the lagoon of Cosa. Table X VII-2. Measurements of partial skeleton of Canis domesticus found in Spring House well.

310 310 311 316 317

Frequently Cited References and Abbreviations

T h r o u g h o u t this book, frequently cited references are abbreviated as listed here. Abbreviations of ancient authors cited are largely those accepted by the Oxford Classical

REFERENCES

AA Archdologischer Anzeiger. AJA American Journal of Archaeology. AJP American Journal of Philology. Almagro I Almagro, M. Las Necropolis de Ampurias, I. Monografias Ampuritanas, 3. Barcelona, 1953. Andren Andren, A. Architectural TerracottasfromEtruscan Italic Temples. 2 vols. Lund, 1940. AntAfr Antiquites Africaines. AntJ Antiquaries' Journal. Anziani Anziani, D. "Cosa-Portus Cosanus-Portus HerculisSuccosa-Orbetello dans 1'antiquite." MelRome 30 (1910) pp. 373-395. ArchCl Archeologia Classica. ArchEph Archaiologtke Ephemeris. ArchEspArq Archive Espanol de Arqueologia. AttiPontAcc Atti della Pontificia Accademia Romana di Archeologia. AWeIt Antike Welt. Bace Bace, E. J. "Cosa: Inscriptions on Stone and BrickStamps." Doctoral dissertation, University of Michigan, Ann Arbor, 1983. Bailey Bailey, D. M. Catalogue of Lamps in the British Museum. II. Roman Lamps Made in Italy. London, 1980. Baldacci 1967-1968 [1969] Baldacci, P. "Alcuni aspetti dei commerci dei territori cisalpini." Atti del centre studi e documentazione sull' Italia romana 1, 1961-68. Milan, 1969, pp. 7-50. Baldacci 1969 [1972] Baldacci, P. "Le principal! correnti del commercio di

Dictionary.

anfore romane nella Cisalpina." I problemi della ceramica romana di Ravenna, della Valle padana e dell'alto Adriatico. Atti del convegno internazionale. Ravenna, 1012maggio 1969. Bologna, 1972, pp. 103-131. Baldacci 1972 Baldacci, P. "Importazioni cisalpine e produzione apula." Baldacci, P., et al. Recherches sur les amphores romaines. Rome: Ecole frangaise de Rome, 1972, pp. 7-28. BAntFr Bulletin de la Societe nationale des antiquaires de France. Barker Barker, G. Landscape and Society. Prehistoric Central Italy. New York, 1981. Bass, History Bass, G. F., ed. A History of Seafaring Based on Underwater Archaeology. New York, 1972. BCH Bulletin de conespondance hellenique. BdA Bolletino d'Arte. BdI Bullettino dell'Istituto di Corrispondenza Archeologica. Beltran 1970 Beltran Lloris, M. Las dnforas romanas en Espana. Zaragoza: Institution Fernando el Catolico, 1970. Beltran 1978 Beltran Lloris, M. Ceramica romana. Zaragoza, 1978. Beltran 1980 Beltran Lloris, M. "El comercio del aceite en el valle del Ebro a finales de la Republica y comienzos del Imperio romano." Blazquez Martinez 1980, pp. 187224. Benoit 1961 Benoit, F. L'epave du Grand Congloue a Marseille. Gallia, supp. 14. Paris, 1961. Blackman 1982-1 Blackman, D. J. "Ancient Harbours in the Mediterranean. Part 1." IJNA 11 (1982) pp. 79-104. Blackman 1982-2 Blackman, D. J. "Ancient Harbours in the Mediterranean. Part 2." IJNA 11 (1982) pp. 185-211. Blake 1947 Blake, M. E. Ancient Roman Construction in Italy from the Prehistoric Period to Augustus. Washington, 1947.

XXIl

REFERENCES AND ABBREVIATIONS

Blake 1959 Blake. M. E. Roman Construction in Italy from Tiberius through the Flavians. Washington, 1959. Blake 1973 Blake, M. E. Roman Construction in Italy from Nerva through the Antonines. Ed. D. T. Bishop. Philadelphia, 1973. Blazquez Martinez 1980 Blazquez Martinez, J. M., et al. Produccion y comercio del aceite en la antiguedad (I Congreso International). Madrid: Universidad Complutense, 1980. BMBeyrouth Bulletin du Musie de Beyrouth. BMC Mattingly, H. Coins of the Roman Empire in the British Museum. 6 vols. London, 1923-1967. Boethius and Ward-Perkins Boethius, A., and Ward-Perkins, J. B. Etruscan and Roman Architecture. Baltimore, 1970. Bohlen Bohlen, D. Die Bedeutung der Fischerei fiir die antike Wirtschafi. Hamburg, 1937. Bonnjbb Bonner Jahrbiicher. Bourgeois and Gazda Bourgeois, J., and Gazda, G. "Geologic and Sedimentologic History of the Roman Harbor and Fishery Site (273 B.C. to 3rd Century A.D.), Cosa, Italy." AJA 86 (1982) p. 256. Boyce A. A. Boyce. "The Harbor of Pompeiopolis: A Study in Roman Imperial Ports and Dated Coins." AJA 62 (1958) pp. 67-78. Bradford Bradford, J. Ancient Landscape: Studies in Field Archaeology. London, 1957, pp. 227-231. Broneer Broneer, O. Corinth. IV, 2. Tenacotta Lamps. Cambridge, 1930. Bronson and Uggeri Bronson, R. C , and Uggeri, G. Isola del Giglio, Isola di Giannutri, Monte Argentario, Laguna di Orbetello. Florence, 1970. Broughton 1951-1952 Broughton, T. Robert S. The Magistrates of the Roman Republic. 2 vols. New York, 1951-1952. Broughton 1972 Broughton, T. Robert S. "Oil-producing Estates in Southern Spain." Vestigia 17: Akten des VI Intemationalen Kongresses fiir griechischen und lateinischen Epigraphik. Munich, 1972, pp. 475-476. Broughton 1974 Broughton, T. Robert S. "Some Notes on Trade and Traders in Roman Spain." Polis and Imperium. Studies in Honour of Edward Togo Salmon. Toronto, 1974, pp. 11-30. Broughton 1980 Broughton, T. Robert S. "Oil-producing Estates in Roman Baetica." Revista de la Universidad Complutense {Homenaje a Garcia y Bellido, 4) 28, no. 118 (1980) pp. 153-160.

Brown 1951 Brown, F. E. "Cosa I. History and Topography." MAAR 20 (1951) pp. 1-113. Brown 1980 Brown F. E. Cosa. The Making of a Roman Town. Ann Arbor, 1980. Brown et al. Brown, F. E.; Richardson, E. H.; and Richardson, L. Cosa II. The Temples of the Arx. MAAR 26 (1960). Bruno, Will, and Schwarzer Bruno, V. J.; Will, E. L.; and Schwarzer, J. "Exploring the Gulf of Talamone." Archaeology 33 (1980) pp. 35-43. BSA British School at Athens, Annual. BSR British School of Archaeology at Rome, Papers. BullComm Bullettino della Commissione archeologica comunale di Roma. Buttrey Buttrey, T. V. "Cosa: The Coins." MAAR 34 (1980) pp. 1-153, pis. 1-10. Callender 1965 (also referred to as "Callender") Callender, M. H. Roman Amphorae. London, 1965. Cambi 1976 Cambi, N . "Spanish Amphorae Found near Split." RCRFA 16 (1976) pp. 115-124. Carandini and Settis Carandini, A., and Settis, S. Schiavi e Padroni nell'Etruria Romana. Bari, 1979. Cardarelli 1924, 1925 Cardarelli, R. "Confini fra Magliano e Marsiliana; fra Manciano e Montauto Scerpenna Stachilagi; fra Tricosto e Ansedonia; fra Orbetello e Marsiliana; fra Port'Ercole e Monte Argentario (28 dicembre 15082 marzo 1510)." Maremma 1 (1924) pp. 131-142, 155186, 205-224; 2 (1924) pp. 2-36, 75-128, 147-213. Carettoni Carettoni, G. G. "Saggi nell'interno della Casa di Livia." NSc (1957) pp. 72-119. Casson, SSAW Casson, L. Ships and Seamanship in the Ancient World. Princeton, 1971. Cerda 1980 Cerda Juan, D. La nave romano-republicana de la Colonia de Sant Jordi (Ses Salines—Mallorca). Palma de Mallorca, 1980. Charlesworth Charlesworth, M. P. Trade Routes and Commerce in the Roman Empire. Cambridge, 1924. ChronEg Chronique d'Egypte. CIL Corpus Inscriptionum Latinarum. Q Classical Journal. Colls et al. 1977 Colls, D., et al. L'epave Port-Vendres II et Ie commerce de la Betique a Pepoque de Claude. Archeonautica 1. Paris, 1977.

REFERENCES AND ABBREVIATIONS Colls and Lequement 1980 Colls, D., and Lequement, R. "L'epave Port-Vendres II: nouveaux documents epigraphiques." Blazquez Martinez 1980, pp. 177-186. Colonna Colonna di Paolo, E., and Colonna, G. Castel d'Asso. Rome, 1970. Comfort, AJA Comfort, H. "A Preliminary Study of Late Italian Si­ gillata." AJA 40 (1936) pp. 437-451. Comfort, EAA, supp. Comfort, H. "Terra Sigillata." EAA, supp. (1973) pp. 803-835. Corcoran, "Roman Fishponds" Corcoran, T. H. "Roman Fishponds." Classical Bul­ letin 35 (1959) pp. 37-43. Cotton Cotton, M. A. The Late Republican Villa at Posto, Francolise. London, 1979. Cousteau 1954 Cousteau, Jacques-Yves. "Fish Men Discover a 2,200-year-old Greek Ship." National Geographic 105 (1954) pp. 1-34. Curtis Curtis, R. I. "The Production and Commerce of Salted Fish Products (Garum and Salsamenta) in the Roman Empire." Aufstieg und Niedergang der romischen Welt, II, 11, forthcoming. CVA Corpus Vasorum antiquorum. D'Ancona D'Ancona, U. "Fishing and Fish Culture in BrackishWater Lagoons." FAO Fisheries Bulletin 7, no. 4 (1954) pp. 1-24. D'Arms 1980 D'Arms, J. H. "Senators' Involvement in Commerce in the Late Republic: Some Ciceronian Evidence." The Seaborne Commerce of Ancient Rome: Studies in Ar­ chaeology and History. Ed. J. H. D'Arms and E. C. KopfT. MAAR 36 (1980) pp. 77-89. D'Arms 1981 D'Arms, J. H. Commerce and Social Standing in Ancient Rome. Cambridge, Mass., 1981. DarSag Daremberg, Ch., and Saglio, E. Dktionnaire des antiquites grecques et romaines. Paris, 1877-1919. De Angelis 1959 De Angelis, R. Fishing Installations in Brackish Lagoons. General Fisheries Council for the Mediterranean. Studies and Reviews 7. Rome: FAO, 1959. De Angelis 1960 De Angelis, R. Mediterranean Brackish Water Lagoons and Their Exploitation. General Fisheries Council for the Mediterranean. Studies and Reviews 12. Rome: FAO, 1960. Dechelette Dechelette, J. Les vases cέramiques omis de la Gaule romaine. 2 vols. Paris, 1904. Del Rosso Del Rosso, R. Pesche e peschiere antiche e modeme nell'Etruria marittima. Florence, 1905.

XXlIl

Deneauve Deneauve, J. Lampes de Carthage. Paris, 1969. Dennis Dennis, G. The Cities and Cemeteries ofEtruria. 2 vols. 1848. Rev. ed. Rome, 1968. DialAr Dialoghi di Archeologia. Dragendorff Dragendorff, H. "Terra Sigillata." Bonnjbb 96-97 (1895) pp. 18-155. Dressel 1879 Dressel, H. " D i un grande deposito di anfore rinvenuto nel nuovo quartiere del Castro Pretorio." BullComm 1 (1879) pp. 36-112, 143-196. Dressel 1889 (referred to as "Dressel") Dressel, H. Inscriptiones Urbis Romae Latinae. Instrumentum Domesticum. CIL XV, 2. Berlin, 1889. Du Plat Taylor Du Plat Taylor, J. Marine Archaeology. New York, 1966. Dyson, JFA Dyson, S. L. "Settlement Patterns in the Ager Cosanus: The Wesleyan University Survey, 1974-1976." JFA 5 (1978) pp. 251-268. Dyson, Utilitarian Pottery Dyson, S. L. Cosa: The Utilitarian Pottery. MAAR 33 (1976). EAA Enciclopedia dell'arte antica, classica e orientate. Rome, 1958-1965. ESAR An Economic Survey of Ancient Rome. Ed. Tenney Frank. 5 vols. Baltimore, 1933-1940. Euzennat Euzennat, M. "Ancient Marseille in the Light of Re­ cent Excavations." AJA 84 (1980) pp. 133-140. FA Fasti Archaeologici. Fairbridge, "Quaternary Sedimentation" Fairbridge, R. W. "Quaternary Sedimentation in the Mediterranean Region Controlled by Tectonics, Paleoclimates and Sea Level." The Mediterranean Sea. Ed. D. J. Stanley. Stroudsburg, Pa., 1972, pp. 99113. FAO Food and Agricultural Organization of the United Nations. FHG Miiller, C. Fragmenta Historicorum Graecorum, 18411870. Flemming 1969 Flemming, N. C. "Archaeological Evidence for Eustatic Change of Sea Level and Earth Movements in the Western Mediterranean during the Last 2,000 Years." Geological Society of America, Special Paper 109 (1969) pp. 1-125. Frost 1963 Frost, H. Under the Mediterranean. Englewood Cliffs, N.J., 1963. Garnsey, Hopkins, and Whittaker Trade in the Ancient Economy. Ed. P. Garnsey,

XXlV

REFERENCES AND ABBREVIATIONS

K. Hopkins, and C. R. Whittaker. Berkeley and Los Angeles, 1983. Genito Genito, M. C. Lucerne fittili delle Collezioni del Museo Civico Archeologko di Bologna. Bologna, 1977. GH etruschi GH etruschi in maremma. Popolamento e attivita productive. Ed. M. Cristofani. Milan, 1981. Goudineau 1968 Goudineau, C. La ceramique aretine Hsse. MelRome, supp. 6, no. 4 (1968). Goudineau 1970 Goudineau, C. "Note sur la ceramique a engobe interne rouge-pompeien ('pompejanisch-roten Platten')." MelRome 82 (1970) pp. 159-186. Grace 1961 Grace, V. R. Amphoras and the Ancient Wine Trade. Princeton: American School of Classical Studies at Athens, 1961. Harris Harris, W. V. Rome in Etruria and Umbria. London, 1971. Hatzfeld 1912 Hatzfeld, J. "Les Italiens residant a Delos." BCH 36 (1912) pp. 10-218. Hayes Hayes, J. W. Late Roman Pottery. London, 1972. Helbig4 Helbig, K. F. W. Fiihrer durch die offentlichen Sammlungen klassischer Altertumer in Rom. Ed. H. Speier, 4th ed., 4 vols. Tubingen, 1963-1972. Heres Heres, G. Die romischen Bildlampen der Berliner Antiken-Sammlung, Berlin, 1972. Hermet Hermet, F. La Graufesenque. Paris, 1934. Hesnard 1977 Hesnard, A. "Note sur un atelier d'amphores Dr. 1 et Dr. 2-4 pres de Terracine." MelRome 89 (1977) pp. 157-168. Hesnard 1978 Hesnard, A. "Les amphores." Tchernia, A., et al. L'epave romaine de la Madrague de Giens (Var). Gallia, supp. 34. Paris, 1978, pp. 33-46. Houston Houston, G. W. "The Administration of Italian Seaports during the First Three Centuries of the Roman Empire." MAAR 36 (1980) pp. 157-171. IJNA International Journal of Nautical Archaeology. ILS Dessau, H. Inscriptiones Latinae Selectae, 1892-1916. Incitti 1977 Incitti, M. "Le anfore." Caretta, L., et al. Introduzione alio studio delta ceramica romana di eta imperiale proveniente dal Colosseo. Rome, 1977, n.p. JFA Journal of Field Archaeology. JHS Journal of Hellenic Studies.

JoIy JoIy, E. Lucerne del Museo di Sabratha. Rome, 1974. Ji?S Journal of Roman Studies. Kahler Kahler, H. Der grosse Fries von Pergamon. Berlin, 1948. Kapitan Kapitan, G. "I relitti di Capo Graziano (Filicudi): scoperte dalla spedizione NACSAC nel 1968." SicA 10:34 (1977) pp. 40-53. Karmon Karmon, Y. Ports Around the World. New York, 1980. Klumbach Klumbach, H. "Das verbreitungsgebiet der spatitischen Terra Sigillata." RGZM 3 (1956) pp. 117-133. Lamboglia, Albintimilium Lamboglia, N . GH scavi di Albintimilium e la chronologia della ceramica romana. Bordighera, 1950. Lamboglia, "Ceramica campana" Lamboglia, N . "Per una classificazione preliminare della ceramica campana." Atti del I Congresso internazionale di Studi Liguri (1950). Bordighera, 1952, pp. 139-206. Lamboglia 1952 Lamboglia, N . "La nave romana di Albenga." RStLig 18 (1952) pp. 131-236. Lamboglia 1955 Lamboglia, N . "Sulla cronologia delle anfore romane di eta repubblicana (H-I secolo A.C.)." RStLig 21 (1955) pp. 241-270. Lamboglia 1958, 1963 Lamboglia, N. "Nuove osservazioni sulla 'terra sigillata chiara.' " RStLig 24 (1958) pp. 257-330; 29 (1963) pp. 145-212. Laporte 1980 Laporte, J.-P. "Les amphores de Tubusuctu et 1'huile de Mauretanie Cesarienne." Bulletin archeologique du Comite des Travaux Historiques et Scientifiques, n.s. 1214 (1976-1978). Paris, 1980, pp. 131-157. Laubenheimer 1977 Laubenheimer, F. "Ateliers de potiers de Sallelesd'Aude. Fouilles de sauvetage 1977." Bulletin de la Societe d'Etudes Scientifiques de I'Aude 77 (1977) pp. 8390. Lavizzari Pedrazzini, Mariana Lavizzari Pedrazzini, M. P. "Terra sigillata decorata." La ceramique sigillee italique et tardo-italique. Les fouilles de Mariana, Corse, 7. Bastia, 1980, pp. 15-40. Lavizzari Pedrazzini, Pisani Dossi Lavizzari Pedrazzini, M. P. La terra sigillata tardo-italica a rilievo nella collezione Pisani Dossi del Museo Archeologko di Milano. Milan, 1972. Lehmann-Hartleben Lehmann-Hartleben, K. Die antiken Hafenlagen des Mittelmeeres. KHo 14 (1923). Lenzi Lenzi, F. Iporti della maremma toscana. Monografia storica dell'antichita nella penisola italiana, Ministero della marina. Rome, 1905.

REFERENCES AND ABBREVIATIONS Lequement 1980 Lequement, R. "Le vin africain a l'epoque imperiale." AntAfr 16 (1980) pp. 185-193. Levi Levi, D. "Escursione archeologica nell'Agro Cosano." StEtr 1 (1927) pp. 477-485. Lewis 1972 Lewis, J. D. "Controlled Airlifting in Sand-filled Harbours." IJNA 1 (1972) pp. 169-170. Lewis 1973 Lewis, J. D. "Cosa: An Early Roman Harbour." Marine Archaeology, pp. 233-259. Loeschcke, "Haltern" Loeschcke, S. "Keramische Funde in Haltern." Mitteilungen der Altertumskommission fiir Westfalen 5 (1909) pp. 101-322. Loeschcke, Lampen Loeschcke, S. Lampen aus Vindonissa. Zurich, 1919. Loeschcke, "Petrisberg" Loeschcke, S. "Alteste romische Keramik vom Petrisberg uber Trier." TrZ 14 (1939) pp. 29-112. Loeschcke 1942 Loeschcke, S. In Veroffentlichungen aus dem stadtischen Museum JUr Vor- una Fruhgeschichte Dortmund. II: Das Romerlager in Oberaden. Ed. Christoph Albrecht. 2: Die romische una die belgische Keramik. Dortmund, 1942. Lugli Lugli, G. La tecnica edilizia romana. II. Rome, 1957. Luni I Scavi di Luni. Relazione preliminare delle campagne di scavo 1970-1971. Ed. A. Frova. Rome, 1973-1974. Luni II Scavi di Luni. Relazione della campagne di scavo 19721973-1974. Ed. A. Frova. Rome, 1977. Lythgoe Lythgoe, J., and Lythgoe, G. Fishes of the Sea. New York, 1975. MAAR Memoirs of the American Academy in Rome. Manacorda, Athenaeum Manacorda, D. "Considerazioni sull'epigrafia della regione di Cosa." Athenaeum 57 (1979) pp. 73-97. Manacorda 1977 (Ostia) Manacorda, D. "Anfore." Ostia IV, pp. 116-266, 277283, 359-383. Manacorda 1977 (Pompeii) Manacorda, D. "Anfore spagnole a Pompei." L'instrumentum domesticum di Ercolano e Pompei. Rome, 1977, pp. 121-133. Manacorda 1978 (referred to as "Manacorda") Manacorda, D. "The Ager Cosanus and the Production of the Amphorae of Sestius: New Evidence and a Reassessment." JRS 68 (1978) pp. 122-131. Manacorda 1980 Manacorda, D. "L'ager Cosanus tra tarda Repubblica e Impero: forme di produzione e assetto della proprieta." MAAR 36 (1980) pp. 173-184. Manacorda 1981 Manacorda, D. "Produzione agricola, produzione

XXV

ceramica e proprietari nell'ager Cosanus nel I. a.C." Societa romana e produzione schiavistica. II. Merci, mercati e scambi nel Mediterraneo. Ed. A. Giardina and A. Schiavone. Ban, 1981, pp. 3-54, 263-274. Marine Archaeology Marine Archaeology. Colston Papers, 23. Ed. D. J. Blackman. London, 1973. Marsh Marsh, F. B. A History of the Roman World from 146 to 30 B.C. 3rd rev. ed. London, 1971. MASCA Museum Applied Science Center for Archaeology, University of Pennsylvania, Philadelphia, Pa. Mayet Mayet, F. Les ceramiques a parois fines dans la peninsule iberique. Paris, 1975. McCann 1971 McCann, A. M. "Underwater Survey of the Etruscan Port of Populonia." Muse 5 (1971) pp. 20-22. McCann 1979 McCann, A. M. "The Harbor and Fishery Remains at Cosa, Itzly." JFA 6 (1979) pp. 391-411. McCann, Roman Sarcophagi McCann, A. M. Roman Sarcophagi in the Metropolitan Museum of Art. New York, 1978. McCann, StHarbArch I McCann, A. M. "The Roman Port and Fishery of Cosa: A Center of Trade in the Late Roman Republic." StHarbArch I, pp. 115-156. McCann, RCRFA McCann, A. M. "The Portus Cosanus: A Center of Trade in the Late Republic." RCRFA, forthcoming. McCann and Lewis 1970 McCann, A. M., and Lewis, J. D. "The Ancient Port of Cosa." Archaeology 23 (1970) pp. 200-211. McCann, Bourgeois, and Will McCann, A. M.; Bourgeois, J.; and Will, E. L. "Underwater Excavations at the Etruscan Port of Populonia." JFA 4 (1977) pp. 275-296. McCann and Will 1984 McCann, A. M., and Will, E. L. Review of The Seaborne Commerce of Ancient Rome: Studies in Archaeology and History (ed. J. H. D'Arms and E. C. Kopff. MAAR 36 [1980]) in AJA 88 (1984) pp. 92-95. McKay McKay, A. G. Houses, Villas and Palaces in the Roman World. Ithaca, N.Y., 1975. Meiggs, Ostia Meiggs, R. Ostia. Oxford, 1960. MelRome Melanges d'archeologie et d'histoire de I'Ecolefrancaisede Rome. Merciai 1910 Merciai, G. Mutamenti avvenuti nella configurazione del litorale tra Pisa e Orbetello dal pliocene in poi. Pisa. 1910. Merciai 1919 Merciai, G. "Sulle condizione fisiche del litorale etrusco tra Livorno e Civitavecchia." StEtr 3 (1929) pp. 347-358.

XXVl

REFERENCES AND ABBREVIATIONS

MittKairo Mitteilungen des deutschen Instituts fur agyptische Altertumskunde, Kairo. Moevs, AJA Moevs, M.T.M. "New Evidence for an Absolute Chronology of Decorated Late Italian Sigillata." AJA 84 (1980) pp. 319-327. Moevs, "Italo-Megarian Ware" Moevs, M.T.M. "Italo-Megarian Ware at Cosa." MAAR 34 (1980) pp. 161-227. Moevs 1973 Moevs, M.T.M. The Roman Thin Walled Pottery from Cosa (1948-1954). MAAR 32 (1973). MontAnt Monumenti Antichi. Morel Morel, J. P. Ciramiqut a vemis noir du Forum Romain et du Palatin. MilRome, supp. 3 (1965). Mori Mori, A. "Problemi dell'insediamento umano e della vita marittima nell'antica Etruria centrale." Bollettino della Societd Geografica Italiana, ser. 9, 4 (1963) pp. 343-355. Moriarty Moriarty, C. Eels: A Natural and Unnatural History. New York, 1978. Muckelroy 1980 Archaeology Under Water: An Atlas of the World's Submerged Sites. Ed. K. Muckelroy. New York and London, 1980, pp. 166-177. MiinchGeogAbh Munchner Geographische Ahhandlungen. NSc Notizie degli Scavi di Antichita. OCD Oxford Classical Dictionary. Oxford, 1970. Oleson, StHarArch I Oleson, J. P. "Herod and Vitruvius: Preliminary Thoughts on Harbour Engineering at Sebastos." StHarArch I, forthcoming. Oleson, Water-Lifiing Devices Oleson, J. P. Greek and Roman Mechanical Water-Lifiing Devices: The History of a Technology. Toronto, 1984. Ostia I Ostia I. Le Terme del Nuotatore. Scavo dell'ambiente IV. StMisc 13 (1967-1968). Ostia II Ostia II. Le Terme del Nuotatore. Scavo dell'ambiente I. StMisc 16 (1968-1970). Ostia III Ostia III. Le Terme del Nuotatore. Scavo degli ambienti III, VI, VII. StMisc 21 (1969-1970, 1970-1971, 19711973). Ostia IV Ostia IV. Le Terme del Nuotatore. Scavo dell'ambiente XVI e dell'area XXV. StMisc 23 (1972-1973, 19751977). Oswald Oswald, F. Index of Figure-Types on Terra Sigillata.

Annals of Archaeology and Anthropology, supp. 23. Liverpool, 1936. Oswald and Pryce Oswald, F., and Pryce, D. An Introduction to the Study of Terra Sigillata. London, 1920. Oxe, ARR Oxe, A. Anetinische Reliefgefasse vom Rhein. Materialien zur romische-germanischen Keramik, 5. Frankfurt am Main, 1923; Bonn, 1968. Oxe and Comfort Oxe, A., and Comfort, H. Corpus Vasorum Arretinorum: A Catalogue of the Signatures, Shapes and Chronology of Italian Sigillata. Bonn, 1968. PAAR American Academy in Rome, Papers and Monographs. Pallares Pallares, F. "Terra Sigillata non decorata." La ceramique sigillee italique et tardo-italique. Les fouilles de Mariana, Corse, 7. Bastia, 1980, pp. 2-14. Palma and Panella 1968 Palma, B., and Panella, C. "Anfore." Ostia I, pp. 97116. Panella 1970 Panella, C. "Le Anfore." Ostia II, pp. 102-156. Panella 1972 Panella, C. "Annotazioni in margine alle stratigrafie delle Terme ostiensi del Nuotatore." Baldacci, P., et al. Recherches sur les amphores romaines. Rome: Ecole franqaise de Rome, 1972, pp. 69-106. Panella 1973 Panella, C. "Appunti su un gruppo di anfore della prima, media e tarda eta imperiale (secolo I-V d.c.)." Ostia III, pp. 463-633. Panella and Fano 1977 Panella, C , and Fano, M. "Le anfore con anse bifide conservate a Pompei: contributo ad una loro classificazione." Methodes classiques et methodes formelles dans I'etude des amphores (Actes du colloque de Rome, 27-29 mat 1974). Rome: Ecole franqaise de Rome, 1977, pp. 133-177. Panella 1980 Panella, C. "Retroterra, porti, e mercati: l'esempio dell'ager Falemus." MAAR 36 (1980) pp. 251-259. Pascual Guasch 1962 Pascual Guasch, R. "Centros de production y difusion geografica de un tipo de anfora." Congreso National de Arqueologia VII, Barcelona 1960. Zaragoza, 1962, pp. 334-345. Peacock 1974 Peacock, D.P.S. "Amphorae and the Baetican Fish Industry." AntJ 54 (1974) pp. 232-243. Peacock 1977 Peacock, D.P.S. "Recent Discoveries of Roman Amphora Kilns in Italy." AntJ 57 (1977) pp. 262-269. PECS The Princeton Encyclopedia of Classical Sites. Ed. R. Stillwell. Princeton, 1976. Pelichet 1946 Pelichet, E. "A propos des amphores romaines trouvees a Nyon." Zeitschrifl fir schweizerische Archaeologie

REFERENCES AND ABBREVIATIONS and Kunstgeschichte (Revue suisse d'art et d'archeologie) 8 (1946) pp. 189-202. Peman 1959 Peman, C. "Alfares y embarcaderos romanos en la provincia de Cadiz." ArchEspArq 32 (1959) pp. 169173. Pemp Pemp, R. Drei Wasserhebewerke Pompejis. WurzburgAumiihle, 1939. Poidebard, Tyr Poidebard, A. Un grand port disparu: Tyr. Recherches aeriennes et sous-marines, 1934-6. Bibliotheque archeologique et historique 29. Paris, 1939. Poidebard and Lauffray Poidebard, A., and Lauffray, J. Sidon. Amenagements antiques du port de Saida. Beirut, 1951. Pongratz Pongratz, E. "Historische Bauwerke als Indikatoren fur kiistenmorphologische Veranderungen (Abrasion und Meeresspiegelschwankungen) in Latium—FeIdbegehung und Luftbildauswertung." MiinchGeogAbh 4 (1972) pp. 1-144, including maps and charts. Ponsich 1974, 1979 Ponsich, M. Implantation rurale antique sur Ie bas-Guadalquivir. 2 vols. Madrid: Casa de Velazquez, 1974, 1979. Ponsich and Tarradell Ponsich, M., and Tarradell, M. Garum et industries antiques des salaison dans la Mediterranee occidentale. Paris, 1965. ProcAmerPhilSoc Proceedings of the American Philosophical Society. Purpura 1975 Purpura, G. "Alcuni rinvenimenti sottomarini lungo Ie coste della Sicilia Nord-Occidentale." SUA 8:28-29 (1975) pp. 57-84. QDAP Quarterly of the Department of Antiquities in Palestine. RA Revue archeologique. Radcliffe Radcliffe, W. Fishing from the Earliest Times. Chicago, 1974. RAECE Revue archeologique de I'Est et du Centre-Est. RAjr Revue africaine. Rathbone 1981 Rathbone, D. W. "The Development of Agriculture in the 'Ager Cosanus' during the Roman Republic: Problems of Evidence and Interpretation." JRS 71 (1981) pp. 10-23. Raveggi Raveggi, P. "Orbetello antica e moderna." Maremma 8, fasc. 1-2 (1933) pp. 26-96. RCRFA Ret Cretariae Romanae Fautorum Acta. RE Pauly, A.; Wissowa, G.; and Kroll, W. Real-Encyclo-

XXVIl

padie der klassischen Altertumswissenschafi. Stuttgart, 1893-. REA Revue des etudes anciennes. Remesal Rodriguez 1977-1978 Remesal Rodriguez, J. "La economia oleicola betica: nuevas formas de analisis." ArchEspArq 50-51 (19771978) pp. 87-142. Remesal Rodriguez 1980 Remesal Rodriguez, J. "Reflejos economicos y sociales en la produccion de anforas olearias beticas (Dressel-20)." Blazquez Martinez 1980, pp. 131-153. RendNap Rendiconti della R. Accademia di Archeologia, Lettere ed Arti, Naples. RendPontAu Atti della Pontificia Accademia Romana di Archeologia, Rendiconti. RevPhil Revue de philologie, de litterature et d'histoire anciennes. RGZM Romische-germanische Zentralmuseum, Mainz, Jahrbuch. Rickman, StHarbArch I Rickman, G. E. "Towards a Study of Roman Ports." StHarbArch I, forthcoming. Risanamento Risanamento e protezione dell'ambtente idrobiologico delle lagune di Orbetello. Orbetello, 1977. RivIstArch Rivista del R. Istituto d'Archeologia e Storia dell'Arte. Robinson 1959 Robinson, H. S. Pottery of the Roman Period. The Athenian Agora 5. Princeton: American School of Classical Studies at Athens, 1959. Rodenwaldt and Lehmann Rodenwaldt, E., and Lehmann, H. "Die antiken Emissare von Cosa-Ansedonia, ein Beitrag zur Frage der Entwasserung der Maremmen in etruskischer Zeit." Sitzungsberichte der Heidelberger Akademie der Wissenschaflen 1(1962) pp. 3-31. Rodriguez-Almeida 1972 Rodriguez-Almeida, E. "Novedades de epigrafia anforaria del monte Testaccio." Baldacci, P., et al. Recherches sur les amphores romaines. Rome: Ecole franqaise de Rome, 1972, pp. 107-241. Rodriguez-Almeida, 1974-1975 [1977], 1978-1979 [1981] Rodriguez-Almeida, E. "BoIh anforari di Monte Testaccio." BullComm 84 (1974-1975 [1977]), pp. 199248; 86 (1978-1979 [1981]), pp. 109-135. Rodriguez-Almeida 1980-1 Rodriguez-Almeida, E. "El monte Testaccio, hoy: Nuevos testimonios epigraficos," and "Alcuni aspetti della topografia e dell'archeologia attorno al monte Testaccio." Blazquez Martinez 1980, pp. 57-130. Rodriguez-Almeida 1980-2 Rodriguez-Almeida, E. "Vicissitudini nella gestione del commercio dell'olio betico da Vespasiano a Severo Alessandro." MAAR 36 (1980) pp. 277-290. Roman and Rancoule 1977 Roman, Y., and Rancoule, G. "Les amphores de Ses-

XXVlIl

REFERENCES AND ABBREVIATIONS

tius de La Lagaste (Aude) et de sa region." Revue archeologique de Narbonnaise 10 (1977) pp. 247-261. RomMitt Mitteilungen des deutschen archdologischen Instituts, Rotnische Abteilung. Rostovtzeff, SEHRE Rostovtzeff, M. The Social and Economic History of the Roman Empire. 2 vols. Oxford, 1957. Rouge Rouge, J. Recherches sur ['organisation du commerce maritime en Mediterranee sous {'empire remain. Paris, 1966. RStUg Rivista di Studi Liguri. Salmon Salmon, E. T. Roman Colonization under the Republic. London, 1969. Santangelo Santangelo, M. L'Antiquarium di Orbetello. Rome, 1954. Schafer Schafer, J. "Zur Erforschung antiker Hafenanlagen." In Mansel'e Armagan. Melanges Mansel, Ankara, 1974, pp. 663-678. Schioler Schioler, T. Roman and Islamic Water-Lifting Wheels. Odense, 1973. Schmiedt, Atlante Schmiedt, G. Atlante aerofotografico delle sedi umane in Italia. Parte seconda, Ie sedi antiche scomparse. Florence, 1970. Schmiedt, Il livello antico Schmiedt, G. Il livello antico del mar tirreno. Testimonianze dei resti archeologici. Florence, 1972. Scranton, Kenchreai I Scranton, R.; Shaw, J. W.; and Ibrahim, L. Kenchreai. Eastern Port of Corinth. I. Topography and Architecture. Leiden, 1978. Scullard, Etruscan Cities Scullard, H. H. The Etruscan Cities and Rome. Ithaca, N. Y., 1967. Scullard, History Scullard, H. H. A History of the Roman Worldfrom753 to 146 B.C. London, 1972. Shaw Shaw, J. W. "Greek and Roman Harbourworks." Bass, History, pp. 88-112. SkA Sicilia archeologica. Stenico Stenico, A. "Ceramica arretina a rilievo e terra sigillata tardo-italica." .RCRFA 2 (1959) pp. 51-62, figs. 1-4. StEtr Studi Etruschi. StHarbArch I Studies in Harbour Archaeology, Proceedings of the First International Workshop on Mediterranean Harbours of Antiquity. Center for Maritime Studies, University of Haifa and the Caesarea Ancient Harbor Excavation Project, June 24-28, 1983. Ed. A. Raban, BAR 257 (1985).

StMisc Studi Miscellanei, Seminario di Archeologia e Storia dell'Arte Greca e Romana dell'Universita di Roma. StRom Studi Romani. Sutri I Duncan, G. C. "A Roman Pottery Near Sutri." BSR 32 (1964) pp. 38-88. Sutri II Duncan, G. C. "Roman Republican Pottery from the Vicinity of Sutri." BSR 33 (1965) pp. 134-176. Szentleleky Szentleleky, T. Ancient Lamps. Budapest, 1949. TAPS Transactions of the American Philosophical Society. D. M. Taylor Taylor, D. M. "Cosa: Black-Glaze Pottery." MAAR 25 (1957) pp. 65-193. L. R. Taylor Taylor, L. R. Local Cults in Etruria. PAAR 2 (1923). Tchernia 1968-1970 Tchernia, A. "Premiers resultats des fouilles de juin 1968 sur l'epave 3 de Planier." Etudes classiques 3 (1968-1970) pp. 51-82. Tchernia 1971 Tchernia, A. "Les amphores vinaires de Tarraconaise et leur exportation au debut de l'Empire." ArchEspArq 44 (1971) pp. 38-85. Tchernia and Zevi 1972 Tchernia, A., and Zevi, F. "Amphores vinaires de Campanie et de Tarraconaise a Ostie." Baldacci, P., et al. Recherches sur les amphores romaines. Rome: Ecole frangaise de Rome, 1972, pp. 35-67. Tchernia and Villa 1977 Tchernia, A., and Villa, J. P. "Note sur Ie materiel recueilli dans la fouille d'un atelier d'amphores a Velaux (Bouches-du-Rhone)." Methodes classiques et methodes formelles dans I'etude des amphores (Actes du colloque de Rome, 27-29 mat 1974). Rome: Ecole frangaise de Rome, 1977, pp. 231-239. Tchernia 1978 Tchernia, A., et al. L'epave romaine de la Madrague de Giens (Var). Gallia, supp. 34. Paris, 1978. Tchernia 1980 Tchernia, A. "Quelques remarques sur Ie commerce du vin et les amphores." MAAR 36 (1980) pp. 305312. Tchernia 1983 Tchernia, A. "Italian Wine in Gaul at the End of the Republic." In Garnsey, Hopkins, and Whittaker, pp. 87-104. TechCult Technology and Culture, Chicago. Testaguzza Testaguzza, O. Portus. Rome, 1970. Thevenot 1954 Thevenot, E. "La marque d'amphore 'Sesti.' " RAECE 5 (1954) pp. 234-243. Thiel Thiel, J. H. A History of Roman Sea-Power Before the Second Punic War. Amsterdam, 1954.

REFERENCES AND ABBREVIATIONS ThLL Thesaurus Linguae Latinae. Toynbee Toynbee, A. J. Hannibal's Legacy. London, 1965. TrZ Trierer Zeitschrifi. Uenze 1958 Uenze, O. Friihromische Amphoren als Zeitmarken im Spatlatene. Marburg-Lahn, 1958. Vegas 1965 Vegas, M. "Spatkaiserzeitliche Keramik aus Pollentia (Mallorca, Spanien)." Bonnjbb 165 (1965) pp. 108-140. Vegas 1973 Vegas, M. Cerdmica comun romana del mediterraneo occidental. Barcelona, 1973. Vegas 1975 Vegas, M. "Die augustische Gebrauchskeramik von Neuss." Novaesium VI Limesforschungen 14. Berlin, 1975. Venerosi-Pesciolini Venerosi-Pesciolini, G. "Una memoria del secolo XVI sul padule di Ansedonia." Maremma 6 (1931) pp. 15-28. VjesDal Vjesnik za Arheologiju i Historiju Dalmatinsku. Von Vacano 1961 Von Vacano, O. W. "Odipus zwischen den Viergespannen." RomMitt 68 (1961) pp. 9-63. Von Vacano 1969 Von Vacano, O. W. "Die Figurenanordnung im Giebelrelief von Telamon." RomMitt 76 (1969) pp. 141161. Widemann et al. 1978 Widemann, R, et al. "Analytical and Typological Study of Gallo-Roman Workshops Producing Amphorae in the Area of Narbonne." Archaeo-Physika 10 (1978) pp. 317-340. Will 1956 Will, E. L. "Les amphores de Sestius." RAECE 7 (1956) pp. 224-244. Will 1962 Will, E. L. "Latin-stamped Amphoras in the Eastern Mediterranean Area." Year Book of the American Philosophical Society (1962) pp. 647-650. Will 1970 Will, E. L. "Les timbres amphoriques latins." Bruneau, P., et al. L'llot de la Maison des Comediens. Exploration archeologique de Delos 27. Paris, 1970, pp. 383-386. Will 1977 Will, E. L. "The Ancient Commercial Amphora." Archaeology 30 (1977) pp. 264-270. Will 1979 Will, E. L. "The Sestius Amphoras: A Reappraisal." JFA 6 (1979) pp. 339-350. Will 1982-1 Will, E. L. "Greco-Italic Amphoras." Hesperia 51 (1982) pp. 338-356. Will 1982-2 Will, E. L. "Ambiguity in Horace, Odes 1.4." Classical Philology 77 (1982) pp. 240-245.

XXlX

Will 1982-3 Will, E. L. "Diversification in Roman Industry: Some Ceramic Evidence." Archaeological Institute of America Abstracts 7 (1982) p. 46 (AJA 87, 1983, p. 269). Will 1983-1 Will, E. L. "Exportation of Olive Oil from Baetica to the Eastern Mediterranean." Blazquez Martinez, J. M., et al. Production y comercio del aceite en la antiguedad (II Congreso International). Madrid: Universidad Complutense, 1983, pp. 391-440. Will 1983-2 Will, E. L. "The Spargi Wreck: A Reconsideration." Archaeological Institute of America Abstracts 8 (1983) p. 32 (AJA 88, 1984, p. 264). Will 1984 Will, E. L. "Ahnlichkeiten zwischen Stempeln auf Amphoren und auf arretinischen Gefassen." RCRFA 23, 24 (1984), pp. 9-11. Williams Williams, P. F. de Coetlogon. "Roman Harbours." IJNA 5 (1976) pp. 73-79. Zancani Montuoro Zancani Montuoro, P., and Brown, F. E. "Il faro di Cosa in ex-voto a Vulci?" RivIstArch 3, series 2 (1979) pp. 5-29. Zevi 1966 Zevi, F. "Appunti sulle anfore romane." ArchCl 18 (1966) p. 208-247. Zevi 1967 Zevi, F. "Anfore istriane ad Ostia (nota sul commercio istriano)." Atti e memorie della societa istriana di archeologia e storia patria 15 (1967) p. 21-31. Zevi and Tchernia 1969 Zevi, F., and Tchernia, A. "Amphores de Byzacene au bas-Empire." AntAfr 3 (1969) p. 173-214.

ABBREVIATIONS OF CATALOGUE G R O U P S

A AD BG C CO D F G H K L LT M MC PC RG S T TW W

Roman Amphoras Architectural Decoration Black Glaze Coarse Ware Coins Domestic Ware Faunal Remains Glass Human Bones Kitchen Ware Lamps Lagoon Temple Decoration Metal Miscellaneous Ceramics Portus Cosanus Red-Gloss Pottery Shell Remains Tortoise Remains Roman Thin Walled Pottery Wood

REFERENCES AND ABBREVIATIONS OTHER ABBREVUTIONS USED IN CATALOGUES

ATh a.s.l. b.s.l. CF D FC FG H L LS MPD

Average Thickness above sea level below sea level Capitolium Fill Group, town oi Cosa excavations Diameter Forum Cistern Group, town of Cosa excavations Forum Gate Group, town of Cosa excavations Height Length Late Shops Group, town of Cosa excavations Maximum Preserved Diameter

MPH MPL MPW MTh PC PD SH Th W V-D 16-IV 22-11

Maximum Preserved Height Maximum Preserved Length Maximum Preserved Width Maximum Thickness Portus Cosanus Pottery Dump Group, town of Cosa excavations Spring House, Cosa lagoon excavations Thickness Width Quadrant V-D Group, town of Cosa excavations Room 16, Level IV, of Atrium Publicum (Group), town of Cosa excavations Room 22, Level II beside Atrium Publicum (Group), town of Cosa excavations

Preface

The purpose of the collaborative team of archaeologists, scientists, and engineers in writing this book is to present a complete picture of the archaeological remains thus far discovered in the early Roman port and lagoon fishery of Cosa (Italy), in the context of the port's ancient landscape and of the trade and industry that gave it life from the third century B.C. until the third century A.D. This study also supplements the history of the hill town of Cosa, under investigation since 1948 by the American Academy in Rome through excavations directed by Frank E. Brown. As the center of the commercial life of the area during the peak of its prosperity in the later second and first centuries B.C., the Portus Cosanus is further linked to the wider history of the Ager Cosanus, an area joined to Rome some 140 kilometers to the east by the Via Aurelia. The recent surveys of the Ager Cosanus and the excavations of the villa at Le Colonne by Stephen L. Dyson have established both the changing fortunes of the colonial farms and the importance of this scenic area for the wealthier class, which built large villa estates there in the late Republic and early Empire. In addition, the excavation of the villa at Sette Finestre by Italian and English archaeologists under the direction of Andrea Carandini will allow a fuller evaluation of the role that this beautiful maritime region of modern Tuscany, known as the Maremma, played in the chapters of earlier Roman history. The port of Cosa first came to my attention when I participated in the land excavations at the town site of Cosa in the spring of 1965. The need for an underwater investigation of the extensive, visible archaeological remains, as well as for a further land survey, was apparent. This was undertaken with General John David Lewis, U.S. Army, in the fall of 1965. Excavations both on land and under water were begun in the port and ancient lagoon area in the summer of 1968 and continued in 1969, 1970, and 1972. Since that time, smaller research teams of both archaeologists and scientists have continued to work at the site every summer. In 1970, 1971, and 1972 the Lerici Foundation in Rome, directed by Richard E. Linington, made a magnetic and drilling

survey of the now silted ancient lagoon area. In 1978 further cores were taken in connection with the mineralogical study of the amphora. In 1975 John Peter Oleson of the University of Victoria, British Columbia, re-excavated and restored some of the wooden water-lifting machinery discovered in 1972. The underwater excavations, directed by Lewis, took place largely in 1968 and 1969, with the assistance of Robert L. Hohlfelder, University of Colorado; Ross Young, University of Missouri; Stanford Low; Commander William Robinson, U.S. Navy; and Scott Guthrie. Student assistants in the land excavations of 1969 were Carl Berkowitz and Loeta Smith from the University of Missouri. Participating in the extensive mechanical excavation of the lagoon fishery in 1972 were: Elaine K. Gazda, University of Michigan; John P. Oleson and Martha I. Oleson; Robert D. Taggart; and Giovanni Uggeri, University of Florence. This difficult work could not have been accomplished without the collaboration of my colleague Giovanni Uggeri and the skill, loyalty, and hard work of the team of Italian workmen, so ably supervised by Ivo Piccolini. The main architectural plans, originally executed by David van Zanten, Northwestern University, have been revised for publication by J. F. Warren of I. M. Pei and Partners and John Stubbs of Beyer, Blinder and Belle with assistance from Mark Stankard, Cornell University. The reconstruction drawings of the Spring House in its various phases were also executed by Warren, who made additional maps and supervised the architectural presentation as a whole. The theoretical plans and sections of the site in its three architectural periods were drawn by Warren and drafted by Stankard. A reconstruction drawing of the main phase was executed by Kathleen K. Borowick. The plan of the Tagliata was made by Gazda, as were the section drawings of the lagoon trenches made during the excavations of 1972 (drafted by Stubbs), with the collaboration of geologist Joanne Bourgeois of the University of Washington. The section drawings from the 1968 and 1969 excavations, and the drawings of most of the finds, were done by Elizabeth F. Solomon. Addi-

XXXIl

PREFACE

tional drawings of finds have been contributed by Borowick, Mareile Fenner, Gazda, and Martha Joukowsky of Brown University. Elizabeth Lyding Will, University of Massachusetts, Amherst, undertook the photography and drawings of the amphora stamps. The photography both on land and underwater was done by this writer, and Barbara Bini of Rome photographed the finds. Aerial photographs were taken by Julian and Eunice Whittlesey. Further photographic help was provided by Philip Evola. Scientific studies have gone hand in hand with the archaeological research. In 1972 Robert Raikes and Desmond P. Carr, hydrologists from Raikes and Partners, Rome, studied the sea level and geomorphology of the site. An extensive geological survey of the area and a sedimentological study of the ancient lagoon were made by Bourgeois in 1974, coinciding with Gazda's research on the materials and techniques of construction. Further studies of the lagoon soils were conducted by Richard Linington, and other analyses were made by Isaac Barshad with Brian Viani of the University of California, Berkeley. The study of the human bone material has been contributed by J. Lawrence Angel of the Smithsonian Institution; that of the faunal material is by Brian Hesse of the University of Alabama, Birmingham, and Paula Wapnish of the Smithsonian Institution. The tortoise and seashell remains have been identified by Charles R. Crumly of the Smithsonian Institution and by William Old of the American Museum of Natural History. Raffaello Trigila and Domenico Cozzupoli of the University of Rome have provided a petrographical report on the tuff samples and a mineralogical analysis of some amphora sherds and clay deposits in the Cosa area to determine provenance of these materials. Jelle de Boer of Wesleyan University has contributed further mineralogical analyses of the amphoras and local clays and also kindly shared with us the unpublished results of his geological surveys of the Cosa area for the Wesleyan University archaeological teams. Bourgeois has acted as editor of all the scientific reports. The archaeological study of the vast amphora material was begun by Will in 1974 and 1975, with an additional summer at the site in 1978. The study of the glass was made in 1976 by David Frederick Grose, University of Massachusetts, Amherst, in conjunction with his study of the glass from the town site of Cosa. Lionel Casson has served as consultant to the project, generously sharing with the team over the years his wealth of knowledge about ancient ships and navigation. Luc Long, conservateur, Direction des Recherches Archeologiques

Sous-Marine, Marseilles, has kindly contributed a prepublication summary of his research on the two shipwrecks found off the Grand Congloue rock near Marseilles. Material from the Portus Cosanus excavations will be exhibited in the Cosa Museum within the town site of Ansedonia, now opened by the American Academy in Rome and under the direction of the Superintendent of Antiquities of Etruria, Dr. Francesco Nicosia. The project could not have been accomplished without substantial financial support. The foundations and institutions that have provided the bulk of our funds are the National Endownment for the Humanities, the Atlantic Foundation, the Kress Foundation, the University of Missouri, and the American Council of Learned Societies. The Lerici Foundation furnished a magnetic and drilling survey, and the University of Pennsylvania, with Elizabeth Ralph, extended magnetic and resistivity coverage. The Whittlesey Foundation contributed aerial photographic coverage. Additional funds for Oleson's research were provided by Florida State University and the University of Victoria, British Columbia. The Forsyth Fund, University of Michigan, aided Gazda's studies. Among the individual benefactors of the project whom I would like particularly to thank are Rensselaer Lee, Richard H. McCann, Robert D. Taggart, and Robert Walker. Our work would not have been possible without the generosity of the Venturini family, upon whose land the port and fishery lie. For permission to undertake the excavation we gratefully acknowledge the Directorate General of Archaeology and Fine Arts of Italy and the then Superintendent of Antiquities of Etruria, Guglielmo Maetzke. Finally, I would like to express my thanks to the American Academy in Rome, under whose auspices the work has been accomplished, and in particular to Professor Frank E. Brown for his original encouragement and interest. To one and all who have so generously and devotedly contributed their various talents, time and money, it is impossible for me to express adequately my appreciation. In my research on the pottery, Howard Comfort, Cleo R. Fitch, Maria Teresa Marabini Moevs, Ann Reynolds Scott, and Russell T. Scott have generously shared with me their expert knowledge. In the preparation of the manuscript I am grateful to Lorin A. Driggs for her careful typing and to Margaret Aspinwall, and Douglas Preston of the American Museum of Natural History for their fine editorial help. I am again deeply indebted to the

PREFACE

National Endowment for the Humanities for an additional research grant for the final preparation of the manuscript, architectural drawings, and photographs. To Princeton University Press, whose early contract agreement encouraged and stimulated us to finish, we owe our very special thanks. We also wish to thank warmly the American Academy in Rome and the University of Michigan, Horace H. Rackham School of Graduate Studies, for supplementary funding for the publication. The manuscript was completed in December 1982. Bibliography after that date does not enter the text discussion, although material to June 1984 is included in footnotes.

XXXlU

Without the continued support of a wonderful husband, Robert D. Taggart, and my immediate family, Mr. and Mrs. Richard H. McCann and Mrs. Dorothy McCann Preston and her family, who have sustained me through long years of work and research, this book could not have been completed. We hope that our many friends who have wished the project well over the years will be rewarded by its outcome. Anna Marguerite McCann Director and Editor, Cosa Port Excavations

T H E R O M A N P O R T A N D F I S H E R Y OF C O S A

Introduction: A Perspective ANNA MARGUERITE M C C A N N

A history of ancient harbors has yet to be written. The only comprehensive study, by Karl Lehmann in 1923,' predated the aqualung and underwater archaeology and relied largely upon the ancient literary sources combined with travelers' reports of the then visible surface remains. Lehmann catalogued some three hundred harbor sites in the Greek and Roman world, but at that time only a very few harbor sites had been studied by archaeologists or excavated. Nevertheless, Lehmann's pioneering book established ancient ports as a distinct and significant area for scholarly research and recognized the evolution of harbor design as a major contribution of ancient architecture. Since the 1920s the accelerated pace of scientific advances has brought to the archaeologist vast new areas of knowledge for potential

use, as well as new tools and techniques for research. The development in 1942 of SCUBA—the SeIfContained Underwater Breathing Apparatus—by Jacques-Yves Cousteau, Frederic Dumas, and Emile Gagnan opened the sea to exploration. The sea's abundance of natural resources as well as archaeological remains has consequently become an international concern, requiring new organizations and laws for its supervision and protection. 2 The first international congress on underwater archaeology, organized by the late Professor Nino Lamboglia of Italy,3 was held in Cannes in 1955, and others have followed.4 An International Journal of Nautical Archaeology appeared in 1972, and now several centers offer graduate-level study of marine archaeology.5 This field has emerged as a serious and specialized area

1 Lehmann-Hartleben. Full citations for shortened references listed throughout this book are given in Frequently Cited References and Abbreviations. 2 For example, see J. D. Lewis, "The Deep Sea Resources," Naval War College Revue, June 1969, pp. 130-151, with further bibliography. Also note the formation of the United Nations Law of the Sea Conference m 1973, composed of representatives from about 150 nations (New York Times, March 9, 1981, p. Al; March 19, 1981, p. A23). On results of the convention in Jamaica, see E. M. Borgese, "The Law of the Sea," Scientific American 248 (1983) pp. 42-49. For archaeology, see F. Pallares, "20 anni di archeologia sottomarina in Liguna e nel Mediterraneo," Liguria 10 (1970) pp. 3-10; Pallares, "L'archeologia sottomarina in Italia e nel Mediterraneo," Il Mare, Istituto geografico De Agostini di Novara (1971) pp. 3-30; G. F. Bass, Archaeology Under Water (New York, 1966), reviewed by A. M. McCann in Archaeology 21 (1968) pp. 72-73. 3 Club Alpin sous-marine, Rapport du Premier Congres International d'Archeologie Sous-Marine (Cannes, 1955); Du Plat Taylor, pp. 190-201. 4 Actes du II' Congres International d'Archeologie Sous-Marine, Albenga, 1958 (Bordighera, 1961); AtH del IH Congresso Internazionale di Archeologia Sottomarina, Barcellona 1961 (Bordighera, 1971). The fourth congress, held in Nice in 1970 (A. M. McCann, "IVth International Congress of Underwater Archeology," Archaeology 24 [1971] pp. 273-274), and the fifth at Lipari in 1976 unfortunately remain unpublished. The sixth congress, which took place in April 1982, at Cartagena, Spain, also awaits publication. Professor Lambogha's untimely death in 1977 was a major loss to this field of research. See A. M. McCann, "Necrology: Professor Nino Lamboglia," JFA 4 (1977) pp. 264-265. In the U.S. the late John Huston, through the Council of Underwater Archaeology

(which he formed), also made a major contribution to this field by bringing together underwater archaeologists, scientists, and divers at a number of international conferences in the 1960s and early 1970s. See A. M. McCann, "Underwater Archaeology Conference in Miami," Archaeology 20 (1967) pp. 302-303. Other conferences have also taken place; for example, see Marine Archaeology. Both Huston and Lamboglia had a vision of an international organization and of cooperation among countries and specialists; since their deaths no one organization for underwater archaeology has emerged to pursue these goals. In the U.S. underwater research is reported at the meetings of both the Archaeological Institute of America and the Society for Historical Archaeology. 5 Graduate work in the field of underwater archaeology as a separate discipline only just began in the 1970s. The Institute of Nautical Archaeology at Texas A & M University, founded and directed by G. F. Bass, offers a master's degree in marine archaeology. The Institute's field work concentrates largely on excavations of ancient shipwrecks. For their most recent work see: G. F. Bass, "The Shipwreck at Serge Liman, Turkey," Archaeology 32 (1979) pp. 36-43; F. H. van Doorninck, "An 11th Century Shipwreck at Serge Liman, Turkey: 1978-1981," IJNA 11 (1982) pp. 7-11; J. R. Steffy, "The Reconstruction of the 11th Century Serge Liman Vessel," IJNA 11 (1982) pp. 13-34; G. F. Bass and F. H. van Doorninck et al., Yassi Ada: A Seventh-Century Byzantine Shipwreck (College Station, Texas, 1982). A new graduate program has been set up at East Carolina University in Greenville, North Carolina, by G. Watts to study American maritime history and underwater archaeology. In France, the Institut d'Archeologie Mediterraneene at Aix-en-Provence has been active over the last several decades in fieldwork. For some of their latest results excavating a Roman shipwreck, see Tchernia 1978. An

4

I N T R O D U C T I O N : A PERSPECTIVE

for research, involving the established methods of scholarship but also requiring additional scientific knowledge and techniques for the excavation, recovery, and preservation of artifacts from their aquatic environment. In these pioneering days of underwater exploration, archaeologists have focused largely upon ancient shipwrecks, often brought to their attention by chance finds of local fishermen.6 The lure of the earliest ancient ship or a sunken cargo of bronze art treasures is always there, tempting all underwater archaeologists. Furthermore, diving on shipwrecks is usually in deep, clear, clean water—both glamorous and dangerous. The public media have also been fascinated by these discoveries and lent their financial support. On the other hand, diving in ancient harbors, often in shallow, polluted water with poor visibility, is less alluring. A port and its artifacts, which can represent many centuries rather than a small time period, require different methods and techniques of excavation. But it is only through the study of harbor sites that we can hope to obtain certain kinds of new information about the ancient world. For example, from the historical point of view, some questions still to be answered are: Where were the ancient ports and why were they located where they were? What role did they play in war

and in peace? What products were traded? What made a port economically successful? Who built them and controlled them? Our understanding of ancient technology also remains far from complete. An evaluation of the Roman contribution to port engineering and hydraulics is essential. How did the Romans, the most sophisticated of all ancient engineers, resolve the problem of function and water circulation so necessary for the prevention of sedimentation? What role did Roman military engineers have in these solutions?7 When and how did they come to use hydraulic concrete made with pozzolana mortar, which sets and endures in salt water?8 The discovery of this revolutionary building material was one of the major breakthroughs for modern technology. Without it, no modern harbor could have been constructed. For the field of economic history, port studies, with their wealth of amphora remains and other artifacts of trade, constantly reveal new evidence about the importing and exporting of goods and about common ancient business practices.9 Urban Rome in the late Republic and Empire was completely dependent upon ports and ships to supply food for its enormous populace. For the history of ideas, it is often to ports that we must turn to trace the first seeds of new influences. For example, many of the Eastern mystery religions, brought

Institute of Maritime Archaeology has also been founded at the University of St. Andrews, Fife, Scotland. In Israel, a Center for Maritime Studies at the University of Haifa was founded in 1972 by E. Linder, now directed by A. Raban. Extensive research projects are in progress at several harbor sites in Israel, including Atlit, Akka, Dor, and now Caesarea. See E. Lmder and O. Laenhardt, "Recherches d'archeologie sous-marine sur la cote mediterraneene d'Israel," RA (1964) pp. 47ff.; E. Linder, "La ville phenicienne d'Athlit-a-t-elle eu l'un des plus anciens ports artificiels de Mediterranee?" Archeologia 17 (1967) pp. 25-29; A. Raban and R. L. Hohlfelder, "The Ancient Harbors of Caesarea Maritima," Archaeology 34 (1981) pp. 56-60. Further references for Caesarea cited in n. 67 below. Several museums for underwater finds, largely from shipwrecks, now exist in the Mediterranean countries. Among them are the museums at Marseilles, Antibes, Albenga, Cartagena, Kyrenia (Cyprus), Haifa, and most recently, Caesarea. The Anthropology Museum in Monaco has a special section on underwater finds, and the Bardo Museum in Tunis houses the great finds from the Mahdia shipwreck. For a recent summary of some of the scholarly developments in underwater archaeology, see G. F. Bass, "Marine Archaeology: A Misunderstood Science," Ocean Yearbook 2 (1980) pp. 137-152. 6 For example, see G. F. Bass et al., "Cape Gehdonya: A Bronze Age Shipwreck," TAPS 57, no. 8 (1967) pp. 1-177, reviewed by A. M. McCann in AJA 74 (1970) pp. 105-106; M. L. Katzev, "Resurrecting the Oldest Known Greek Ship," National Geographic 137 (June 1970) pp. 840-857; George F. Bass, Donald A. Frey, and Cemal Pulak, "A Late Bronze Age Shipwreck at Kas, Turkey," IJNA 13 (1984) pp. 272-279. For additional bibli-

ography see Bass, History; Casson, SSA W; Du Plat Taylor. 7 On port engineering and hydraulics, see: F. Pellati, "L'ingegneria idraulica ai tempi dell' impero," Quademi dell' impero. La scienza e la tecnica ai tempi di Roma imperiale 12 (1940) pp. 3-22; Blake 1947; Blake 1959; R. S. Rowe, Bibliography of Rivers and Harbors and Related Fields in Hydraulic Engineering (Princeton, 1953); R. G. Goodchild, "Harbors, Docks and Lighthouses," in C. Singer et al., A History of Technology (Oxford, 1956) II, pp. 493ff.; R. J. Forbes, Studies in Ancient Technology, vol. 1 (Leiden, 1964); Lugli; H. F. Cornick, Dock and Harbour Engineering, vol. IV: Construction (London, 1962); P. Brunn, Port Engineering (Houston, 1973); J. G. Landels, Engineering in the Ancient World (London, 1978). For further works on ancient technology and hydraulics, see the references cited in chapter V by J. P. Oleson. For the most recent survey of ancient harbors, including their technology, see Blackman 1982-1, which contains an excellent bibliography, and Blackman 1982-2. These articles were published after this book was in production. Also see the information on ancient harbors in Muckelroy 1980, pp. 166-177. Muckelroy suggests that Rome had a corps of harbor engineers who were sent out around the Mediterranean. Such an organized system would explain standardized building techniques and rapid construction of harbor works. Muckelroy assumes that there must have been "training establishments" and "manuels of engineering methods" for harbor engineers (p. 176). For an excellent survey of ports, both ancient and modern from a geographer's point of view, see Karmon. 8 See Gazda in chapter seven and McCann, chapter eighteen. 9 For summary of recent developments and literature, see Will 1977-1, pp. 264-270, and further references in chapter nine.

I N T R O D U C T I O N : A PERSPECTIVE

5

by sailors and travelers to Italy, had their first life in the port communities and then spread throughout the Empire to lay the foundations for the Christian era.10 Thus, ancient harbors and the life surrounding them lie at the very heart of ancient civilizations, particularly the Roman one that is our focus here. Their study constitutes what scientists term "basic research,"11 for information about them is needed to fill the gaps that still exist in our understanding of this our classical heritage in both the traditional and recent sense of the term. Harbor research has suffered from historical neglect owing both to the lack of necessary technology and to the preoccupation with the study of cities and what were considered major artistic monuments. The major studies on Roman architecture do not yet even include ports as an architectural category.12 But now we are in a position to correct this situation with the help of the sciences. Indeed, it is the obligation of the serious archaeologist to do so. Since this book was completed, an important step forward for harbor studies has occurred. The first international workshop on "Mediterranean Harbours of Antiquity" took place in June 1983, attended by underwater archaeologists, ancient historians, geographers, and geologists. Hosted by the Center for Maritime Studies of the University of Haifa and the Caesarea Ancient Har-

bor Excavation Project in Israel, the conference affirmed ancient ports as a distinctive subject for interdisciplinary research. More workshops are planned with the hope that port studies will now receive and profit from the attention they deserve. In the fifty years since Lehmann's first publication of the subject, harbor archaeologists have learned that what appeared simple in neat, clear designs is actually much more complex. It also follows that what is more complex is probably all the more worth researching. Theoretical plans can differ radically from the actual archaeological remains found underwater. Furthermore, geologists and geomorphologists can now tell us much more about what causes changes in sea level over the millennia and in the retreat or extension of coastlines—essential aspects of any harbor study. Because of Italy's predominantly sandy shoreline, sedimentation is a frequent problem in both ancient and modern harbors. Analysis of sediments by soil specialists (pedologists) and interpretation by sedimentologists can yield information often not recorded by archaeological data.13 Today, advanced equipment can be used in archaeological and geophysical prospecting: drilling equipment for obtaining deep soil samples, magnetometers, and resistivity, sonar, and radar instruments. 14 A broader and deeper interpretation of any harbor or land site may now be obtained by tapping

10 For example, note the history and spread of the popular Isis cult through port communities; J. Ferguson, The Religions of the Roman Empire (Ithaca, N.Y., 1970) pp. 23-26; R. E. Witt, Isis in the Graeco-Roman World (Ithaca, N.Y., 1971); F. Dunard, "Le culte d'Isis dans Ie bassin oriental de la Mediterranee," Etudes priliminaires aux religions orientales dans I'empire romaine 26 (Leiden, 1973); F. Solmsen, Isis Among the Greeks and Romans (Cambridge, Mass., 1979). For the exciting discovery of an Isis sanctuary at Kenchreai, see Scranton, Kenchreai I, pp. 53-90; L. Ibrahim, R. Scranton, and R. Brill, Kenchreai: Eastern Port of Corinth, vol. II: The Panels of Opus Sectile in Glass (Leiden, 1976). At Kenchreai remains have been found of a Christian shrine as well, reminding us that Christianity in its early days spread first into port communities, for example, Corinth, Athens, Ephesus, Nea Paphos, Caesarea. See Paul's letters in the New Testament, particularly 1 Corinthians and 2 Corinthians, Ephesians, and Acts of the Apostles. For Paul's departure from Kenchreai see Acts 18:18 and for the church there, Romans 16:1. (Cf. O.F.A. Meinardus, St. Paul in Greece [Athens, 1973].) Already in the sixties Paul found Christians at Pozzuoli (Acts 28:13-14). 11 For example, see Lewis Thomas, "The Planning of Science," in Thomas, The Lives of a Cell (New York, 1974) pp. 134-140. 12 For example, Boethius and Ward-Perkins. See forthcoming, StHarbArch I, proceedings of conference on ancient harbors in Caesarea, Israel, 1983. 13 On sea level changes, see, for example, Fairbridge, "Quaternary Sedimentation"; Flemming 1969; Pongratz, pp. 1-144; D. J. Blackman, "Evidence of Sea Level Change in Ancient Harbors and Coastal Installations," in Marine Archaeology, pp. 115-137; C. Delano Smith, "Coastal Sedimentation, Lagoons and Ports in Italy," Papers in Italian Archaeology, vol. I: The Lancaster Seminar,

British Archaeological Reports, supp. series 41, i (Oxford, 1978) pp. 25-33. For discussion of sea-level change at Cosa with further references see Bourgeois below, chapter two and nn. 13, 14, and 19. On problems of sedimentation and archaeology, see, e.g., K. W. Butzer, Environment and Archaeology, an Ecological Approach to Prehistory (Chicago, 1971); G. Rapp, "The Archeological Field Staff: The Geologist," JFA 2 (1975) pp. 229-237; Geoarchaeology: Earth Science and the Past, ed. D. A. Davidson and M. L. Shackley (London, 1976); B. G. Gladfelter, "Geoarchaeology: The Geomorphologist and Archaeology," American Antiquity 42 (1977) pp. 519-538; F. A. Hassan, "Sediments in Archaeology: Methods and Implications for Paleoenvironmental and Cultural Analysis," JFA 5 (1978), pp. 197-213; Hassan, "Geoarchaeology: The Geologist and Archaeology," American Antiquity 44 (1979) pp. 267-270; Encyclopedia of Geoarchaeology, ed. R. W. Fairbridge and A. S. Gilbert (Stroudsburg, Pa., in preparation). Increasing participation of geologists in archaeological endeavors is also reflected by the growth of the Archaeological Geology Division of the Geological Society of America to a membership of more than 350 geologists. For reconstruction of ancient coastal sites in Greece and Turkey, see recent studies by Prof. John Kraft of the University of Delaware (J. Hardin, "Troy Revisited," Enquiry, University of Delaware, Spring 1984, pp. 12-20). I am grateful to Homer Thompson for bringing Kraft's work to my attention. 14 For example, see reports in the MASCA Newsletter of the Applied Science Center for Archaeology, The University Museum, University of Pennsylvania, and in Archaeometry, Bulletin of the Research Laboratory for Archaeology and the History of Art, Oxford University; R. E. Linington, "Techniques Used in Archaeological Field Surveys," PhilTransRoySocLondA 269

6

I N T R O D U C T I O N : A PERSPECTIVE

the knowledge of scientists in their specialized fields. Silted harbor sites often involve the problems of land archaeology. Backhoes, bulldozers, and irrigation pumps can now conquer the mud and silt. In the excavation of the underwater remains the air lift, or a type of suction hose attached to a compressor on the surface, has become standard equipment since its first use in 1952 on the Roman shipwrecks at Grand Congloue Island off Marseille.15 For the harbor excavations at Cosa, engineer John D. Lewis designed equipment especially for digging in sandy harbor sites.16 Aerial photography for archaeological research, developed since the Second World War, offers a new dimension for interpretation. 17 The underwater remains of ancient harbors, as well as geomorphic evidence of coastline changes, are often discernible only through photographs taken from helicopter, balloon, airfoil, bipod, and now even by spacecraft.18 A. Poidebard, in the first strictly archaeological and scientific study of a harbor site at Tyre in 1934193619 and later at Sidon (begun in 1945),20 was the first to use aerial photography for port research in combination with underwater survey and excavation. In the past two decades Giulio Schmiedt has made a major contribution to this field through his collection and publication of aerial photographs of ancient ports 21 and fisheries (peschiere)22 along the

Italian coastline. Most recently, infrared photography has been used to detect sites from the air,23 and new methods of photogrammetry have been developed for recording in three dimensions artifacts found underwater. 24 The problem of dating ancient harbors is a particularly ticklish one, because underwater recovery of stratigraphy is not possible with the same degree of accuracy as in a land excavation. The underwater archaeologist needs all the help he can get from scientific analysis. For all organic materials and especially for the study of wooden artifacts, such as ship remains, docks, equipment, and cofferdams, radiocarbon (carbon-14) dating is now a regular practice.25 Tree-ring chronologies from various parts of Europe are now being collected and interrelated to aid in the growing field of dendrochronology. 26 Furthermore, research on the conservation of wooden artifacts has been developing rapidly. In the restoration of the water-lifting device from the Cosa fishery, Oleson used the acetone-rosin impregnation process,27 but other methods employing polyethylene glycol and freezing techniques are also in wide use.

(1970) pp. 89-108; Linmgton, Lo scavo nella zona Laghetto della necropoli della Banditaccia a Cerveteri (Rassegna di studi del civico museo archeologico e del civico gabinetto numismatico di milano), Notizie dal chiostro del monastero maggiore, fasc. 25, 26 (1980). Also see R. E. Linington, chapter fourteen. 15 See Du Plat Taylor, pp. 66-76. 16 Lewis 1972. See also chapter three, Appendix. 17 Bradford; L. Deuel, Flights into Yesterday (New York, 1969). 18 F. H. Goodyear, Archaeological Site Science (London, 1971) pp. 245-247; J. H. Whittlesey, "Tethered Balloons for Archaeological Photos," Photogrammetric Engineering 36 (1970) pp. 181186; J. W. Myers and E. E. Myers, "The Art of Flying: Balloon Archaeology," Archaeology 33 (1980) pp. 33-40; D. J. Hambhn, "Aerial Views for Probes under the Sea," The Etruscans (New York, 1975) pp. 147-153; M. Joukowsky, A Complete Manual of Field Archaeology: Tools and Techniques of Field Work for Archaeologists (Englewood Cliffs, N.J., 1980) pp. 435-436. On photography from spacecraft, and most recently, by radar, see, N. M. Short et al., Mission to Earth: Landsat Views the World (Washington, D.C., 1976) pis. 241-245 (Italy); C. Elachi, "Radar Images of the Earth from Space," Scientific American 247, no. 6 (1982) pp. 54-61. 19 Poidebard, Tyr. Also see Frost 1963, pp. 65-87; Frost, "Recent Observations on the Submerged Harbourworks at Tyre," BMBeyrouth 24 (1971) pp. 103-109. 20 Poidebard and Lauffray. Also see Frost 1963, pp. 88-95; Frost, "The Offshore Island Harbour at Sidon and other Phoenician Sites in the Light of New Dating Evidence," IJNA 2 (1973) pp. 75-94. 21 Schmiedt, Atlante; G. Schmiedt, Antichi porti d'ltalia (Florence, 1975); Schmiedt, "Contribution of photo interpretation to the reconstruction of the geographic-topographic situation of the

ancient ports in Italy," Tenth Congress of International Society of Photogrammetry, Lisbon, Sept. 7-19, 1964 (Lisbon, 1964). 22 Schmiedt, Il livello antico. 23 "Archaeological Prospecting. Infrared Photography: Detection of Cultural Anomalies," MASCA Newsletter 8, no. 1 (Dec. 1972) p. 3; E. S. Hirsch, "Infrared Photography and Archaeology: Painted Floors at Gournia," Archaeology 28 (1975) pp. 260-266. 24 For exahiple, "Underwater Photogrammetry," MASCA Newsletter 5, no. 2 (Nov. 1969) p. 2. On land techniques, see S. A. Turpin, R. P. Watson, S. Dennett, and H. Muessig, "Stereophotogrammetric Documentation of Exposed Archaeological Features," JFA 6 (1979) pp. 329-337; J. Whittlesey, "Photogrammetry for the Excavator," Archaeology 19 (1966) pp. 273-276. 25 For bibliography on carbon-14 dating, see Joukowsky, Complete Manuel (cited in n. 18) pp. 600-602. 26 E. K. Ralph, H. N. Michael, and M. C. Han, "Radiocarbon Dates and Reality," MASCA Newsletter 9, no. 1 (1973); D. J. Schove, "Tree-Ring Teleconnections m Europe," MASCA Newsletter 11, no. 1 (1975); P. I. Kuniholm and C. L. Striker, "Dendrochronological Investigations in Greece 1977-1980," Athens Annals of Archaeology 14 (1981) pp. 230-234; Dendrochronology in Europe: Principles, Interpretations and Applications to Archaeology and History, ed. J. Fletcher, British Archaeological Reports, International Series 51 (Oxford, 1978); M.G.L. Baillie, Tree-Ring Dating and Archaeology (Chicago, 1982); P. I. Kuniholm and C. L. Striker, "Dendrochronological Investigations in the Aegean and Neighboring Regions, 1977-1982,"JFA 10 (1983), pp. 411-420. 27 See J. P. Oleson, chapter five. See also H. McKerrell, E. Roger, A. Varsanyi, Studies in Conservation 17 (1972) pp. 111125. 28 A. O. Shepard, Ceramics for the Archaeologist, Carnegie Institution 609 (Washington, D . C , 1956); Ceramics and Man, ed. F. R.

Mineralogical testing, trace element analysis, and dating by thermoluminescence (TL) have revolutionized the analysis of pottery in just the past decades.28 It is now possible to discover with more

I N T R O D U C T I O N : A PERSPECTIVE

7

certainty where the ancient trading centers were, when they existed, and how, through a network of routes, they linked the Mediterranean civilizations together. The most common type of pottery found in an ancient port is the amphora—the large clay container used for shipping commodities in bulk, such as wine, fish products, fruits, nuts, and olive oil. Study of these commercial, mass-produced jars as a separate class of pottery that can be dated has emerged only during the last fifty years, led by archaeologist Virginia Grace of the Athenian Agora excavations, who devised a typology and a dating system for Greek amphoras. The same has most recently been done for Roman amphoras by Elizabeth Lyding Will, who has formed a new classification that revises and expands Heinrich Dressel's of 1889. As important as Dressel's pioneering study has been for this field, his typology is now incomplete and outdated. More than thirty years of research on Roman amphoras throughout the Mediterranean world and Europe yielded a vast body of material, which Will arranges into twenty-four categories with their subgroups, thirteen of which appear in this volume. Her closely defined typology, based on dated contexts of the Athenian Agora excavations, allows a much more precise dating of port sites, and new material, continually being found, can be absorbed more accurately. Moreover, Will's leading studies on Roman trading patterns have revealed how this class of pottery may be used to interpret economic history. Stamps are another important aspect of amphora studies. They provide new names of ancient entrepreneurs, whose personalities come alive as their business practices become known. Even the history of one family enterprise can be traced over centuries, as Will has first done for the Sestii of

Cosa. All this mass of amphora information is now being stored on computers to serve both the archaeologist and the economic historian. More work in this extensive field is needed for full understanding of harbor sites and of the far-reaching waves of commerce and men that gave them life.29 Silted port sites often also yield floral and faunal remains that can be identified and dated by botanists30 and zoologists.31 What was ancient man's relationship to his physical environment and how and why did his relationship change? For our particular study of Cosa, what did the ancient Cosan eat and especially how did he catch, raise, and preserve the fish which, next to bread, was his key staple? Why did he stop developing the fishing lagoon? Was this due to natural or human causes? Unfortunately, at the time of the coring survey in Cosa's ancient lagoon, the Lerici Foundation did not practice flotation, so that some of these questions remain unanswered. However, others can be asked. How did the ancient Cosan live and carry on the everyday management of life's activities in a busy port and fishery? The archaeologist can now obtain further help from the anthropologist and sociologist, as well as the ancient historian, in trying to answer these important queries. The practice of "total retrieval" at archaeological sites, only begun in the last decades, is essential for achieving this broader view of ancient man's whole environment and has been used here. During the scientific revolution, the goals of archaeological research have inevitably shifted away from the older, purely artistic or historical approaches to ones that seek to reconstruct the background of the past in its physical, anthropological, and cultural aspects and to explain why events occurred as they did.32 Port research must also share in considering these ques-

Matson (Chicago, 1965); E. G. Ehlers, "Thermoluminescent Dating of Ceramic Materials," Archaeology 28 (1975) pp. 98-101; C. W. Beck, "Archaeometric Clearinghouse: Specialist Seminar in Thermoluminescence Dating," JFA 7 (1980) pp. 259, 461-465 (a list of periodicals in the field of archaeometry as a whole). For reports on some of the most recent work in this field, see Abstracts: The Twenty-first Symposium for Archaeometry, May 18-22, 198i (Upton, N.Y., 1981). For further bibliography on pottery analysis, see Joukowsky, Complete Manuel (cited in n. 18) pp. 589593. 29 For general remarks on amphoras, see Will 1977-1. For her earliest study of the Sestii see Will 1956 and more recently Will 1979, Will 1982-1, and Will 1982-2. Her definitive catalogue of Roman amphora types is shortly forthcoming in the Athenian Agora series. Compare Dressel 1889. For other revisions of Dressel's forms, see particularly the studies of Lamboglia 1955, Calender 1965, Zevi 1966, Panella 1973, and Manacorda 1977. For further references on Roman amphoras see Will, chapter nine, below. 30 For recent examples of the use of botany in archaeology, see D. S. Brose, "Botanical Archaeology at the Norman P Site in

Ohio," JKA 2 (1975) pp. 293-305; R. W. Dennell, "Archaeobotany and Early Farming in Europe," Archaeology 31 (1978) pp. 813; W. F. Jashemski, The Gardens of Pompeii (New Rochelle, N.Y., 1979) pp. 257-265 (on the use of pollen analysis in her excavations of the gardens of Pompeii). 31 Note the foundation of the International Council for Archaeozoology in 1976. C. W. Beck, "Archaeometric Clearinghouse," JFA 5 (1978) pp. 361-370; Beck, "Taxonomists: Fauna and Flora," JFA 6 (1979) pp. 485-487. 32 On these new directions in archaeology, see particularly C. Renfrew, "The Great Tradition Versus the Great Divide: Archaeology as Anthropology?" AJA 84 (1980) pp. 287-298; J. Wiseman, "Archaeology in the Future: An Evolving Discipline," AJA 84 (1980) pp. 279-285; Wiseman, "Conflicts in Archaeology: Education and Practice,"_/R4 10 (1983) pp. 1-9; and Barker. Also note, for example, F. Rainey, "Science and Archeology," Archaeology 27 (1974) pp. 10-21; D. P. Williams, "As a Discipline Comes of Age: Reflections on Archaeology and the Scientific Method," Archaeology 29 (1976) pp. 229ff.; S. Struever and J. Carlson, "Koster Site: The New Archaeology in Action," Archaeology 30 (1977) pp. 93-101; Tecnologia economica e societa nel

8

I N T R O D U C T I O N : A PERSPECTIVE

tions, and the attempt is made here, as far as our evidence allows, to bring harbor studies into the mainstream of this "new archaeology" with its em­ phasis on cultural ecology. Many more studies of ancient harbor sites along these broader lines are needed before the richness of our classical heritage can be fully understood. This single study of the earliest Roman harbor thus far identified is but one link in a chain that has only just begun to be formed, and we do not claim to have all the answers even for this one. It is clear from our excavations at the port and fishery site of Cosa that there is more to be learned, and future excavations are projected. But it is important at this point to build up a bank of information for port archaeolo­ gists to draw upon and to publish what we have found and know now—lest it perish. Besides Tyre and Sidon already mentioned, a few of the other major harbors of antiquity have received the attention of archaeologists. The first large-scale survey of harbor remains was made in 1916 at the port of Alexandria by G. Jondet, 3 3 the chief engineer of ports and lights in Egypt. Alexandria was the ma­ jor center of commerce from the east in Hellenistic and Roman times, and its lighthouse was considered one of the wonders of the ancient world upon which mondo rotnano, Atti del convegno di Como, 27, 28, 29 settembre 1979 (Como, 1980); J. Pouilloux, "Archaeology Today," AJA 84 (1980) pp. 311-312. 33 G. Jondet, "Les ports submerges de l'ancienne ile de Pharos," Memoires de I'Institut Egyptien 9 (1916). For contributions to port studies before 1916, besides Lehmann-Hartleben, see particularly M. de Ja Blanchere, "Le port de Terracine," Melanges d'archiologie et d'histoire 1 (1881) pp. 322-348; R. T. Gunther, "Earth Move­ ments in the Bay of Naples," Geographical Journal 22, no. 2 (1903) pp. 121-149; Gunther, "The Submerged Greek and Roman Foreshore near Naples," Archeologia 58 (1903) pp. 499-560; Monografia storica dei porti dell' antichita nella peninsola italiana, C. Mirabello, ed. (Rome, 1905); Monografia storica dei porti dell' antichita nell' Italia insulare, Mirabello, ed. (Rome, 1906); P. Dubois, Pouzzoles antique (Paris, 1907) pp. 249-268; A. Georgiades, Les ports de la Grece dans Vantiquite (Athens, 1907); V. Chapot, "Seleucie de Pierie," Μέηιοίη5 de la societe des antiquaires de France 6 (1906) pp. 203-212; W. H. Yates, "The Ancient City and Port of Seleucia Pieria," Museum of Classical Antiquities 6 (1852) pp. 111-131. For articles on ancient ports in general, with further bibliography A. v. Gerkan, "Meereshohen und Hafenanlagen im Altertum," Festschrift W. Dorpfeld zum 80. Geburtstag (1933). Reprinted in E. Boehringer, ed., Von antiker Architektur und Topographie: Gesammelte Aufsdtze von Armin von Gerkan (Stuttgart, 1959) pp. 139142. L. Saville, "Ancient Harbours," Antiquity 15 (1941) pp. 208232; P. R. Mouterde, "Les ports anciens de Mediterranee orientale," in Poidebard and Lauffray, pp. 11-30; Frost 1963, pp. 65114; Du Plat Taylor, pp. 160-189; R. A. Yorke and M. F. Dallas, "Underwater Surveys of North Africa, Yugoslavia and Italy," Underwater Association Report (1968) pp. 21-34; H. Frost, "Ancient Harbours and Anchorages in the Eastern Mediterranean," UNESCO, Underwater Archaeology: A Nascent Discipline (Paris, 1971) pp. 95-114; Shaw, pp. 88-112; E.S.P. Ricotti, "I porti della zona di Leptis Magna," RendPontAcc 45 (1972-1973) pp. 75-103;

others were modeled. Also in 1916, a study of the remains of the Greek port of Delos was published 34 by J. Paris in conjunction with land excavations by French archeologists. The Romans made Delos a free port in 166 B.C.; thereafter it became the inter­ national center for trade throughout the Aegean. Ex­ cavations at the Roman harbor of Leptis Magna in ancient Tripolitania were begun in 1924 by the Ital­ ian archaeologist R. Bartoccini and continued again 35 from 1953 to 1958. The now silted port of Frejus (Forum Julii) in southern Gaul, one of Augustus's 36 major naval bases, was excavated and studied. Likewise, Ostia, the silted ancient port of Rome at the mouth of the Tiber River, has been the subject of a series of studies and excavations: by G. Lugli 37 and G. Filibeck in 1935, followed by R. Meiggs in 38 I960, V. Scrinari (1960, 1963, 1971, and 1979),39 and by the engineer O. Testaguzza (1970).40 In Greece, both land and underwater excavations have been carried out by R. Scranton and his American team at Kenchreai, the eastern port of Corinth on the Saronic Gulf.41 Five volumes of these important excavations have already appeared, with several more to come. Beginning in the 1950s, articles have also recorded both underwater surveys using SCUBA reconJ. Fryer, "The Harbour Installations of Roman Britain," in Marine Archaeology, pp. 261-273; Schafer, pp. 663-678; Williams, pp. 7379; K. Muckelroy, Maritime Archaeology (Cambridge, 1978) pp. 75-84, 255-267 (bibliography). For the most recent complete sur­ vey and bibliography on ancient harbors since Lehmann-Hartle­ ben see Blackman 1982-1 and Blackman 1982-2. 34 J . Paris, "Contributions a 1'etude des ports antiques du monde grec, II (Les etablissements maritimes de Delos)," BCH 40 (1916) pp. 1-73. 35 R. Bartoccini, Il porta romano di Leptis Magna (Rome, 1958). 36 A. Donnadieu, Frejus, Ie port militaire du Forum Julii (Paris, 1935); P. A. Fevrier, "Documents commentes. Plans anciens de Frejus et d'Antibes," Gallia 17 (1959) pp. 207-213. 37 G. Lugli and G. Filibeck, Il porto di Roma imperiale e I'agro portuense (Rome, 1935). 38 Meiggs, Ostia, pp. 51-82. 39 V. Scrinari, "Strutture portuali relative al 'port di Claudio' messo in luce durante ι lavori dell'Aeroporto Intercontinentale di Fiumicino (Roma)," Rassegna dei lavori pubblici 7 (1960) pp. 173190; Scrinari, "U port di Claudio ed osservazioni sulla tecnica del conglomerato cementizio presso i Romani," L'industria italiana del cemento 33 (1963) pp. 527-538; Scrinari, "Il 'Portus Claudii' e ι piu recenti ritrovamenti nella zona di Fiumicino," Atti del III Congresso Internazionale di Archeologia Sottomarina, Barcellona 196i (Bordighera, 1971) pp. 215-224; Scrinari, Le navi del porto di Clau­ dio (Rome, 1979). 40 Testaguzza. 41 Scranton, Kenchreai I\ Ibrahim, Scranton, and Brill, Kenchreai II (cited in n. 10); R. Hohlfelder, Kenchreai. Eastern Port o/Corinth, vol. Ill: The Coins (Leiden, 1978); B. Adamsheck, Kenchreai. East­ ern Port of Corinth, vol. IV: The Pottery (Leiden, 1979); H. Wil­ liams, Kenchreai. Eastern Port of Corinth, vol. V: The Lamps (Lei­ den, 1981).

I N T R O D U C T I O N : A PERSPECTIVE

naissance and excavations at scattered harbor sites throughout the Mediterranean. A sampling can be mentioned here. The port of Narbonne in ancient Gaul was investigated by M. Guy in the early 1950s.42 Following Poidebard's approach, Guy combined the use of air photography, land survey, and some diving. The Aegean island of Crete early received attention from the British School in Athens, led by J. Leatham and S. Hood, who surveyed underwater remains of several Roman harbors in 1955.43 Investigations of the submerged Greek ports of Apollonia in Cyrenaica were published in 1959 and 1961 with a preliminary map drawn by N. C. Flemming. 44 The first attempt at a systematic plan of a large submerged area was made in the Bay of Baiae adjacent to the ancient port of Puteoli, the major port of Italy in the late Republic. This work was first undertaken by A. Maiuri45 in 1958 and continued by Lamboglia with his underwater research vessel, the Daino, in 1959 and I960.46 In northern Italy, the Roman ports in Istria, Aquileia,47 and Ravenna48 have been studied and published by several Italian and French scholars. The late German underwater archaeologist H. Schlager, with J. Schafer and D. J. Blackman, documented several ancient ports in Asia Minor and Greece: Kyme in Aeolis,49 Phaselis in Ly42

M. Guy, "Les ports antiques de Narbonne," RStLig 21 (1955) pp. 213-240. 43 J. Leatham and S. Hood, "Sub-marine Exploration in Crete, 1955," BSA 53-54 (1958-1959) pp. 263-280. 44 N. Flemming, "Underwater Adventure in Apollonia," Geographical Magazine 31 (1959) pp. 497-508; Flemming, "Apollonia Revisited," Geographical Magazine 33 (1961) pp. 522-530. Also see Du Plat Taylor, pp. 168-178. 45 A. Maiuri, "L'esplorazione archeologica sotto marine di Baia," Actes du II' Congres International d'Archfologie Sous-Marine, Albenga, 1958 (Bordighera, 1961) pp. 108-112. 46 N. Lamboglia, "Inizio dell'esplorazione di Baia sommersa (1959 e I960)," Atti del III Congresso Internazionale di Archeologia Sottomarina, Barcellona, 1961 (Bordighera, 1971) pp. 225-252. 47 G. Brusin, "Il porto fluviale," in Aquilea e Grado (Padua, 1956) pp. 94-98; A. Degrassi, "I porti romani deU'Istria," AttiMemVenezia 57 (1957) pp. 24-81; M. M. Roberti, "Il porto romani di Aquileia," Atti del Congresso Internazionale di Studi sulle Antichita di Classe, Ravenna, 1967 (Ravenna, 1968) pp. 383-395. 48 R. Chevallier, "A la recherche des ports antiques de Ravenne," RBelgPhil 41 (1963) pp. 92-109; Chevallier, "Representations cartographiques anciennes du port de Ravenne," Felix Ravenna 3-4 (1972) pp. 55-60; Studi storici, topograficai e archeologia sul 'Portus Augusti' di Ravenna e sul territorio classicano (Faenza, 1961). 49 J. Schafer and H. Schlager, "Zur Seeseite von Kyme in der Aeolis," AA (1962) pp. 41-54. 50 J. Schafer and H. Schlager, "Phaselis. Zur Topographie der Stadt und des Hafengebietes," AA (1971) pp. 542-561; D. J. Blackman, "The Harbours of Phaselis," IJNA 2 (1973) pp. 355364; H. Schlager et al., Phaselis: Beitrage zur Topographie und Geschichte der Stadt und ihrer Hafen (Istanbuler Mitteilungen, Beiheft 24) 1981. In honor of the late Helmut Schlager, besides the archaeological remains, this latest thorough study includes a geo-

9

cia,50 Anthedon in Boeotia,51 and Larymna in Locris.52 In the Peloponnesos, the American archaeologist M. H. Jameson excavated both on land and under water at the ancient port of Halieis from 1962 to 1968.53 The island harbor at Aegina,54 site of the famous Greek sanctuary, has been studied by P. Knoblauch, who also reports on the harbor of Side55 in ancient Pamphylia along the southern coast of Turkey. The Roman harbor of Sarepta on the coast of modern Lebanon was surveyed by J. B. Pritchard in 1969 and 1970.56 In France, M. Euzennat and F. Salviat have continued well the work of the late F. Benoit at the silted port of Marseilles.57 General underwater surveys are also adding knowledge. In Greece, N. C. Scoufopoulos and J. G. McKernan surveyed the port of Gythion in the Peloponnesos, 58 and R. S. Carter has located the ancient port of Aperlae on the Lycian coast of Turkey. 59 In North Africa, Flemming has contributed a valuable study of ancient harbor remains, including those of Thapsus in Tunisia, where he mapped the extensive Roman breakwater. 60 R. A. Yorke and D. P. Davidson have surveyed the Roman ports of Algeria and Tunisia, and R. Hohlfelder has made a preliminary survey of the Roman ports of Baetica in southern Spain.61 logical and sea-level report with good bibliography on this subject as well as on ancient ports and ships in general. 51 H. Schlager, D.J. Blackman, and J. Schafer, "Der Hafen von Anthedon mit Beitragen zur Topographie und Geschichte der Stadt," AA (1968) pp. 21-102. 52 J. Schafer, "Beobachtungen zu den seeseitigen Mauern von Larymna in der Locris," AA (1967) pp. 527-545. 53 M. H. Jameson, "Excavations at Porto Cheli and Vicinity, Preliminary Report, I. Halieis, 1962-1968," Hesperia 38 (1969) pp. 311-342. Compare F. Frost, "The 'Harbor' at Halieis," SfHarbArch I, forthcoming, who convincingly argues for a relocation of the harbor site. 54 P. Knoblauch, "Neuere Untersuchungen an den Hafen von Agina," Bonnjbb 169 (1969) pp. 104-116. 55 P. Knoblauch, Die Hafenanlagen und die anschliessenden Seetnauern von Side (Ankara, 1977). 56 J. B. Pritchard, "The Roman Port at Sarafund (Sarepta). Preliminary Report on the Seasons of 1969 and 1970," BMBeyrouth 24 (1971) pp. 39-56; Pritchard, Recovering Sarepta, A Phoenician City (Princeton, 1978). 57 F. Benoit, "L'evolution topographique de Marseille. Le port et l'enceint a la lumiere des fouilles," Latomus 31 (1972) pp. 5470; M. Euzennat and F. Salviat, "Marseille retrouve ses murs et son port grecs," Archeologia 21 (1968) pp. 5-17; Euzennat, pp. 133-140. 58 N. C. Scoufopoulos and J. G. McKernan, "Underwater Survey of Ancient Gythion, 1972," IJNA 4 (1975) pp. 103-116. 59 R. S. Carter, "The Submerged Seaport of Aperlae, Turkey," IJNA 7 (1978) pp. 177-185. 60 N. C. Flemming, Cities in the Sea (New York, 1971) pp. 152167; see pp. 162-163 for Thapsus. 61 R. A. Yorke and D. P. Davidson, "Roman Harbours of Algeria," Underwater Association Report (1969) pp. 8-21; R. A. Yorke, "Les ports engloutis de Tripolitaine et de Tunisie," Archeologia 17

10

I N T R O D U C T I O N : A PERSPECTIVE

Underwater archaeological research has increased along the Tyrrhenian coast of Italy during the last fifteen years. An underwater survey of the ancient Tuscan shoreline, following the routes of the ancient Maritime Itineraries, was begun by this author in 1965. In addition to the work at the harbor site of Cosa, this author undertook underwater excavations at the Etruscan ports of Populonia and Pyrgi. 62 The latter were accomplished with the collaboration of the Centro Sperimentale di Archeologia Sottomarina in Albenga, directed by the late Professor Lamboglia. Continuing with this survey, V. J. Bruno and a team of divers investigated the Etruscan port of Telamon in 1973,63 and E. Shuey mapped the underwater remains at Gravisca, the port of Tarquinia.64 Recent investigations further south include a restudy of the ancient port installations along the Tiber in Rome itself by F. Castagnoli and A. M. Colini, the port of Minturnae on the Garigliano (Liris) river by Brother S. D. Ruegg, and the port of Pontia by V. J. Bruno. 65 (1967) pp. 18-24. R. Hohlfelder, "The Ports of Roman Baetica: A Preliminary Reconnaissance,' JFA 3 (1976) pp. 465-468. 62 For Populonia: McCann 1971, pp. 20-22; McCann, Bourgeois, and Will, pp. 275-296; N. Lamboglia, "Campagna di scavo italo-americana nelle acque di port Baratti (Populonia)," Forma Maris Antiqui 10 (1973-1974) pp. 56-61. For Pyrgi: J. P. Oleson, "Underwater Survey and the Excavation in the Port of Pyrgi (Santa Severa), 1974," JFA 4 (1977) pp. 297-308; N. Lamboglia, "Campagna di richerche sottomarine italo-americana sul port etrusco di Pirgi (Santa Severa) (16-31 luglio 1974)," Forma Maris Antiqui 10 (1973-1974) pp. 61-65. 63 Bruno, Will, and Schwarzer, pp. 34-43. 64 E. Shuey, "The Gravisca Port Survey," INA Newsletter 6, no. 2 (1979) pp. 5-6; Shuey, "Underwater Survey and Excavation at Gravisca, the Port of Tarquinia," BSK 49 (1981) pp. 17-45; Shuey, StHarbArch I, forthcoming. Cf. B. Frau, Graviscae. Il port antico di Tarquinia e Ie sue fortificazwni (Rome, 1981), and Frau, GIi porti di Tarquinia (Rome, 1982). For the most comprehensive study of the topography of the area see L. Quilici, "Graviscae," La via Amelia da Roma a Forum Aurelii, Quaderni dell'Istituto di Topografia Antica della Universita di Roma, 4 (Rome, 1968) pp. 107-120. 65 F. Castagnoli, "Installazioni portuali a Roma," MAAR 36 (1980) pp. 35-42; A. M. Cohni, "Il port fluviale del foro Boario a Roma," MAAR 36 (1980) pp. 43-53. S. D. Ruegg, "The Underwater Excavation in the Gargliano River: Final Report 1982. The Roman Port and Bridge at Minturnae, Italy," IJNA 12 (1983) pp. 203-218. For probable location of early Roman harbor at Pontia, see V. J. Bruno and E. L. Will, "The Island of Pontia: A Nautical Survey," Archaeology 38 (1985) pp. 40-47. 66 J. Baradez, "Nouvelles recherches sur les ports antiques de Carthage," Karthago 9 (1958) pp. 45-78; P. Cmtas, Le port de Carthage (Manuel d'archiologie Punique 2) (Paris, 1973); H. Hurst, "Excavations at Carthage 1974. First Interim Report," AntJ 55 (1975) pp. 11-40; Hurst, "Excavations at Carthage 1978-9: Fourth Interim Report," AntJ 59 (1979) pp. 19-49; R. A. Yorke and J. H. Little, "Offshore Survey at Carthage, Tunisia, 1973," IJNA 4 (1975) pp. 85-102; Yorke and Little, "Offshore Survey of the Harbours of Carthage," IJNA 5 (1976) pp. 173-176; D. P. Davidson, J. H. Little, and R. A. Yorke, "Ancient Harbours of Carthage: Survey and Excavation," Progress in Underwater Science 2 (1977) pp. 103-109; L. A. Stager, "Carthage 1977: The Punic and Roman

At the present time, two major Roman harbors are being excavated in the Mediterranean. Carthage, the Punic site in North Africa, has had the concentrated attention of an international team since the early 1970s.66 They have conducted both land excavations in the now silted ancient harbors as well as an underwater survey along the coast. In Israel, the Roman port of Caesarea is being excavated also by a collaborative, international team from Israel, USA, and Canada, using over 100 volunteer divers.67 In Italy, the ancient silted port and city of Luni at La Spezia is now being surveyed by a combined team of British and Italian scholars.68 Some smaller ports are also now receiving attention. A Roman port on the Red Sea known today by its Arabic name, Quseir al-Qadim, has been under study since 1978 by D. Whitcomb and J. H. Johnson. 69 The Late Bronze Age port of Dor in Israel is also being excavated by A. Raban and a team from Haifa University. 70 Boston University and the University of Alexandria are investigating the Roman and Byzantine port of MaHarbors," Archaeology 30 (1977) pp. 198-200; Stager, "Excavations at Carthage," Oriental Institute Annual Report (1976-1977) pp. 34-40; ibid. (1978-1979) pp. 52-59; Blackman 1982-1, pp. 79-80; Blackman 1982-2, pp. 204-206, figs. 10, 11. 67 The Caesarea Ancient Harbor Excavation Project (C.A.H.E.P.) is under the direction of Avner Raban, Center for Maritime Studies, University of Haifa with Robert L. Hohlfelder, University of Colorado at Boulder, John Peter Oleson, University of Victoria, and Robert L. Vann, University of Maryland. For publications thus far on this important project see above n. 5 and: A. Flinder, "A Piscina at Caesarea—a Preliminary Survey," Israel Exploration Journal 26 (1976) pp. 77-80; A. Raban and E. Linder, "Caesarea, the Herodian Harbour," IJNA 7 (1978) pp. 238-243; R. L. Hohlfelder and J. P. Oleson, "Sebastos, the Harbor Complex of Caesarea Maritima, Israel: The Preliminary Report of the 1978 Underwater Explorations," in Oceanography: The Past, ed. M. Sears and D. Merriman (New York, 1980) pp. 765779; A. Raban, "Recent Maritime Archaeological Research in Israel," IJNA 10 (1981) pp. 287-308; Raban, "The Siting and Development of Mediterranean Harbors in Antiquity," in Oceanography: The Past, pp. 750-764; E. Linder and A. Raban, "Israel, Caesarea," IJNA 10 (1981) pp. 157-160; R. L. Hohlfelder, "Caesarea beneath the Sea," Biblical Archaeology Review 8, no. 3 (1982) pp. 42-47; J. P. Oleson, "The Caesarea Ancient Harbor Excavation Project: 1982 Season," Israel Exploration Journal 33 (1983) pp. 119-121; J. P. Oleson, "Underwater Excavations in the Harbor of Caesarea Maritima: 1982 Campaign," Archaeological Institute of America Abstracts 7 (1982) pp. 24-25; j . P. Oleson, StHarbArch I, forthcoming; R. L. Hohlfelder, J. P. Oleson, A. Raban, and R. L. Vann, "The Caesarea Ancient Harbor Excavation Project (C.A.H.E.P.), Preliminary Report on the 1980-83 Seasons," JFA 11 (1984) pp. 281-305. 68 Luni I; Luni II; Bryan Ward-Perkins, "Luni: The Prosperity of the Town and Its Territory," Archaeology and Italian Society: Prehistoric, Roman and Medieval Studies, Papers m Italian Archaeology II, British Archaeological Reports, International Series, 102 (Oxford, 1981) pp. 179-180. On the ancient marble quarries, see Bace, p. 138, n. 120. 69 D. Whitcomb and J. H. Johnson, "Egypt and the Spice Trade," Archaeology 34 (1981) pp. 16-23. 70 A. Raban, "Some Archaeological Evidence for Ancient Maritime Activities at Dor," SEFUNIM 4 (1981) pp. 15-26; Raban,

I N T R O D U C T I O N : A PERSPECTIVE

11

rea on Lake Maryut in Egypt. 71 Harbor surveys have continued in Greece, recently on the island of Paros, directed by G. Papathanassopoulos and D. Schilardi with the help of P. F. Johnston. 72 All of these studies of ancient harbors to date have concentrated on identifying the physical remains: breakwaters, concrete and wooden piers, wharves, slipways and shipsheds, mooring stones, anchors, and the like. Important as it is to document all these remains, the ancient historian Geoffrey Rickman recently noted in a paper entitled "Towards a Study of Roman Ports" 73 that they tend "to obscure the view" of what a port is all about, namely, a place where people interact, a focal point and terminal for commerce and trade involving problems also of geography, geology, history, and economics. This viewpoint is shared by the geographer Yehuda Karmon, who writes in his recent comprehensive study Ports Around the World, "A port cannot be regarded as an isolated phenomenon, but as part of the political, social and economic life of a region." 74 Our study of one Roman harbor, besides presenting the archaeological remains, brings into focus some of

these concerns at least for the period of the late Republic. Our approach came about naturally through the collaboration of archaeologists, historians, architects, and scientists who over the course of the past eighteen years have stimulated each other with shared ideas. It is deeply believed that this interdisciplinary approach is the way to study ports in the future. As the work on ancient harbor sites progresses, with increasing numbers of archaeologists receiving the necessary training in both land and underwater fieldwork, there is hope that the ancient ports of the Mediterranean world can be documented before they are destroyed by modern pillage, pollution, and exploitation for marinas.75 Many ancient port remains have already been obliterated by modern harbor works. If ancient ports are not studied in the next few decades, a replacement to Lehmann's book may never be written. Science has given us new techniques and tools. Let us use them to recover this important sphere of our classical past in its fullest perspective before its contribution is lost for today's world and for future generations.

"Recent Maritime Archaeological Research in Israel," IJNA 12 (1983) pp. 229-241. 71 K. M. Petruso and C. Gabel, "The Ancient Harbor at Marea near Alexandria, Egypt," Archaeological Institute of America Abstracts 5 (1980) p. 23; Petruso and Gabel, "Marea: A Byzantine Port in Northern Egypt," Working Papers, no. 62, African Studies Center, Boston University; Petruso and Gabel, "Marea: A Byzantine Port on Egypt's Northwestern Frontier," Archaeology 36, no. 5 (1983) pp. 62ff. 72 P. F. Johnston, "Ancient Harbor Survey of Paros, Greece," Archaeological Institute of America Abstracts 5 (1980) p. 13. 73 Rickman, StHarbArch I, forthcoming. Rickman makes a dis-

tinction between a "harbor" as an area of partially enclosed water and a "port," which is a terminal for marine facilities and goods, part of a whole system of services for trade. He follows the definitions of A. De F. Quinn, Design and Construction of Ports and Maritime Structures (New York, 1961) pp. 71ff. Rickman concludes his provocative study with a plea for "an imaginative cooperation between historians and archaeologists" in order that "Roman ports will then take their proper place in the changing history of Roman government and society." 74 Karmon, p. 7. 75 G. F. Bass, "Treasure-Hunting Divers," JFA 6 (1979) pp. 104-107.

Part I. The History, Geography, and Architectural Remains

Text Fig. 1-1. Hypothetical reconstruction of the Portus Cosanus and its fishing lagoon during the late second and first centuries B.C. (Architectural Phase 2). Reconstruction is based on the archaeological evidence as well as the following sources: lighthouse, Zancani Montuoro, pp. 521 (Fig. VII-3); tuna watch tower above the Tagliata, Aelian, De natura animalium 15.5.6 (chapter one, n.17); Spring House complex with aqueduct, J. P. Oleson, chapter five and reconstruction by J. F. Warren (Fig. V48); fish farm and trapping areas, De Angelis 1959 (Fig. VII-4) and modern facilities at Orbetello (Nassa), particularly Del Rosso, p. 471 (Text Fig. VII-3; Figs. VII-4, 7); cane huts in lagoon after modern huts at Lago di Burano; lagoon temple, Brown et al., figs. 108-109 (Figs. VI-I, 2); fortified hill town of Cosa, Brown 1980, figs. 68, 73. E. L. Will designed plan for amphora factory with kilns in central harbor area, after R. Hampe and A. Winter,

Bei Topfem una Topferinnen in Kreta, Messenien una Zypem (Mainz, 1962) pis. 18, 20, 21, 23, and 29. A. M. McCann designed plan of saltery and garum facility to north of amphora factory, after Ponsich and Tarradell, fig. 36 (Cotta). For the crane on Pier 3, Aqueduct terminal, and market area see D. Macaulay, City. A Story of Roman Planning and Construction (Boston, 1974) pp. 21, 47, and 60. Lionel Casson has kindly reviewed all drawings of merchant ships and fishing craft. See illustrations in chapter eight and also reconstruction of Kyrenia merchant ship, S. W. and M. L. Katzev, "Last Harbor for the Oldest Ship," National Geographic 146 (Nov. 1974) pp. 622-623. McCann and Warren are responsible for the final overall concept. Compare plan and sections by Warren, Figs. VII-10, 11, and 12. Reconstruction drawing by Kathleen K. Borowick.

Chapter I. The History and Topography ANNA MARGUERITE M C C A N N

The Roman port of Cosa was located 138 km northwest of Rome on the Tyrrhenian coast of ancient Etruria (modern Tuscany) (Map 1). In antiquity this area was one of the richest and most varied in all Italy. Its wealth of natural resources included minerals from volcanic mountain ranges, timber from vast forests now largely destroyed, rich farm lands, and fishing both along the coast and in once extensive coastal lagoons. An easy day's sail from Rome, Portus Cosanus enjoyed the protection of one of the few promontories along this predominantly sandy coastline (Color Fig. 1; Fig. 1-1). The limestone headland stands out as the key vantage point, commanding in ancient times the coastal zone between Etruscan Vulci to the south and Telamon and Populonia to the north (Map 2). On either side, rich, brackish fishing lagoons communicate with the sea across narrow sand barriers (Fig. 1-2). These coastal lagoons are unique to this area of the Tyrrhenian seaboard and, along with those of the Po valley on the Adriatic, still provide Italy with one of its most valuable natural resources. Furthermore, in days of coastal navigation, when lights to steer by and lookouts were essential, the promontory of Cosa would have been a necessary point of reference for ships moving westward around the Argentario peninsula up the coast to Gaul and Spain and to the islands of Sardinia and Corsica. Control of such a point would also have been vital to any naval defense or commercial trading system. Along with Portus Herculis on the opposite peninsula of the Argentario, Portus Cosanus provided the best available anchorage between Portus Lunae (La Spezia) to the north and Portus Caietae (Gaeta) to the south. Un-

doubtedly these natural topographical advantages were major reasons for the establishment of the maritime Latin colony of Cosa and its port in 273 B.C., following Rome's conquest of the Etruscan cities of Vulci and Volsinii in 280 B.C.

1 Velleius Paterculus 1.14.7. Trans. F. W. Shipley (London, 1955). "At Cosam et Paestum abhinc annos ferme trecentos Fabio Dorsone et Claudio Canina consulibus. . . ." Compare Livy Periocha 14; Pliny NH 3 51. 2 Livy 22.11.6. Trans. B. O. Foster (London, 1929). ". . . Litterae ab urbe allatae sunt naves onerarias commeatum ab Ostia in Hispaniam ad exercitum portantes a classe Punica circa portum Cosanum captas esse. Itaque extemplo consul Ostiam proficisci iussus navibusque quae ad urbem Romanam aut Ostiae essent

completis milite ac navahbus sociis persequi hostium classem ac litora Italiae tutan." 3 Livy 30.39.1-2. Trans. F. G. Moore (London, 1949). "Claudium consulem, profectum tandem ab urbe, inter portus Cosanum Loretanumque atrox vis tempestatis adorta in metum ingentem adduxit. Populonium mde cum pervenisset stetissetque ibi, dum religuum tempestatis exsaeviret, Ilvam insulam et ab Ilva Corsicam, a Corsica in Sardiniam traiecit."

ANCIENT SOURCES

273 B.C. (Velleius Paterculus) Colonies were established at Cosa and Paestum in the consulship of Fabius Dorso and Claudius Canina, three hundred years before the present date.' 217 R. c. (Livy) . . . a dispatch was delivered from the City, announcing that ships of burden with supplies from Ostia for the army in Spain had been captured by the Punic fleet off the port of Cosa. Accordingly, the consul was ordered to set out at once for Ostia and, manning such ships as were at Rome or Ostia with soldiers and naval allies, to pursue the enemy's fleet and protect the coasts of Italy.2 202 B.C. (Livy) Claudius, the consul, had at last left the city when a very violent storm, which he encountered between the harbour of Cosa and the Portus Loretanus, caused him great alarm. Then after reaching Populonium and lying at anchor there until the rest of the storm should abate, he crossed over to the Island of Elba and from Elba to Corsica, from Corsica to Sardinia.3

16

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

77 B.C. (Rutilius Namatianus) Amid the traces of his camp our conversation weaves again the tale of Lepidus in headlong flight to Sardinia; for 'twas from Cosa's shore that Rome, following the lead of valiant Catulus, drove off the foes of her own blood. 4 49 B.C. (Caesar) On his [Caesar's] arrival there [Gaul] he learns that Vibullius Rufus, whom he had captured at Corfinium and dismissed a few days before, had been dispatched by Pompeius; also that Domitius [L. Domitius Ahenobarbus] had gone to seize Massilia with seven merchant-vessels which he had requisitioned from private persons at Igilium and in Cosanum, and had manned with his own slaves, freedmen, tenants. . . . 5 49 B.C. (Cicero) Domitius [L. Domitius Ahenobarbus], I hear, is * Rutilius Namatianus 1. 295-298. Trans. J. W. Duff and A. M. Duff (London, 1954). "Inter castrorum vestigia sermo retexit Sardoam Lepido praecipitante fugam; litore namque Cosae cognatos depulit hostes virtutem Catuli Roma secuta ducis." 5 Caesar BCiv. 1.34. Trans. A. G. Peskett (London, 1914). " Q u o cum venisset, cognoscit missum a Pompeio Vibullium Rufum, quern paucis ante diebus Corfinio captum ipse dimiserat; profectum item Domitium ad occupandam Massiliam navibus actuarns septem, quas Igilii et in Cosano a privatis coactas servis, libertis, colonis suis compleverat. . . ." 6 Cicero Au. 9.6.2. Trans. E. O. Winstedt (London, 1918). "Domitius, ut audio, in Cosano est, et quidem, ut aiunt, paratus ad navigandum, si in Hispaniam, non probo, si ad Gnaeum, laudo. . . ." 7 Cicero Au. 16.4.4. Trans. E. O. Winstedt (London, 1918). " N a m et ipse et Domitius bona plane habet dicrota, suntque navigia praeterea luculenta Sesti, Buciliani, ceterorum." I owe this reference to Will. See her association of this passage with Lucius Sestius Quirinus, son of Publius Sestius, and her discussion in Will 1979, p. 349. Following Will, John D'Arms uses this passage as evidence for the involvement of the Sestii in trade; D'Arms 1980, p. 83, review by McCann and Will 1984; cf. D'Arms 1981, pp. 55-62, review by E. L. Will, Archaeology 36 (1983) p. 77. Just what kind of ships were supplied by Lucius Sestius can be ques­ tioned. Lionel Casson (personal correspondence) believes that the whole tone of Cicero's passage implies that "navigia . . . lucu­ lenta" were warships carrying men-at-arms. Will, on the other hand, believes that merchant ships adapted for war use and loaded with supplies for the army must have accompanied warcraft, and she would interpret "navigia . . . luculenta" accordingly. Although Cicero does not mention the port οι Cosa in this passage, he does refer to Publius Sestius' villa at Cosa in another letter to Atticus (15.27.1). In the light of the overwhelming ar­ chaeological documentation for the connection of the Sestii and Domitii Ahenobarbi with the port of Cosa, this passage is in­ cluded here as related evidence for the port's use in this period. 8 Strabo 5.2.8. ". . . μετά δέ το Ποπλώνιον Κόσσαι πόλις μι­ κρόν υπέρ της θαλάττης· §στι δ' έν κόλπφ βουνός υψηλός, έφ' ου τό κτίσμα· υπόκειται δ ' Ηρακλέους λιμήν και πλη­ σίον λιμνοθάλαττα και παρά την άκραν την υπέρ τοΰ κόλπου θυννοσκοπεΐον. ακολουθεί γάρ ό θύννος ού τη βαλάνψ μόνον, άλλα και τη πορφύρα παρά γην, άρξάμενος άπό της έξω θαλάττης μέχρι και Σικελίας." I am grateful to Oleson for the

at Cosa, and ready it is said to sail. If it is to Spain, I do not approve, but if to Pompey, he has my praise. 6 44 B.C. (Cicero) For both he [Brutus] and Domitius [L. Domitius Ahenobarbus] have quite good two-banked gal­ leys, and there are also some good ships belonging to Sestus [L. Sestius], Bucilianus, and others. 7 Augustan (Strabo) After Populonia comes Cosa, a city a little above the sea. For there is in the bay a high hill upon which is the settlement. The port of Herakles lies opposite, a lagoon is close by, and along the promontory over the bay is a tuna watchpost. Along the shore the tuna follow not only the acorns but also the purple fish, beginning their course at the outer sea and going even as far as Sicily.8 translation used here. For an alternate translation of this passage see Brown 1951, p. 13, in which "υπόκειται" is translated as "opposite." According to H. G. Liddell, R. Scott, H. StuartJones, Greek-English Lexicon, 9th ed. (Oxford, 1940), the verb means "to lie under" or "below." This seems to me the most accurate translation: from the Cosa hilltop, Port'Ercole is clearly below. Why Strabo does not mention a port at Cosa we cannot say. The archaeological evidence from the port excavations indi­ cates that the Portus Cosanus did decline sometime in the late first century B.C. It may be that in Strabo's time Portus Herculis was the port in the area used for regular coastal traffic and the port of Cosa served largely for fishing or was out of use. Del Rosso, pp. 61-70, uses this passage from Strabo to identify a tuna watch on the Argentario, namely at modern Porto Santo Stefano. He further associates Porto Santo Stefano with the Ind­ iana portus, or port where tunas were collected and caught (cetaria), mentioned in the Maritime Itineraries of the third century A.D. (It. Ant. Marit. 499.7-8, ed. O. Cuntz, 1929). Cf. RE 9, 2 (1916), col. 1249, s.v. "Incitaria portus" (Philip). For the meaning oi cetaria, see ThLL, "cetaria," and D. Du Cange, Glossarium Me­ diae et Infimae Latinitatis (London, 1833), II, "cetaria." For specific association of cetaria with tuna fish see, for example, Horace Sat­ ires 2.5.44. Tuna fishing has long been associated with Porto Santo Stefano, and the harbor was a center for catching tuna until the late nineteenth century, when naval ships destroyed the port as a fishing ground. Del Rosso's identification of Incitaria portus with Porto Santo Stefano seems very possible from the sequence of harbors mentioned (499.6-8, ed. O. Cuntz, 1929): ab Amine portum Herculis a portu Herculis Incitaria, portus, ab Incitaria Domitiana, positio. The "Domitiana positio" has been associated with Santa Liberata (Santangelo, p. 86) about 10 km to the east of Porto Santo Ste­ fano, where remains of a large villa with fishponds extending into the sea are still visible. Cf. Bruno, Will, and Schwarzer, pp. 4142. Strabo's passage, however, seems clearly to refer to a watch within the gulf of Cosa near the lagoons of Orbetello. (Cf. San­ tangelo, p. 84, who also interprets Strabo's passage as referring to a tuna watch at Cosa.) From Cosa's promontory one has an admirable view of the open sea, not available from Port'Ercole, which faces east. Porto Santo Stefano, furthermore, is actually

17

I: HISTORY AND TOPOGRAPHY

A. D. 16 (Tacitus) Clemens by name, he was the slave of Agrippa Postumus . . . and since in the meantime the execution had been carried out, he fell back on a more ambitious and precarious scheme; purloined the funeral ashes; and sailing to Cosa, a promontory on the Etrurian coast, vanished into hiding until his hair and beard should have grown: for in age and general appearance he was not unlike his master.9 Third Century A.D. and later (Itineraria Antonini Augusti. Itinerarium Maritimum) LIKEWISE BETWEEN SARDINIA AND ITALY

. . . the island of Igilium [Giglio] from Cosa 90stades. 10 (Cosmographia Anonymi Rauennatis) Likewise, near the Mare Gallicum are cities called 1 Punicum 9 Armenta 2 Castro novo 10 Ad novas 3 Centum cellis 11 Succosa 4 Minium 12 Cosa 5 Tuvelari 13 Ad portum 6 Gravisca Cossam 7 Martha 14 Albilia 8 Forum Aurelii 15 Telamone" (Guidonis Geografica) 1 Punicum 2 Castrum novum 3 Centum cellis

4 Mimum 5 Tabellaria 6 Gravisca

over 14 km to the north and west of the northern fishing lagoon of Orbetello on the northern exposure of the Argentario peninsula. Of course, there may well have been other tuna watches along the Tyrrhenian coast unnoted by Strabo. 9 Tacitus Ann. 2.39. "Postumi Agnppae servus, nomine Clemens, . . . atque interim patrata caede ad maiora et magis praecipitia conversus furatur cineres vectusque Cosam, Etruriae promunturium, ignotis locis sese abdit, donee crinem barbamque promitteret: name aetate et forma haud dissimili in dominum erat." 10 It.Ant.Marit. 513.4; 514.2 (ed. O. Cuntz, 1929). "ITEM INTER SARDINiAM ET ITALiAM: . . . insula Igilium a Cosa stadia X C . " 11 Rav.Cosm. 4.32 (ed. J. Schnetz, 1940). "Item iuxta Mare Gallicum est civitas que dicitur . . ." 12 Guidonis Geografica 34 (ed. J. Schnetz, 1940), "Portum Augustii." 13 See above, notes 2 and 3. 14 See above, note 4. B. Maurenbrecher inferred from this passage in his commentary on Sallust, Historiarum Reliquiae. II. Fragmenta, 1.82 (Leipzig, 1893), that it was to Cosa that Marcus Aemilius Lepidus, consul in 78 B.C., fled in 77 B.C. after his defeat in Rome by Catulus and that it was therefore at the Portus Cosanus that he embarked his remaining troops for Sardinia, where he died. His army later joined Sertorius in Spain. It should be noted, however, that with the exception of Rutilius Namatianus, the ancient literary sources that describe these events do not men-

7 8 9 10 11 12 13 14 15

Martha Forum Aurelii Armenta Ad novas Subcosa Cosa Ad portum Cose Albinia Tehmona

16 17 18 19 20 21 22 23

Asta Umbrona Salembrum Ardea Malliana Populonia Vadis Voliternis Celsinum' 2

The Portus Cosanus, as distinct from the fortified hill town of Cosa found in 273 B.C. (Map 3; Fig. I3), is first recorded in Livy's account of the Second Punic War in the last two decades of the third century B.C. 1 3 During the restless and violent final years of the Roman Republic in the first century B.C., the harbor may have been used in 77 B. c by Lepidus to embark his remaining forces for Sardinia after his defeat by Pompey near Cosa. 14 Later in the first century Caesar and Cicero both record the use of Portus Cosanus by nobles raising fleets or engaging in trade. 15 The geographer Strabo provides evidence for a fishing industry there in the late first century B.C., for he notes a watch for tuna fish on Cosa's promontory, only one of three watches he identifies along the western Tyrrhenian coast of Italy.16 Tunas were one of the major food sources in the ancient Mediterranean. The eagerly awaited schools of huge fish migrated eastward in the months of May to October from the Pillars of Hercules to warmer waters to lay their eggs. 17 Where there was a watch for tuna, there must also have been a port with a beach tion Cosa or its port (Plutarch Sertorius 15.1-2; Appian BCiV. 1.13.108; Julius Exsuperantius 6). The exact location of Lepidus' battle with Pompey in Etruria and his place of embarkation for Sardinia must remain uncertain. Cosa, though a good possibility, is only Maurenbrecher's conjecture. Sallust in the above-mentioned fragment and Servius, In Aeneidem 10.168, do not mention Lepidus in connection with Cosa. Brown 1951, p. 19 and n. 31, however, uses the passage in Sallust and Virgil as evidence for Lepidus' use of the harbor of Cosa. I thank Casson for help in interpreting these passages. Compare discussion by Marsh, pp. 140-142, 427, for further bibliography. 15 See above, nn. 5, 6, 7. 16 The other two cited were at Populonia (Strabo 5.2.6) and Velia (Strabo 6.1.1). 17 Oppian Halieutica 3.630-648; Strabo 1.2.15; Herodotus 1.62. For further ancient sources on tuna fishing, see Curtis, p. 36, nn. 236, 237, and Radcliffe, pp. 99-105. Aelian, in De natura animalium (15.5.6), gives a vivid description of how schools of tunas were caught with nets set out between small fishing boats, each manned with six oarsmen. From a high lookout on the shore, erected from pine trunks, a watchman signaled the direction the tunas were running. (Text Fig. 1-1) Manilius Astronomica (5.656681) describes how the huge fish were killed with knives and dragged in nets to the shore, where they were cut up and salted. From the guts of the fish garum was made in large vats or dolia. I am grateful to Robert Curtis for the reference to Manilius as

18

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

to process the fresh fish dragged upon the shore and a saltery of some kind to preserve them. The last mention of the Portus Cosanus in the ancient literature is in the Maritime Itineraries from the third century A.D. 18 For the details of the history and commercial life of the port and its fishery during its span of life between the third century B. c. and the third century A.D., further written information is largely lacking. The purpose of this book is to contribute to the picture what has been learned from the archaeological, geological, and other scientific evidence obtained from the excavations at the port site and its attached fishing lagoon.

GEOLOGICAL SETTING

For the limestone promontory that once sheltered the Portus Cosanus and today supports the crowded seaside resort colony of Ansedonia, there is no third century B.C. or twentieth century A.D. (Figs. I1,2,3). The currents of the Tyrrhenian Sea that travel up the coast in a northwesterly direction have swept around the point and the prevailing southwesterly winds have rolled the waves onto the beach since immemorial time, keeping in continuous motion the processes of geomorphic and geological evolution. The Cosa promontory, rising to a height of about 114 m a.s.l., is composed of calcare cavernoso, a calcareous brecciated limestone formed in the Triassic period over 190 million years ago (Color Fig. 1; Fig. 1-4).19 This bedrock weathers to terra rossa, a limestone product typical of the Mediterranean area that gives some of the sediments of the site their deep reddish color. In the Quaternary period (about two million years ago) mountains in the area of Lake Bolsena to the north and east (territory of Etruscan Volsinii) were covered by a complex of volcanic rocks and sediments (Map 2), including tuffs and pozzolana. A special feature of the Italian peninsula well as other references and especially for so generously sharing with me his unpublished, forthcoming manuscript, cited here as Curtis. For modern tuna fishing in Italy, using fixed nets, see F. FlorioSartori, "The Tunny Fisheries of Italy," Anglo-Italian Review 6 (1920) 481-486, 552-554. For further discussion of ancient tuna fishing, see P. Rhode, Thynnorum Captura (Leipzig, 1890); S. Ferri, "La pesca del tonno neU'Adriatico," VjesDal 65-67 (19631965) pp. 39-44. Today, the largest of the different tunas, commonly called the blue-fin tuna (Thunnus thynnus L.), is rapidly becoming an endangered species. The Mediterranean now provides only about 10% of the world's total production of its once most important commercial fish. Blue-fins can be more than 10 feet long and weigh up to 1,200 pounds. See Lythgoe, pp. 233235; E. C. Migdalski and G. S. Fichter, The Fresh and Salt Water Fishes of the World (New York, 1976) pp. 278-281.

and its chain of islands, these rocks have from ancient times to the present offered a source of minerals and building materials. The volcanic deposits rich in heavy mineral crystals give the sands transported to Cosa by alongshore or offshore currents their characteristically gray colors. Just to the east of the Cosa promontory lies the harbor site of the Portus Cosanus with its once attached ancient fishing lagoon to the north (Map 3). The fishery site occupies a former barrier-lagoon complex formed after the rapid post-glacial rise in sea level subsided approximately five thousand years ago. Today over two-thirds of the ancient lagoon has been silted over and is used for pasture land, including its westernmost portion in which the ancient Cosa fishery was located (Fig. 1-5). The original long, narrow body of water stretched to the east as far as the Tafone River 20 km away (Map 2). Estimated to be about 800 m at its widest point, it narrowed to a breadth of not more than 240 m in the western end of the lagoon at Cosa, with a depth of about 5 m. This shallow lake of brackish water, gushing with freshwater springs at its Cosan end and rich in vegetation supporting marine life, was connected to the sea by several natural inlets cutting across the sand barrier. It provided the ancient Cosans with their chief resource besides farming: the raising and catching of certain euryhaline species of fish drawn into the warmer, brackish waters of the lagoon to feed and grow and to be caught when instinct led them back to the sea to spawn. Such fishing continues today both to the west (Fig. 1-2) in the lagoons of Orbetello and to the east in the unfilled portion of the ancient Cosa lagoon, Lago di Burano, about 5 km to the east of the Portus Cosanus (Map 2). Lago di Burano is now about 3 km long and less than a meter deep. The chief kinds of fish found in the remaining lagoons and also mentioned by the ancient authors 20 are: members of the eel family, which make up 50 percent of the fish caught in the 18

See above, nn. 10, 11, 12. See the geology report, Bourgeois, chapter two, from which most of the information cited here has been taken; see also Bourgeois and Gazda. The latitude of the Cosa promontory is 42°17'36"E. See nautical chart, Promontorio Argentario, Istituto Idrografico della Marina-Genova (1973) no. 97. For a general description of the Maremma see L. Bortolotti, La maremma settentrionale 1738-1970. Storia di un territorio (Milan, 1976) pp. 15-18. For a general study of the geology and morphology of the Maremma see especially recent publication by A. Ciacci, "L'ambiente naturale," in GIi etruschi, pp. 11-28, with further bibliography. Ciacci's work was not available to me until after this book was in publication. 20 Columella Rust. 12.16-17; Varro, Rust. 3.17.2-10. Also see Radcliffe, pp. 210-214. 19

I: HISTORY AND TOPOGRAPHY

lagoons of Orbetello; gray mullet (especially Mugil cephalus), which are particularly adaptable to low salinity; sea bass (Dicentrarchus labrax); gilthead (Spams auratus); and sole (Solea solea).21 Mullet was particularly prized in antiquity, and large ones fetched exorbitant prices.22 A special variety of fish sauce was made from its liver.23 The site of the ancient Cosa fishery at the western end of the Cosa lagoon lay concealed in the rich muds formed from its gradual silting since Roman times. In these black, organically rich muds and reddish gray sand layers have been found archaeological materials that largely date between 300 B.C. and A.D. 300. This short span of six hundred years, documenting man's active use of the port and its fishery, is but a moment in the context of geological time. Our finds, however, probably represent only the last remaining e\ idence of the site's use since Neolithic man inhabited the area about five thousand years ago. 24 Sediments obtained from an intensive drill study document the history of the site beyond the range of the archaeological evidence. The oldest sediments are the gray sea sands deposited before the barrier lagoon was formed. They are found at depths of less than one meter below present sea level on the lagoon's most northern boundary (Text Fig. XIV-2), on top of limestone bedrock, and at depths of about five meters below present sea level within the central area of the western end of the ancient lagoon (Text Fig. XIV-3). Sometimes showing a brownish tinge when mixed with terra rossa, these gray sea sands are covered by the modern dark barrier sands, abundant in minerals, from the present beach and sand dunes as well as by the brown wash-over sands. The black, organically rich, lagoonal muds that contain the chief archaeological material lie under a layer of red humus (Color Fig. 2; Figs. 11-14, 16, 19). The sedi21

For a report of a recent extensive survey by biologists, chemists, and engineers of the lagoons of Orbetello for the city council of Orbetello, see Risanamento. I am grateful to P. Paradise for obtaining this report for me and to V. Brasola for kindly sharing his knowledge of the Orbetello fishery and providing further photographs. For information on the activities and the following publications of the Fisheries Department of FAO, Rome, I am grateful to Mr. Michael N. Mistakidis and Dr. Pagan: De Angelis 1959; De Angelis 1960; FAO Aquaculture Bulletin 7 (Oct. 1974-Jan. 1975) p. 3; FAO Agriculture Bulletin 8 (Jan. 1977) p. 20; V. Brasola, A. M. Kalfa, A. Cannas, "Esperienze positive di nproduzione artificiale di Mugil cephalus (L.) effettuate nella Laguna di Orbetello," Rivista italiana di piscicoltura e ittiopatologia 14 (Jan.-March 1979) pp. 1-6; G. Alessio, "Riproduzione artificiale e piscicoltura intensiva di specie ittiche marine come possibilita di sfruttamento della laguna di Orbetello (GR)," Ateneo Parmense, acta nat. 12 1,1976) pp. 315332; D'Ancona, with further references. I also wish to thank Dr.

19

ments also record an end of one phase of the lagoon's life during archaeological times. Dark sea sands are found to overlie lighter sands in one of the channel inlets connecting the lagoon to the sea in the harbor area, filling its passage (Figs. 11-17, 18). Without a free flow of water between the lagoon and the sea, the temperature, salinity, and silting control so essential for the life of such a lagoon fishery would be lacking. Furthermore, this channel-filling event appears to have occurred rapidly, fanning its dark sea sands into the water of the open lagoon beyond. What caused this catastrophe? The sediments of Cosa still hold the answer, but a single sudden storm or storms, such as are still characteristic of the tideless Mediterranean today, could have brought in sufficient sand and silt to block the narrow inlet. Between one excavation season and the next, many of the ancient harbor remains at Cosa were completely covered as well as ancient walls along the beach. The problem of sedimentation in the harbors along Italy's predominantly sandy coastline is an age-old one, strikingly documented by the ancient harbor of Rome itself: the Portus at Ostia, once at the mouth of the Tiber River, is now over 4 km from the sea.25 Despite the extensive and sophisticated efforts of ancient engineers at the Portus Cosanus, sedimentation from a storm or a series of storms appears to have been the immediate cause for the sudden demise of the port and its flourishing fishing industry sometime toward the end of the first century B.C. Except for the filling in of the Cosa lagoon, the coastline in this central area of Italy, known as the Maremma, has changed very little over the last two thousand years. Unlike the south, it appears to have remained largely undisturbed by tectonic subsidence.26 The sea has risen no more than a meter and a half and probably less than a meter. C. Lavett Smith and Dr. J. Atz of the Icthyology Department of the American Museum of Natural History for talking with Elaine Gazda and me and for further information on various euryhaline species of fish. Used here for identification are the FAO Identification Sheets for Fishery Purposes. Mediterranean and Black Sea (Fishing Area 37), I, and Lythgoe. 22 Suetonius Tib. 34; Radcliffe, p. 203. 23 See Pliny NH 31.44, for allex, a sauce made from the sediment of garum. 24 Bronson and Uggeri, pp. 1-2, for evidence of Neolithic man on the Argentario and the island of Giannutri. 25 For the most complete recent study of the port of Ostia, see Testaguzza; "Portus," PECS, p. 731 (R. Meiggs), for further bibliography. 26 For example, compare Puteoli (Pozzuoli), where parts of the lower town have sunk ca. 8 m and risen again through bradyseismic action or movement of the earth's crust. "Puteoli," PECS, pp. 743-744 (H. Comfort).

20

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

HISTORICAL TOPOGRAPHY

To the west of the Cosa promontory the dominating mountain peninsula of the Argentario juts 12 km out from the sandy coastline (Map 2). Before historical times it was an island, but today it is connected to the mainland by two long sand bars that enclose a lagoon of over 26 sq. km. A tongue of land extending out from the mainland in the middle of the lagoon supports the modern town of Orbetello (Fig. 1-2).27 Today a causeway connects the isthmus of Orbetello to the Argentario and divides the lagoon in two. As in antiquity, the chief source of livelihood for the inhabitants is the fish that fill the two lagoons. At the present time Orbetello is part of an intensive aquaculture development program that includes a technical school and a hatchery for artificial breeding. 28 The fishing lagoon on the north is separated from the sea by a tombolo, or sand bar, 8 km long, known as the Tombolo della Giannella, and on the south by the Tombolo di Feniglia, 7 km in length. In the past, geologists believed that the northern sand bar was still unformed in historic times, but the archaeological survey by R. Bronson and G. Uggeri brought to light material from both tomboli dating as early as Neolithic times (ca. 50002000 B.C.), indicating that they must have been already completely formed in prehistoric times. 29 In the most recent geological study of the Orbetello lagoons, scientists have allowed for a good deal of mobility and oscillation in the tombolo formation, which are now calculated to shift in cycles of 550 years.30 It may be that the tomboli have experienced such oscillations in their evolution, which would accommodate for periods of detachment from the Argentario peninsula. Other finds from Bronson's and Uggeri's survey document the ongoing life of the tomboli, especially in the Bronze Age (ca. 1600-900 B.C.) and in the Villanovan (ca. 800-700 B.C.), Etruscan (ca. 750-300 B.C.), and Roman periods.31 That Orbetello was once a flourishing Etruscan 27 Raveggi, pp. 26-96; Santangelo; EAA, V, pp. 708-709 (P. Bocci); Schmiedt, Atlante, pi. CXXX; "Orbetello," PECS, p. 653 (P. Bocci Pacini). 28 See above, n. 21. 29 Bronson and Uggeri, p. 2. For period dates used here, see D. Trump, Central and Southern Italy Before Rome (New York, 1966) p. 16. 30 See Risanamento, pp. 21-22. 31 Bronson and Uggeri, pp. 11-14. 32 See M. Cristofani and M. Michelucci, "La valle dell'Albegna," in GH etruschi, p. 97 with bibliography on p. 112; M. Cristofani in La civilta arcaica di Vulci e la sua espansione (Florence, 1977) pp. 246ff.; Santangelo, p. 40; Raveggi, p. 32; "Orbetello," PECS, p. 653 (P. Bocci Pacini).

city is clear from the extensive remains of a large necropolis containing material dating largely from the seventh to the third century B.C.32 However, the ancient name of this center remains uncertain. Various identifications have been suggested for this town, whose present name does not appear in the written records until the ninth century A.D. 3 3 It seems clear from the studies of M. Pallottino that the name Cosa derives from an Etruscan form, *Cusi or *Cusia. 34 A possible identification for the name of earlier Orbetello was offered by Maria Santangelo in 195435 and is followed by Frank E. Brown in his most recent publication on the hill town of Roman Cosa—namely, that Orbetello was originally Etruscan *Cusi or old Cosa. 36 It in turn lent its name to the Roman colony of new Cosa, founded in 273 B.C. with its fortified center built on the top of the promontory of Ansedonia, 11 km to the southeast. Neither the excavations of the town site of CosaAnsedonia37 nor those of its port have yielded any evidence of an earlier Etruscan settlement. Virgil's reference in the Aeneid to "Cosas civitas Tusciae" has been used as proof of the existence of a city of Cosa in Etruria in pre-Roman times. 38 It is also clear from Strabo's later description that Roman Cosa was located on the crest of the promontory, not in the lagoon of Orbetello. 39 Whether the ancient, massive, polygonal walls still enclosing Orbetello on three of its sides are Etruscan or Roman has also been debated (Fig. 1-6). Brown believes them to be Roman on the basis of their structure and technique of workmanship, which is similar to that of Roman Cosa's carefully dressed fortifications on the Ansedonia hilltop (Fig. 1-7). Such polygonal walls are also typical of other Latin colonies.40 Other scholars would associate the walls of Orbetello with the earlier Etruscan city and date them no later than the fourth century B.C., although the present walls may follow the line of an older fortification.41 Further evidence suggestive of a Roman date is their similarity to ancient walls em33 Dennis, II, p. 234; Del Rosso, pp. 305-310; Anziani, pp. 381392; Cardarelli 1925, pp. 3-10; Raveggi, p. 35. 34 "Nomi etruschi di citta," Scritti in onore di Bartolomeo Nogara (Rome, 1937) pp. 344-345, 352, 354, 356; Brown 1951, p. 16. 35 Santangelo, p. 121. 36 Brown 1980, pp. 8, 16. Also accepted by Cristofani and Michelucci, "La valle" (cited above, n. 32). 37 Brown 1951, pp. 102-113. *> Aeneid 10.168. 39 See above, n. 8. 40 Brown 1980, p. 21, and fig. 17. For similar polygonal defense walls, compare the Latin colony of Norba, PECS, pp. 628-629 (L. Richardson). 41 Cardarelli 1925, p. 4; P. Raveggi, "Sulla costruzione delle

I: HISTORY AND TOPOGRAPHY

21

banking channels that cut across the Tombolo di Feniglia at both its southern (Pertuso) and northern (Ansedonia) ends. The northern channel continues to function as an inlet-outlet between the sea and the southern fishing lagoon of Orbetello. These channels have been associated with the Roman period, and the archaeological evidence of Bronson and Uggeri supports this dating. 42 Bronson and Uggeri also identify remains of a Roman fishery and loading platforms along the inner northern shore of the Tombolo di Feniglia. At the southern end of the Tombolo della Giannella, at the mouth of the modern canal leading into the Orbetello lagoon where the Nassa fishery is located today, V. J. Bruno has noted underwater remains of Roman concrete walls and pottery from the first century B.C.43 The Romans therefore developed these fishing lagoons. Improvements to them might also have included reconstruction of the city embankment and fortification walls on the lagoons. It has been surmised that Etruscan Orbetello also had a sheltered harbor in its northern lagoon, assuming that the Tombolo della Giannella was not completely joined at this time to the Argentario. 44 But until further archaeological evidence is brought to light, the question of a seaport for the Etruscan town remains open. Further Roman remains have been found on the Tombolo di Feniglia. A Roman roadbed cuts across

the tombolo from Ansedonia,45 linking the Argentario and its ancient port of Hercules (modern Port'Ercole) with both Cosa and the Via Aurelia Antica, one of the great consular roads. The road's coastal section was probably already built in the second half of the third century B.C. (Map 2; Figs. 1-3, 8).46 As one of the major arteries of the Roman world, this coastal highway connected Rome with Cosa and other littoral colonies—Castrum Novum (264 B.C.), Pyrgi (247? B.C.), Alsium (247 B.C.), Fregenae (245 B.C.), and later Graviscae (181 B.C.), Luna (177 B.C.)—and on to Gaul, the province so important for trade in the late Republic and early Empire, linked directly to the port of Cosa through the Sestius amphora trade. 47 The Portus Cosanus thus enjoyed the best of both land and sea communications to support a lively export and import trade. In the northern lagoon of Orbetello, at the mouth of the Albenga River (ancient Albinia), additional archaeological evidence from Roman times has been found.48 Another ancient anchorage at the mouth of this river leading into the interior of Etruria must be envisaged (Map 2). The river is still navigable for small boats up to Marsiliana. It may be that ancient salt flats—so essential for the preservation of fish in antiquity—may also have been located here, since there is evidence of a large saline or saltworks still functioning on the site in the early sixteenth century,

mura estrusche di Orbetello," StEtr 1 (1933) pp. 413-415; Lugli, pi. xxii, p. 3; Santangelo, pp. 112-126; M. Pallottino, Etruscologia (Milan, 1973) p. 191; Etruscan Cities, ed. F. Coarelli (London, 1975) pp. 118-119; Cnstofani and Michelucci, "Le Valle" (cited above, n. 32). 42 Bronson and Uggeri, pp. 10-14, especially p. 13, sites 90-95. 43 Bruno, Will, and Schwarzer, pp. 32-43. 44 Santangelo, p. 44; Schmiedt, Atlante, pi. CXXX, fig. 4; Coarelli, Cnstofani and Michelucci (cited above, n. 32) p. 97. Etruscan Cities (cited in n. 41); p. 119. 45 Merciai 1929, p. 355; Santangelo, pp. 115-116. For a hoard of second century B.C. Roman coins found on the Feniglia, see Rassegna Numismatica 1 (1904) p. 34; ibid., 2 (1905) pp. 49-51; Memorie dell'Istituto Italiano di Numismatica 2 (1915) pp. 181-185. 46 For the most recent comprehensive study of the Via Aurelia Antica, see G. M. De Rossi et al., La Via Aurelia da Roma a Forum Aureli, Quaderni dell'Istituto di Topografia Antica della Universita di Roma, 4 (Rome, 1968). Cf. earlier studies by H. Nissen, Italische Landeskunde (Berlin, 1902) II, pp. 299ff., 310ff.; S. B. Platner and T. Ashby, A Topographical Dictionary of Ancient Rome, rev. ed. (Rome, 1965) p. 56; M. Lopes Pegna, "Itinera Etruriae," StEtr 22 (1952-1953) pp. 381ff.; G. Radke, "Die Erschliessung Italiens durch die romischen Strassen," Gymnasium 71 (1964) pp. 204-235; Brown 1951, p. 18; Salmon, p. 84; Harris, pp. 163-165, who suggests a date of 144 B.C. De Rossi, pp. 154-155, distinguishes between the earlier Via Aurelia Vetus, which connected Rome's coastal colonies, and the later Via Aurelia Nova, which linked Rome to the inland Etruscan cities. Both roads are documented in the second half of the second century B.C. (De Rossi, p. 154; ClL XIV.3610). Nissen first associated the Via Aurelia Antica with the censor of 241 B.C., C. Aurelius Cotta. Bace, p. 20, n. 22, would date the Via Aurelia Vetus to the third century

B.C. (241 B.C.) and the Via Aurelia Nova to the second century B.C. (144 B.C.). I am grateful to E. Bace for sharing his unpublished Ph.D. thesis on the Cosa inscriptions and brick stamps with me. Portus Herculis is linked to the Via Aurelia Antica on the Peutinger Table, dating about A.D. 500. See L. Bagrow and R. A. Skelton, History of Cartography (London, 1964) pp. 37-38, and K. Miller, Die Peutingersche Tafel (Stuttgart, 1962) segm. IV, 4. 47 The excavations of the famous ancient shipwrecks off the Grand Congloue rock east of Marseille, which represent the beginnings of modern underwater archaeological exploration, have yielded about 1200 amphoras of the Sestius type. See here report by L. Long, Appendix, chapter eight. Will identifies the port of Cosa as their likely place of manufacture. See chapter nine and her recent article with an up-to-date discussion of the Grand Congloue evidence: Will 1979, pp. 339-341. For literary evidence for Roman traders in Gaul in the first century B.C., see Cicero Pro Fonteio 11: "Gaul is packed with traders, crammed with Roman citizens. No Gaul ever does business independently of a citizen of Rome; not a penny changes hands in Gaul without the transaction being recorded in the books of Roman citizens . . ." (trans. N. H. Watts [London, 1931]). For further discussion of evidence for trade links between Rome and Gaul see below, n. 135. 48 G. Schmiedt, "Contribution of Photo Interpretation to the Reconstruction of the Geographic-Topographic Situation of the Ancient Ports of Italy," Tenth Congress of International Society of Photogrammetry. Commission VII (Lisbon, 1964) p. 19, fig. 25. Also see Bronson and Uggeri, p. 12. For the saline in the Middle Ages, see Cardarelli 1925, p. 94. For the discovery of an ancient amphora kiln in the area of Albinia, see Peacock 1977, pp. 266-269. Cf. Will 1979, pp. 348-349 and n. 35.

22

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

and the area today is known as Tre Saline.49 The area warrants further archaeological investigation both on land and under water. Some 22 km north of Orbetello lies the site of yet another ancient harbor, that of Etruscan Telamon (modern Talamone), now completely silted in by the Osa River (Map 2). 50 Telamon is famous in history for the dramatic battle fought nearby in 225 B.C., when the Romans, so massively outnumbered, repelled the invading Gaulic tribes and thus preserved the Romanization of Etruria. 51 Telamon's continued use as a harbor is documented by both the ancient literature and the recent underwater archaeological investigations by Bruno aided by E. L. Will and J. Schwarzer.52 Plutarch writes that during the civil wars in 87 B.C. the Roman consul Marius landed there with his followers from Africa to collect new forces for his struggle against Sulla.53 He succeeded in collecting forty ships and loading them with recruits—freed slaves and dissatisfied farmers—gathered from the area. Telamon was probably later destroyed by Sulla in 82 B.C. along with other towns in Etruria that had given their support to Marius. 54 The original Latin colony served by Portus Cosanus thus comprised an estimated area of about 550 sq. km (213 sq. miles) (Map 2). 55 Its territory contained over 30 km of coastline running from the mouth of the Tafone River in the south at the eastern end of the Cosa lagoon to the Albenga River to the north and west. Also included were the Argentario peninsula with the Orbetello fishing lagoons and the offshore islands of Giglio (ancient Igilium) and Giannutri (ancient Dianium). The northern boundary of 49 For information on ancient and modern salt resources in Italy, see S. J. Lefond, World Salt Resources (New York, 1969) pp. 206-212, with bibliography. I am grateful to Mrs. Beatrice Hopkinson for this reference and for discussing salt resources in ancient times with me. 50 "Telamon," PECS, p. 891 (O. W. Von Vacano); Cristofam and Michelucci, "Le valle" (cited above, n. 32) p. 101, with further bibliography on p. 113. 51 Polybius 2.25-31. 52 Bruno, Will, and Schwarzer, pp. 35-43. 53 Plutarch Marius 41. 54 On the probable destruction date of Telamon, see O. W. Von Vacano in PECS, p. 891. Cf. Harris, p. 207, n. 1, and p. 258. Brown 1980, p. 74, believes that Telamon, along with Cosa, was destroyed by pirates in 67 B.C., a date that appears too late for the destruction of Telamon on the basis of the archaeological evidence: the latest coin dates to 85 B.C. Rather, the demise of Telamon appears to be linked to the widespread destruction by Sulla in the areas of Etruria that had given support to Marius, e.g., Populonia (Strabo 5.223), Clusium (Harris, pp. 210-212; "CIusium," PECS, p. 229 [E. Richardson]), Vetulonia (I. Falchi, "Vetulonia," NSc [1895], pp. 295-296; Harris, p. 258 and n. 4), and possibly Rusellae (A. Mazzolai, Il museo archeologico delta Maremma [Grosseto, 1977] p. 48; M. Michelucci and A. Romualdi, "Per

the original colonial tract was the mountainous range of which the Cosa promontory is the southernmost extension. More accessible to mainland traffic than the Portus Herculis opposite, the Portus Cosanus provided the colony with its closest and best harbor. It was also well protected from the newly conquered Etruscans on land and the marauding pirates on the seas by the fortifications on the hill above. The city's eighteen watchtowers, incorporated into the polygonal battlements of the colonial center (Fig. 1-7), commanded a bird's-eye view over the entire area from Telamon to the Tafone River, ensuring tranquillity for the farmers and fishermen living largely outside the town walls (Map 3). Most important of all for the continued life of the Portus Cosanus is the presence in the harbor area of several gushing springs of fresh water exposed by our excavations in the ancient lagoon area. Fresh water would have been essential for incoming ships as well as for the life of the brackish water fishery established in the lagoon behind the port. Neither the hill town of Cosa nor Portus Herculis enjoys freshwater springs. This abundant water supply at the Cosa port may also have served the nearby community of Succosa located between the port and the Via Aurelia Antica to the north (Map 2; Fig. 1-3). This settlement, of which ample traces are still visible, remains unexcavated, but surveys suggest that it originated after the decline of the hill town in the first century B.C. It has been suggested by Brown that Succosa was also the site of the later Respublica Cosanorum recorded in third-century inscriptions from the hill town. 56 Succosa is mentioned in the una tipologia della ceramica a vernice nera di Roselle," StEtr 42 [1974] pp. 99-110). Also see the numismatic evidence of the abandonment of coin hoards in the area: M. Crawford, Roman Republican Coin Hoards (London, 1969), nos. 258 (Capalbio), 260 (Carrara), 262 (San Miniato), 266 (Montiano), 317 (Tolfa); Crawford, "A Republican Hoard from Capalbio," BSJ? 37 (1967) pp. 1-3. For further bibliography and discussion of this period see Bace, pp. 27-28. 55 Cardarelli 1924, pp. 183ff, and 1925, pp. 17ff, 86-89, 188ff; followed by Brown 1951, p. 17, and Brown 1980, p. 6. On the centuriation of Cosa, see F. Castagnoli, "La centuriazione di Cosa," MAAR 24 (1956) pp. 149-65. 56 For the location of Succosa now accepted by scholars, see O. Benndorf, "Scavi di Orbetello," BdI (1867) pp. 145-148. Succosa is documented in the Tabula Peutingeriana as well as in the Cosmographia Anonymi Ravennatis and the Guidonis Geografica. See above, notes 11, 12, 46, and below, note 57. For other bibliography on Succosa see Cardarelli 1925, pp. 19-20; P. Raveggi, "La Subcosa e il vico Cosano (contributo alio studio dei suburbi e vichi etruschi)," Convegno nazionale etrusco. Firenze 27 aprile-4 maggio, 1926. Atti, 2 (Florence, 1926) p. 106; Raveggi, Maremma 3 (1926-1927) pp. 101; Brown 1951, p. 21. For inscriptions see CIL XI.2633, 2634; Bace, II B7, pp. 84-85 (Cosa inscriptions CB 900 and CC 576), and IIA4 pp. 79-80 for Cosa inscription CF

I: HISTORY AND TOPOGRAPHY

23

Itineraries and located on the Tabula Peutingeriana (Fig. 1-8).57 The site needs further investigation, for its history must be closely linked to the life of the port and fishery, and its location at the juncture of the road link north from the Via Aurelia to the Via Clodia at the Latin colony of Saturnia (183 B.C.) gave it continuing strategic importance for land traffic. What, then, was the specific role of Portus Cosanus and its relationship to the other ports in the area, particularly to Portus Herculis across the bay (Fig. 1-9)? The latter has been thought by some to be the earliest harbor serving the colony, and it has even been identified as the Portus Cosanus, despite the literary evidence to the contrary. 58 Our excavations at the site of the Portus Cosanus have uncovered Greco-Italic amphora sherds dated by Will to the later fourth and earlier part of the third century B.C., establishing the use of the port for imports as early as the founding of the colony in 273 B.C., if not before. Furthermore, Portus Herculis, about 18 km from Cosa, would have been much less convenient for carrying on the everyday commercial life of the colony. Although it was closer for boat traffic to the offshore islands, Portus Herculis was about a day's journey away by ox cart from the town of Cosa. Recent archaeological evidence suggests that an early Etruscan settlement existed near Port'Ercole, but the earliest historical event recorded in the ancient literature for Portus Herculis appears to be the reference to C. Hostilius Mancinus' departure from the harbor in 137 B.C. on his way to his disastrous defeat at Numantia in Spain.59 The Maritime Itineraries of the third century A.D. list Portus Herculis along with the Portus Cosanus, Albinia, and Telamon. 60 But by A.D. 416, when Rutilius Namatianus makes his way back to Gaul, it is for Portus Herculis that his ship heads, not to the Portus Cosanus, which presumably lay deserted.61 Our under-

water survey in Port'Ercole in 1968 produced no visible evidence of ancient harbor works, although amphora sherds and ballast stones were found at the seaward entrance and in the northern anchorage area.62 An ancient breakwater may well lie under the modern one buried beneath the sand. It seems likely from this evidence that the Portus Cosanus was the earliest Roman harbor established in the region and, indeed, the earliest Roman harbor thus far discovered. The development of the Portus Herculis must have followed close behind with the building of the road across the southern tombolo and the coastal Via Aurelia Antica (Fig. 1-8). Both harbors are relatively small by comparison with later Imperial ports, but they must have been used for a variety of maritime needs—first by colonists and navies during the Carthaginian wars, and then by entrepreneurs and villa owners, with their expanding economic demands. Our amphora evidence from Portus Cosanus provides further help in distinguishing the uses of these two neighboring harbors, at least during their prime years in the last two centuries of the Roman Republic. During this period the Sestii clearly controlled the trade from the Portus Cosanus and used the harbor for the export of their products, which were shipped in their containers and probably also in their ships.63 By contrast, Portus Herculis probably served as the port through which imported goods could pass. These uses probably also depended upon weather conditions and shipping routes in an era of coastal navigation. In a southeast (scirocco) storm Port'Ercole is the more protected; the port of Cosa, with its long, sandy beach, would have been ideal for a commercial emporium or for dragging up a large catch of fish. Port'Ercole is the only ancient harbor in the area still in active use today, sheltering the royal and pleasure yachts of Europe alongside small, native fishing boats. Protected on three sides by natural cliffs (Fig.

1501. Cf. V. Saladino, "Iscrizioni del territorio di Cosa," Epigraphica 39 (1977) pp. 142-143; C. L. Babcock, "An Inscription of Trajan Decius from Cosa," AJP 83 (1962) pp. 247-258. Cf. Manacorda 1978, p. 131 and n. 35. 57 See above, nn. 11, 12, 46. Tabula Peutingeriana, segm. Ill, 4, in A. and M. Levi, La "Tabula Peutingeriana" (Bologna, 1978), and segm. IV, 4, in Miller, Die Peutingersche Tafel. Dating from about A. D. 500, the map was copied from a lost original by a monk of Colmar in 1265 and acquired in 1508 by Peutinger. 58 Brown 1980, p. 50 and n. 8. There seems to be no evidence for Brown's assumption that the Portus Cosanus, active during the Second Punic War and documented by Livy, is to be identified during this period with the Portus Herculis on the Argentario. The latter port is always distinct from that of Cosa in the later sources and would appear to have been so from its beginning. For

the history of Port'Ercole, see Lenzi, pp. 222-224; Santangelo, pp. 77-84; Del Rosso, pp. 628-638. 59 A walled structure has been identified to the west of Port'Ercole and is believed to be the remains of one of the few Etruscan settlements on the peninsula; L. Brasini, Resti di costruzioni preromane rinuenuti presso Port'Ercole (Monte Argentario, 1974). For the ancient sources, see Valerius Maximus 1.6.7; Julius Obsequens Prod. Lib. 24. 60 ft. Ant. 293.1; It.Ant.Marit. 499.6-7; Rav.Cosm. 4.32; Guidonis Geografica, 34, 74. 61 Rutilius Namatianus 1.293. 62 A. M. McCann, "Portus Cosanus Daily Journal, June 17JuIy 31, 1968," pp. 67, 68, 70, 71. Search made and reported by S. Low and R. Young. 63 See Cicero Att. 16.4.4, discussed by Will 1979, p. 349.

24

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

1-9), Portus Herculis never suffered sedimentation, the ultimate cause of the destruction of both the Portus Cosanus and the port of Telamon.

T O W N SITE OF COSA

Because the excavations (begun in 1948) of the hilltop site of Cosa are still not fully published, and because the history and fortunes of the port and town are interrelated, a summary of what has been learned about the town of Cosa thus far is included here (Map 3; Fig. 1-3). From our excavations of the port and fishery, it is clear that the harbor's flourishing industries and export trade were the sources of both the town's and the colony's prosperity in the late second and first centuries B.C. The walled town of the colony of Cosa was placed on the summit of the promontory, 114 m above the port to the northwest and 11 km to the southeast of Orbetello, or Old Cosa. Defense, rather than residence, shaped its original design. The massive polygonal fortress less than a mile around and enclosing 32 % acres served both as protector for the port 64 Brown 1951, p. 113; Brown 1980, pp. 15-18. Brown estimates that the original colony had 2,500 families, or about 9,000 free men, women, and children. Compare Salmon, p. 38, who argues for a larger number of families—about 4,000—most of whom lived on their farms. Within the town walls, Brown estimates that about 300 house plots accommodated some 1,100 inhabitants. The fortified town of Cosa was thus a small, compact community serving primarily as a fortress and religious and civic center rather than as a residential area. One may contrast the city of Pompeii for scale: it enclosed some 160 acres within its walls and supported an estimated urban population of about 20,000 ("Pompeii," PECS, pp. 724-726 [L. Richardson]). 65 See Dyson, JFA, pp. 264, 266-268. Compare Rathbone 1981. 66 G. Santi first described the ruins of the town site of Cosa in his Viaggio secondo per Ie due provincie senesi (Pisa, 1798) pp. 126152. The first map of the fortifications, drawn in 1809, was published by G. Micali, Antichi monumenti per seruire all'opera intitolata I'ltalia avanti il dominio dei Romani (Florence, 1821) pi. IV. The fullest description of the site and the topography of the Ager Cosanus, by R. Cardarelli, appeared in Maremma in 1924 and 1925. Also see W. Smith, Dictionary of Greek and Roman Geography (London, 1854) I, pp. 695-696; Dennis, 2, pp. 236-255; Nissen, Italische Landeskunde (cited in n. 46) pp. 310ff.; RE IV, 2 (Stuttgart, 1901) pp. 1666-1667 s.v. "Cosa" (C. Hudson); M. Fasciato, "A travers la Maremme Toscane, Cosa et Roselle," MilRome 58 (1941-1946) pp. 272-286. For further bibliography, see Brown 1951, p. 12, n. 2; and especially Carandini and Settis. 67 F. E. Brown, "Cosa: Exploration in Etruria," Archaeology 2 (1949) pp. 2-10; W. Tongue, "The Brick-Stamps of Cosa," AJA 54 (1950) p. 263; Brown 1951, pp. 5-113; L. Richardson, "Excavations at Cosa in Etruria (1948-1952)," Antiquity 27 (1953) pp. 102-103; Will 1956, pp. 224-244; Santangelo, pp. 51-67; F. Castagnoli, "La centuriazione di Cosa," MAAR 24 (1956) pp. 149165; Bradford, pp. 227-231; L. Richardson, "Cosa and Rome: Comitium and Curia," Archaeology 10 (1957) pp. 49-55; D. M. Taylor; A. von Gerkan, "Zur Datierung der Kolonie Cosa," Miscellanea Libertini (Florence, 1958) pp. 15IfF.; Brown et al.;

below and as lookout over the surrounding territory (Map 3; Fig. 1-7). Its temples, forum, bath, and marketplaces also made it the religious and civic center for the colony. Only a small portion of the colony's estimated twenty-five hundred families could have lived within the walls of the waterless town. 64 The rest must have lived on their allotted land plots, which, along with the sea, provided their means of livelihood. Over one hundred farm sites have been identified in the recent survey of the Ager Cosanus.65 The hill town was linked to the port by two roads, one leading down from the Porta Marina on the northeast and the other, smaller road leading from the postern gate in the southeast wall. Although the location of the town of Cosa has long been known, 66 its excavation by a team of scholars began only in 1948, directed by Frank E. Brown for the American Academy in Rome. 67 Except for the early excavations of the fortification walls and of the Capitolium Temple and Temple D on the arx, the main excavations of the forum area, houses, and finds still await full publication. From the evidence published so far, the following chronological development can be briefly sketched to aid P. MacKendrick, The Mute Stones Speak (New York, 1960) pp. 98-115; F. E. Brown, Roman Architecture (New York, 1961) pp. 19-24; Babcock, "An Inscription" (cited above, n. 56); F. E. Brown, "Scavi a Cosa-Ansedonia, 1956-'66," BdA 52 (1967) pp. 37-41; Brown, "Excavations at Cosa, 1965-1968," AJA 73 (1969) p. 232; V. J. Bruno, "Antecedents of the Pompeian First Style," AJA 73 (1969) pp. 305-317; J. Collins, "A late Antique Shrine of Liber in the Forum at Cosa," AJA 73 (1969) p. 234; Salmon, pp. 29-39; F. E. Brown, "Cosa," EAA, supp. 1970, pp. 263-264; V. J. Bruno, "A Town House at Cosa," Archaeology 23 (1970), pp. 232-241; F. E. Brown, "Incontro di studi su 'Roma e I'ltalia fra i Gracchi e Silla,' " DWAr 4-5 (1971) pp. 326ff.; R. T. Scott, "Cosa 1969-70: The Houses," AJA 75 (1971) p. 213; Moevs 1973; E. L. Will, "The Sestius Amphoras from Cosa," AJA 79 (1975) p. 151; E. L. Will, "New Light on the Sestius Question," AJA 80 (1976) p. 194; "Cosa," PECS, pp. 245-246 (F. E. Brown); Dyson, Utilitarian Pottery; J. Collins, A Late Antique Shrine of Liber Pater at Cosa (Leiden, 1977); J. Collins, The Marble Sculptures from Cosa (Ann Arbor: University Microfilm International, 1978); R. T. Scott, "Pre-Arretine: New Evidence from Cosa," AJA 82 (1978) p. 249; McCann 1979; Will 1979; E. J. Bace, "The Inscriptions of Cosa," AJA 84 (1980) p. 193; Bace; Brown 1980, reviewed by S. L. Dyson, JFA 8 (1981) pp. 79-83; Buttrey; M.T.M. Moevs, "Aco in Northern Etruria: The Workshop of Cusonius at Cosa," MAAR 34 (1980) pp. 231-280, pis. 1-10; Moevs, "Italo-Megarian Ware," pp. 157-227, pis. 1-23; E. L. Will, "The Earliest Roman Amphoras," Archaeological Institute of America Abstracts 6 (1981) p. 44; Will 1982-1; Will 1982-2; C. R. Fitch, "The Lamps of Cosa," Scientific American 247, 6 (1982) pp. 148-160; Bourgeois and Gazda; McCann, StHarbArch I; McCann, RCRFA. Studies in progress on the hill site include: F. E. Brown, "The Forum; Cosa III"; V. J. Bruno, and R. T. Scott, "The Houses: Cosa IV"; V. J. Bruno and A. Laidlaw, "SUNY House"; C. R. Fitch, "Lamps"; D. F. Grose, "Glass"; M.T.M. Moevs, "Arretine and Terra sigillata"; E. L. Will, "Amphoras." For further references on the port site, see below, chapter three, n. 36.

I: HISTORY AND TOPOGRAPHY

in an evaluation of the town's relationship to the port site. Between 273 B.C., the founding of the colony, and 241 B.C., the end of the First Punic War, the fortifications and centuriation of the town and territory were accomplished. Cosa's city plan reveals Etruscan-Italic as well as Greek traditions, both of which became fused into a new Roman pattern of life. A market with a horreum (storehouse) was located near the northwest gate, and a temporary comitium (assembly place) and curia (senate house or municipal council house) were built. 68 Following the end of the First Punic War, a small, single cella temple was erected in the southwest angle of the arx, facing northwest. Identified as the Temple of Jupiter, this earliest enclosed sanctuary of the colony was completely destroyed except for the outlines of its foundations in the bedrock. However, remains of terracotta decorations from the temple were found to have been built into the later Capitolium Temple, which replaced the Temple of Jupiter on the crest of the arx. These decorations, in an early Hellenizing style, were apparently crafted in an Etruscan workshop at nearby Tarquinia.69 Other finds from this early period are few but include some bronze coins dated before the sextantial bronze reduction and the introduction of the denarius (about 212 B.C.) and some Greco-Italic amphora fragments of Will Type la. The coins come from Cosa itself, Rome, Sardinia, Naples, Syracuse, and Marseille, suggesting the range of Cosa's contacts before the Second Punic War.70 Throughout the struggle with Hannibal, Cosa remained loyal to Rome, 71 but the long war depleted the town's manpower. Following the war, Cosa's petition to the mother city for one thousand new colonists was granted in 197 B.C. 72 The rapidly growing commercial needs of the port facility probably also conditioned this request. With their arrival began the most active building program on the hilltop and also the port, which lasted through the century. A master plan for the forum area included eight new buildings, a portico on three sides (dated about 170 B.C.), and a triple arch on the northwest, longitudinal axis. Only one of these buildings, the atrium publicum, has been fully excavated.73 These 68 69

Brown 1980, pp. 18-25. Ibid., pp. 25-26; Brown et al., pp. 19-23, 151-179, 205, 312-

323. 70

Brown 1980, p. 28; Buttrey, pp. 32, 39-40. Livy 27.10.8-9. Livy 32.2.7; 33.24.8-9. Also see Will, chapter nine, introductory text. 73 Brown 1980, pp. 31-36. 71

72

25

utilitarian buildings were laid on bedrock with squared blocks of limestone for foundations and walls of sun-dried brick. Stone masonry walls during this time were used only for temples and permanent civic structures. One such edifice was the Temple of Concord built next to the now enlarged curia on the northeastern side of the long rectangular piazza.74 Its substructures are of limestone polygonal masonry, and its walls consist of coursed rubblework. The last public building erected in the forum area was the basilica, dating about 150 B.C. Columns of both travertine and tuff carried a clerestory. Its walls of random rubblework differ from the earlier coursed rubble masonry and are the earliest example of this type of construction on the hilltop. 75 Such mortared rubblework is considered the forerunner of opus incertum, Rome's earliest type of concrete in which a facing of irregularly shaped stones is used.76 The date of the initial use of concrete, first described by Vitruvius in the late first century B.C.,77 is widely debated. The earliest remains of true concrete work now appear to be in the Portus Cosanus, where extensive structures built of hydraulic concrete using tuff and pozzolana have been found.78 The hill town of Cosa and, particularly, its port thus furnish the history of technology with the first chapter in Rome's experimental use and development of this revolutionary building material. An open area flanked by two cisterns on the southwest side of the forum has been identified by Brown as an outdoor fish market. 79 The bath building outside the forum to the northwest, with its impressive vaulted cistern, circular room, and shaft for a waterlifting apparatus, remains unexcavated. The most complete town known to have survived from the third century B.C., Cosa provides evidence for how an early Roman colony was organized and functioned and offers the earliest surviving examples of a number of basic Roman building types: comitium, curia, atrium publicum, basilica, and monumental triple arch. These architectural forms must have been developed elsewhere, but the loss of earlier examples gives Cosa a particularly important place in the history of Roman architecture. On the arx of Cosa, the sacred hill overlooking 74

IWd., p. 38. Ibid., pp. 56-58. 76 Ibid., p. 57; Boethius and Ward-Perkins, pp. 105-106. 77 Vitruvius 2.8.1-4. 78 Vitruvius 2.6.1-6. See Gazda, chapter four. 79 Brown 1980, p. 36. This identification seems highly unlikely to me. Fresh fish would have been sold in the emporium at the port itself rather than hauled to the hilltop. 75

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

26

the sea on the southwest, the remains of the tall rub­ ble walls of the Capitolium Temple still stand on a Tuscan podium of yellow sandstone. 80 These walls, like others at Cosa, were probably stuccoed and painted. The columns of the pronaos were made of tuff. Examples of the richly painted terracottas from its various periods of decoration have also been found, including figured plaques of Hercules and Ganymede—all suggesting the temple's original colorful profile against the sky. The style of the sculpture shows the special impact of the artistic tra­ ditions of the Greek Hellenistic world upon Rome during this formative period. The architecture of the tripartite temple corresponds to Vitruvius' descrip­ tion of a Tuscan temple 81 and also closely copies the Temple of Jupiter, Juno, and Minerva, which was 82 then still standing on the Capitoline in Rome. Erected about 150 B.C., the Capitolium at Cosa underwent regular refurbishing until the last half of the first century B.C., when it fell into disrepair. It was completely renovated for the last time in the Augustan period. This chronological sequence cor­ responds to the period of the town's active life, dur­ ing which the Capitolium served as the seat of the protecting deities and as a symbol of the town's original link with the mother city even after it re­ ceived Roman citizenship in 90 B.C., thus becoming a municipium along with all the other Latin colonies. A temple of the single cella type was found in the northwestern sector of the arx and is known as Tem­ ple D (Fig. 1-10). A podium of trapezoidal limestone blocks supports walls of coursed rubblework, 83 and an altar with tuff foundations was found in front of the temple. Terracotta decorations, including a frieze of dolphins and marine monsters, have been uncov­ ered along with part of a life-sized terracotta female(?) statue believed to be the cult image. 8 4 These fragments were found scattered around the base of the temple and over the surface of the north slope of the arx. Brown associates Temple D with the Latin goddess Mater Matuta and dates its construction and that of its twin temple overlooking the port to about 170 B.C. Another identification for this temple and its companion overlooking the fishing lagoon will 85 be offered in our discussion below. Of the ninety or so houses scattered over the site, 80

Brown, 1980, pp. 51-56; Brown et al., pp. 49-109, 127-140, 206-284, 332-367; L. T. Shoe, "Etruscan and Republican Roman Mouldings," MAAR 28 (1965) pp. 39-45, pis. HII, pp. 84-88, pi. XXIII. 81 Vitruvius 4.7.1-5; Brown 1980, p. 53. 82 Brown 1980, p. 53 and n. 13. 83 Brown et al., pp. 25-47, 111-118.

only eight near the center have been fully exca­ vated. 8 6 None of these dates before the third quarter of the third century B.C. Two houses, larger than the others, were formed from joining several of the original small house plots. This extensive remodel­ ing occurred from around the turn of the second to the first century B.C. The earlier of the town dwell­ ings has been associated with one Quintus Fulvius because of the name scratched upon two fragments of black-glazed pottery found in the house. Under the floor of the house a hoard of 2,004 Roman Re­ publican denarii was discovered. The silver coins range in date from the earliest denarii of the late third century B.C. until about 72 B.C. T. V. Buttrey dates the beginning of the hoard between 110 and 100 B.C.87 After this time, coins—in mint condi­ tion—were added annually, while the earlier denarii show the wear of circulation, with not all years rep­ resented. The jar of coins, filled to the top, had been placed in a hole and sealed under the floor. The ex­ cavators also report on empty hole of identical size found under the floor in an adjacent bathroom, 8 8 and Buttrey wonders whether there might have origi­ nally been two coin hoards. We do not know how long the coin hoard may have been buried, but Brown uses them as conclusive evidence for dating the abandonment of the town between 70 and 60 B.C.89

The later house—the largest and most elegant one found—measures 17 m χ 42 m and includes a garden adjoining a porticoed dining room. Since the exca­ vations by V. J. Bruno were financed by the State University of New York, it is known as SUNY house. 9 0 Painted wall decorations in the First Style and coin evidence date the house between 90 and 80 B.C. A skeleton found in its cistern indicates a catas­ trophe, which has been dated by the coin hoard be­ tween 70 and 60 B.C., when the house was destroyed by fire along with the other excavated houses. Houses in this excavated area of about two blocks were not rebuilt until the Augustan period. These two larger town houses from the late sec­ ond and early first centuries B.C. reflect the general trend of the whole Ager Cosanus in this period to­ ward more luxurious private dwellings. Three large turreted villas, two of which are under excavation, 84

Ibid., pp. 182-204, 324-330. Brown 1980, pp. 47-50. Compare McCann, chapter six. Brown 1980, pp. 63-75. 87 Buttrey, pp. 79-153. 88 IWd., p. 81. 89 Brown 1980, p. 73. 90 Bruno, "Town House at Cosa" (cited in n. 67). 85 86

I: HISTORY AND TOPOGRAPHY

have been identified in the immediate neighborhood, dating from the first half of the first century B.C.91 Though apparently not comparable in size to the great landed estates using huge slave gangs seen by Tiberius Gracchus in Etruria and reported by Plu­ tarch, 92 they nevertheless attest to the prosperity of the region in the first century B.C. According to Brown, the active life of the fortified town center of Cosa came to an abrupt end between about 70 and 60 B.C. It was abandoned for about forty years and never regained its prosperity. 93 Ar­ chitectural evidence and finds of marble sculpture, coins, and pottery indicate that the town was par­ tially revived under Augustus and the Julio-Claudians. 9 4 Material from the site then becomes scarce after the mid-first century A.D. The lack of civic ac­ tivities is revealed by the rebuilding of the basilica into an odeon or roofed theater under Nero. In the second century a mithraeum occupied the basement of the curia. 95 From the reigns of Marcus Aurelius and Commodus, Buttrey notes a cluster of coins. 96 Fragments of at least seven public inscriptions (and possibly four others) provide evidence for a last re­ vival of civic life in the first half of the third cen­ tury. 9 7 The name of the emperor Caracalla (A.D. 211-217) appears in five of the certain ones, suggest­ ing that it may have been under this emperor that Cosa was revived as the Respublka Cosanorum, doc­ umented from inscriptions spanning the years A.D. 213-251. 98 The most complete inscription from Cosa comes from this period and records that Maximinus Thrax and his son Maximus ordered the restoration of the odeon and other buildings around the forum in A.D. 235. 9 9 The last recorded dedication of the Respublka Cosanorum is to the emperor Decius (A.D. 249-251).100 Evidence that the small municipality continued comes from the last datable inscription from Cosa, now immured in the cathedral in Orbetello, from the reign of Aurelian (A.D. 270-275).101 By the later third century the town appears once again to have been abandoned, sharing in the general decay of Etruria during the crisis of the late Empire. 91

See below, nn. 144-146; Dyson, JFA, p. 261. Tiberius Gracchus 8.7. See also below, n. 131. Brown 1980, pp. 73-75; Brown et al., p. 118. 94 See above, n. 67, especially Collins, Marble Sculptures; Scott, "The Houses"; Buttrey; Bace; Moevs 1973. 95 Richardson, "Cosa and Rome" (cited in n. 67). 96 Buttrey, p. 33. 97 Bace, p. 59. 98 Bace, p. 60. Brown has suggested that the site of the Respu­ blka Cosanorum was actually below the hilltop at Succosa (Brown 1951, p. 21; cf. Manacorda 1978, p. 131). The inscriptions, how­ ever, were found within the hill site. Succosa, in any case, had both an earlier and a later history. Both Cosa and Succosa appear 92

93

27

In the later fourth and fifth centuries the ruins of the old forum were used by an agricultural estate. A Christian church occupied the basilica-odeon in the fourth century, and a shrine of Liber Pater, uncov­ ered at the southeastern end of the forum, shows the continued use of the ancient site as a place of wor­ ship for the pagan cults that lingered on into the fourth and fifth centuries. 102 The sad fate of the an­ cient Roman municipium was noted by the Gallo-Roman poet and traveler Rutilius Namatianus when he sailed up the coast to Gaul in 416 and saw "desolate Cosa's ancient ruins and unsightly walls." 1 0 3 This area of the Maremma enjoyed importance once again in the Middle Ages, when ancient Cosa became known by its modern name, Ansedonia. Castles dotted its hilltops, and bishoprics were es­ tablished at nearby Sovana and Russellae. At Cosa itself, the remains of an important early shrine and church of San Biagio still stand along the road lead­ ing from the Via Aurelia Antica to the port on the edge of the settlement of Succosa. 104 The small vaulted oratory, built over a Roman tomb, appar­ ently dates from as early as the seventh century, and a nave added in the twelfth century was remodeled in the fifteenth. This shrine, which once housed the head of the martyr San Biagio (associated with mir­ acles and legends of Ansedonia), unfortunately re­ mains unexcavated. Under Charlemagne the Civitas Ansedonia with its Portus Feniliae, Orbetello with its fishing lagoons, saltworks, and much of the sur­ rounding territory were given to the Abbey of the Three Fountains (Abbazia delle Tre Fontane) in Rome. A thirteenth-century fresco in the abbey's en­ trance gate documents this event (Figs. 1-11,12). Various castles of the area are pictured, including that of Ansedonia, along with Charlemagne and pontiffs bearing the miraculous head of Saint Anastasio. N o port is shown, but ships carrying warriors attacking the castle appear next to the walls of An­ 105 sedonia. Archaeological remains further document castle sites in the area around Cosa from the eleventh and twelfth centuries, and excavations at the site of in the Peutinger Table, dating about A.D. 500. See above, n. 57. 99 Cosa Inv. C70.523. Bace; pp. 76-78, no. ΙΙΑ3, fig. 15. 100 Cosa Inv. CF 1501. Babcock, "An Inscription" (cited in n. 56). Cf. Bace, pp. 79-80, no. ΙΙΑ4, fig. 16. 101 CiL XI.2636; Bace, p. 62 and n. 13. 102 Collins, Late Antique Shrine (cited in n. 67). 103 Rutilius Namatianus 1.286. 104 Cardarelli 1925, pp. 80-82, 91-92, with illustration and plan opposite p. 91; Santangelo, p. 72; Raveggi, p. 81. 105 Cardarelli 1925, pp. 86-100; Raveggi, pp. 35ff. For the fresco in the Abbey of the Three Fountains, Cardarelli, pp, 96-98, with illustration opposite p. 106; Tre Fontane (Rome, 1967), pp. 15-17, fig. on p. 16.

28

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

Capalbiaccio behind Cosa have established the presence of this fortified hill town between the twelfth and fifteenth centuries.106 On the hilltop of Cosa itself, within the ruins of the ancient Capitolium, a monastery existed around the turn of the millennium. Graves of some of the monks were uncovered when excavations were reopened in 1965. From about 1200 until 1319, when the territory was ceded to the town of Orvieto, the fortifications of Cosa housed local lords—vassals of the Aldobrandeschi family—who were finally overtaken by the Sienese in 1329.107 The name Ansedonia first emerges in papal documents from the Middle Ages, that of Cosa having been forgotten. After the Middle Ages the hill site lay abandoned again, except for shepherds, olive growers, and the famous white Tuscan Chianina cattle, until excavations began in 1948.108 What is the chronological relationship between the fortified town of Cosa and the port and fishery below? The amphora finds from the port site clearly substantiate that the harbor was in use from the founding of the colony in the third century B.C. until the third century A.D. A single jar from the port, Cat. A294, indicates its use as late as the fifth century. 109 The port flourished along with the hill town in the late second and early first centuries B.C. Indeed, the port and its fishery appear to have been the source of the colony's prosperity from its beginning. Following the early first century B.C., however, the port and hill town seem to have developed along separate lines. Our finds give no indication of a cessation of activity in the 60s; excavated material continues into the Augustan period at the end of the century. After a period of relative dormancy in the first century A. D., activity in the second and third centuries A.D. centered around the port's seaside villa. The assumption that the hill town suffered largescale destruction is based upon the evidence of burning in a relatively small sample of excavated houses 106 Cardarelli 1925, pp. 95-96; S. L. Dyson, "Survey Archaeology: Reconstructing the Roman Countryside," Archaeology 34 (1981) pp. 31-37. Excavations were carried out at Capalbiaccio by Dyson in 1977-1980 and are awaiting full publication. 107 Cardarelli 1925, pp. 99ff.; Brown 1951, pp. 21-23. 108 On the name Ansedonia, see Anziani, p. 374, n. 1, and p. 378, n. 1, citing evidence of a papal bull of 1255. The large, white Chianina cattle from the Chiana Valley in Tuscany have an ancient Roman origin (Columella 6.1.2). This largest of all cattle breeds and the earliest documented from Europe, has recently been introduced into North America by Canadian cattlebreeder R. Bronson, as a potential source for improver genes. 109 See Will, chapter nine. 110 Brown 1980, p. 74 and n. 12.

and upon evidence from the forum area (not yet published). In addition, modification to the town walls and the building of a raised embankment on the arx composed of amphoras, masonry, and other remains are offered as evidence of hasty attempts at fortification. The latest coin from this embankment fill is dated 85 B.C.110 However, Brown dates the abandonment of the city by the coin hoard from the house, in which the latest denarii are from about 72 B.C. He further suggests that the cause of the destruction and abandonment of the town was a catastrophic pirate raid from the sea.111 The plundering activities of the Cilician pirates along the Tyrrhenian coastline during the first decades of the first century B.C. are -well documented in the ancient sources.112 The threat to the safety of the seas led finally to Pompey's special commission in 67 B.C. to rid the Mediterranean of this danger, a mission successfully carried out in forty days with the help of his admiral on the Tuscan shorelines, Lucius Gellius. A pirate raid on the hill town of Cosa might well have occurred at this time, but the archaeological evidence for the extent of the town's destruction awaits publication. There is no evidence of such a disaster at the port or in the surrounding villas. Furthermore, the finds from the hill site thus far published do not altogether support the interpretation of total abandonment in the second half of the first century B.C. In her study of the thin walled pottery, M.T.P. Moevs indicates less material from this time but no interruption until the end of thin walled ware in the Claudian-Neronian period. S. L. Dyson's study of the commonware pottery (likewise based on specific and limited deposits) also shows pottery in the forum area continuing at least into the fifth century A. D.113 Will further documents the uninterrupted use of the Sestius amphora jars well into the Augustan period at the end of the first century B.C., based on finds from both the hill site and the port. 114 Certain amphora types from the Empire are also found in quantity on the hill. Finally, one might question the 111

Ibid., p. 74. Plutarch Pompey 24-29; Appian Mith. 92-95; Cassius Dio 36.20-37. H. A. Ormerod, Piracy in the Ancient World (Liverpool, 1924) pp. 227-241. 1,3 Moevs 1973, pp. 21-31, 231-233; Moevs, "Italo-Megarian Ware," pp. 171-180. See Dyson, Utilitarian Pottery, p. 87, for a pottery dump (Deposit 5) found below the west wall of the town between towers 8 and 9, dating from ca. 110-100 to early in the last quarter of the first century B.C.; for dating of deposits from the first through third centuries A.D., see pp. 115 and 139; for a forum cistern (Deposit 8) with material from the late fourth and early fifth centuries, see p. 161. 114 Will 1979; Will, "SUNY House: The Roman Amphoras," unpublished abstract, May 1, 1980. 112

I: HISTORY AND TOPOGRAPHY

use of a single coin hoard as conclusive evidence for the date of a presumed total destruction of the site. The many possible circumstances that could explain the burial of the completed hoard under a sealed floor do not allow firm conclusions. Until these important excavations are fully published, allowing scholars to evaluate all the evidence, the chronology of the hill site of Cosa unfortunately must remain incomplete. Additional excavation in largely unexplored areas of the town site, such as the northeastern acropolis and the forum baths, would also be useful for a full reconstruction of its history. Until this is done, the Portus Cosanus and its fishery must be viewed as a separate site.

THE HISTORICAL BACKGROUND

The historical life of the Portus Cosanus begins with the initial phase of Rome's colonization of Italy. From the end of the Latin War in 338 B.C. 1 1 5 until the establishment of the Citizen colony at Luna in 177 B.C.—an event that brought all of continental Italy under Rome's control—no colonies were planted anywhere except along Italy's Tyrrhenian and Adriatic coasts. The former received priority, and Cosa was one of a series of promontories along the Tyrrhenian seaboard upon which early colonies were founded:116 Circeii (393 B.C.), Antium (338 B.C.), Tarracina (329 B.C.), and the offshore island of Pontia (312 B.C.). The immediate historical event that precipitated Cosa's birth in 273 B.C. was the takeover in 280 B.C. of the Etruscan city states of Vulci and Volsinii and the subsequent confiscation by Rome of one-third of their territory, including much of their coastline.117 The shores of Etruria had already attracted Dionysius of Syracuse in the earlier fourth century. Now, as Rome began to reach out for trade beyond its shores in the third century, this 1,5

Salmon, pp. 45-48, with map, fig. 5. For a full and readable account of the early period of Rome's colonization of Italy with further bibliography, see Salmon, pp. 55-111. Cf. Harris. One illustration of the high value Rome placed upon the Tyrrhenian coastline appears in the first treaty with Carthage (348 B.C.), in which the only cities named by Rome are five coastal ones: Lavinium, Ardea, Antium, Circeii, and Tarracina. Toynbee, p. 131; Scullard, History, p. 137, and pp. 434-435 for discussion of dates of the treaties with Carthage and bibliography. 117 Fasti Triumph. Capit., Inscriptiones Italiae, xiii, i, pp. 73, 545. The location of the Etruscan port of Vulci remains uncertain. Because Pliny mentions "Cosa Vulcentium" (NH 3.5.5.)—Cosa of the Vulcians—its port has been thought by some to have been at Cosa. See discussion in L. Banti, The Etruscan Cities and Their Culture (Berkeley, 1973) p. 85. Cf. Scullard, Etruscan Cities, p. 124, and A.M.S. Moretti, "Vulci," in GIi etrusche, pp. 53-72. 116

29

coastline was vulnerable to Punic Carthage, which dominated the western Mediterranean. The ensuing struggle for power between Rome and Carthage resulted in three Punic Wars, beginning in 264 B.C. and ending only in 146 B.C. with the destruction of Carthage. In the same year Rome's growing interference in eastern Mediterranean trade was capped by the total destruction of Corinth, the last of the great Greek port cities to come under Roman control. Rome thus became the leading world power and the center for interlocking eastern and western trade routes around the Mediterranean. At the time of its founding, Cosa was the northernmost maritime settlement in Rome's defense and commercial trading system along the Tyrrhenian seaboard. The southernmost defense point was Cosa's sister colony of Paestum, planted in the same year 400 km away at the mouth of the SeIe River on the site of the Greek city of Poseidonia (Map I). 118 Cosa, the fifteenth Latin colony, was the first established in Etruscan territory;119 Paestum bordered land that had been under Greek control since the sixth century B . C The Greek colonies of Magna Graecia had been brought under Rome's power only in 275 B . C , when the defeat of Pyrrhus of Epirus ended the era of Greek rule in southern Italy. Both Cosa and Paestum combined a commercial harbor with the coast guard function of the original Citizen colonies. Whether Cosa, like Paestum, acted with the Greek cities as naval allies to Rome (socii navales), supplying men and ships for the newly created fleets of the Punic Wars, is not known. 120 The first Roman fleet of 120 ships was built in sixty days in 260 B.C. and was manned by crews taught to row on land in wooden frameworks. 121 The period from that time until Carthage's defeat in 146 B.C. constitutes one of the most amazing chapters in all history—how a basically agricultural people, living in a land singularly For traces of an ancient seaport near Regisvilla that may well have been Vulci (3 km south of the mouth of the Fiora River) and an ancient road leading to Vulci, see G. M. De Rossi, "La Via Aurelia del Marta al Fiora," Quaderni dell'Istituto di Topografia Antica della Universita di Roma 4 (1968) pp. 144-152. 118 For bibliography on Paestum, see PECS, pp. 663-65 (W.D.E. Coulson); Salmon, p. 63 and n. 113; Schmiedt, Atlante, pis. LXH-LXIII. The ancient port of Paestum, presumably located at the mouth of the SeIe River and now silted over and about one-half mile from the sea, remains unexcavated. 119 VeIIiUS Paterculus 1.14.7; Livy Periochae 14; Pliny NH 3.5.51. 120 Livy 26.39.5. Thiel, p. 34, n. 91, argues that Cosa and other Latin colonies furnished naval contingents. 121 Polybius 1.20.9-10, 1.20.13-16, 1.21.1-3; Pliny NH 16.192; ESAR, I, pp. 61ff.; Thiel, pp. 46-47, 66-67, 171-206.

30

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

lacking in natural resources, emerged as the chief naval and commercial power of the Mediterranean to defeat the fleets and long seafaring traditions of Carthage and Greece. These years witnessed the astonishingly rapid development of Rome's seapower, which it used to change the course of western civilization. These were also the first years in the life of the colony and port of Cosa on the Tuscan coast. In part, Rome's spectacular success can be attributed to the careful and systematic planning of coastal colonies and ports for protection from and as commercial outposts against its enemies. These colonies ultimately became the means for absorbing Etruscans, Greeks, and Carthaginians into the Latin way of life and served as the foundations of the Roman Empire. The importance of the Portus Cosanus for the colony of Cosa is apparent in its first issue of coinage— the rare half-litra bronzes. On the obverse is the head of Mars, god of war and of agriculture; on the reverse is the protome of a bridled horse posed over a dolphin (Figs. 1-13,14,15).122 Buttrey notes that while the Mars/horse's head type is drawn from earlier south Italian coinage (the earliest RomanoCampanian didrachm), "the dolphin is peculiar to the bronze, to identify Cosa as a port." 123 He dates this Cosa coin, Type I, shortly after the founding of the colony in 273 B.C.124 A further discussion of the meaning of these first coins of Cosa will be offered later.125

THE ECONOMIC AND SOCIAL BACKGROUND

Rome's mastery of the Mediterranean world resulted in far-reaching economic and social changes unforeseen in the early life of the Portus Cosanus. A senatorial, officeholding aristocracy emerged whose wealth was based largely on extensive landholdings 122 Buttrey, pi. 1, A, B, C, D, E (top); pi. 2, top four rows of coins, excluding top left. 123 Ibid., p. 22. 124 Ibid., p. 25. 125 See McCann, chapter six, on the lagoon temple. 126 ESAR, I, pp. 109ff.; Marsh, pp. 5-9; Plutarch Tiberius Gracchus 8.1-3. Cf. below, n. 130. 127 For example, Polybius (31.25) writes that Cato denounced the extravagant imports coming into Rome, citing as an example a jar of pickled fish from Pontus that sold for 300 denarii. 128 See the recent exhibition catalogues prepared by John WardPerkins and Amanda Claridge, Pompeii A.D. 79 (Boston, 1978) III, and II, p. 219, for further bibliography on these important sites. 129 For this period see Toynbee's comprehensive study, particularly II, p. 332. F. Rossi, "La situazione economica dell'Etruria dai Gracchi agli inizi deU'Impero," Bollettino delta Societa Storica Maremmana 27 (1973) pp. 9-41.

and commerce. 126 This noble class had grown rich through the booty and profits of war and commerce, which included vast numbers of slaves from conquered lands to work their estates. Noble Romans exposed to the culture and wealth of the Greek Hellenistic world created a demand for new luxury products and acquired a taste for things Greek.127 The last two centuries of the Roman Republic saw the most rapid absorption of Hellenism into Italian soil, and an influx of foreign craftsmen and artists brought new traditions and knowledge. Rome adorned itself with the famous art treasures of the Hellenistic east, a taste that continued into the Empire, as evidenced by the great private collections found in the villas at Pompeii and Herculaneum. 128 With the Second Punic War, the occupation and devastation of southern Italy by Hannibal started in motion the sweeping changes and tensions that eventually led to the Social and Civil Wars of the first century B.C.129 During this period, many small farmers and herdsmen were forced to leave their lands and seek the protection of the larger towns, never to return. The resulting decline in smaller farms and freeholders led to a concentration of land into large holdings by the noble class. Although some freedmen were employed on these large estates, slaves were the major source of labor on the great latifundia.130 Plutarch vividly describes this condition in his life of Tiberius Gracchus.131 He writes that it was Tiberius' experience of seeing large slave gangs working the latifundia in Etruria when he was traveling up the Via Aurelia to Numantia in northern Spain in 137 B.C. that prompted his later land reforms. This shift away from small, independent farms worked by freeholders to a system of large estates primarily using slave labor allowed the mass production of products for export and the development of more sophisticated technology.

130 On slavery in the Roman world, see M. I. Finley, Slavery in Classical Antiquity (Cambridge, 1960); Scullard, p. 353, with further bibliography; W. Harris, "Towards a Study of the Roman Slave Trade," MAAR 36 (1980) pp. 117-140. Cf. Rathbone 1981, who argues also for the extensive use of free peasant labor in the Ager Cosanus at this time. Our study of the Portus Cosanus further suggests that the demand for new colonists in the early second century B.C. and the need for increased labor, particularly in the first century B.C., cannot be explained by agrarian needs alone. Any study of the labor force of the Ager Cosanus must take into account the demands of the commercial center at the port of Cosa and in particular the Sestius pottery and fishing activities in the lagoons of Cosa and Orbetello. Rathbone seems unaware of the previous studies of the port or its amphora material in McCann 1979 and Will 1979. 131 Plutarch Tiberius Gracchus 8.7. Cf. D. B. Nagle, "The Etruscan Journey of Tiberius Gracchus," Historia 25 (1976), pp. 487489; Harris, pp. 202-208; Brown, 1980, p. 71.

I: HISTORY AND TOPOGRAPHY

31

Along with the emergence of a landed aristocracy in the late Republic, a new class of businessmen, known as "equestrians," appeared.132 Their wealth was largely built upon handling government contracts, leasing state properties, collecting taxes and port dues, banking, and other private enterprise. In the age of the Gracchi, the equestrians also gained in political power. 133 The influence of this commercial class upon government is clearly illustrated in the career of Cicero, the most gifted statesman of his day and the first eques to be elected consul (in 63 B.C.). Both of these wealthy upper classes could also draw upon the resources of a rapidly developing urban society with its center at Rome. It became fashionable for the wealthy to maintain both city and country homes. Cicero, for example, owned eight country villas in addition to his home on the Palatine.134

The last two centuries of the Republic also witnessed an enormous expansion of Roman trade and commerce. 135 A true money economy gradually emerged, and by the end of the Second Punic War Roman coinage held a monopoly. 136 Although trading policies were slow in forming, Rome's premise from the beginning of its maritime expansion that all ports were open to free trade and that the sea and all its resources were held in common by all men137 greatly encouraged commerce on the seas and the use of maritime resources. Furthermore, the elimination of piracy by Pompey in 67 B.C. brought needed security to the Mediterranean.138 Though transport by ship was costly and susceptible to natural disaster, it was usually quicker and cheaper than land conveyance and was the preferred means for the great grain trade during the late Republic and Empire.139 The urbanization of Italy in the late Republic

132 Marsh, pp. 316-318. On the handling of government contracts by equestrians, see Polybius 6.17 and ESAR, I, p. 149. 133 Marsh, pp. 32-67. 134 J. H. D'Arms, Romans on the Bay of Naples (Cambridge, Mass., 1970); McKay, pp. 67, 102, 115 and n. 190 for further bibliography. 135 Charlesworth, pp. 2-13; ESAR, I, pp. 201ff; L. Breglia, Le Antiche rotte del Mediterraneo (Rome, 1966); Rouge, pp. 201-211; Rouge, Ships and Fleets of the Ancient Mediterranean (Middletown, Conn., 1981), pp. 99-118; A. Tchernia, "Les amphores romames et l'histoire economique," journal des savantes 65 (1967) pp. 216234; A. J. Parker, "The Evidence Provided by Underwater Archaeology for Roman Trade in the Western Mediterranean," Marine Archaeology (London, 1973) p. 366; Broughton 1974, pp. 1130; T. Robert S. Broughton, "Remarks on the Roman Empire and Mediterranean Trade," Archaeological News 8 (1979) pp. 9596; L. Casson, Ancient Trade and Society (Detroit, 1984). For the great number of Roman traders in Gaul during Cicero's time, see his Pro Fonteio 11-12, and above, n. 47. Other archaeological evidence for increased trade between Rome and Gaul in the late Republic is the survey of ancient shipwrecks along the French Mediterranean coastline of 1975 when twenty-nine out of one hundred and three wrecks were noted containing "Dressel 1" amphoras. See R. Lequement and B. Liou, "Les epaves de la cote de Transalpine," Cahiers Ligures de prehistoire et d'archeologie 24 (1975) pp. 76-82. For recent discussions of trade between Rome and Gaul see C. Goudineau, "Marseilles, Rome and Gaul from the Third to the First Century B.C.," in Garnsey, Hopkins, and Whittaker, pp. 76-86; Tchernia 1983. The chief flaw in Goudineau's and Tchernia's interpretation of the Republican wine trade as based on barter rather than on a monetary system stimulated by supply and demand is their failure to account for the great Sestius enterprise of Cosa. Also see below, Will, chapter nine, introductory text, n. 6. Furthermore, there is literary evidence for some monetary exchange for wine in the second century B.C. in Plutarch, Vit. C. Gracch. 2.5, cited by Will in her forthcoming article, "Amphoras and Trade in Roman Sardinia," Studies in Sardinian Archaeology II: Sardinia in the Mediterranean, ed. M. S. Balmuth. While we do not know how extensive money exchange was, it is clear that a monetary system was in place and at least in limited use. (See McCann, StHarbArch I, n.48) Exports and imports also need not be in balance. For example, in the massive grain import trade under the Empire, we know that ships regularly sailed in ballast from Puteoli to Alexandria, where they were loaded for their return voyage (Strabo 17.793; Casson, SSAW, p. 298, n. 8).

There is also archaeological evidence that ships often sailed under capacity for legs of their voyage, for example, the seventh century A. D. ship from Yassi Ada, rated at 60 tons, was wrecked when carrying a load of just over 27 tons (J. R. Steffy in G. F. Bass and F. H. van Doormnck, et al., Yassi Ada: A Seventh-Century Byzantine Shipwreck [College Station, Texas, 1982] p. 86. I am grateful to Casson for the above references). Thus, the picture of trade in wine from Italy exclusively for metals and slaves from Gaul in the late Republic as presented by Tchernia appears distorted in the light of other literary and archaeological evidence. The improvement of the harbor facilities on the Tiber in the first quarter of the second century B.C. also attests to Rome's expanding trading activities: F. Castagnoli, "Installazioni portuali a Roma," MAAR 36 (1980) pp. 35-42; A. M. Colim, "Il porto fluviale del Foro Boario a Roma," MAAR 36 (1980) pp. 43-53. 136 Scullard, History, pp. 347-351. 137 For Roman concepts of free trade, see Rouge, pp. 540ff. For rights to the resources of the sea, see: Dig. 41.1.1: "Therefore all animals which are captured on land, on sea, or in the air, that is to say, wild beasts and birds, as well as fish, become the property of those who take them"; Inst. lust. 2.1.2: "All rivers and ports are also public, and therefore the right of fishing in a harbor or in a stream is common to all." (Trans. S. P. Scott [Cincinnati, 1932].) Concerning the rights of the shore of the sea, see: Dig. 41.1.14: "Whatever anyone builds upon the shore of the sea will belong to him; for the shores of the sea are not public like the property which forms part of the patrimony of the people but resembles that which was formed in the first place by Nature, and has not yet been subjected to the ownership of anyone. For their condition is not dissimilar to that of fish and wild animals, which, as soon as they are taken, undoubtedly become the property of him under whose control they have been brought." (Trans. S. P. Scott [Cincinnati, 1932].) For a discussion of the Roman origins of modern laws governing the sea as well as fishing, see P. Fenn, The Origin of the Right of Fishery in Territorial Waters (Cambridge, Mass., 1926) pp. 3-32. 138 Plutarch Pompey, 24-29; Appian Mith., 92ff., documenting piracy on the Etruscan coast; Ormerod, Piracy (cited in n. 112). 139 On the risks of marine loans and Cato's activities in this area, see Plutarch Cato, 21.6; cf. E. Gabba, "Riflessioni antiche e moderne sulle attivita commerciali a Roma nei secoli H e I A.C.," MAAR 36 (1980) pp. 91-102. On the use of ship versus land transport for the grain trade, see G. E. Rickman, "The Grain Trade Under the Roman Empire," MAAR 36 (1980) p. 262. For

32

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

brought with it the development of the mother city. To feed its populace Rome had to rely upon imported products, especially grain for the bread doles.140 The major trading centers of the Italian peninsula inevitably shifted from the north—which had supplied timber, iron, armor, and men for the Punic Wars141—to the south. The maritime colony of Puteoli (modern Pozzuoli), created in 194 B.C., became the gateway for the Alexandrian grain fleet and for ships bringing other products from the east.142 The main period of Puteoli's prosperity extended from 166 B.C., when Delos became a free port, until the early first century A.D., when Claudius built the port of Ostia at the mouth of the Tiber, but the colony remained active throughout the period of the western Roman Empire. In Imperial times other large ports were constructed along the Tyrrhenian coast. Besides rebuilding the port of Ostia, Trajan (A.D. 98-117) established the large naval and commercial harbor at Centumcellae (modern Civitavecchia) on the Tuscan coast.143 The need for further maritime centers in ancient Etruria had vanished. With the conferring of Roman citizenship on all of Italy south of the Po following the Social War of 90 B.C., the colonial period of Cosa's life was over.

the Republic and the first century of the Empire. Small farms and holdings gave way to larger villas; four of the largest were located within the immediate area of Cosa. Andrea Carandini's excavation at one of these, Settefinestre,145 and Daniele Manacorda's research146 confirm these findings. The main period of activity for the vineyards and olive groves at Settefinestre was from about 75 B.C. through 100 A.D., with the greatest expansion occurring between 25 and 100 A.D. Likewise, Dyson's excavations of the villa at Le Colonne indicate that its greatest prosperity came during the early Empire. In the third century the villas in the Ager Cosanus declined, reflecting the general rural crisis of the late Roman world, perhaps also encouraged in the Maremma by malarial conditions caused by the lagoonal swamps. This coastline remains infamous in literature for its unhealthy qualities,147 which were not corrected until the construction of modern drainage facilities. Today, as in the early Roman Empire, the coast is dotted with the elaborate seaside villas of affluent Romans.

A land survey and excavations at two of the major villa sites in the area have revealed that the Ager Cosanus and the Portus Cosanus were affected by these developments. The land survey by Stephen Dyson144 has brought to light a changing pattern in the Ager Cosanus settlements in the last century of

For the Portus Cosanus, trade began with the earliest days of the colony in the second quarter of the third century B.C., judging from the Greco-Italic amphora sherds found in the harbor area.148 Of the total number of amphora fragments found on the site, Will estimates that 9 percent are Greco-Italic and can be dated from the late fourth through the late second century B. c., especially from the first half of the second century. The greater part of the amphora material from the whole site—about 75 per-

A.H.M. Jones's calculations on cheapness of shipping grain in the Mediterranean versus ox cart conveyance, see his book, The Later Roman Empire (Oxford, 1964) pp. 821-823. 140 For new evidence from Pompeii on dealers in the grain trade, see L. Casson, "The Role of the State in Rome's Grain Trade," MAAR 36 (1980) pp. 21-33. For an excellent study of the grain trade during the Empire, see Rickman, "Grain Trade" (cited in n. 139) pp. 261-275; Rickman, Roman Granaries and Store Buildings (Cambridge, 1971); Rickman, The Corn Supply of Ancient Rome (Oxford, 1980), review by N. Purcell in JRS 71 (1981) pp. 197-198. Compare P. Garnsey, "Grain for Rome," in Garnsey, Hopkins, and Wittaker, pp. 118-130. 141 On supplies provided by the cities of Etruria in 205 B.C. for Scipio Africanus' army and fleet, see Livy 28.45.15-18. On sources for Rome's timber, see R. Meiggs, "Sea-borne Timber Supplies to Rome," MAAR 36 (1980) pp. 185-196. 142 "Puteoli," PECS, pp. 743-744 (H. Comfort). 143 For Ostia, see above, n. 25; "Centumcellae," PECS, pp. 212-213 (E. Richardson); Schmiedt, Atlante, pi. CXXXIII. 144 Dyson, JFA, pp. 251-268; Dyson, "Settlement Reconstruction in the Ager Cosanus and the Albenga Valley: Wesleyan University Research, 1974-1979," Archaeology and Italian Society: Prehistoric, Roman and Medieval Studies, British Archaeological Reports, International Series, 102 (Oxford, 1981) pp. 269-274; Dyson, "Survey Archaeology: Exploring the Roman Countryside," Archaeology 34 (1981) pp. 31-37.

145 Carandini and Settis, with full bibliography on the area and its villas; A. Carandini, "Il vigneto e la villa del fondo di Settefinestre nel Cosano: un caso di produzione agricola per il mercato transmarine" MAAJ? 36 (1980) pp. 1-10; L. Quilici and S. Quilici Gigli, "Ville dell'agro Cosano con fronte a torrette," RivIstArch 3, 1 (1978) pp. 11-64; Rathbone 1981. 146 Manacorda 1978, pp. 122-131; Manacorda, Athenaeum, pp. 73-97; Manacorda, "Questioni cosane. Produzione agricola, produzione ceramica e proprietari nell'ager Cosanus nel I secolo a . C , " Forma di produzione schiavistica e tendenze della societa romana: II a.C.-IId.C. Un Caso di sviluppo precapitahstico, Pisa 4-6 gennaio 1979, Rome, Istituto Gramsci (this unpublished article was not available to me); Manacorda 1980, pp. 173-184; Manacorda 1981, pp. 3-54. 147 Rutilius Namatianus complains about the "marshy smell" of this coastline in the early fifth century (1.280). He further adds (1.285) that a plague of rats was the cause of Cosa's abandonment. Brown 1951, p. 20 and n. 40. suggests that malaria was the actual cause, brought on by the silting up of Cosa's coastal lagoon. On malaria in antiquity, see McKay, p. 524, n. 10, and particularly Ciacci, "L'ambiente naturale" (cited in n. 19) pp. 25, 26, and n. 34. Ciacci believes malaria as a cause for the decline at least of the Etruscans is overexaggerated. 148 McCann 1979, pp. 392-393. Statistics and material on the Sestii are drawn from Will's research presented here and in her other publications listed above in Frequently Cited References.

I: HISTORY AND TOPOGRAPHY

cent—comes from the late second through the first century B.C. Of the total number of amphora fragments bearing stamps, 86 percent can be identified with one group of early Roman amphoras associated with the name of their manufacturers, the Sestii, from the stamps upon the jars (Color Figs. 3,4). The Sestii probably also manufactured the earlier amphoras found at Cosa. This is an unprecedented concentration of a single type of jar to the exclusion of other contemporary types at any one place. Will has concluded from this evidence that the Cosa port area was probably the chief place of manufacture for the famous Sestius amphora jars, which constituted a virtual monopoly in the western Mediterranean during the late Republic. Sustained by the continuing support of the Sestii, the port of Cosa now emerges as a dominant force in the commercial activities both of the Ager Cosanus and of the whole western Mediterranean for almost two hundred years. The last period of activity in the commercial life of the harbor of Cosa occurred in the second and third centuries A.D., after a period of relative dormancy in the first century A. D. An extensive seaside villa was apparently built in the second half of the first century A.D. along the beach at the port. Remains of its magazines and quays still extend into the sea (Map 6). D. Levi, in his survey of the area in 1927, noted additional remains that we did not find: a hypocaust, an apsidal structure, steps, and storage magazines.149 Though the villa is still unexcavated, finds indicate that it underwent several building periods and continued in use into the third century A.D. During this period of the harbor's life, the fishery in the Cosa lagoon apparently was no 149

On the site of the villa at the Tagliata, see Benndorf, "Scavi di Orbetello" (cited in n. 56); E. Galli, "Antiche vestigia nel dominio cosano dei Domizi Ahenobarbi," Historia 1 (1927) pp. 2224; P. Raveggi, "Orbetello-Ritrovamenti archeologici nel territorio Cosano," NSc 3 (1927) pp. 204-210; Levi; R. Raveggi, "Ville imperiali romane nell'agro cosano," Maremma 8 (1933) pp. 5-6; Santangelo, p. 70; Dyson, JFA, pp. 261, 267, no. 78. Although Dyson places this villa in his late Republican group, he notes that the bulk of material comes from the first and second centuries A. D. Our survey of the visible architectural remains of the villa site revealed nothing earlier than the Imperial period. Gazda's study of the beach walls indicates at least two different periods for their construction, beginning in the second half of the first century B.C. See below, chapter four. Also note brick stamps from the reign of Trajan, previously found at the Tagliata (Bace, p. 149, A5). The ancient remains mentioned by Levi in his 1927 survey of the site may well have been destroyed in the building of the modern parking area along the beach and the hotel U Pescatore on the northwestern bank of the ancient lagoon at its southern end. Levi also noted a vast amount of colored marble fragments, painted stucco with geometric designs, some column fragments, clay antefixes, and palmette decoration. Pottery sherds included impasto, Arretine, and black glaze wares as well as unidentified brick

33

longer functioning, at least on a large scale. Walls from the villa (Wall O) block one of the major inlet/ outlet channels leading from the sea into the lagoon between polygonal Walls A and B (Fig. IV-21). Who was the owner of this extensive and elaborate maritime villa? A funerary inscription of the late first or early second century A. D., found on the villa site and recently published by Manacorda, identifies an Imperial freedman of Claudius or Nero, a certain Clemens, who was procurator or manager of the villa.150 This evidence implies that the villa was at that time Imperial property. At least two other Imperial holdings are documented for the Cosa area. Suetonius records that Vespasian was brought up by his paternal grandmother, Tertulla, on her estates at Cosa.151 Also, it is assumed that the large property of the Domitii Ahenobarbi in the Cosa area was passed on to Nero, their descendant. Nero also honored the hill town of Cosa with the odeon built into the old basilica in the forum (Fig. VII-8).152 The still unknown owner of the port villa appears to have used part of the lagoon fishery for his own table, following a favorite pastime of the rich, whose elaborate development of pisciculture provoked the ridicule of Roman authors from the late Republic and early Empire. 153 The structure abutting the earlier embankment wall of the western lagoon (in Area YZ), identified here as a fountain house, appears to date from early Imperial times and was used into the third century (Maps 7, 8).154 What happened to the Sestii and their property at Cosa during Imperial times? We do not know, but Will has traced the history of this plebeian noble stamps. From inside the cut of the Spacco della Regina he identified an aeneolothic ax head of yellowish green serpentine found by a shepherd. These finds were given to the Orbetello Museum and the Archaeological Museum in Florence. Also in the Orbetello Museum, from the area of the Tagliata, is a Hellenistic statuette of the goddess Ceres seated and holding a cornucopia (Fig. HI-Il). 150 Manacorda, Athenaeum, pp. 86-87, fig. B, no. 18. Another fragmentary inscription from the port site is recorded by Manacorda, Athenaeum, p. 97, no. 31 and Bace p. 209, bearing the name "Ulixses," probably that of a slave. Bace suggests a date in the second century A. D. for the latter inscription. 151 Suetonius Vespasian 2.1. 152 Two inscriptions found in the basilica in the forum at Cosa are probably to be restored as dedications to Nero. Cosa Inv. CC 876 a-h, CD 259 a, b, and CD 260; CC 875, Bace, pp. 75-76, no. IIA2 and p. 80, no. HBl. 153 For example, see Cicero Att. 1.19.6, 1.20.3, 6.20.3; Columella Rust. 8.16, 8.17; Corcoran, "Roman Fishponds," pp. 3743; T. H. Corcoran, "Fish Treatises in the Early Roman Empire," CJ 59 (1964) pp. 271-274. 154 See Gazda, chapter four, and Gazda and McCann, chapter seven.

34

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

family back to the first half of the second century B.C. or late third century and into Augustan times.155 The elder Lucius Sestius was tribune around 91 B.C. His son Publius first married the daughter of C. Albinius, a senator before 60 B.C. and perhaps the owner of the Albiniana pottery factory of the early first century B.C., known from amphora stamps. A possible ancient pottery kiln has recently been identified at nearby Albinia (modern Albenga).156 A close friend of Cicerco, Publius was eloquently defended by him in 56 B.C.157 Publius' son, the younger Lucius, became consul suffectus in 23 B.C. Will believes that he was the owner of the "splendid ships" ("navigia luculenta") referred to by Cicero.158 The wealth of this important Cosan family thus apparently derived from shipping, potteries, wine, and probably garum, the popular fish sauce of antiquity regularly made from the guts of fish as a byproduct of a saltery.159 Will has identified amphora Type 5 and perhaps Type 24a as garum containers, and she connects them with the Sestius pottery that also produced amphoras of Types Id and perhaps other forms of Greco-Italic jars, 4a, and 4b. 160 Will further notes the repeated stamp (SEX DOMITI) on Type 24a and suggests the interesting possibility that this stamp may refer to the famous family of the Domitii Ahenobarbi, who also owned large estates in the Cosa area.161 Perhaps the powerful Domitii Ahenobarbi took over the Sestius factory sometime in the later first century B.C., or at least jointly controlled it. Although we have no proof that either of these important families was involved with the actual building and financing of the fishery and port at Cosa during its floruit, the evidence of the amphoras points to the Sestii as the first controllers of this port and its export industries. They would have had both the interest to develop the extensive harbor and fish-

ery complex and the means to employ engineers who were familiar with the latest materials, techniques, and machinery. These include the first dated use of hydraulic concrete (probably originally developed in the area of Puteoli), composed of pozzolana and tuff, and the oldest known wooden, geared water-lifting apparatus. The water-lifting device further reveals the sophisticated level that pumping technology had reached by the late second century B.C. when it was adapted at Cosa to serve in a largescale hydraulic project for the fishery.162 It is against this framework of social, economic, and technological change that the detailed history of the Portus Cosanus and its fishery must be viewed. Its greatest days of prosperity were closely linked to the fortunes of the famous Sestius family of Cosa, who were the key force for its continuity, strength, and wealth. The site provides new proof that as early as the second century B.C., the Portus Cosanus was a focus for trade in the western Mediterranean and nobles were actively engaged in it. In fact, the Sestius family probably developed the first truly Roman amphoras sometime after the First Punic War. Will's Type Id, discussed here in chapter nine, is found more frequently at Cosa than at any other site in the Mediterranean and was one of the first commercial amphoras produced by the Romans for international trade. The evolution of the Sestius amphora jars further reveals the development of largescale manufacturing, diversification, merging, and organized methods of distribution that included monopolistic practices. It has been widely assumed that Roman nobles of the later Republic were not interested in overseas trade, owing partly to evidence from the Lex Claudia of 218 B.C., which limited to a capacity of three hundred amphoras the size of ship that a senator could own. 163 The Sestius finds from

155 See Will, chapter nine, and Will 1979, pp. 347-350. Also for the Sestii factory in Rome, see M. Steinby, "I bolli laterizi degli Antiquari del Foro e del Palatino," Atti delta Accademia Nazionah iei Lincei. Classe di scienze tnorali. Memorie 7, ser. 8, fasc. 3 (1974) p. 107. 156 Will 1979, p. 348, n. 34, and p. 349, n. 35; Peacock 1977, pp. 266-267. 157 Cicero Pro Sestio. See E. S. Gruen's discussion in The Last Generation of the Roman Republic (Berkeley, 1973) pp. 300-304, which is also an excellent study of this period as a whole. 158 Will 1979, p. 349. Cicero Au. 16.4.4. 159 Phny NH 43.93; Manilius Astronontica 676-681. McCann 1979, p. 403, n. 22, p. 405, n. 27. The basic study on the garum industry is Ponsich and Tarradell. For the latest study on garum see Curtis and below, n. 197. 160 See Will, chapter nine, for discussion of these forms as well as Type 16 (Cat. A271) and miscellaneous fragments Cats. A302 and A303. Will uses these fragments made of the typical Sestius clay as evidence for a possible revival on a small scale of the Sestius factory somewhere in the area but not at the port in the latter

first century A.D., after the demise of Type 4b in the second half of the first century B.C. 161 Ibid. 162 See J. P. Oleson, chapter five; J. P. Oleson, "The Spring House and Water Lifting Device in the Roman Port of Cosa," AJA 80 (1976) p. 194 (abstract bound separately); Oleson, WaterLifiing Devices, pp. 201-204, 360-380. ' " Livy 21.63.3. Scullard, pp. 169, 309; ESAR, I, pp. 103, 407; D'Arms 1980, p. 77. Instead, the law appears to reflect the growing power of the equestrian merchant class, who feared competition from senators already actively involved in commerce in the late third century B.C. For discussion of the meaning of this law, see C. Wikander and O. Wikander, "Republican Prosopography: Some Reconsiderations," Opuscula Romana 12 (1979) pp. 10-12. Cf. M. Cary, History of Rome (London, 1960) p. 261 and n. 11, who suggests that the law was actually an invention of a later annalist. For recent comments on nobles and trade, see Will 19772, pp. 294-295; Will 1979, esp. pp. 347-350; Will 1982-2. Cf. D'Arms 1981 and Will's review in Archaeology (cited in n. 7).

I: HISTORY AND TOPOGRAPHY

the Portus Cosanus provide overwhelming evidence to the contrary. The Sestii were among the pioneering individuals of the late Republic who, in building their own vast fortunes, first developed for the Roman and western world economic practices still in wide use today and laid a network of trade routes linking Italy to her western neighbors. With the establishment of the Empire the active commercial life of this relatively small but uniquely important harbor and fishing complex was over. Transformed by the Imperial maritime villa on its shore, as well as Succosa, the settlement founded between the port and the Via Aurelia, the port is now used only for import. The site thus reflects in microcosm the economic, sociological, and technological developments of the macrocosm of which it was an integral part.

PORT AND FISHERY MANAGEMENT

The excavations of the port of Cosa and its attached fishery have posed problems for which neither the archaeological remains nor the ancient sources have thus far yielded definite answers. Questions naturally have arisen concerning the ancient control and management of the excavated complex. Who built, at considerable expense, the large harbor works with its attached fishery complex? Who controlled them, and how were they managed from day to day? Was the Roman state involved at any time, and to what extent did the municipality of Cosa exert authority? What was the relationship between private individuals, including the fishermen, and the public sector? Who profited from its trade and resources? Answers to some similar questions about the handling of the huge grain trade during the Imperial period have been offered.164 But what applies to the management of grain—the essential imported staple for the Roman people—does not necessarily apply to other products. Likewise, at the great Imperial harbors of Ostia and Puteoli through which the grain fleets passed one might well expect to find Imperial administrative officials not present at the smaller harbors of Italy. It is best, therefore, to look for evidence first in 164

See Rickman, "Grain Trade" (cited in n. 139) pp. 261-275. C. M. T. Mommsen and P. Krueger, eds., Corpus Iuris Civilis, 16th ed. (1954); C.M.T. Mommsen, ed., Digesta Iustiniani, 2 vols. (1866, 1870); The Civil Law, trans. S. P. Scott, 17 vols. (Cincinnati, 1932); Fenn, Right of Fishery (cited in n. 137) pp. 332. 166 Di?. 1.8.pr. and 1.8.1; Dig. 18.1.51; Dig. 41.1.1; Seneca Ben. 4.28.3. 167 Di?. 41.1.14.pr. and 41.1.14.1; Dig. 43.8.2.9; Inst. lust. 165

35

the earliest documents defining Rome's attitude toward the sea and its resources. The earliest codification of Roman maritime law is found in the Corpus Iuris Civilis of Justinian, which laid the foundations for all later western jurisprudence in this area.165 A sixth-century compilation of works of earlier jurists, Justinian's Institutiones and Digesta reflect long-established beliefs. It seems safe to assume that the views expressed therein also applied to Cosa of the Republic. From the beginning, Rome viewed the sea and its shore—to the high-water mark of winter tides— and the fish in the sea as held in common by all men, res communes, and thus not owned by the state.166 Fishing rights, ius piscandi, included the right to store and repair nets and even to construct shelters on the shore.167 If an individual constructed pilings in the sea and built upon them, they were recognized as the property of the individual, res privatae.168 But it also followed: "Nothing shall be constructed in the sea, or on the shore of the same, by which a harbor, anchorage, or the course of navigation may be obstructed," res publicae;169 and "AU rivers and ports are also public, and therefore the right of fishing in a harbor or in a stream is common to all," ius gentium.170 Thus, jurisdiction over harbors as res publicae ultimately lodged in the state, but the state could not claim ownership. AU Romans were free to use port facilities and to catch the resources of the sea as res nullius. Nevertheless, individuals could build and claim ownership over structures in a port or along the shore where they could bring in and clean their fish.171 Despite the free status of the sea, the state did control certain harbors and apparently some fisheries in various parts of the Roman world. 172 The extensive excavations at Portus (Ostia), the Imperial port of ancient Rome, have produced special evidence of its management by the state. G. W. Houston, in a study of the administration of Imperial harbors, notes the presence of Imperial officials at Ostia, first quaestors and later procurators, who supervised both the port and the grain trade flowing up the Tiber to the capital city.173 In the absence of a merchant marine, additional help in the administration of these activities came from the collegia, or commercial guilds, and 2.1.2; Inst. lust. 2.1.5. 168 Di?. 41.1.30. 169 Di?. 43.12.17. 170 Inst. lust. 2.1.2. 171 For a summary of this evidence, see Fenn, Right of Fishery (cited in n. 137) pp. 13-16. 172 Curtis, pp. 47-48; Bohlen, pp. 44-51. 173 Houston, pp. 157-171; Rouge, pp. 210-211.

36

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

their officers. A similar administrative system may also have existed for the port of Puteoli,174 which received the grain ships from Alexandria during the late Republic and early Empire before the port of Ostia was built by Claudius and Trajan. For other Italian ports, however, Houston found no evidence of Imperial officials of any kind, other than military personnel. He concludes that Ostia and Puteoli are atypical from the point of view of harbor administration and that most ports, at least during the early Empire, were left in the hands of local administrators. One therefore may assume that the building of most Roman harbors, as well as the continued repair of their facilities, was largely carried out by local governments or private citizens. In cases where we know of Imperial construction, such as Ostia, the actual building of the facilities was probably contracted for by private individuals or companies.175 We know that this was the case for the Republican port at Tarracina, which, according to Livy, the censor Marcus Aemilius Lepidus had built with state funds in 179 B.C., and for a harbor and bridge built over the Tiber in the second century B.C.176 Actual evidence for municipalities building their own harbor works is slight, but Houston suggests on the evidence of wax tablets discovered near Pompeii that the late Republican port at Puteoli may have been so financed.177 These documents appear to show that the city of Puteoli owned onshore grain warehouses that it rented to private individuals. The city may also have been involved in the building of facilities for unloading the large grain ships. In certain cases there is also evidence that ports were built by private individuals. For example, an inscription from Parenzo on the Adriatic documents that a certain T. Abudius Verus built or repaired a mole in the late first century A. D.178 From this combined evidence, Houston concludes that the work of building Imperial ports, keeping them in repair, and managing com174

Houston, pp. 162-163. Ibid., p. 164. 176 Livy 40.51.2, 40.51.4. 177 Houston, p. 164; L. Bove, "A proposito di nuove tabulae Pompeianae," RendNap 44 (1969) pp. 25-51; Casson, "The Role of the State" (cited in n. 140) pp. 26-29; cf. Rickman, "Grain Trade" (cited in n. 139) p. 270. 178 Houston, p. 164; CIL V.328; S. Panciera, "Porti e commerci nell'Alto Adriatico," Antichita Alto Adriatkhe, II (Aquileia e VitaMa) (Udine, 1972), p. 82. 179 Houston, p. 164. 180 ESAR, II, p. 335; M. C. Besta, "Pesca e pescatori nell'Egitto greco-romano," Aegyptus 2 (1921) pp. 67ff.; J. Dumont, "La peche dans Ie Fayoum hellenistique. Tradition et nouveautes d'apres Ie papyrus Tebtynis 701," Chronique d'Egypte, 52 (1977) pp. 125-142. 175

mercial activities "was ordinarily accomplished through a combination of Imperial subvention, local administration, and private initiative."179 It seems reasonable to assume a similar situation for the early years of Roman port building during the Republic. What, then, can be said about the ownership and management of Roman fisheries, which were usually located in port areas? State control over natural resources was not the general rule in the Roman world, but evidence from certain areas indicates that the state received revenues from fisheries. Under the Ptolemies the state controlled the rich fishing industry of Egypt. 180 Not only Egypt's coastal waters but also the Nile and its marshes and lakes abounded with fish of all kinds that supplied a lively export trade. Under the Roman Empire the state continued to receive considerable income from this resource, either directly or by leasing fishing rights to companies and private individuals.181 Papyri further document the existence of private hatcheries. In some cases, temples controlled the fishing rights in nearby lakes, and certain fish were considered sacred.182 For example, the Temple of Artemis at Ephesus, whose properties included both the fishing lagoon and the salt pans, exercised authority over fishing.183 There is some additional scattered evidence for state control of fishing rights and fisheries in certain areas of the western Roman world. An inscription found at Leeuwarden in the Netherlands is of particular importance. 184 It.identifies a Q. Valerius Secundus as manceps (contractor) of a societas (company) that purchased fishing rights from the state for the Atlantic waters in that area. On the Atlantic coast of Brittany, in the bay of Douarnenez, where a number of Armorican salteries have been found, a similar state connection may have existed.185 It has also been argued that the famous fish industries of Spain, like its silver mines in the second century B.C., operated under state ownership. 186 Of special interest in this respect is the famous garum sociorum, "fish sauce of 181 O. Bates, "Ancient Egyptian Fishing," Harvard African Studies 1 (1917) p. 209; ESAR, II, pp. 374-378. 182 J. H. Breasted, Ancient Records of Egypt (Chicago, 19061907), II, p. 883; IV, p. 148; IV, p. 283; Rostovtzeff, SEHRE, II, pp. 688-689, with additional bibliography. 183 ESAR, IV, p. 679; Strabo 14.1.26. 184 Curtis, p. 48 and n. 326; CIL XIII.8830. 185 Curtis, p. 48 and n. 327; R. Sanquer, "Une nouvelle lecture de !'inscription a Neptune trouve a Douarnenez (Finistere) et !'Industrie du garum amoricain," Annates de Bretagne 80 (1973) pp. 216ff. 186 Andre Piganiol, in the preface to Ponsich and Tarradell, p. viii; Curtis, pp. 13-25, 48-49, for garum production and organization in Spain. On ownership of Spanish mines see Broughton 1974, pp. 11-16, with further bibliography.

I: HISTORY AND TOPOGRAPHY

the allies," made in New Carthage and praised by Pliny and other ancient authors. 187 The many inscriptions (tituli picti) on amphora jars, identifying both the contents and the producers, 188 further prove the widespread use of garum in the western Mediterranean in the first two centuries of the Empire. The industry, however, may have been in operation earlier and was apparently still functioning in the late fourth century. A connection with the state for such a long-lived societas is certainly a possibility and has been assumed by Tenney Frank and others.189 On the other hand, R. I. Curtis, in his extensive study of both ancient salteries and the garum industry, points out that the exact nature of the socii at New Carthage is not actually known. R. Etienne has suggested that the company was formed instead for the management of the state saltworks, so essential for any fishing industry before the days of refrigeration.190 In any case, there is no evidence that other fisheries and salteries in Baetica or even the largest installation in the western Mediterranean at Lixus in neighboring Mauretania were owned and operated by the state. Whether the fish-processing business in Spain was under private or state management, it was a highly organized commercial business. One fish product manufactured in Spain was even shipped to Puteoli for bottling.' 91 Roman municipalities sometimes owned fisheries and salteries. Cyzicus on the Sea of Marmora, one of the first areas, according to Strabo, to develop pisciculture, received revenues from its fisheries in the first century B.C.192 The Digest of Justinian also includes laws governing the leasing of fishing rights from municipalities, indicating that this practice must have been well established in other areas of the Roman world. 193 For the most part, however, both the literary and the archaeological evidence indicates 187

Pliny NH 31.94; Martial 13.102. See particularly the recent excavations at Ostia in "The Bath of the Swimmer" and the chronology worked out there for Spanish fish sauce amphoras, Panella 1968, 1970, 1972, and 1973. 189 ESAR, V, p. 292. 190 R. Etienne, "Garum sociorum," Latomus 29 (1970) pp. 297313; Curtis, p. 49. 191 T. H. Corcoran, "Roman Fish Sauces," CJ 58 (1963) p. 209; Aelianus NA 13.6. 192 Bohlen, pp. 47-48; Curtis, p. 48. 193 Di?. 43.14.7. 194 Corcoran, "Roman Fishponds," pp. 37-43; D'Arms, Romans on the Bay of Naples (cited in n. 134) pp. 40-43, 135-138. 195 Cicero Att. 1.18; 6.20.3; Pliny NH 9.81.172; Aelianus NA 8.4; 12.30; Plutarch DSA 8.4; Corcoran, "Roman Fishponds," p. 43 and n. 14. For a summary of the literary evidence, also see J.M.C. Toynbee, Animals in Roman Life and Art (London, 1973) pp. 209-212. 196 T. H. Corcoran, "Fish Treatises in the Early Roman Empire," CJ 59 (1964) pp. 271-274; T. H. Corcoran, "Roman Fish188

37

that the vast majority of fisheries and salt establishments were privately owned and operated. Indeed, the ancient sources are full of the names of wealthy Italian piscinarii who owned elaborate, artificially constructed fishponds connected to their seaside villas.194 Many amusing anecdotes ridicule these villa owners, who cared more for their fishponds than for the affairs of state; some even decorated their pet fish with gold earrings and wept upon their deaths.195 These wealthy maritime villa owners, particularly those associated with the region of ancient Campania in the late second and first centuries B.C., primarily enjoyed raising fish for their own tables and amusement. Yet the many fish treatises from the time of the late Republic and early Empire indicate that pisciculture was widespread and that many engaged in it for commercial profit.196 Fresh fish, pickled fish, dried fish, and salted fish provided the Roman people and the army with their key source of nourishment besides bread. Furthermore, the various fish sauces—garum, liquamen, allex, and muria— satisfied the palate with salty flavor.197 That Roman pisciculture was highly developed as well as lucrative is clear from the ancient writers on the subject. The chief sources are Columella, who wrote his De Re Rustica198 about A. D. 60-65, Varro199 (116-27 B.C.), and Pliny the Elder200 (A.D. 23-79). The stocking and feeding of fish was common practice, as was the transportation of fish spawn from the sea for hatching. 201 In the late first century A.D., when Juvenal complained that the seas off the Campanian coast were "fished out," wrasse, a common shallow-water fish inhabiting rocky reefs and sea grass, were brought from the eastern Mediterranean to stock the waters. 202 Exorbitant prices were paid for certain species offish sold in Rome. 203 Columella continually emphasizes the profitable aspects of fish ermen," Classical World 56 (1963) pp. 97-102; Radcliffe, pp. 63297. 197 Pliny NH 31.42.87; Curtis, pp. 2-4; Corcoran, "Roman Fish Sauces" (cited in n. 191) pp. 204-210; R. Zahn, "Garum," RE (1912) p. 7, cols. 842-849; M. Besnier, "Salsamentum," DarSag IV, 2, pp. 1022-1025; P. Grimal and T. Monod, "Sur la veritable nature du 'Garum,' " REA 54 (1952) pp. 27-38; F. Benoit, "Industrie de pecherie et de salaison," RStLig 18 (1952) pp. 237-307; C. Jardin, "Garum et sauces de poisson de l'antiquite," RStLig 27 (1961) pp. 70-96; R. I. Curtis, "In Defense of Garum," CJ 78 (1983) pp. 232-240. 198 Trans. E. S. Forster and E. H. Heffner (London, 1954) bk. 8.16-17. 199 De Re Rustica, trans. W. D. Hooper (London, 1934) bk. 3. 200 NH, trans. H. Rackham (London, 1956) bks. 9.76.1669.81.172, 31.43-44. 201 Columella Rust. 8.16.1-2. 202 Juvenal 5.92-96; Pliny NH 9.62; Macrobius Sat. 3.16.10. 203 Large mullets were particularly prized: Pliny NH 9.31; Suetonius Tib. 34; Radcliffe, p. 203.

38

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

raising. He advises that "the fishpond is also a source of profit which the head of a household can gain from his country estate" and that anyone owning land near the sea "should establish a source of revenue from the sea."204 The many representations of fishing in Roman art, particularly in mosaics, reveal the central place of fishing in a world bounded by the sea and also illustrate the different methods of catching fish from the shore with lines and from boats with nets, baskets, and amphoras. 205 The method most particularly developed by the Romans for raising fish, and the one that is of special interest here, was the artificially constructed tank or pond. Columella206 and Varro207 distinguish between two types of fishponds—the fresh and the salt. Freshwater ponds are associated with poorer farmers and were the earliest type, found in the volcanic lakes of Bracciano, Bolseno, Vico, and Piedi di Luco. By Varro's time freshwater fish were considered clearly inferior to those raised in salt water. Saltwater ponds were developed at great expense, particularly by the wealthy owners of the maritime villas in Campania in the second and first centuries B.C. Pliny the Elder208 writes that Licinius Murena invented such saltwater ponds sometime before the Marsic War (91-89 B.C.).209 Some of his ponds raised the prized murenae of the eel family, and thus he received his cognomen. 210 About the same time, L. Sergius Orata obtained his nickname from raising giltheads (Spams auratus) and was famous for his construction of the first oyster beds in the bay of Baiae.211 Archaeological remains along the Tyrrhenian coast document the elaborateness of some of these piscinae, which extend out into the sea in complicated designs with compartments for different species of fish.212 Varro, in fact, complains that his generation, "with the same extravagance with which 204

Rust. 8.16.6. Radcliffe, pp. 74-89, 219-221; Toynbee, Animals in Roman Life (cited in n. 195) pp. 212-215; G. Lafaye, "Piscatio et Piscatus," DarSag IV, pp. 489-494; also see Casson, chapter eight. 206 Rust. 8.16-17. 207 Rust. 3.17.2-4. Cf. Aulus Gellius NA 2.20.6, 2.20.7; Plautus True. 32-39. 208 Varro Rust. 3.17.2-3; D'Arms, Romans on the Bay of Naples (cited in n. 134). Cf. descriptions of fishponds operated by the Greeks, from whom the Romans may first have learned about their construction: Diodorus Siculus 11.25.4, fishpond built by the Agrigentines for their tyrant Gelon in the fifth century B.C.; Pliny NH 8.17.44, fishponds of Alexander the Great. 209 Pliny NH 9.80.170. 210 Varro Rust. 3.3.10. 2,1 Pliny NH 9.79.168. Some of his oyster beds were even artificially heated to stimulate growth. 212 Schmiedt, Il liuello antico; L. Jacono, "Note di archeologia marittima," Neapolis 1 (1913) pp. 354-371; L. Jacono, "Pescina in littore constructai," NSc 21 (1924) pp. 333-340. Note especially 205

it extended the boundaries of its warrens, has thrust fish-ponds to the sea, and has brought into them whole schools of deep-sea fish."213 Besides this type of saltwater pond immersed directly in the sea, either cut out of the bedrock or constructed with concrete and pozzolana, the ancient sources tell of brackish ponds set back from the ocean but connected to it by channels. In this way fresh and salt water mixed to control the circulation and temperature so necessary for pisciculture.214 Near Naples the volcanic lakes of Lucrinus and Avernus in the coastal area of Baiae and Puteoli were so developed, and indeed, this whole area of the Campanian coast was famous in antiquity for its fish, oysters, and luxurious seaside villas.215 Varro writes about the elaborate fishpond of L. Licinius Lucullus and his brother M. Lucullus: But while he [Licinius Lucullus] was building near Baiae he became so enthused that he allowed the architect to spend money as if it were his own, provided he would run a tunnel from his ponds into the sea and throw up a mole so that the tide might run into the pond and back to the sea twice a day from the beginning of the moon until the next new moon, and cool off the ponds. 216 Such an arrangement is strikingly similar to the complex found at Cosa. Varro further writes that upon Lucullus' death his fishponds were sold for an enormous sum. 217 Fishpond owners raised fish in large quantities, judging from Pliny's report that Gaius Hirrius furnished Caesar with six thousand eels for his triumphal banquets. 218 Thus these piscinarii of Campania gained nicknames, wealth, and fame—preserved for us in the lively accounts of the ancient authors. At the elaborate fish tanks of the early Imperial villa at Astura: Schmiedt, Il livello antico, pp. 108-120; F. Castagnoli, "Astura," StRom 11 (1963) pp. 637-644; F. Piccarreta, Astura, Forma Italiae, Regio I, vol. 13 (Florence, 1977); L. Quilici, "Il problema di Torre Astura," Italia nostra 75-76 (1970) pp. 18-21. Cf. Varro Rust. 3.17.4 for ponds with compartments like paint boxes for keeping varieties offish separate. For an Imperial fish preserve, see Oppian Halieutica 1.60-70. For further bibliography on piscinae throughout the Mediterranean, see G. Lafaye, "Viviers," DarSag V, pp. 959-962; K. Schneider, "Piscina," RE 20 (1950) cols. 1783-1790; Scranton, Kenchreai I, pp. 25-35. 2,3 Rust. 3.3.10. 214 Columella Rust. 8.17. 2,5 D'Arms, Romans on the Bay of Naples (cited in n. 134). 216 Varro Rust. 3.17.9; Pliny NH 9.80.170. 217 JiHJt. 3.2.17. Marcus Cato, who took over the guardianship of Lucullus' ponds, sold the fish for 40,000 sesterces. 2,8 NH 9.81.172. Pliny also writes that the estate of Gaius Hirrius sold for 4,000,000 sesterces because of its fishponds.

I: HISTORY AND TOPOGRAPHY

no point, however, do the ancient writers indicate that the state controlled this profitable occupation in Italy. Fish raising continued into the late Empire, when Macrobius tells of fishponds at Rome. 219 Likewise, the extensive manufacture of fish products—salsamenta (salted fish) and garum—appears to have been in the hands of private individuals, either alone or in partnership (societas), with the possible exception of the famous and expensive garum sociorum from New Carthage. 220 Garum was known in the Greek world in the fifth century B.C. but is not documented in the Roman world before the time of Cato (234-149 B.C.), when it was considered a luxury.221 The name garum does not appear in Latin literature until the mid-first century B.C. in Varro's De Lingua Latina.222 By this time, Italy must have been producing its own varieties of fish sauces. Four are mentioned in the ancient sources, but their distinction is not always clear. Garum was the main product and is often used interchangeably with liquamen. Allex, made from the sediment of garum and muria, was evidently a byproduct of liquamen.223 In ancient times no part of the fish was wasted. Garum was made by soaking the guts offish in brine and allowing them to ferment in the sun for two or three months. 224 The final product was a liquid, probably quite thick. It could be diluted with wine, vinegar, oil, or water and given additional flavor by 219

Sat. 3.15.7. See references cited in n. 186 above. 221 Curtis, p. 35; Cato De Agricultura 58. 222 Curtis, pp. 35-36; Varro De lingua Latina 9.40.66. 223 Curtis, pp. 2-4, with further bibliography on garum. On Allex see, R. I. Curtis, "Negotiatores Allecarii and the Herring," Phoenix 38 (1984) pp. 147-158. 224 The earliest Roman writers describing the process of garum production are Manilius in the reign of Tiberius (Astronomica 5.664-681) and Pliny the Elder, who died in A.D. 79 (NH 31.9395). I am grateful to Curtis for these references. 225 NH 31.43.94. 226 De Re Coquinaria. Curtis, p. 3. 227 Pliny NH 31.44.96-97. 228 I. M. Mackie, R. Hardy, and G. Hobbs, "Fermented Fish Products," FAO Fisheries Reports, no. 100 (Rome, 1971) pp. 1-54. I am grateful to R. G. Thomas of FAO for this reference. 229 NH 31.43.94, 31.44.95. 230 MaHiIiUS Astronomica 5.676-681. For dolia found at the site of the saltery injavea (Alicante), see G. Martin, "Las pesquerias romanas de la costa de Alicante," Trabajos de arqueologia dedicados a D. Pio Beltran, Papeles del laboratorio de arqueologia de Valencia, no. 10 (Valencia, 1970), pp. 139-153, fig. 2; G. Martin and D. Serres, La factoria pesquera de Punta d I'Arenal y otros restos romanos de Javea (Alicante) Servicio de investigacion prehistorica. Serie de Trabajos varies, no. 38 (Valencia, 1970). Although presenting evidence of large and small vats and dolia, Martin and Serres conclude surprisingly that not garum but a similar tuna product was produced at the factory at Punta de I'Arenal (ibid., pp. 104-7). Cf. Curtis, p. 17 and n. 96, who argues that garum was a product of the factory and made in the vats and dolia. He cites the evidence of Manilius that all parts of the tuna were used and placed 220

39

adding honey, herbs, and spices. Pliny tells of a garum liquor intended for drinking, 225 but its main use was obviously as the chief condiment for Roman foods, satisfying the desire for a pungent, salty taste. Apicius, in his ancient cookbook, cites over 350 recipes using garum. 226 The ancient authors also relate its use as a cure-all applied internally and externally for both humans and animals. Garum was recommended for dysentery as well as constipation, sciatica, and ulcers. It was also advised for the healing of burns, dog or crocodile bites, and the itch in sheep.227 The closest modern equivalent for this liquid, salty fish condiment appears to be the Vietnamese nuoc-mam sauce also made from the discarded parts offish. 228 Garum was made from many different varieties of large and small fish and crustaceans. Pliny mentions particularly the mackerel (Scomber scombrus), used in the Spanish garum sociorum. Other fish named include tuna (Thunnus thynnus), mullet (Mugil cephalus), wolf fish (Anarchichas lupus), picarel (maenae), and various small fry.229 Different species could also be mixed. Manilius in the earlier first century A.D. gives a particularly vivid description of the processing of garum made from tuna, which were cleaned directly on the shore and the guts placed in large dolia or vats.230 Excavated ancient salteries, found mostly in southern Spain231 but also in North in dolia in the preparation of fish sauce. It is perhaps significant also to note here the dolia jars found in the harbor at Populonia, where Strabo also cites a tuna watch (5.2.6). Could these too be remains of a garum installation on its shores? On underwater excavations at Populonia, see McCann, Bourgeois, and Will, p. 278 and n. 14; also A. M. McCann, "Le ricerche della missione italoamericano (1974) nell'antico porto di Populonia e nelle acque di Pirgi," Atti del Congresso Internazionale di Archeologica Sottomarina, Lipari, Italy, 1976, awaiting publication. On site of Populonia as a whole see most recently M. Martelli, "Populonia," in GIi etruschi, pp. 153-174. 231 For a basic survey of this material, see Ponsich and Tarradell, who include both Spain and Africa. The most recent survey of ancient Mediterranean salteries, from which much of this material has been drawn, is Curtis, pp. 13-25. For discussion of the best preserved Spanish saltery at Baelo (Bolonia) in Cadiz, see Curtis, p. 16; P. Paris et al., Fouilles de BeIo (Paris, 1923) I, pp. 169-186; Ponisch and Tarradell, pp. 85-88; P. MacKendrick, The Iberian Stones Speak (New York, 1969) p. 208; M. Ponsich, "A propos d'une usine antique de salaisons a BeIo (Bolonia-Cadix)," Melanges de la Casa de Velazquez 12 (1976) pp. 69-79. For other Spanish salteries see: A. Garcia Y. Bellido, "La industria pesquera y conserva espanola en la antiquedad," Investigacion y progreso 13 (1942); M. Sotomayor, "Nueva factoria de salazons de pescado en Almunecar (Granada)," Noticiario arqueologica hispanico 15 (1971) pp. 147-178; R. Thouvenot, Essai sur la province romaine de Bitique (Paris, 1940) pp. 540-542. Also see Peacock 1974. For the salteries in Portugal, see Maria Luisa Estacio Da Veiga Affonso Dos Santos, Arqueologia romana do Algarve, vol. I (Lisbon, 1971); Antonio Mesquita De Figueiredo, "Ruines d'antiques etablissements a salaisons sur Ie littoral sud du Portugal," Bulletin hispanique 8 (1906) pp. 109-121; F. Pellati, "I monument! del Por-

40

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

Africa232 and Gaul,233 include such facilities for the making of garum. None, however, dates before the Augustan period and no saltery has thus far been discovered in Italy, although the ancient sources document several. There may have been one in Vibo as early as the fourth century B.C., and apparently there was a fishery at Tarentum in the second century B.C. where a purple dye factory continued into the late Empire. 234 The dye was made from purpleproducing shellfish (murex), and dye works were a regular part of salteries where shellfish were available. The ancient authors further mention that Velia,235 Beneventum, 236 Thurii, 237 Puteoli,238 Cumae, 239 and Pompeii 240 supported fish-processing industries, which would indicate the presence of salteries. Likewise, Strabo's observance of tuna watches at Populonia, Cosa, and Velia is evidence for the existence of salting and garum facilities in these ports. 241 Inscriptions on amphora jars indicate that Rome imported garum from the cities of Antium, Pompeii, and Puteoli.242 There is also evidence that Pompeii exported its garum to Delos and to

Athens in the first century A.D.243 The archaeological remains of the fishery at the Portus Cosanus and Will's identification of amphoras of Type 5 and perhaps 24a as garum containers on the basis of their close relationship to the Spanish garum jars from which they derive provide the earliest known evidence of the export of garum from Italy in the late second and first half of the first century B.C. The traditional theory that all narrow-necked amphora jars held wine needs reexamination. Garum, as described by the ancient authors, was clearly a liquid and could have easily been poured into the narrower jars. The manufacture of garum was also considerably more profitable per jar than wine. As Pliny commented: "Scarcely any other liquid except unguents has come to be more highly valued, bringing fame even to the nations that make it."244 Pliny reports further that the famous garum sociorum sold for 1,000 sesterces for two congii or 4,000 sesterces for one amphora while even the choicest of wines did not sell for more than 400 sesterces in 89 B.C.245 Some types of garum, however, must have been

togallo romano," Historia 5 (1931) pp. 214-216; A. I m , "As ruinas de tanques de saiga de peixe encontradas recentemente em Olhao," lndustria portuguesa 273 (1950) pp. 727-732; O. Da Veiga Ferreira, "Algumas consideragoes sobre fabircas de conservas de peixe da antiguidade encontradas em Portugal," Archive de Beja 23-24 (1966-1967) pp. 123-134. 232 Ponsich and Tarradell, pp. 9-77; Curtis, pp. 25-30, with full bibliography; A. Jodin, Les etablissements du roi Juba II aux lies Purpuraires (Mogador) (Tanger, 1967) pp. 68-71, 173-180, 256-261. Lixus is the largest installation so far excavated in the western Mediterranean and the only African city to strike a coin with the representation of a tunny on its reverse: Ponsich and Tarradell, p. I l l ; M. Ponsich, "Perennite des relations dans Ie circuit du Detroit de Gibraltar," Aufstieg una Niedergang der romischen Welt, vol. II, part 3 (Berlin, 1975) p. 675. For the site of Lixus as a whole, see "Lixus," PECS, p. 521 (M. Euzennat). The best preserved saltery in North Africa is at Cotta; Ponsich and Tarradell, pp. 5568. Other excavated sites include Cartenna (M. Leglay, "Cartennae," FA 9 [1956] p. 377); Tipasa (W. Schleiermacher, "Tipasa," Germania 43 [1965] pp. 165-168; J. Baradez, "Nouvelles fouilles a Tipasa," Libyca 9 [1961] pp. 92-96; Baradez, Tipasa, ville antique de Mauretanie [Alger, 1952] pp. 39-40; Baradez, "Quatorze annees de recherches archeologiques a Tipasa (1948-1961)," RAfr 105 [1961] p. 235, for expansion of saltery to include baths); Neapolis (J. P. Darmon, "Niapolis," Africa 2 [1968] pp. 271-83); Sullecthum (L. Foucher, "Note sur l'industrie et Ie commerce de salsamenta et du garum," Actes des 93' congres national des societes savantes, Tours, 1968, Section d'archeologie [Paris, 1970] pp. 17-21); Rusicade (J. Lassus, "Chronique l'algerienne en 1956," Libyca 5 [1957] p. 147). The industrial complex of Rusicade includes cistern, basins, and channels. 233 Curtis, pp. 30-35, with full bibliography. The most extensive archaeological work has been done around the Bay of Douarnenez on the coast of Brittany, where over fifteen ancient fish salteries have been found: P. Merlat, "Ploare-Douamenez," Gallia 15 (1957) pp. 186-190; R. Sanquer and P. Galliou, "Garum, sel et salaisons en Armorique gallo-romaine," Gallia 30 (1972) pp. 199223; R. Sanquer, "Bretagne: Douarnenez," Gallia 33 (1975) pp. 349-350; 35 (1977) pp. 355-360; Sanquer, "L'inscription a Neptune" (cited in n. 185) pp. 215-236; P. Galliou, "Les industries de

salaisons en Amorique romaine," Caesarodunum 10 (1975) pp. 141147. For the villa at Villepey-le-Reydissard near Frejus, which may have had provisions for producing garum, see A. Donnadieu, "Le fouilles des ruines gallo-romames de Villepey (Villa Podia) pres Frejus (Forum JuIn)," Memoires de I'institute des fouilles de Provence et des Pnalpes 1 (1926-1929) pp. 129-149; A. Grenier, "Le garum de Frejus," REA 34 (1932) pp. 291-292; P. A. Fevrier, "Les recents travaux archeologiques effectues a Frejus (Var)," BAntFr (1976) pp. 26-27. For remains of fish installations within the Gulf of Lyons, see F. Benoit, "L'economie du littoral de Ia Narbonnaise a l'epoque antique: Le commerce du sel et les pecheries," RStLig 25 (1959) p. 104; Benoit, "L'archeologie sous-marine en Provence," RStLig 18 (1952) pp. 290-297; Benoit, "Les abbayes du sel. L'heritage antique du delta au moyen-age," Delta 3 (1961) pp. 17-32; Benoit, Recherches sur I'Helttnisation du Midi de la Gaule (Aix-en-Provence, 1965) pp. 210-211. 234 Curtis, p. 35 and n. 233; Pliny NH 31.73.84, 9.137 (Tarentum). For the purple dye industry in Tarentum, see Curtis, p. 26 and p. 37, n. 250. 235 Strabo 6.1.1 (Velia). 236 Pliny NH 32.19 (Beneventum). 237 Pliny NH 31.94 (Thurii). 238 Phny NH 35.45 (Puteoli). 239 Strabo 5.4.4. (Cumae). 240 Pliny NH 31.94 (Pompeii). For nearby salt pits at Herculaneum, see Columella 10.135-136; A. Maiun, "Note di Topografia pompeiana, II. Le 'Salinae Herculeae,' " RendNap 34 (1959) pp. 79-81. 241 Strabo 5.2.6 (Populonia), 5.2.8 (Cosa), 6.1.1 (Velia). Cf. Aelinaus NA 13.16, who says that the people of Liguria and Italy engaged in catching tuna. 242 See Curtis, pp. 36-37, nn. 239, 246, 247; CIL XV.4712 (Antrum); CIL XV.4687-4688 (Puteoli). Also, Athenaeus writes that Antium exported fish to Rome (6.224c). CIL XV.4686 (Pompeii). 243 Will, in her forthcoming publication on Roman amphoras in the Athenian Agora series, will cite evidence of Umbricii Scauri jars at the Athenian Agora as well as at Delos. 244 Pliny NH 31.43.94. 245 Pliny NH 31.43.94. Compare the prices of wine, which varied greatly according to quality, type, and period. We know that

I: HISTORY AND TOPOGRAPHY

41

sold cheaply and in quantity, judging from the extensive export trade from southern Spain, which appears from the evidence now known to have largely controlled the market during the first two centuries of the Empire. 246 This entire subject needs further study, supported by systematic exploration of the Italian seacoast for remains of fishing industries. Little is known about the actual organization of the fish-processing industry in Italy. Our best evidence comes from Pompeii, cited by Pliny as one of the chief centers, along with Clazomenae and Leptis, for garum production in the Roman world during the first century A.D.247 Although no saltery has yet been discovered in Pompeii, the many tituli picti on garum containers giving the names of private individuals indicate that this business was privately operated and not controlled by the city or state. The most frequently repeated family name on the jars is that of the Umbricii Scauri, who apparently grew rich from the garum industry, judging from the remains of their impressive family tomb and home. 248 Other evidence from Pompeii is significant to a study of the fishing industry. An election notice indicates that an organization of fishermen existed which urged the support of an aedile.249 Such guilds of fishermen and fish merchants were apparently common and are documented in cities such as Ostia, New Carthage, Ephesus, Cyzicus, and Rome. 250 Inscriptions also record combined associations such as the Corpus Piscatorum et Urinatorum251 in Rome, a union of fishermen and salvage divers who re-

covered lost goods fallen overboard, inspected ship hulls, and helped in port construction. Probably this was the same guild that held annual games in Rome. 252 Whether individually or in associations, it is evident that both fish manufacturers and fishermen could earn a profitable living and in turn influence local affairs by campaigning in elections and establishing games. What does all this scattered evidence suggest for the Portus Cosanus and its fishery? No inscriptional evidence from either the port or the town site has been found that suggests state control of these facilities at any time. Nor has any evidence been found indicating municipal involvement. It does seem likely, however, that during at least the early life of the colony the municipal government exercised some jurisdiction over Cosa's rich lagoons, a jurisdiction that extended over the lagoons surrounding Orbetello today. The municipality may have also provided funds for the construction of the port, at least in its earliest period. But by the late second and first centuries B.C., as the archaeological evidence indicates, the Sestii enjoyed a monopoly over trade in the port of Cosa.253 Perhaps they obtained a franchise for the fishing rights in the lagoon from the new municipium (a self governing town whose inhabitants had Roman citizenship) established in 90 B.C. Although none of the excavated jars of the various Sestius types preserves an inscription identifying its contents with garum or other fish products, the jars' close association with the port fishery is suggestive

in 89 B.C. the censors set a maximum price for the best imported wines at 400 sesterces (100 denarii) for one amphora of 26 liters (Diod. 37.3; Pliny NH 14.95). Cheaper wines, such as that produced for a mass market by the Sestii, sold for much less. During the second and also the first centuries B.C., it is estimated that these cheaper wines cost only about 20 sesterces (5 denarii) per amphora (ESAR, I, pp. 93, 403-404). According to Pliny (NH 14.56), wine first began to bring good prices about 150 B.C. after Delos became a free port m 166 B.C., and precisely during the first active trading period of the port of Cosa under the Sestii. Pliny further informs us that the vintage dates on amphoras were first marked in 121 B.C. (NH 14.94), another indication of increased wine trade in the second century B.C. See Will's discussion below in chapter nine, under Type 4. For possible related economic reasons for this heightened time of trade in the late Republic, see K. Hopkins, "Taxes and Trade in the Roman Empire (200 B . C - A . D . 400)," JRS 70 (1980) pp. 101-125. On Gaul's importation of wine from Latium, see L. C. West, Roman Gaul: The Objects of Trade (Oxford, 1935) pp. 172, 181. Cf. Tchernia 1983. On Spain's importation of Italian wine (Falernian) see L. C. West, Imperial Roman Spain: The Objects of Trade (Oxford, 1929) pp. 85-92. According to Pliny (NH 14.56), the price of "Opimian Falernian" in 121 B.C. was 100 sesterces per amphora. 246 Curtis, pp. 18, 38. See particularly the evidence of remains of Spanish amphoras at Monte Testaccio: ESAR, III, pp. 184-185; R. Etienne, "Les amphoras du Testaccio au III·= siecle," MelRome

61 (1949) pp. 151-181. Zevi 1966, pp. 231-233, argues that Dressel amphora types 7-13 contained fish sauce and salted fish products from Baetica in Spain. 247 Pliny NH 31.94. 248 Curtis, pp. 39-43, writes that the name of Aulus Umbricius Scaurus or that of his family, slaves, or freedmen appears on 31 percent of the fish sauce containers found in Pompeii and Herculaneum. On identification of the garum shop of this family see Curtis, "The Garum Shop of Pompeii," Cronache Pompeiane 5 (1979) pp. 5-23; Curtis, "The Salted Fish Industry of Pompeii," Archaeology 37, no. 6 (1984) pp. 58, 59, 73-75; Curtis, "A Personalized Floor Mosaic from Pompeii," AJA 88 (1984) pp. 557-566. 249 Curtis, p. 39 and n. 264; ClL IV.826. 250 Meiggs, Ostia, p. 267; Corcoran, "Roman Fishermen" (cited in n. 196) pp. 98-99. The salt industry at Ostia, however, was controlled by the state (Livy 29.37.3) and given out to contractors. On Roman contracts in general, see Polybius 6.17. For fishermen's guilds in New Carthage see ClL 11.2959. For the powerful fishermen's guild at Ephesus, see Strabo 14.1.26; ESAR, IV, p. 645; Rostovtzeff, SEHRE, II, p. 689; Bohlen, p. 49. On guilds at Cyzicus and the city's early coinage with representations of tuna fish, see ESAR, IV, p. 626; F. W. Hasluck, "Inscriptions from the Cyzicus Neighborhood," JHS 24 (1904) p. 32, no. 43. 251 See the references in J. P. Oleson, "A Possible Physiological Basis for the Term Urinator 'Diver,' " AJP 97 (1976) pp. 22-29. 252 Ovid Fasti 6.237. 253 See Will, chapter nine.

42

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

in itself.254 We also know from Cicero that the Sestii owned ships in the first century B.C., important evidence for their involvement in the transportation of their products and for their need for port facilities.255 It seems very likely that this wealthy family, whose amphoras today still blanket the whole site to the exclusion of other contemporary Roman amphora types, had a hand in the financing of the extensive harbor and fishery complex constructed during the second and earlier first centuries B.C. Their possible use of the Cosa lighthouse for one of their amphora stamps is particularly suggestive in this respect (Cat. A79, Figs. IX-122, 123, 124) as well as the painted inscription CO/SES on the fragment of a Type 5 amphora neck found in the Athenian Agora excavations (Color Fig. 4 and see discussion in chapters nine and nineteen). Whether the actual fishing rights in the lagoon were leased from the municipal government or whether the fisheries were under private ownership, at least during the time of the Sestii, remains a question still to be solved. Finally, although further excavation alone can prove whether a saltery and garum establishment once existed as part of the whole fishing complex at Cosa, what we have uncovered thus far is extremely suggestive. Further architectural remains were found on the eastern embankment of the ancient lagoon to the east of Wall PE (pier 9 and Wall S), as yet not fully excavated (Maps 7, 8). It is also significant that the excavated Aqueduct running from the Spring House eastward across the lagoon connects with pier 9 in this eastern area of the ancient lagoon. Essential to any saltery and garum establishment was an abundant source of fresh water to clean both the fish and the facilities. The odor from garum was indeed a particular source for comment among the ancient writers. 256 If such a saltery did exist—as seems likely on the basis of the large holding and propagation tanks for fish that we discovered to the west and in 254

See Will's discussion in chapter nine. Cicero Att. 16.4.4, and above, n. 7; Will 1979, p. 349. 256 Corcoran, "Roman Fish Sauces" (cited in n. 191) p. 204; Martial 6.93, 7.94, 11.27.1-2; Horace Sat. 2.4.66. 257 See nn. 230 and 231 above and chapter eleven. Doha Cats. MC2, MC3, MC4, Manilius Astronomha 5.664-681. 258 See above, n. 233 (villa at Villepey-le-Reydissard); Curtis, p. 27 (villa and saltery at Mogador), p. 33 (villas at Trequer and Douarnenez), pp. 34, 59. Villas also have been found at Tipasa and Cotta. 259 Curtis, pp. 18, 46; Ponsich, "Une usine antique de salaisons a BeIo" (cited in n. 231) pp. 70-71; Ponsich, "Perennite des relations dans Ie circuit du Detroit de Gibraltar" (cited in n. 232) pp. 676-677. 260 See above, n. 232 (Rusicade in Algeria); Curtis, p. 26 and n. 162, for cisterns at Cartenna and Leglay, "Cartennae" (Cartenna); Jodin, Les etablissements (Mogador); Foucher, "Note" (Sullec255

view of the building of the Spring House, with its elaborate water-lifting machinery, which must have served some purpose besides providing fresh water for ships—this area to the east of the modern drainage canal would appear to be the most likely place. Although none of the rectangular vats commonly used for the making of garum in the known ancient fish factories of North Africa and southern Spain (for example, at Baelo) has as yet been found in the area of the Cosa port, many fragments of dolia have been discovered all over the site. These large storage jars are mentioned by Manilius as one of the containers used for the making of fish sauce and are found in some of the ancient salteries (for example, at Javea and Tipassa).257 During the Imperial period, the life of the port and its fishery was clearly connected with the maritime villa built on its shores. The archaeological evidence collected by Curtis for salteries in Spain, North Africa, and Gaul from this period may be relevant to Cosa. Salteries and garum factories were commonly attached to large villas in which the owner or manager lived.258 Labor was supplied by slaves, freedmen, fishermen, and possibly migrant workers who followed the migrating packs of tuna during the spring season.259 Such factories were invariably placed near the mouth of a river or another source of fresh water and near a salt mine or salt marsh. Cisterns have also been found at some salteries.260 Fisheries also obtained salt from the evaporation of sea water in salt pans. Many salteries contained furnaces to speed up the fermenting process necessary for garum production, or they used the heating facilities of a nearby bath.261 AU salteries were located near a source of fresh fish and often had their own holding tanks. Some also had pottery kilns nearby.262 It is significant to our study that many of these elements common to a salting and garum establishment have been already found at the thum); and Ponisch and Tarradell, p. I l l (Cotta). 261 Curtis, pp. 27 and 33. Furnaces for the most part are lacking in the Spanish salteries but are a regular feature of African fish installations (for example, at Cotta, Tahadart, and Mogador). Tipasa and Sullecthum had baths and salteries, and at Tipasa the saltery was expanded to include a bath (destroyed in the third century A.D.). See above, n. 232. Furnaces are also found with Armorican salteries. 262 Curtis, pp. 18, 59. For example, amphora kilns have been found at Gades and Carteia; Peacock 1974, pp. 232-243. A furnace found attached to a saltery at Sanlucar de Barrameda in Cadiz may have also served as a pottery kiln; M. E. Guerrero, "Sanlucar de Barrameda (Cadiz); fabrica de salazon romana en la Algaida," Noticiaro arqueologica hispanico 1-3 (1952) p. 127. For a possible ancient kiln in the Cosa area at Albinia, see Peacock 1977, p. 268, and Will 1979, pp. 348-349, n. 35. Also Will below, chapter nine, Type 16.

I: HISTORY AND TOPOGRAPHY

43

site of the Portus Cosanus. Other elements may be there, awaiting future excavation. Whether the owner of the Imperial villa in the Portus Cosanus used the much reduced facilities of the fishing lagoon for commercial profit in the first and second centuries, as perhaps the Sestii did before him, cannot be said. The reduction in amphora sherds from the Imperial period implies the continued use of the harbor site into the third century, but on a greatly diminished scale. The greater variety of imperial amphora types indicates the absence of a monopoly controlled by one manufacturer, as in the days of its floruit in the late Republic under the Sestii. Today man's oldest industry continues in the rich

fishing lagoons of Orbetello. These lagoons are now under the joint management of a fishermen's cooperative and the commune of Orbetello. The profits from the fishing industry are divided 70 percent for the fishermen and 30 percent for the municipality.263 Such a fishermen's union shows the continuation of the ancient societas that protects man's natural rights, iure naturale, over the resources of the sea that are still considered res nullius. Perhaps a similar combination of a fishermen's guild, controlled by the Sestii, and the municipal government of ancient Cosa jointly administered the local fisheries, at least during their most productive years in the last two centuries of the Roman Republic.

263 B. Leoni, in Covegno per il risanamento, la valorizzazione e I'utilizzazione delta laguna di Orbetello, 8-9 marzo 1975 (Orbetello, 1979) pp. 39-41. I am grateful to P. Paradise for material on the fisheries and lagoons of Orbetello, including also Risanamento. For the economics of modern fisheries in general, see R. Hannesson, Economics of Fisheries (Oslo, 1978), and for fish farming in general, see P. H. Milne, Fish and Shellfish Farming in Coastal Waters, rev.

ed. (London, 1979). For fishing in brackish water lagoons in the Mediterranean, see De Angelis 1959; De Angelis 1960; R. De Angelis, Brackishwater Aquaculture in the Mediterranean Region, Studies and Reviews 52 (Rome: FAO, 1973); Aspects of Brackish Water Fish and Crustacean Culture in the Mediterranean, Studies and Reviews 57 (Rome: FAO, 1980).

Chapter II. Geography and Geology JOANNE BOURGEOIS

Where people live close to the land, local geology and geography have a powerful impact on their manner of life. Such was the case in the colonial town of Cosa and in its port and fishery, where most building materials were locally derived. It is indeed remarkable that some materials for the hydraulic structures in the port and fishery were imported from a considerable distance.1 The Romans were cognizant of Italy's geologic activity—its volcanoes, hot springs, and earthquakes—which at times erupted with drastic impact. Living on the shores of the Tyrrhenian Sea, the Romans at Cosa would furthermore have witnessed the sometimes violent interaction of land and sea. The promontory on which the town of Cosa was built provided formidable protection from storms, waves, and currents, but the coastal fishing lagoon below was a fragile environment subject to rapid change. The nature of the countryside around Cosa, known as the Ager Cosanus, is now being elucidated by detailed archaeological and geologic studies.2 A geologic and sedimentologic history of the fishing lagoon and harbor area at Cosa is the specific purpose of this report. 3 But in order to understand the ways in 1 See Gazda, chapters four and seven, which discuss the importation of tuff and volcanic ash to construct the underwater portions of the concrete walls. Most of the other walls in the port area and on the hill are constructed of the local bedrock, calcare cavernoso (see text). 2 Dyson, JFA, pp. 251-268. Specific geologic investigations have been carried out by J. de Boer, J. Cohrenwend, R. Thomas, and L. Davis (see Dyson nn. 1 and 13). Also see recent study of Ciacci in GIi etruschi (cited in chapter one, n. 19). 3 This report is based on one month of field work in 1974, as well as detailed analysis and study of records and samples from the Cosa harbor area excavations. I would like especially to thank Elaine Gazda, who was on the site at the time to familiarize me with it and with the excavation records. Professors S. L. Dyson and J. de Boer of Wesleyan University provided useful advice and information on the geography and geology of the region. 4 An excellent general reference on this subject is Barker, esp. chapter two, "The Natural Environment." Also see J. Ward-Perkins, Landscape and History in Central Italy (Oxford, 1965); WardPerkins, "Etruscan Tombs, Roman Roads, and Medieval Villages: The Historical Geography of Southern Etruria," Geographical Journal 128 (1962) pp. 389-405; C. Vita Finzi, Archaeolog-

which the inhabitants of Cosa altered this environment, it is first necessary to explore briefly the geologic and geographic setting of the Ager Cosanus. 4

GENERAL SETTING

Geology Cosa, in the province of Tuscany on the Tyrrhenian Sea (Maps 1, 2), is similar geologically and topographically to the prominent Monte Argentario to the northwest and to other headlands along the northern Tyrrhenian coast. These promontories lie within the structural and physiographic region of the Anti-Apennines, 5 which in the vicinity of Cosa is called the Colline Metallifere, or Tuscan Metallifere, because of the ore mineralization such as is found on Elba.6 In the broad sense, Cosa lies near the southern limit of the structural belt known as the Northern Apennines, which is characterized by large-scale nappes—bodies of folded rock displaced long distances and thrust over other rocks. 7 These structures reflect the complex and tectonically active history of ical Sites and Their Setting (London, 1978). The importance of geologic studies to archaeologic investigations is becoming more and more widely recognized (see n. 13 in McCann, Introduction). See G. Rapp Jr., and J. A. Gifford, "Archaeological Geology," American Scientist 70 (1982) pp. 45-53; D. C. Kamilli, "Archaeological Geology," Geotimes 27 (1982) pp. 18-19. 5 Barker. 6 G. Marinelli, "Le intrusione terziare dell'Isola d'Elba," Societa Toscana di Scienze Naturali, Atti ser. A (Pisa), 66 (1959) pp. 50253; Carta Geologica d'ltalia, Foglio 126 (1969); V. Bortolotti and P. Passerini, "Magmatic Activity," Sedimentary Geology 4, no. 3/ 4 (1970) pp. 599-624; S. Santini, "Le ricerca infrastrutturale nella Toscana metallifera," Industria Mineraria (Rome) 28 (1977) pp. 7786; P. Zuffardi, "Italian Ore Deposits in the Context of Global Tectonics of the Tethys Region," International Association on the Genesis of Ore Deposits. Proceedings of the 4th Symposium 4, no. 2 (1977) pp. 509-514. 7 L. Carmigniani, G. Giglia, and R. Khgfield, "Structural Evolution of the Apuane Alps: An Example of Continental Margin Deformation in the Northern Apennines," Journal of Geology 86 (1978) pp. 487-504; Geology of Italy, ed. G. Squyres (Tripoli, 1974).

II: GEOGRAPHY AND GEOLOGY

45

Italy within the Mediterranean basin.8 Using platetectonics models, which describe the movement of rigid, lithospheric continental and oceanic plates around the surface of the earth, we can characterize the Mediterranean as an area of complex movements of small continental plates between the larger plates of Europe and Africa.9 A late Cenozoic history of the region includes the following events. The Northern Apennines are believed to have formed in Miocene times, when Corsica collided with Italy, an event we can compare on a different scale to the Himalayas, formed when India collided with Asia.10 In Pliocene times the Tyrrhenian basin opened up between Corsica and Italy. Since Pliocene times a belt of volcanoes has developed to the east and south of Cosa (Roman Comagmatic Series).11 During the last two million years (the Quaternary period), the area has been affected by repeated rise and fall of sea level.12 From Pleistocene times to the present the coast of Italy and other coastlines around the world have been subject to eustatic (worldwide) changes in sea level caused by the growth and retreat of continental glaciers during the Quaternary ice age.13 During maximum glaciation, sea level was about 130 m lower than at present because much of the world's ocean waters were locked up in glacial ice. On the other hand, during minimal periods of glaciation, when the Antarctic and Greenland icecaps were smaller, for example, sea level was higher than at present.14 On the tectonically active coast of Italy, uplift or subsidence of the land could also have caused local changes in sea level. In addition, erosion

or deposition can cause changes in the position of the shoreline, as at Talamone, where a bay has been filled in since Roman times, 15 or as at Populonia, where the coastline has retreated due to beachcliff erosion since Etruscan times. 16 Often we can determine only relative changes in sea level, without knowing whether they were caused by eustatic changes, local changes, or both, and it is important to realize that changes in coastline configuration by erosion or deposition may occur with no change in sea level.

8 K. J. Hsii, "Tectonic Evolution of the Mediterranean Basins," The Ocean Basins and Margins, vol. 4A, The Eastern Mediterranean, ed. A.E.M. Nairn et al. (New York, 1977) pp. 29-75; H. Laubscher and D. Bernoulli, "Mediterranean and Tethys," ibid., pp. 1-28; W. Alvarez, "The Application of Plate Tectonics to the Mediterranean Region," Implications of Continental Drifl to the Earth Sciences, ed. D. H. Tarling and S. K. Runcorn (New York, 1973) pp. 893-908; J. de Boer, "Paleomagnetic Indications of Megatectonic Movements in the Tethys," Journal of Geophysical Research 70 (1965) pp. 931-944. 9 Hsii, "Tectonic Evolution" (cited m n. 8); Laubscher and Bernoulli, "Mediterranean and Tethys" (cited in n. 8); W. Lowrie and W. Alvarez, "Paleomagnetic Evidence for Rotation of the Italian Peninsula," Journal of Geophysical Research 80 (1975) pp. 1579-1582. 10 R. Kligfield, "The Northern Apennines as a Collisional Orogen," American Journal of Science 279 (1979) pp. 676-691; G. Sestini, ed., "Development of the Northern Apennines Geosyncline," Sedimentary Geology 4, no. 3/4 (1970) pp. 203-642. " M. Boccaletti and P. Manetti, "The Tyrrhenian Sea and Adjoining Regions," The Ocean Basins and Margins, vol. 4A, The Eastern Mediterranean (cited in n. 8), pp. 149-200; P. DiGirolamo, "Geotectonic Settings of Miocene-Quaternary Volcanics in and around the Eastern Tyrrhenian Sea Border (Italy) as Deduced from Major Element Geochemistry," Bulletin Vokanologique 41

(1978) pp. 229-250; L. Civetta et al., "Eastwards Migration of the Tuscan Anatectic Magmatism Due to Anticlockwise Rotation of the Apennines," Nature 276 (1978) pp. 604-606. 12 Fairbridge, "Quaternary Sedimentation," pp. 99-113, and p. 104, Fig. 1 for suggested expanded geologic time scale for the Quaternary period, the last two millon years. 13 R. W. Fairbridge, "Eustatic Changes in Sea Level," Physics and Chemistry of the Earth, ed. L. H. Ahrens et al. (New York, 1961) IV, pp. 99-185; Fairbridge, "World Sea Level and Climatic Changes," Quaternaria 6 (1962) pp. 111-134; A. L. Bloom, Atlas of World Sea Level Curves (New York, 1976); Quaternary Coastlines and Marine Archaeology, ed. P. M. Masters and N. C. Flemming (New York, 1983). Sea-level changes may also be caused by changes in the geoid (the shape of the earth); see, for example, Earth Rheology, Isostasy and Eustasy, ed. Nils-Axel Morner (New York, 1980). 14 Fairbridge, "Quaternary Sedimentation"; R. W. Hey, "Quaternary Shorelines of the Mediterranean and Black Seas," Quaternaria 15 (1971) pp. 273-284. 15 Bruno, Will, and Schwarzer, pp. 34-43. 16 See section by Bourgeois in McCann, Bourgeois, and Will, pp. 288-293. 17 Ente Maremma, La Riforma Fondiaria in Maremma (Rome, 1966).

Geography From Tuscany to Campania, the coastline of Italy is characterized by long, sandy, crescent barrier islands, beaches, and dunes that connect rocky promontories such as Cosa and the Argentario. Landward of these modern beaches is a coastal plain with lagoons, swamps, and older barriers, beaches and dunes that formed when sea levels were higher or when the land was lower. Scattered inland hills may once have been coastal promontories similar to the hill of Cosa today. Much of the coastal plain near Cosa has been altered in the twentieth century by the Ente Maremma, a program of land reform that reclaims coastal marshlands by filling them in and introducing modern, mechanized agriculture.17 The natural geography and environment of Cosa are significantly influenced by the adjacent sea, which not only provides a means of transportation and a source of food for Cosans but also moderates the local climate. As a result of general circulation,

46

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

the prevailing longshore current along the Tyrrhenian coast flows northward at velocities of one to two knots. 18 Local winds cause variations in this current; tidal ranges are negligible in the Mediterranean (average range 25-30 cm) and have little effect on coastal currents. The strongest storms, which occur in the fall and winter, come from the southeast (scirocco) and southwest (libeccio) and may last two to three days. 19 Other strong winds may come from the west and northwest. In the Cosa region, storms and wind-generated waves coming from the southeast and southwest have the most impact on the coastline, which is protected by the Promontorio Argentario to the northwest (Map 2). There are no major rivers near the Cosa harbor site. The drainages of the Albegna and Osa to the north are separated from Cosa by the Promontorio Argentario, and the smaller Tafone and Fiore rivers are nearly 20 km to the south (Map 2). Only short, small streams drain into the modern Lago di Burano, several kilometers away from Cosa, and only during the rainy season. This drainage pattern is important to the history of the Cosa site for two reasons. First, the lack of surface sources of fresh water near Roman Cosa led to the exploitation of freshwater springs. 20 Second, the absence of any source of fluvial sediments at the western end of the Cosa lagoon allowed the lagoon to remain unfilled for centuries.21

Thus, the winters are mild and wet and the summers hot and dry. Soil types in the Cosa region depend on the subsurface material and on the climate.23 On the sandstones and shales of the highlands, a clay-based regosol develops—a poor soil composed principally of broken-down bedrock. In places on limestone bedrock residual soils (rendzinas) have developed and support good grazing land and marginal agriculture. Elsewhere on limestone bedrock is found the terra rossa, or red earth, a severe residual soil that usually develops in pockets and fissures within the bedrock. Terra rossa, composed of reddish-brown ferruginous silty clay, is characteristic of the coastal Mediterranean climatic belt. The richest soils in the region are those that develop on Pleistocene and modern alluvium, but these areas often require artificial drainage techniques in order to make them arable.

The coast of Tuscany enjoys a climate of the temperate Mediterranean type, dominated in the fall and winter by the Mediterranean cyclone, with average northerly winds and storms from the south, which bring most of the year's rain, about 0.70 m annually.22 The Apennine mountain chain protects the coast from cold Alpine air masses moving down from the northeast. In the summer the Mediterranean anticyclone prevails, with associated weak westerly winds; the summer anticyclone is a divergent air mass that produces very little moisture.

LOCAL GEOLOGY

The bedrock and Quaternary geology of the Cosa area are illustrated in Figs. II—1, 2. 25 Although the bedrock geology in the vicinity is complex, at Cosa it consists of only one formation: the Triassic-age calcare cavernoso, a calcareous to dolomitic limestone. In places this formation is well layered; elsewhere it is highly brecciated, probably as the result of cavern formation and collapse shortly after its original deposition.26 This bedrock weathers typically to terra

18 H. Lacombe and P. Tchernia, "Caracteres hydrologiques et circulation des eaux en Mediterranee," The Mediterranean Sea, ed. D. J. Stanley, (Stroudsburg, Pa., 1972) pp. 25-36; C. Bartolini et al., "Studi di geomorphologica costiera, III, Il Tombolo di Feniglia," Societa Geologka ltaliana, Bollettino 96 (1977) pp. 117-157. 19 See Pongratz for information on this coastline, and V. J. Bruno, "The Mystery of the Etruscan Coastline," Archaeology 26 (1973) pp. 198-212. 20 See McCann, chapter one; J. P. Oleson, chapter five. 21 See discussion in this chapter of coastal geomorphology and historical development of the coastline. For a general discussion of the impact of fluvial sediment supply to coastal environments, see C. Vita Finzi, "Supply of Fluvial Sediment to the Mediterranean During the Last 20,000 Years," The Mediterranean Sea (cited in n. 18) pp. 43-46.

22 G. T. Trewartha, An Introduction to Climate (New York, 1968); Barker; Pongratz; Bruno, "Etruscan Coastline" (cited in n. 19); Risanamento. 23 Barker. 24 Barker; Pongratz. 25 R. Signorini, Note illustrative delta Carta Geologka d'ltalia, Foglio 135: Orbetello (Rome, 1967). The bedrock that makes up the Cosa promontory may extend beneath the breakwater area as a natural, wave-cut platform (J. de Boer, personal communication, 1981). The Romans may have taken advantage of this natural platform. See McCann, chapter three, on the underwater excavations. 26 There is some disagreement concerning the origin of these breccias. For an alternative explanation suggesting the collapse of solution-produced caverns following intense deformation, see

The vegetation in coastal Tuscany reflects these climate and soil conditions. 24 The coastal region is characterized by the holm oak (an evergreen or live oak) mixed with olive, oleander, carob, mastic, and Aleppo pine. In areas where the forests have disappeared or have been destroyed by human intervention, a dense growth of low, bushy plants called garrigue or macchia prevails.

II: GEOGRAPHY AND GEOLOGY

rossa.27 Northeast of Cosa the bedrock consists of a tectonically deformed complex of Paleozoic sandstone and conglomerate (Verrucano), Triassic carbonates (calcare cavemoso), and Mesozoic to Tertiary siliciclastic rocks, including the Galestri and Palombini formations (a Cretaceous sequence of silty and calcisilty rocks) and the Macigno (a sequence of mid-Tertiary siltstones and sandstones). Farther east, at Tuscania and Civitavecchia,28 these older rocks have been intruded and covered by an immense complex of Quaternary-age, acidic-volcanic rocks and sediments, including tuffs, welded tuffs, tufflavas, rhyolite, trachyte, phonolite, and at least one occurrance of pozzolana ash.29 These rocks are particularly rich in feldspars (especially sanidine, also andesine, labradorite), augite (a pyroxene), and leucite; less abundant are aegirine and biotite. Accessory minerals include magnetite, olivine, apatite, rutile, zircon, and sodalite. The composition of these volcanic rocks and of the other types of bedrock in the Cosa area is important for determining the sources of sediments deposited in the Cosa fishing lagoon and harbor area and for establishing provenance of Roman structures and artifacts.30

Faults and Karst Development Since the Pliocene, the region around Cosa has been characterized by an extensional tectonic regime, resulting in numerous small, normal faults trending north-northwest/south-southeast, with associated conjugate fractures oblique to the faults.31 In the bedrock around Cosa there are no major faults, but the limestone is strongly fractured and jointed, producing a blocky appearance. Ground V. Bortolotti, P. Passerini, and M. Sagri, "The Miogeosynclinal Sequences," Sedimentary Geology 4 (1970) pp. 341-444; also see Signorini, Carta Geologka d'ltalia (cited in n. 25). 27 See the description in this chapter of sediment sources. 28 A. Alberti, M. Bertini, G. L. del Bono, G. Nappi, and L. Salvati, Note illustrative della Carta Geologka d'ltalia (alia scala 1:100,000) Foglio 136: Tuscania; Foglio 142: Civitavecchia (Naples, 1970). 29 Pozzolana ash was used in the concrete structures of the port and fishery of Cosa, but it appears to have been imported from the area of Pozzuoli. See Gazda, chapter four, n. 5. 30 See discussion in this chapter on interpretation of the sedimentary sequence; Limngton, chapter fourteen; Barshad and Viani, chapter nine; reports by Trigila and de Boer, chapter sixteen. 31 Signorini, Carta Geologka d'ltalia (cited in n. 25); A. Mori, "I fenomem carsici dell'Orbetellano e del Capalbiese," R. Societa Geografica Italiana, Rome, Memorie 17 (1932) pp. 117-196. 32 See above, n. 20. 33 See Gazda, chapter four, and Map 6. 34 See Gazda, chapter seven, on the reconstruction of the fishery.

47

water traveling along these fractures has dissolved the limestone, widening the fractures and producing some small caverns. This fracturing and dissolution process (karstification) had two important consequences for the Cosa fishing lagoon and harbor areas. First, the fractures were conduits for ground water, which surfaced around the edge of the lagoon at points of contact between the limestone and the lagoonal sediments. Several large freshwater springs are situated in the Cosa harbor and lagoon area.32 Second, three of the karstified fractures in the harbor area—the Spacco della Regina, Piccolo Spacco, and Tagliata33—were adapted wholly or in part by the inhabitants of Cosa for use in connection with the lagoon fishery.34 Although it is difficult to determine bedrock offset across the Spacco della Regina, it has been suggested that some displacement or fault activity has occurred along its length. 35 Evidence for this fault includes the presence of a major spring within the Spacco and the presence of brecciated limestone, which may be fault-related or dissolution-related. Large fallen blocks within the Spacco della Regina may have been dislodged by an earthquake of unknown age. There is also evidence that some structures in the town of Cosa and in the lagoon may have been damaged by earthquake, 36 but the postulated earthquake or earthquakes apparently occurred after occupation ceased.37 The Tyrrhenian region, including Naples and Rome, experiences earthquakes of a magnitude 6 or greater once every 50 ± 25 years.38 Most of these earthquakes would not have affected the Cosa area, and there is no documented evidence as yet that suggests a damaging earthquake in the Cosa region during the period of Roman occupation.39 35

J. de Boer, personal communication, 1981. J. de Boer suggests that the triple arch at the northwestern entrance to the forum in the town was damaged by an earthquake, causing it to fall in large blocks rather than gradually crumbling away due to weathering (Brown 1951, pp. 73-75, figs. 67-69). Other evidence for this earthquake is the nature of the arch collapse—toward the outside rather than toward the center. In the lagoon pieces of Wall YZ and the Spring House may have fallen as the result of an earthquake. Cf. Gazda, chapter seven, reconstruction of the fishery. 37 No occupational materials are found above the fallen structures. Evidence from the geometry of fallen walls on the hill suggests the epicenter of this earthquake was in the Adriatic region (J. de Boer, personal communication, 1981). 38 J. de Boer, personal communication, 1981. 39 There is, however, apparently some evidence at the Villa of Le Colonne excavated by S. L. Dyson in the Ager Cosanus, of a damaging earthquake which he would date in the late second or early third century A.D. (unpublished information, kindly provided by Dyson). Until this evidence is published and its exact nature known, it cannot be used for drawing conclusions here. 36

48

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

Quaternary Sediments The rock formations mentioned above are the sources for Quaternary sediments at Cosa, which include alluvium, detritus, panchina, red sands, and, to be discussed in more detail, Quaternary marine sediments, lithified eolian sands, lagoonal sediments, and modern beach and dune sands. Terra rossa is also Quaternary in age (Figs. H-I, 2). Approximately 10 to 20 m above present sea level is a Pleistocene barrier-lagoon complex. The Pleistocene barrier (sd in Fig. H-I) is marked by partially lithified sand dunes, or eolianites, consisting of crossbedded sand and silt. F. E. Brown, in his initial topographical survey of the area,40 mistakenly stated that these lithified sand dunes were found along the present coastline, where they would have separated the fishing lagoon from the open sea. Rather, they form the landward side of the Holocene-age barrierlagoon complex and do not occur in the immediate vicinity of the harbor and fishing lagoon area, which is instead flanked by limestone bedrock on the north and west (Figs. II-2, 15).41 Behind the ancient barrier (sd) are found ancient lagoonal (tp) and marine terrace (Qm) deposits (Fig. H-I). The modern barrier complex comprises two deposits: the present-day beach, dune, and washoversand complex (s in Fig. II-l), and lagoonal sediments (tp) south of the lithified dunes (sd). The Lago di Burano is the yet-unfilled part of this lagoon. This barrier probably formed about 5000 B. P. (before present) when the rapid postglacial rise in sea level decreased markedly. 42 As can be seen in Figs. II-l, 2, and Map 7, the lagoon at the harbor site, when open, would have been very narrow—about 60 m wide except within the channel area (between Walls P and PE)43—because it is flanked on its north side by calcare cavemoso bedrock. Although one might expect that this narrow lagoon would have filled in relatively soon after barrier formation, sedimentologic evidence indicates that it was not yet filled in 40

Brown 1951, pp. 1-113. These eolianites resemble the building stone used for the Temples of the Arx on the hill. 42 Fairbridge, "Quaternary Sedimentation"; Fairbridge, "Eustatic Changes"; Fairbridge, "World Sea Level and Climatic Changes" (cited in n. 13); J. C. Kraft, R. B. Biggs, and S. D. Halsey, "Morphology and Vertical Sedimentary Sequence Models in Holocene Transgressive Barrier Systems," Coastal Geomorphology, ed. D. R. Coates, S.U.N. Y. Publications in Geomorphology (Binghamton, N. Y , 1973) pp. 321-354. 43 See Gazda, chapter four, and Maps 7, 8. 44 See discussion in this chapter on coastal geomorphology and historical development of the coastline. 45 F. P. Gulliver, "Shoreline Topography," Proceedings of the American Academy of Arts and Sciences 34 (1899) pp. 149-258; Mer41

Roman times from causes.44

either natural or

artificial

COASTAL GEOMORPHOLOGY AND HISTORICAL DEVELOPMENT OF THE COASTLINE

Cosa is situated on a bedrock promontory with a barrier-lagoon complex to the east, where the harbor site is located. To the west of Cosa, the promontory of the Argentario, composed of the same bedrock as the Cosa promontory, would be an island except for the famous double (nearly triple) tombolos that connect it to the mainland (Map 2; Fig. 1-2). During periods of higher sea level the Argentario would have been an island, and the history of the development of its tombolos has been the focus of much study. 45 The spit on which the modern town of Orbetello is located must have formed before the two tombolos—Tombolo della Giannella to the north and Tombolo di Feniglia to the south (Map 2)—because open ocean circulation would have been necessary to deposit the spit. The older age of the Orbetello spit is also established by the presence there of beachrock (panchina; Qp on Fig. II-l), a cemented beach deposit formed within the intertidal zone and thus produced during a former period of higher sea level.46 The antiquity of the spit can also be documented by the presence of an Etruscan necropolis and of a polygonal sea wall of undetermined age.47 The rare double tombolos that connect the Promontario Argentario to mainland Italy have been noted by geologists and coastal geomorphologists since the late nineteenth century. 48 In fact, the Italian word tombolo was adopted into geomorphologic vocabulary to define any subaerial bar that connects an island and a mainland, a usage suggested by F. P. Gulliver in 1899.49 Based on their morphology, Gulliver suggested that the Orbetello spit formed first, ciai 1910; B. Lotti, Geologia della Toscana (Rome, 1910); D. Johnson, Shore Processes and Shoreline Development (New York, 1919); Merciai 1929; A. D'Arrigo, Richerche sul regime dei litorale net Mediterraneo (Rome, 1936); Bronson and Uggeri, pp. 1-14; Bartolini, "Studi" (cited in n. 18); A. Brambati et al., "Aspetti sedimentologici delle lagune—fenomeni de erosione dei tomboh," in Risanamento, pp. 18-25, 121-132. 46 J. Bourgeois, "Beachrock," Encyclopedia of Sedimentology, ed. R. W. Fairbridge and J. Bourgeois (Stroudsburg, Pa., 1978) pp. 44-45. 47 Bronson and Uggeri. Also see McCann, chapter one, who accepts a Roman date for these fortification walls. 48 See above, n. 45. 49 Gulliver, "Shoreline Topography" (cited in n. 45).

II: GEOGRAPHY AND GEOLOGY

the Tombolo della Giannella next, and the Tombolo di Feniglia last. Since Gulliver's paper, the tombolos have been discussed principally with respect to archaeological and historical evidence for their development. In the early twentieth century several scholars50 suggested that the formation of the tombolos was largely historical. G. Merciai pointed out, however, the presence of an ancient road on the Tombolo di Feniglia and suggested that the tombolo had existed at least since Etruscan times. 51 On the evidence of a seventeenth-century map of ancient Italy, Merciai believed that at least half of the Tombolo della Giannella had been formed by Roman times but that the Oibetello lagoon was still open to the north. Most recent investigators have followed this reasoning. However, based on evidence obtained in an archaeological survey of the tombolos in 1970, R. Bronson and G. Uggieri asserted that the entire area—both tombolos and the Orbetello spit—were formed before Neolithic times. 52 On the other hand, G. Schmiedt published aerial photographs of the coast in 1970 which suggest that the Tombolo della Giannella was only half formed in ancient times. 53 The evidence of an ancient road on the Tombolo di Feniglia, probably of Roman date, as well as of artificial inlets believed to be of Roman construction does support the hypothesis that this tombolo was complete in Roman times. 54 Indirect evidence suggests that the Tombolo della Giannella was not then complete. Scattered archaeological sites found by Bronson and Uggieri would not contradict the existence of one or more inlets. V. J. Bruno has suggested that the Orbetello lagoon was a single, navigable body of water in Etruscan times, probably with several entrances to the sea.55 The polygonal sea wall at Orbetello may have been necessary to protect it from waves approaching through openings in the Tombolo della Giannella.56 In a study of the modern Orbetello lagoon, A. Brambati and others made a significant observation.57 In noting twentieth-century changes in the morphology of the tombolos, they pointed out that 50

Especially see Merciai 1910; Merciai 1929; Lotti, Geologia della Toscana; Johnson, Shore Processes (cited in n. 45). si Merciai 1910 and Merciai 1929. 52 Bronson and Uggeri. 53 Schmiedt, Atlante, pi. CXXX. 54 Brambati in Risanamento; Bartohni, "Studi" (cited in n. 18). See also McCann, chapter one. 55 Bruno, "Etruscan Coastline" (cited in n. 19). 56 Bocci Pacini, as cited in Bruno, Will, and Schwarzer. 57 Brambati in Risanamento. 58 Frescoes (1580-1583) by A. Danti in the Vatican, Gallena delle Carte Geografiche, Rome.

49

the tombolos could have been complete at some time in past centuries and breached at others. Thus, published maps that reconstruct the ancient Italian coastline may be inaccurate if they are based on the then existing coastal morphology. Lagoon ofCosa Less attention has been paid to the lagoon that lay east of the town of Cosa in Roman times (Laguna di Cosa, Map 2). Two important questions must be answered: How extensive was the lagoon then? Was there an inlet at its western end? These questions can be answered by both historical and archaeological evidence. In the sixteenth century the Italian geographer-artist A. Danti painted at least two maps of the Cosa area (Figs. II-3, 4). 58 Both show that the lagoon of Cosa was much longer than its modern remnant, the Lago di Burano, and extended to the base of the hill of Cosa. One map shows an inlet at the western end of the lagoon (Fig. II-3) and one does not (Fig. II4). Curiously, the Tombolo della Giannella is shown as complete on one map (Fig. II-3) and incomplete on the other (Fig. II-4). A seventeenth-century map (Fig. II—5) by the German geographer and historian Philipp Cluver, reproduced by G. Merciai in 1910,59 inaccurately illustrates the lagoon of Cosa as an estuary (a river mouth) rather than as a lagoon behind a barrier. The map clearly shows an inlet beneath the hill at Cosa and accurately depicts the Tombolo della Giannella as incomplete. 60 A much more accurate representation of the coast is the early nineteenth-century map by G. Inghirami (Fig. II-6),61 which shows the tombolos in 1830 much as they are today. But the lagoon of Cosa (Lago di Burano on map) still remained largely open in 1830 except for fluvial deposits at its western end, where no inlet is shown. The inlet in the center of the lagoon, however, is in the same position as in the most recent maps of the area (Map 2), although more extensive marshes and lagoonal areas are indicated, particularly north of the Cosa lagoon and near 59

Oliver (Cluverius), Italia Antiqua (Batavorum, 1624) II, 1, map between pp. 418 and 419; reproduced in Merciai 1910, with some variations. 60 Other maps showing the Lago di Burano running up to Ansedonia with outflow near the Torre dell'Ansedonia (Tagliata) include those in the Vatican by Jacomo Oddi, 1636-1637; Giacomo Filippo Ameti, Foglio 3, 1696; and Giuseppe Morozzo, Fogh 3 e 4, ca. 1791 (information provided by R. E. Limngton). Also see eighteenth-century map of N. Samson and P. Mortier, reproduced by Ciacci in GIi etrusche, p. 17, fig. 2. 61 G. Inghirami, Carta Geometrka della Toscana (Florence, 1830).

50

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

the mouth of the Albenga River. These marshy areas may have filled in naturally since 1830, although it is likely that they were finally reclaimed by the Ente Maremma. G. Schmiedt used a very similar base map in his reconstruction of the ancient coastline of this area.62 By 1859, when the modern drainage canal was dug, we know that the lagoon was somewhat less extensive.63 But even into the earlier twentieth century, the lagoon was still much longer than it is today. More recently, it has been filled in by melioration projects, except for the modern Lago di Burano, which continues to be used as a fishery.64 The lagoon of Cosa probably remained open for such a long time because there was no local source of fluvial sediments and possibly because the inlet kept water flushing through the area. Compare all these maps with aerial views of Cosa in 1944 (Figs. 1-2, 3, 4; II—7) and a view from the hilltop over the ancient lagoon area today (Fig. II-8).

INTERPRETATION OF THE SEDIMENTARY SEQUENCE

In order to expand our knowledge of the history of the Cosa harbor and lagoon and to produce a reconstruction of the area during its occupation, a more detailed study of the sediments was undertaken.65 This endeavor included a magnetic susceptibility survey and a program of drilling (Map 9), 66 trench-wall description,67 and soil-sample analysis.68 The results of these studies and a general discussion of the sedimentology of the site are presented in later chapters. My purpose here is to interpret the history of the harbor and lagoon area by examining the sequence of sediments deposited therein. The general setting of the western end of the lagoon of Cosa (Map 2) and the surrounding topography is best illustrated by an aerial photograph (Fig. II-7). The Via Aurelia runs east-west through the upper middle section of the photograph. On the present coastline, the former barrier island that separated the Cosa lagoon from the open sea is made up of beach (white area) and dune (dark, vegetated area) deposits. The former lagoon (Laguna di Cosa, Map 2) is underlain by lagoonal muds vegetated by 62 Schmiedt, Atlante, pi. CXXX; Schmiedt, Il livello antico, pp. 25-30. 63 Rodenwaldt and Lehmann, pp. 3-31. 64 See Gazda and McCann, chapter seven. 65 Compare Linington, chapter fourteen, and Barshad and Viani, chapter fifteen. A preliminary report on the sedimentology of the harbor area was presented by the author: J. Bourgeois, "Sedimentation in an Ancient Harbor—Portus Cosanus, Italy,"

lush grass (dark gray). Note also how narrow the Laguna di Cosa becomes at its western end, the site of the fishing lagoon. It is clear that the conditions at the far western end were quite different from those in the rest of the lagoon. Here, the lagoon becomes much narrower and is flanked on the north and west by limestone bedrock rather than by Pleistocene dunes (Figs. 11-11, 15). North of the Via Aurelia the older dune deposits (sd in Fig. H-I) are easily recognized by their hummocky appearance. An oblique air photograph (Fig. II-2) illustrates in more detail the setting and sediments that underlie the Cosa site. Before discussing the sedimentary sequence, I shall consider the various sources for the sediments. Sediment sources There are two major sources for the sediments that compose the harbor and lagoon sequences at Cosa. The most immediate source is the terra rossa, a red, earthy weathering product of the calcareous bedrock that flanks the lagoon and harbor site (Color Fig. 2, and Fig. II-2). Sediments described in the excavation reports as red or reddish brown or as containing red components must have been at least partially derived from slopewash from the west or north side of the harbor. Slopewash was probably not as significant during Roman occupation as at present, however, when the slopes have been largely deforested. The second major source of sediments is the dune, beach, and offshore sand south and east of the site. Sands similar to the beach sands extend at least one kilometer offshore, where they grade into sandy silts and finally to silty muds. 69 The sands are rich (often extremely so) in heavy minerals, which may make up 50 percent or more of their composition. The predominant heavy mineral is augite, a dark green to black pyroxene; magnetite, aegirine, and biotite are usually present in only minor amounts. Fresh, unweathered sanidine (volcanic feldspar), in addition to partially weathered quartz, other feldspars, and lithic fragments, is abundant in the light-mineral fraction. These sands vary in color from light to dark gray, depending on the percentage of heavy Geological Society of American Abstracts with Programs 6/7 (1974) p. 663. An updated report was later presented in 1981; see Bourgeois and Gazda. 66 See Linington, chapter fourteen. 67 See Gazda, chapter four. 68 See Barshad and Viani, chapter fifteen. 69 D'Arrigo, Richerche sul regime dei litorale (cited in n. 45).

II: GEOGRAPHY AND GEOLOGY

minerals, which are dark-colored. These gray sedi­ ments derived from beach and offshore areas are thus easily distinguished from reddish sediments de­ rived from the hillsides. The quartz, rounded feld­ spar, and lithic fragments probably come from local sources—older Quaternary deposits and, ultimately, the complex of Mesozoic and Tertiary rocks imme­ diately east of Cosa. Although no local source of augite or of sanidine is known, a major source is available in the volcanic province to the southeast. Thus, these minerals were transported alongshore and, as the barrier formed, were brought in from offshore, where they had been distributed during pe­ riods of lower sea level. Lagoonal muds may have derived from any one of these sources as well as from the terra rossa whose iron oxides (red) could be reduced to iron sulfides (black) in a stagnant la­ goonal environment. Types of Sediments We may distinguish several types of sediments, genetically and lithologically, in the vertical and hor­ izontal sedimentary sequences at Cosa (Text Figs. Π­ Ι, 2, 3, and 4). 7 0 The deepest and thus oldest sedi­ ments beneath most of the site are dark gray and light gray sands, which were deposited as nearshore sands before the barrier developed. They occur beneath a surface that slopes generally seaward (Text Figs. II2, 3). The sands occur less than one meter below present sea level at the back of the ancient lagoon or north flank of the fishery site (Text Fig. II-4; Figs. 11-10, 11, and 12), where they lie above the bedrock and below alluvium; these deposits may represent an ancient beach. Light gray sands are found at a depth of 3-5 m b.s.l. at most of the drill-hole sites (Text Figs. II-2, 3, and 4). 7 1 Near the bedrock-bounded edges of the site (Fig. 11-15) the oldest sands may be described as reddish brown or brown sands because an element of terra rossa has been admixed. Modern barrier sands are presumably similar to the ancient sea-floor sands (light gray sands), but they were not extensively drilled. The dark gray sands in 1 drill holes 9 and 10 (east end of section A-A ; Text Fig. II-2) represent back-barrier sands. They are 72 overlain by brown washover sands. The barrier surface today from the shore north to the lagoon area consists of: beach sands, generally well-rounded, 70 These drawings are generalized, based on trench-wall exca­ vation records and on drilling survey records. 71 See Linington, chapter fourteen. 72 See N . Kumar, "Modern and Ancient Barrier Sediments: New Interpretation Based on Stratal Sequence in Inlet-Filling

51

well-sorted, coarse- to medium-grained sands com­ posed of augite, sanidine, quartz, feldspar, and lithic fragments, and minor amounts of magnetite and other heavy minerals; dune sands, fine-grained, very well sorted, very dark sands (80 percent augite with magnetite and other minerals in minor amounts); and washover sands, brown sands of the same average composition of the beach and dune sands but con­ taining organic debris that causes the brown color. Washover sands are deposited when storms overtop the barrier and deposit washover fans on the lagoon side; thus the lagoonward barrier shoreline has a lobate appearance (Fig. II-7). Channel sands occur between polygonal Walls P and PE (Figs. 11-17, 18) and in an area of approxi­ mately 150 sq m beyond the northern end of the eastern polygonal Wall PE (Fig. II-2). These sands are gray to gray-brown, are often rich in heavy min­ erals, and resemble beach and nearshore sands. In comparison with beach and dune sands the channel sands are finer-grained, contain more admixed silt, and commonly have higher sanidine/augite ratios than the dune and beach sands. The heavy minerals, which are iron-rich, account for the strong positive magnetic anomaly found by the Lerici survey in the 150 sq m area (Map 9 and Text Fig. II-l). 7 3 The ab­ sence of red color in the channel sands suggests that they were deposited after construction of the polyg­ onal Wall P, which would have prevented an admix­ ture of red slopewash. The sands abut the walls, also confirming their deposition after wall construction. Furthermore, these sands contain very little archae­ ological material, perhaps indicating rapid deposi­ tion. The presence of a large amount of gray sand beneath the archaeological level suggests that a nat­ ural inlet did exist in the area and that the walls may have been built to stabilize it. There is also evidence that the Romans used gray sand behind some of the walls in the lagoon to stabilize them (Figs. 11-17, 18). 74 Lagoonal muds are black, organic-rich muds (silt, clay, and sand) found both at depths of 2-4 m in profiles A-A1 (Text Fig. II-2) and C-C 1 (Text Fig. II-4) toward the center of the lagoon and above the channel sands in the fishery area (Text Fig. II-3, Color Fig. 2). The latter muds contain archaeologi­ cal material. The former are principally older than Roman occupation and are unrelated to the harbor Sands and on Recognition of Nearshore Storm Deposits," New York Academy of Sciences, Annals 220 (1973) pp. 245-340. 73 See Linington, chapter fourteen. 74 See Gazda, chapter seven.

S.L.O

DK. GRAY SAND •

—

-

*

Α-ARCHAEOLOGICAL MATERIAL O-SHELL DEBRIS

RED SAND

LT. GRAY SAND

IO 49

50

51

53

52

Text Fig. II-2. Generalized west-east cross-sections A-A'. Arabic numerals are drill-hole locations.

B' "ALLUVIUM

S.L.O-

Zi GRAY SAND "

S

^

f—

I

YZI

,^.

GRAY SAND d

ALLUVIUM "^BLACK MUD

RED SANO 5

IV(XIV)

22

2-4

Text Fig. II-3. Generalized west-east cross-section B-B'. Roman numerals and YZl are trench locations (Map 8). Arabic numerals are drill-hole locations.

Text Fig. II—1. Index map for cross sections illustrated in Text Figs. II-2, 3, and 4. See Map 9 for detailed record of drill-hole locations and Linington, chapter fourteen. See Map 10 for locations of trench sections. Contours are in meters a.s.l. Areas labeled m were areas with high magnetic susceptibility; the 1-m contour at the end of Wall PE approximately outlines an area underlain by sand derived from offshore and possibly from coastal dunes, rich in heavy minerals.

C

LT. BROWN SAND i

WATER

LT. 8R0WN SAND \

C ΤΤΤ-ΤΠ

gLACK SAND_ fcgj^MUD» '

DARK GRAY SAND 14

13

12/7

Il

i0

20

U2

40

Text Fig. H-4. Generalized north-south cross-section C - C , constructed principally from drill-hole information to illustrate stratigraphy of the lagoon; details near the walls have therefore been omitted.

BEDROCK

II: GEOGRAPHY AND GEOLOGY

and fishery sites. Other deposits of mud, clay-rich sediments, and mixed sand, silt, and mud are noted on the cross-sections (Text Figs. II-2, 3, and 4) on trench sections (Figs. 11-14, 16, and 19), and on drillhole records elsewhere in this volume. 75 Humus—clay, silt, and rich vegetal matter—overlies most of the lagoonal sequence (Fig. 11-13). Alluvium—red earth and fragments of the local bedrock—is found overlying the archaeological level around the periphery of the site (Figs. 11-11, 12).

INTERPRETATION OF THE SEDIMENTARY SEQUENCE

At the time of the occupation of Cosa, the barrier, which probably formed about five thousand years ago, would have been widened by washover deposition. Archaeological and sedimentological evidence indicates that the barrier width extended approximately from its present shore position to the end of the eastern polygonal Wall PE. Strong evidence indicates that there was an open lagoon north of the barrier, as deep as 3-5 m, based on the depth of the pre-archaeological lagoonal muds recovered in drilling.76 Whether or not a natural inlet existed before the building of the two polygonal embankment walls is difficult to determine. Several lines of evidence, however, suggest that it was open: old lagoonal muds are not found abutting the wall structures (Text Figs. II-2, 3; Fig. 11-16; Figs. 11-17, 18), that is, there appears to have been a constant source of sand in this area; and red sands are found rather than mud or alluvium beneath the gray channel sands (Text Figs. II—2, 3; Fig. 11-18). Dark gray sands in the 150 sq m magnetic-high area (Fig. II-2; Text Fig. II—1) are deposited over lagoonal muds (Text Fig. II-4) suggesting that they were deposited after the channel walls were built. If a natural inlet did exist, it may have been active only during certain seasons. During the fall and winter rainy season the lagoon surface would have risen and may have flowed through an outlet (the postulated "inlet"). Wave action caused by scirocco winds or currents from the southeast may have kept an inlet open from time to time during the drier months. The minimal tides in the Mediterranean are not strong enough to keep an inlet open year round. J. D. Lewis's observations of tidal fluctuations during excavations in the 75

See Gazda, chapter four, and Linington, chapter fourteen. See Lmington, chapter fourteen. 77 Taken from daily readings; see unpublished excavational notebook of J. D. Lewis, "Underwater Notebook, 1969," p. 187. 76

53

Cosa port indicated an average tidal change of 0.55 m in the summer months of 1968 and 1969.77 During quiet, inactive periods, therefore, a small bar of sand would form across the seaward mouth of the inlet. Such is the case at the modern inlet to the lagoon fishery at Lago di Burano. 78 Today this inlet is barred by sand during the summer but is open and used as an outlet during the winter. A small delta at the northern, lagoonward end of the modern Burano channel is evidence for its role as an inlet during storms or maximum tides. This situation, however, is not exactly analogous to that at the Cosa harbor site, where the bedrock promontory obstructs the passage of water moving northward along the coast. If a natural inlet-outlet did exist across the barrier at the ancient site of the port of Cosa at the western end of the lagoon, as seems likely, the construction of a fishery within the lagoon north of the port would demand its stabilization. A constant exchange of fresh and salt water between the sea and the lagoon would be needed to regulate both salinity and temperature for the raising of fish in its brackish waters. 79 That the gray sand (with no red component) between the polygonal embankment walls of the lagoon (P and PE) lies 5 m b.s.l. suggests that the ancient Cosans may have dug out at least part of this channel area connecting the western end of the lagoon to the sea (Map 7). Yet the building of a large breakwater in the harbor area at the same time would have compounded the sedimentation problem of the natural inlet by trapping water traveling along the shore (Map 6). Sediments would have been deposited in the harbor itself or carried through the channel into the lagoon. Evidence for this process is the thick fill of gray sands between the polygonal embankment Walls P and PE (Figs. 11-17, 18). This sedimentologic evidence suggests that a natural, channel-filling event occurred at the Portus Cosanus as a consequence of artificial structures. The duration and date of this event cannot be derived from the available geological evidence. Nevertheless, the lack of archaeological material in these gray filling sands and the presence of the fish-tank walls within the broad channel area suggest that this filling event happened rapidly, perhaps within one year or even as a result of one storm. But the possibility of artificial filling of the channel cannot be excluded. The dominance of augite in the modern dunes and its relatively lesser abundance in the inlet 78

G. Manfredi, personal communication; Mr. Manfredi took the author and Gazda on a tour of the Lago di Burano in the summer of 1974. See also Gazda, chapter seven. 79 See Gazda, chapter seven.

54

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

sands suggest that windblown sand, which leaves the heavier grains behind, may have played a supplementary role in the filling of this channel area. The archaeological level—that level with the most sherds and other artificial materials—is found in the sandy muds, lagoonal muds, and alluvium above the gray sand level (Fig. 11-19), evidence that further supports the hypothesis that the channel filled in rapidly to a depth of 1-2 m and thus at no time was used as a ship channel (Figs. 11—17, 18). In earlier studies of the Portus Cosanus, 80 the rock-cut channels in the cliffs at the western end of the harbor—the Tagliata and Spacco della Regina— were regarded as desilting devices for the harbor, controlled by sluice gates. This interpretation is possible only if one can reconstruct a connection between the Tagliata and Spacco and the harbor area in antiquity. If such a connection by sluice gates can indeed be reconstructed, the Tagliata and Spacco could have served as an exit channel for the northwestward currents traveling up the coast, thus preventing the deposition of their sediments in the harbor or lagoon. However, further study of modern lagoonal fisheries in the area and a re-examination of the bedrock still present below the modern cement wall between the ancient quarry site and the harbor cause McCann and her team now to suggest that the northern end of the Tagliata and the quarry were cut off from the harbor area by a natural bedrock barrier.81 Unfortunately, the modern cement constructions in this area have prevented excavation and a definitive answer. A sluice gate connection at this point in the harbor, though still a possibility, is unlikely in view of the identification of the ancient fishery behind the barrier. Lewis, in his study of the breakwater in the harbor, offers solutions to the critical problem of harbor sedimentation. 82 He suggests that the broad and loose form of the breakwater, angled away from the waves, was specifically designed to allow the offshore currents to scour and pass through the harbor area. In ancient times the breakwater was apparently tied directly to the cliffs on the west, thus precluding a circulation channel at this juncture. The fact remains that this ancient breakwater, which orginally functioned above the ancient sea level, is now in places about 3 m b.s.l.—that is, about 2 m below ancient sea level. Lewis attributes this anomaly to rock dispersion through continuous wave action over the centuries and erosion of the sand from un80 Brown 1951; Rodenwaldt and Lehmann; Mori, pp. 343-355; McCann and Lewis 1970, pp. 200-211. 81 See Gazda, chapter four. 82 Lewis 1973, pp. 233-259; also see McCann, chapters three

der the breakwater itself. The outer line of the larger breakwater blocks is not disturbed. The archaeological level of the ancient harbor floor now lies at a depth of 1.00-2.10 m below the present ocean floor.83 This diminishment of the effectiveness of the breakwater at Cosa and the gradual reduction in the depth of the anchorage area behind it may have occurred, at least in part, within the span of the Roman era, accounting for the curtailed use of the harbor during the Imperial period, as indicated by the archaeological evidence. One structure in the harbor area that certainly appears to have been designed by the ancient engineers specifically to aid in the silting problem is the funnel-shaped channel formed by beach Walls A and B (Map 6). Such a funnel at the exit of this channel, whose eastern embankment is formed by Wall PE, would help prevent strong, sediment-laden currents from entering the channel into the fishing lagoon. The shape allows water to enter much more easily at the wider end; water accelerates as it is constricted through the funnel, thus keeping the area scoured and free of sediment. Walls A and B thus form a constricted channel that would allow water to leave the lagoon area but not to enter it.84 If all the elements described were functioning properly, the gray sand sedimentation would have been controlled. However, the archaeological evidence suggests that the port was not used extensively following the period of the Sestius Type 4b amphora jars in the late first century B.C. It seems probable, then, that the greatly diminished use of the port and fishery in the later three centuries of its life during the Imperial period (from the first to the third century A.D.) can be partly linked to the sedimentological evidence presented here. It may be that sedimentation in the harbor and fishery areas occurred near the end of the first century B.C., thus reducing the uses of the harbor and fishery for export purposes.

EVIDENCE FOR SEA-LEVEL CHANGE

Except for the filling in of the lagoon with sediment, the configuration of the coastline at Cosa, unlike that of many areas of Italy,85 has not changed appreciably since Roman times. One to two meters of sand have been deposited in the harbor area around the breakwater, but exposed villa walls along and seven, reconstruction of port. 83 See McCann, chapter three, on the underwater excavations. 84 See Gazda, chapter four. 85 C. Palagiano, "Documents geographiques sur la variation des

II: G E O G R A P H Y A N D G E O L O G Y

55

the beach indicate that there has also been some local erosion. Because the coast at the harbor site has been altered, paved, and built upon extensively since Roman times, much of the archeological and sedimentologic evidence for change has been masked. As noted previously, the Quaternary period is characterized by worldwide eustatic fluctuations of sea level, brought about by repeated advance and retreat of continental ice sheets. These sea-level oscillations are well represented in Italy by marine terraces—shallow marine sediments deposited during periods of high sea level and now exposed along the coast of Italy86—and the names given to these terraces are derived from areas of Italy. There is also offshore evidence of periods of lower sea level.87 Because of tectonic changes in land elevation, these terraces do not always occur at the same levels from place to place, making correlation difficult. The ancient barrier-lagoon complex northeast of the Cosa lagoon is probably Tyrrhenian III in age (approximately 100,000 years old). 88 Further evidence of sealevel change is a series of wave-cut notches of indeterminate age about 2 m above present sea level on the Cosa and Argentario promontories. 89 Another indication of change in the area is the presence of panchina, interpreted as beachrock, found on the Orbetello peninsula.90

At Cosa we are primarily concerned with the last major rise in sea level, which occurred between ten thousand and five thousand years ago, in the HoIocene epoch. Opinions vary as to the nature of the Holocene rise in sea level—whether it was a smooth, steady rise or a fluctuating one. 91 Numerous studies—geomorphological, geochemical, paleontological, archaeological, and geophysical—have been conducted to determine the nature of this curve. N. Flemming surveyed the evidence for sea-level change in ancient harbor sites of the western Mediterranean, averaged the results, and concluded that there has been no change in sea level due to eustasy since the year zero (1950 B.P.).92 Rather, he attributed all changes to tectonism, i.e. local earth movements. On the other hand, E. Pongratz surveyed Roman fish tanks and other wall structures along the coast of Latium south of Cosa and concluded that a 1.0-1.5 m eustatic rise in sea level has occurred since the year zero. 93 All the curves record lower sea levels around 2000 B.P. (the time of Christ), so that a eustatic change in sea level since the end of the Roman Republic is acceptable to many investigators of sealevel changes, although others argue that no eustatic change can be measured in the last five thousand years. At Cosa, Flemming 94 (based on Brown) 95 deter-

cotes Italiennes dupuis la prehistorique jusqu'a nos jours," 23rd International Geological Congress, Proceedings 1 (1976) pp. 397-400; see also Schmiedt, Atlante and Il livello antico. 86 Fairbndge, "Quaternary Sedimentation"; also H. G. Richards and R. W. Fairbridge, "Annotated Bibliography of Quaternary Shorelines," Philadelphia Academy of Natural Sciences, Special Publications 6 (1965) pp. 113-139 (on Italy); H. G. Richards, "Annotated Bibliography of Quaternary Shorelines," ibid., , supp. 1, vol. 10 (1970) pp. 146-156 (on Italy); ibid., supp. 2, vol. 11 (1974) pp. 125-132 (on Italy). 87 A. G. Segre, "Linee di riva sommersa e morfologica della piattaforma continentale italiana relative alia transgressione marina versiliana," Quatemaria 11 (1967) pp. 1-14. 88 Signormi, Carta Geologica d'ltalia (cited in n. 25); also see above, n. 14; F. P. Bonadonna, "Studi sul Pleistocene del Lazio III," Geologica Romana 6 (1967) pp. 121-135; A. Lazzarotto, R. Mazzanti, and F. Mazzoncini, "Geologia del Promontario Argentario (Grosseto) e del Promontario del Franco (Isola del Giglio-Grosseto)," Societa Geologica Italiana, Bollettino 83 (1964) pp. 3-122; A. C. Blanc, "Una spiaggia pleistocenica a 'Strombus bubonius' presso Palidoro (Roma)," R. Accademia Nazionale dei Lincei, Rendiconti 23 (1936) pp. 200-204. 89 Merciai 1910; Merciai 1929; A. M. Radmilli, "Una grotta con riempimento e fauna fossile sul litorale di Ansedonia (Grosseto)," Universita degli studi di Roma—Istituto di geologia e paleontologia, Publicazione 17 (1955) pp. 3-6; Lazzarotto et al., "Geologia" (cited in n. 88). 90 Merciai 1910; Signorini, Carta Geologica d'ltalia (cited in n. 25); Bourgeois, "Beachrock" (cited in n. 46). 91 See, e.g., Fairbridge, "Quaternary Sedimentation," and references cited in n. 13, above. Some scientists argue that there has been no eustatic change in sea level since 5000 B.p., or that it is impossible to measure. See, for example, J. A. Clark, W. E. Far-

rell, and W. R. Peltier, "Global Changes of Post Glacial Sea Level: A Numerical Calculation," Quaternary Research 9 (1978) pp. 265287; N. C. Flemming, "Holocene Eustatic Changes and Coastal Tectonics in the Northeast Mediterranean: Implications for Models of Crustal Consumption," Royal Society Philosophical Transactions 289 (1978) pp. 405-458; J. A. Clark, "A Numerical Model of Worldwide Sea Level Changes on a Viscoelastic Earth," Earth Rheology, Isostasy and Eustasy (cited in n. 13) pp. 525-534; W. S. Newman, L. F. Marcus, R. R. Pardi, J. A. Paccione, and S. Tomecek, "Eustasy and Deformation of the Geoid: 1000-6000 Radiocarbon Years B.P.," Earth Rheology, Isostasy and Eustasy, pp. 555-567. For summary of recent views see Masters and Flemming, Quaternary Coastlines (cited in n. 13) pp. 602-603. For chart of eight different postulated eustatic, sea-level curves see Nils-Axel Morner, "Eustatic Changes During the Last 20,000 Years and a Method of Separating the Isostatic and Eustatic Factors in an Uplifted Area," Palaeogeography, Palaeoclimatology, Palaeoecology 9, no. 1 (1971) pp. 153-181 and p. 160, fig. 5. Compare R. Berger, "Sea Levels and Tree-Ring Calibrated Radiocarbon Dates," in Masters and Flemming, Quaternary Coastlines (cited in n. 13) p. 54, fig. 2. Note that sea level was probably lower during early Roman times but was rising. Compare particularly curves of Fairbndge (1961), Shephard (1963), and Morner (1969). 92 Flemming 1969; Blackman, "Evidence of Sea Level Change" (cited in Preface, n. 13 above) pp. 115-137; Delano Smith, "Coastal Sedimentation" (cited in Preface, n. 13 above) pp. 2533; Flemming, "Holocene Eustatic Changes," (cited in n. 91); Masters and Flemming, Quaternary Coastlines (cited in n. 13); Bloom, Atlas (cited in n. 13). 93 Pongratz. 94 Flemming 1969, pp. 30, 31. 95 Brown 1951.

56

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

\

mined that the sea level has not changed since the year zero. On the other hand, G. Schmiedt,96 based on excavational evidence presented by McCann and Lewis,97 concluded that sea level has risen about 1 m at Cosa since its most active period during the Republican era. In a more detailed study 98 Lewis gives further evidence for a rise of between 0.5 m and 1.0 m since Roman times, conclusions verified by the present study. Some of the significant levels used in this evaluation will now be discussed. The measurements here are taken from Lewis's data from the harbor area in 1968 and 1969 and from J. P. Oleson's and D. van Zanten's work on the Spring House and lagoon structures in 1972. 1. Rock cuttings for sluice gates found in the northern end of the Tagliata are at present sea level (Fig. IV-42). These cuttings are in the form of vertical grooves for sliding boards of a wooden sluice gate. In order for such a gate to have functioned effectively, Lewis points out, the ancient water level must have been below the present level by at least 0.5-1.0 m. These grooves extend down more than 1 m below the present sand fill. 2. In a test trench (Sl) made in the Tagliata Canal leading north into the lagoon from the quarry (Text Fig. II-5; IV-I, 30), bedrock was found at 1.36 m b.s.l. For this channel to have functioned effectively in the exchange of sea water with the brackish water of the lagoon, the ancient sea level could not have been below this reading. This measurement thus gives us the maximum amount of change in sea level possible since the lagoon's use as a fishery. 3. In the fishing lagoon itself, the Spring House platform is 0.53 m a.s.l. at its highest level and 0.28 m a.s.l. for the main platform area (Figs. 11-19; V4). This concrete platform is bisected by a terracotta drain leading from the spring's collecting basin on its western side to the lagoon on the east. Below the drain, an irregular lower step extends at a depth of 0.44 m b.s.l. During the excavations, a terracotta pan tile was found still in place on top of this lower step below the exit of the drain. Apparently it had been placed there to protect the surface of the concrete from water erosion. For this drain from the spring to have functioned with its lower step, it must have been above the ancient sea and lagoon levels. An estimated lower sea level of about 1 m would have provided better protection for the fresh water of the spring that sat in a collecting basin cut into the bedrock to a depth of 1.70-2.05 m below present sea level. 96 Schmiedt, Atlante, pi. CXXX; Schmiedt, Il livello antico, pp. 25-30. 97 McCann and Lewis 1970; McCann 1979, pp. 391-411.

TAGLIATA CANAL LOOKING NORTH

\ WESTERN LmESTONE CLIFF

y - W A T t R LEVEL

\

BEDROCK^,

/

t

0

5

IM

Text Fig. II-5. Section Sl, Tagliata Canal, looking north. See Fig. IV-I for location.

4. The foundations for Wall P south of the Spring House platform are at a depth of 0.42 m b.s.l. Presumably, but not necessarily, these foundations would have been constructed above sea level. 5. The tops of the fish-tank walls in the lagoon vary. Wall N of the southern tank is 0.57 m a.s.l. at its northern end in trench IB (Map 8); Wall V and Wall W of the northern fish tank are at 0.53 m b.s.l.; and Wall U in trench U l is 0.50 m b.s.l. To have contained fish, all these walls must have been above sea level, about 0.50 m being sufficient. A change of about 1 m since their use is consistent with these measurements. In conclusion, the evidence of the archaeological structures combined with the wave-cut notches found on the seaward side of the Tagliata at approximately 1 m b.s.l. indicate a sea-level rise of not more than 1 m but at least 0.50 m since the use of the port and fishery in Roman times. This change in sea level may have been caused by a eustatic rise in sea level brought about by melting of glacial ice; tectonic subsidence; subsidence due to compaction of sediments; or a combination of these processes. Most of the structures mentioned above are built into the bedrock or are on sand, which compacts very little, probably eliminating compaction as an explanation. The observation that Tyrrhenian and earlier terraces in this region maintain their horizontal level for long distances99 and the close agreement with Pongratz's survey100 of the coast just to the south of Cosa are evidence against tectonic subsidence, but larger-scale crustal deformation may be the 98

Lewis 1973. Bonadonna, "Studi sul Pleistocene" (cited in n. 88). ioo Pongratz.

99

II: GEOGRAPHY AND GEOLOGY

cause. A principally eustatic rise in sea level may have occurred since Portus Cosanus enjoyed its most active period of life during the last two centuries of the Roman Republic, but it is difficult to attribute the change definitively to changes in ice volume versus changes in the geoid (shape of the earth). The date and speed of this sea-level rise are not determinable from our data at Cosa.

SUMMARY AND CONCLUSIONS

The sedimentary sequence in the harbor and lagoon of Cosa represents the complex interaction of coastal processes and human intervention. Despite the complex geology of the area surrounding Cosa, two distinctive sources of sediments can be defined: slopewash (red) and coastal sands (gray). Evidence for Quaternary sea-level changes are evident in the Cosa area; structures in the harbor and fishery areas, active until about the year zero, show evidence for a sea level about one-half to one meter lower than present. A possible succession of events to account for the sedimentary sequence is as follows: 1. Barrier formation, probably about 5000 B. P. created a narrow lagoon east of the Cosa promontory. 2. If any natural channel existed between the sea and the lagoon at its western Cosan end, the flow would have been predominantly outward; the lagoon would have been slowly filling with slopewash (red) and washover sands (brown-gray); lagoonal muds would be deposited in the central area. 3. Once an artificial channel was established across the barrier into the lagoon and the breakwater was

57

built, sand would have been contributed from the sea by: (a) scirocco-driven waves and water piling up against the cliffs and spilling into the lagoon; (b) longshore drift, which is from south to north in this region (though not strong, the current nevertheless may have been diverted into the channel by the bedrock cliff and the harbor breakwater); and (c) tides, although very weak. In the winter the flow of water in the various channels connecting the lagoon to the sea would have been primarily seaward due to the heavy rains. This flow would have tended to keep the channels clear but would have created problems because the harbor structure would have trapped the sediments. Besides the engineering devices already discussed, these problems could also have been alleviated by controlling the water in the channels with sluice gates, for which there is abundant archaeological evidence. It may be that the Romans chose to isolate the lagoon from the harbor completely by blocking the natural entrance and controlling the flow of water through completely artificial channels. Unfortunately, modern cement construction at the western end of the harbor has prevented excavation of this critical zone. A direct connection between the port and the quarry via the rock-cut channels in the cliffs appears to be unlikely based on archaeological evidence combined with evidence of a bedrock ledge in this area. Evidence for a change in sea level since Cosa's active period is compelling. We estimate that there has been a rise of about 1 m in sea level since Republican Roman times, based on archaeological and secondary geological evidence. This change in sea level appears to have been caused by eustasy or possibly by broad tectonic down warping.

Chapter III. Modern Scholarship and the Land and Underwater Excavations ANNA MARGUERITE M C C A N N

MODERN SCHOLARSHIP

The ancient remains of the Portus Cosanus and its fishery have always been visible. The massive concrete piers rising above the broad rock breakwater, along with the dramatic channels—the Tagliata and the Spacco della Regina—cut into the western cliffs, dominate the landscape today as they did in antiquity (Color Fig. 1). The long polygonal embankment walls of the ancient fishing lagoon behind and the remains of the buttressing walls of the Spring House and its cistern have also always been above ground even when the rest of the fishery was silted over (Map 7). Deprived of its identity in the Middle Ages, Cosa was known as Ansedonia and its port as the Portus Finiliae.1 Its location has been debated, although most historians now agree that the site of the medieval Portus Finiliae is the same as the Roman Portus Cosanus, just to the east of the Cosa promontory at the modern anchorage known as the Tagliata. F. Lenzi,2 however, followed by M. Fasciato,3 located both ports at the northernmost tip of the Cosa promontory in the angle where it joins the eastern end of the great Tombolo di Feniglia (Map 2). In locating the Roman Portus Cosanus here, Lenzi based his argument largely on the discovery in this area of a Roman Republican coin hoard, and Fasciato cited ruins of an oil press and magazines, which she considered remnants of an ancient port. It should be recalled, however, that the canal still connecting the sea to the interior fishing lagoon of Orbetello at this point appears to date to Roman times, and one would expect evidence of the Romans' use of this region for fishing and agriculture. Our land and underwater survey in the area did not produce 1

CardareUi 1925, pp. 20, 86. Lenzi, pp. 224-225; F. Lenzi, "Un ripostiglio di monete consolari e la localita del Porto Cosano," Rassegna Numismatica 2 (1905) pp. 49-51. 3 Fasciato, "A travers la Maremme Toscane" (cited in chapter 2

any visible remains of harbor constructions from any period. Moreover, the unfavorable navigational position of the Feniglia at the head of the Cosa promontory, fully exposed on the south and west, argues against its suitability for a port. If there was a landing place here in antiquity or the Middle Ages, it must have served only small craft used for fishing in the lagoon or local traffic across the bay to Port'Ercole. During the Italian Renaissance of the fifteenth and sixteenth centuries the port at the Tagliata appears to have been in active use for local agricultural trade, largely in grain (Map 6). 4 In fact, the Medicis considered the port valuable enough to warrant the building of a square watch tower (still standing) in the late sixteenth century to guard local commerce from pirating. In a letter dating from the second half of the sixteenth century an unknown local entrepreneur in grain and cattle urged Cosimo I (1519-1574), the Grand Duke of Tuscany, to construct a tower at the port itself. This tower was apparently under construction already, since the existing Spanish tower, the Torre di S. Pancrazio, above the Cosa promontory gave inadequate protection to the commerce below. 5 The tower was part of a large reclamation project for the whole port and lagoon area, which was sketched in detail. The letter describes plans for improving the port and reclaiming the swamp of Ansedonia for agriculture and fishing. The writer further urged the construction of ironworks to smelt the ore coming from Elba. Existing magazines for grain and cisterns are noted, as well as a freshwater spring used by ships (called "La Fonte de la Reina") within "gunshot" of the Tagliata. This spring is probably the one released during our excavations in one, n. 66) p. 277. 4 CardareUi 1925, p. 168 and n. 3. 5 Venerosi-Pesciolini, pp. 15-28; Archivio Stato Firenze, Carte Strozziane, serie 1, vol. 110, c. 309-329.

Ill: MODERN SCHOLARSHIP AND EXCAVATIONS

59

the Spring House on the western bank of the ancient lagoon. We do not know whether Cosimo I replied or how much of this extensive project besides the tower was actually carried out. Remains of a roadbed and yard that must have serviced the Torre della Tagliata have been found on the surface of the modern parking lot to its southwest. Although no remains of furnaces or iron slag have thus far been found in the port area, a thick layer of carbon was found in seven beach probes just below the surface of the modern parking lot and above the Roman construction level during our excavations in 1972 (Figs. Ill—1; IV-I). This burned level ranged in thickness from about 0.45 m to 1.45 m and covered a wide area, which suggests that some industry of post-Roman date was carried on at the port site, such as charcoal burning or the smelting of ore (see J. de Boer, chapter sixteen). During our excavations of the harbor, moreover, the remains of a ship loaded with iron ore were discovered. Carbon-14 dating of its timbers suggests a probable date in the late sixteenth century. This barge may document the last commercial use of the port of Cosa. 6 The tower at the Tagliata has a longer history (Fig. II-9). Bought in 1919 by Giacomo Puccini, the famous Italian opera composer, for a hunting and fishing retreat, it is also known today as the Torre Puccini. The tower is presently owned by the Venturini family, who have built their own seaside retreat upon the ruins of the ancient Roman villa. A modern restaurant and bathing beach to the east, La Strega, make further use of the ancient Roman villa walls. The silted-over lagoon of Cosa now serves as pasture for the large, white Chianina cattle of the Maremma. Just when this westernmost portion of the lagoon of Cosa became completely silted in is not certain. Our sedimentation evidence and that of the cartographers indicate that it was not completely filled until sometime during the last century. 7 That at least part of the port area was still active for fishing up into the nineteenth century is documented by an inscribed marking stone found in our beach excavations. 8 It reads: "LIMITE TRA LA PESCA MARITTIMA E FLUVIALE" (Fig. III-2). Although the date of the inscription is unknown, it

likely comes from the time when the modern canal leading from the Lago di Burano to the ancient Tagliata channel was built in 1859 as an overflow drain from the lake. By this time, then, the silting in of the western portion of the ancient Cosa lagoon must have been completed. The "Pesca Fluviale" would be coming from the fresh waters of the Lago di Burano (into which the Chiarone and Tafone rivers empty) to exit to the sea at the port of Cosa. The earliest description of the ancient ruins of the Portus Cosanus is that of G. Santi, published in 1798, followed by that of the English traveler G. Dennis, who visited the site in the 1840s.9 Their attention largely focused upon the channels and caverns in the western limestone cliffs, whose association with fishing had been forgotten. This interest is not surprising, for the partly natural, partly manmade clefts of the Tagliata and the Spacco della Regina are still the most dramatic aspects of the ancient site (Figs. III-3, 4, 5, and 6). The Spacco, with its interior cave, has also attracted the attention of modern geographers and geologists (Fig. III-7).10 Today the Tagliata appears to be largely man-cut; however, much of both the Tagliata and the Spacco were originally natural fractures formed millions of years ago and subsequently widened by karstic processes (dissolution of limestone). The date of man's transformation of these fractures and their original functions puzzled both Santi and Dennis, and they continue to intrigue modern investigators. Whereas Santi identified the Tagliata as an ancient bath, Dennis called the Spacco the "Bagni della Regina"—both undoubtedly repeating the Spanish name used for similar structures by the sea.11 Dennis actually entered the Spacco and wrote a vivid description of its impressive interior cave:

6 See Lewis, chapter eight, Appendix. Carbon-14 analysis yielded a MASCA corrected date of A.D. 1588. 7 See Bourgeois's discussion in chapter two. 8 Found August 3, 1972, in Beach Probe IV. "Portus Cosanus Daily Journal, 1972," p. 85 (A. M. McCann). An identical marker, undoubtedly placed at the same time, can still be seen on the eastern side of the hill of Talamonaccio where the Osa River empties into the sea. 9 Santi, Viaggio (cited in chapter one, n. 66) pp. 126-152; Dennis, II, p. 245.

10 Mori, "I fenomeni carsici" (cited in chapter two, n. 31) p. 169; Mori, p. 349, n. 10. Studies of the Spacco have also been made by geologists J. Bourgeois and J. de Boer; the latter kindly provided many observations about the Spacco presented here. De Boer's study of the cave was made in conjunction with Wesleyan University's survey of the Ager Cosanus in 1974-1976 and again in 1980. 11 For example, note the fish tank at Javea, called the Bafios de la Reina, discussed in Gazda and McCann, chapter seven; for references see McCann, chapter one, 230.

You enter a long cleft in the rock, sixty or seventy feet deep, and on one side perceive a huge cave, within which is a second, still larger, apparently formed for baths; for there are seats cut out of the living rock—vivo sedilia saxo—but all now in utter ruin. The place, it has been remarked, recalls the grotto of the Nymphs, described by Virgil; but popular tradition has peopled it with demons, as says Faccio degli Uberti,—

60

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

A quotation follows from Fazio degli Uberti's fa­ mous fourteenth-century poem, Il Dittamondo, in which the then desolate Maremma and Ansedonia are described and the cave in the Spacco is referred to: La e ancor dove fu Lansedonia: La έ la cava, dove andara a'torma Si crede il tristo, overo Ie demonia. [There is also the place where Ansedonia was, There is the cave where it is believed That the bad ones (criminals) or the devils gather.] 1 2 Uberti's poem is an important document of the lost fame of the Spacco cave and the memory of its use in the late Middle Ages as a hiding place. It would have been especially suited for pirates coming in from the sea. Its ancient use as a place of refuge is also reflected in its modern name, said to derive from the legend of a local queen who fled into its cave for sanctuary after saving the city. One enters from the ancient quarry at its northeastern end, where a narrow passageway running for about 25 m leads to the large interior cavern (Map 6). This partly natural, partly scarped space measures about 20 m in length with a maximum width of about 8 m. A skylight partly cut by man lights the cave at its southern end from a height of about 30 m. An additional skylight has been cut in the passageway leading to the cave from the northeast (Fig. III-8). In our survey the rock-cut seats seen by Dennis in the cave could not be found, but cuttings for a wooden stairwell leading down from the skylight were ob­ served. A freshwater spring located along the eastern side of the cave about 1 m a.s.l. is now pumped through a pipe up the skylight to serve a modern villa above. At an earlier time, the whole cavern may have been used as a cistern reached by the steps; water could have been drawn and passed by buckets through the skylight. Nor is the spring within the Spacco cave the only one existing in the cliffs. Dur­ ing recent construction along the southern side of the quarry basin another freshwater source emerged, perhaps the ancient exit of the spring within the cave. Only excavation within the cavern can resolve this and other questions concerning the Spacco's use.

12 Fazio degli Uberti, Il Dittamondo, III. 9.34, ed. G. Corsi (Ban, 1952) p. 209. Translation by M.T.M. Moevs. Il Dittamondo was written about 1360. See J. Burckhardt, The Civilization of the Renaissance in Italy (New York, 1955) p. 108. 13 Compare Cardarelli 1925, p. 22, n.l; Brown 1951, pp. 9096; Bradford, pp. 227-231; R. F. Paget, "Ancient Ports of Cu­

lt does now seem clear, however, that in Roman times the Spacco never served as a sluice for sea water to desilt the harbor or to replenish the fishing lagoon as has previously been suggested. 13 No evi­ dence within the Spacco, such as wave notches or borings of marine animals, indicates a sea level along its walls, nor is there evidence that the cleft was deep enough to have been used as a water channel in an­ tiquity. Provided with its own fresh water supply, it still makes an ideal hiding place, as Uberti's poem revealed. Otto Benndorf reported the first archaeological investigation in the port area in 1867. 14 He writes of the survey made by Francesco Marcelliani to estab­ lish the location of Succosa along the Via Aurelia Antica behind the fishing lagoon. Benndorf also noted a wall extending to the sea, not found by Den­ nis or identified in our survey. In the port area, he described two or three large mounds of amphoras, which he compared to Monte Testaccio, the hill in Rome formed from discarded amphoras from its an­ cient emporium or commercial center. Although amphora fragments still covered the site of the Portus Cosanus at the time of our excavations, the large mounds had disappeared. BenndorPs report pro­ vides further evidence of an ancient, large emporium at the Cosa port. The earliest detailed study of the port area with maps of the Spacco and the Tagliata (Figs. III-9, 10) is Raffaele Del Rosso's work on the ancient and modern fisheries of Etruria, published in 1905. 15 Del Rosso was critical of the earlier travelers' romantic interpretations, and he offered the first functional explanation of the Tagliata as a channel associated with fishing and connected to an ancient fishing la­ goon behind. Noting the cuttings for various sluice gates in the walls of the Tagliata, he concluded that the exit of the fresh water from the lagoon into the sea invited fish to enter and could be controlled, a necessary requirement for the attraction of fish. The sea water coming in with waves and tides could like­ wise be regulated. He regarded the open seaward portion of the Tagliata along with the second tunnel, the Tagliata Piccola, as a saltwater fishery (Map 6). Even though Del Rosso credited the Etruscans with the construction of the Tagliata, he anticipated many of our present conclusions. mae," JRS 58 (1968) pp. 159-160; McCann and Lewis 1970, pp. 201-202; Schmiedt, Il livello antico, pp. 25-31; Etruscan Cities (cited in chapter one, n. 41) p. 128. 14 O. Benndorf, "Scavi di Orbetello," BdI (1867) pp. 147-148. 15 Del Rosso, pp. 71-97, 289-331.

Ill: MODERN SCHOLARSHIP AND EXCAVATIONS

61

Del Rosso also separated the function of the Spacco from that of the Tagliata and the fishery. Intrigued with its large interior cavern, he believed that it had witnessed ancient religious rites associated with the Cabiri cult.16 However, we now know (particularly from excavations in Samothrace) that this mystery cult had its center in the Greek world; there is no evidence for its penetration into Etruria in Etruscan or Republican Roman times. 17 Nevertheless, Del Rosso's association of the Spacco with religious purposes should not be lightly dismissed until excavations within the cave can be made. D. Levi's find of an aeneolithic ax head in the cave during his exploration of the Maremma in 1927 may suggest cult worship. 18 The only other evidence for the ancient religious life of the port area is a marble statuette of Ceres from the third or second century B.C., which is said to come from the area of the Tagliata and is now in the museum at Orbetello (Fig. IH-Il). 19 A cult of Demeter-Ceres would of course be very appropriate for a cave near the sea that is associated with a spring and a temple to Poseidon. 20 Grottoes and nymphaea were also usual features of Roman maritime villas,21 and the Spacco may well have been so used during Imperial times by the owners of the villa at the port. 22 Following Del Rosso's study, the ports of the Ager Cosanus attracted the attention of D. Anziani in 1910.23 He focused upon the literary sources and their evidence for the identification and location of the Portus Cosanus and the Portus Herculis. Because Livy mentions only the Portus Cosanus, and because the Portus Herculis is documented only in the later Imperial sources, Anziani concluded that the ancient port of Cosa and Hercules were one and the same, located at the site of the modern harbor of

Port'Ercole on the Argentario peninsula (Map 2; Fig. 1-9). This assumption was continued by L. R. Taylor24 and later adopted by F. E. Brown, 25 who believes that the port of Hercules was the earliest port in the area and served the colony of Cosa in the third century B.C. before the port at the Tagliata was built. To my knowledge, the earliest literary evidence for the use of the Portus Herculis is in 137 B.C., when C. Hostilius Mancinus is cited as embarking from there for Spain.26 Moreover, our underwater survey of the harbor of Port'Ercole and the adjacent coastline produced amphora material no earlier than the second century B.C. and continuing through the fifth century A.D.27 The amphoras come from a wide variety of areas in the Mediterranean: Greece, Italy, Spain, and North Africa. Although the Portus Herculis may have been in use as early as the third century B.C., no archaeological or literary evidence has yet been found to prove this hypothesis. On the other hand, our excavations at the port site of the Tagliata have brought to light some Greco-Italic fragments that may be dated as early as the late fourth or early third century B.C.28 Thus, the Portus Cosanus and its spring appear to have been in limited use at least as early as the founding of the colony in 273 B.C., and there is no reason to place the port of Cosa on the opposite Argentario peninsula, even in its earliest colonial years. R. Cardarelli, in his extensive historical and topographical study of the Ager Cosanus published in 1924 and 1925, reasserted the location at the Tagliata of both the Portus Cosanus and the later medieval port of Ansedonia, the Portus Finiliae.29 He rightly emphasized the inaccessibility of both the Portus Herculis on the Argentario and the presumed ancient

16 Del Rosso, pp. 82-87. The Etruscan mirror illustrated opposite p. 82 and identified as a scene of a Cabiri cult initiation is clearly rather a death scene, perhaps showing Achilles killing Troilas. I wish to thank M.T.M. Moevs for this suggestion. 17 N. Lewis, Samothrace. I. The Ancient Literary Sources (New York, 1958); J. Ferguson, The Religions of the Roman Empire (Ithaca, N. Y., 1970) pp. 122-123; L. R. Taylor, p. 246, for the cult of the Dioscuri, who became associated with the Cabiri under the Empire. There seems to be no evidence for either cult in the Cosa area during the late Republic. 18 Levi, pp. 477-478. The ax head is now in the Archaeological Museum in Florence. 19 Santangelo, p. 17, fig. 5. 20 For the mystery cult of Demeter at Eleusis see G. Mylonas, Eleusis and the Eleusinian Mysteries (Princeton, 1961). For the worship of Ceres in Etruria, which apparently also included some form of mystery worship, see L. R. Taylor, pp. 39, 42, 77, 99; for a cult of the Nymphs at Clusium, L. R. Taylor, p. 178; for a cult of Neptune at Vetulonia, L. R. Taylor, pp. 171ff. 21 McKay, pp. 115-128, and especially Sperlonga, pp. 127-128. 22 This suggestion has also been offered by P. Raveggi, "Ritro-

vamenti archeologie nel terratorio Cosano," NSc ser. 6, vol. 3, fasc. 5 (1927) p. 209. 23 Anziani, pp. 373-395. 24 L. R. Taylor, p. 171. 25 Brown 1980, p. 50. 26 Valerius Maximus 1.6.7; Julius Obsequens Prod.Lib. 24. Cf. McCann, chapter one, nn. 58, 59. 27 "Portus Cosanus Daily Journal, 1968," pp. 67-68, 70-71 (A. M. McCann). The 20 amphora fragments found in this survey have been identified by E. L. Will and include the following: 1 Greek fragment; 4 Greco-Italic fragments from the second century B.C.; 1 complete Greco-Italic jar from the second century B.C. (not found by us and now at Le Rocce Hotel at the Tagliata); 3 fragments of Will Type 4b, first century B.C.; 2 fragments Dressel Type 8 from Spain, first century A.D.; 3 fragments Will Type 18 from North Africa, second century A.D.; 4 fragments of third century A.D. types from North Africa; one fragment of North African type, fifth century A.D. 28 Cosa fragments: Cats. A2, A15, A21. See Will, chapter nine. 29 Cardarelli 1925, p. 20, n. 1.

62

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

port of Etruscan Old Cosa in the northern Orbetello lagoon. The port of Hercules in antiquity would have been a day's journey by ox cart to the town site of Cosa and the Via Aurelia. Cardarelli further suggested that the lagoon of Cosa, or, as he identified it, the Lake of Vulci, was accessible to ships, which used the channel of the Tagliata as an entranceway along the western cliffs.30 Though uncertain of the date and use of the Spacco, he proposed that it was also intended as a sluice connection with the lake, a project that was abandoned when the more effective Tagliata was cut, perhaps even as early as Etruscan times. In 1948 the American Academy in Rome, under the direction of F. E. Brown, began excavations on the hill site of Cosa. As part of his topographical study published in 1951, Brown discussed the port and lagoon area.31 It was the first detailed description of the site with a surveyed map since Del Rosso's study of the Tagliata and Spacco in 1905. In his interpretations of the visible structures, Brown adopted many of Cardarelli's earlier theories. He located the port of Cosa within the lagoon behind the sand dunes, reached by a ship channel running along the limestone cliffs following the northern course of the channel of the Tagliata (Fig. Ill—6). Brown likewise interpreted the Spacco and the Tagliata as sluices for the desilting of the ship channel and interior port, and he believed that the Spacco was the earlier of the two. For the outer harbor basin, Brown reconstructed two parallel jetties supported by arched substructures. He noted three of the remaining five concrete piers and suggested the reconstruction of seven additional ones. His map of the harbor area also shows the temple on the northern hill above the lagoon (Map 7) and the remains of the

Imperial villa along the beach. He suggested that during the period of the villa the Tagliata sluice was remodeled and vaulted to be used for fish ponds or boat houses. Brown's theories have influenced the studies of later scholars, including the initial research of the present excavators.32 Continued study of the excavational material over the past years, however, has lead us to the very different conclusions published here. In his 1980 publication on the town site of Cosa, 33 Brown made unauthorized adaptations to our maps of the port area to further some of his early theories, despite the archaeological evidence to the contrary. In the early 1960s the Portus Cosanus attracted the interest of geographers as well as that of historians. E. Rodenwaldt and H. Lehmann34 disputed Brown's theory of an inner harbor in the lagoon of Cosa on the basis of hydraulics and a morphological investigation of the area. They noted the use of the Tagliata channel today as an outlet for the Lago di Burano and believed that both the Tagliata and the Spacco were used in antiquity primarily as drainage canals for a large-scale land reclamation project. They suggested that the Spacco was the first sluice constructed—probably in pre-Roman times—along a natural fissure that was later destroyed by an earthquake. Thus doubting the existence of an inner harbor, they were also skeptical of the evidence for any large outer anchorage. Furthermore, they saw no functional connection between an outer harbor and the channels of the Tagliata and Spacco. They suggested rather that the ancient port of Cosa was located in the Orbetello lagoon or on the Tombolo di Feniglia.

30 Cardarelli 1925, pp. 18, 20, 22; Cardarelli 1924, p. 218 Cf. Raveggi, pp. 81-83, who follows Cardarelli's ideas in his interpretation of the Tagliata but is uncertain of the use of the Spacco and its interior cave. 31 Brown 1951, pp. 89-96, map, pi. II. 32 Bradford; McCann and Lewis 1970; Schmiedt, Il livello antico, pp. 25-31. 33 Brown 1980, pp. 49-51, 58-60. The maps of the port area (figs. 59 and 77), which are based upon those made during our excavations by architect David van Zanten, were reproduced without our knowledge and without acknowledgment. Fig. 77 shows Professor Brown's reconstruction of the harbor, which, in the opinion of the excavators and present authors, is not in accord with the archaeological evidence. He represents nine piers angled over a narrow breakwater. The thorough excavations of the harbor in 1968 and 1969 revealed only five piers and no evidence that any others were ever built. Also, his reconstruction of a narrow jetty is not in accord with the archaeological remains. Compare our Map 6. Unfortunately, Professor Brown does not distinguish between the actual remains and his hypothetical reconstruction, which is drawn in heavy black.

Professor Brown gave an additional paper on the "Ports and Fisheries of Cosa" in 1978 at a conference at the American Academy in Rome on The Seaborne Commerce of Ancient Rome: Studies in Archaeology and History, ed. J. H. D'Arms and E. C. Kopff. MAAR 36 (1980), review by McCann and Will 1984. This paper, unfortunately, used inaccurately the research and conclusions of the present scholars, who had shared their material in trust and confidence with him. Despite our request to both Professor Brown and Professor John H. D'Arms, director of the American Academy at that time, that this paper not be published in view of the errors it contained, and despite their written agreement not to do so, Professor Brown did publish his paper within an article by Paola Zancani Montuoro in RivIstArch ser. 2, 3 (1979) pp. 21-23, listed here in Frequently Cited References and Abbreviations under Zancani Montuoro. Apart from the fact that Professor Brown's paper does not acknowledge those who directed, financed, excavated, and studied the material he used, it is most unfortunate that both in his text and in his plans of the harbor he inaccurately presents the actual archaeological remains. 34 Rodenwaldt and Lehmann, pp. 3-31.

Rodenwaldt and Lehmann's discussion prompted the Italian geographer A. Mori to offer his interpre-

Ill: MODERN SCHOLARSHIP AND EXCAVATIONS

tation of the port site in 1963.35 He returned to the theory of an inner harbor and the location of the Roman Portus Cosanus within the lagoon of Cosa. An earlier Etruscan harbor of Old Cosa, he believed, was located in the northern Orbetello lagoon, entered today by way of the Nassa fishery at the western end of the Tombolo della Giannella, which was not yet attached to the Argentario peninsula. For Mori, then, the function of the Tagliata and Spacco channels of the Roman Portus Cosanus was twofold: for draining the Lago di Burano when the waters of the lake had risen, and for flushing the ship channel and inner port with sea water when the waters were high and the winds blew from the south and southwest. Noting the extensive size of the Cosa lagoon in antiquity and its probable use also for fishing, he rejected Rodenwaldt's and Lehmann's hypothesis of an ancient land reclamation project. Repeatedly explored but never excavated, the Portus Cosanus and its ancient lagoon demanded excavation. The site came to my attention when I was a fellow in classical studies at the American Academy in Rome and had the opportunity to participate in the reopening of the excavations at the town site of Cosa in 1965. A project was formed at that time to survey the ancient harbor sites along the Tuscan coastline that were documented in the ancient Maritime Itineraries. In the fall of 1965, I began this survey using SCUBA equipment at the presumed site of the Portus Cosanus at the Tagliata with J. D. Lewis. Underwater excavations were begun in the summer of 1968 and continued in 1969 under the direction of Lewis. Concurrent with the underwater excavations, digging on land was begun in the silted lagoon 250 m behind the port (Maps 4, 5). Underwater reconnaissance was also made along the shore between the Tagliata and the Tomboli di Feniglia, as well as at other port sites in the area: San Clementino (ancient Graviscae), Sco. La Fresca, Port'Ercole (ancient Portus Herculis), and Talamone (ancient Telamon) (Map 2). But the main focus of our attention was at the Portus Cosanus, whose harbor basin, encompassing about 25,000 sq m, was surveyed and 35

Mori, pp. 343-355. A. M. McCann, "Excavations of the Ancient Port at Cosa, 1968," AJA 73 (1969) pp. 241-242; McCann and Lewis 1970; J. D. Lewis, "Excavations of the Ancient Port at Cosa," AJA 74 (1970) p. 199; Lewis 1972, pp. 169-170; Lewis 1973, pp. 233-259; A. M. McCann, "Excavations at the Roman Port of Cosa, 1972," AJA 77 (1973) p. 220; McCann, "Excavations at the Roman Port of Cosa, 1972," IJNA 2 (1973) pp. 199-200; J. P. Oleson, "The Spring House and Water Lifting Device in the Roman Port of Cosa," Abstracts, AIA (1975) p. 19; McCann 1979, pp. 391-411; Will 1979, pp. 339-350; Bourgeois and Gazda (1982); Oleson, Water-Lifiing Devices (1984); A. M. McCann, "Excavations at the 36

63

a contour map drawn (Map 6). E. and J. Whittlesey made aerial photographs over both the port and the lagoon areas with an electric-driven Hasselblad camera suspended from a hydrogen balloon at heights varying from 15 to 200 m (Figs. IV-2 to 5, 18, 26, 27). Further study and mapping of the channels in the cliffs and the breakwater were made in 1970 by McCann, R. L. Hohfelder, J. P. Oleson, and J. F. Warren, and in connection with the study of the materials and techniques of construction, Gazda reexamined all the visible harbor remains in 1972 and 1974. General reports on both the underwater and land excavations have already been published.36 The detailed results of all this work are presented below.

THE UNDERWATER EXCAVATIONS

The Breakwater and Its Extensions The ancient breakwater, now completely submerged, was traced eastward from the limestone cliffs for a distance of ca. 110 m, at which point it bends north for a defined rock line of about 70 m (Map 6; Figs. IV-2, 3). Probing within the visible sand pockets of the breakwater indicated continuous rock below the surface, establishing that the original breakwater had a maximum width of about 70 m, even allowing for the shifting of rock over the centuries, which is especially noticeable along its northwestern portion. The sand pockets seen today in the breakwater must have been produced by its modern use as a quarry for building material. The four visible concrete piers extending in a southerly direction out from the beach were measured (Fig. Ill—12). Probes around Piers 1 and 2 to a depth of 2.50 m were made, but no foundations were reached. Along the western side of Pier 1 a trench exposed impressions of six wooden upright planks used in the building of a single walled cofferdam—proof that the first method recommended by Vitruvius for underwater concrete construction was practiced at Cosa (Fig. III-13).37 Pier 4, now awash, was located ca. 36 m to the south and east of Pier 3. An addiRoman Port of Cosa," Actes du IV' International Congres d'Archaologie Sousmarine, Nice 1970, awaiting publication; McCann, "The Portus Cosanus: An Early Roman Port and Fishery," VI Congreso International de Arqueologia Submarina, Cartagena, Spain, 1982, awaiting publication; McCann, StHarbArch I, forthcoming; McCann, RCRFA, forthcoming. 37 Vitruvius 5.12.1-3. Compare the evidence from the concrete mole in the Roman Imperial harbor of Side in Pamphylia; H. Schlager, "Die Texte Vitruvs im Lichte der Untersuchungen am Hafen von Side," Bonnjbb 171 (1971) pp. 150-161. Schlager correctly points out that Vitruvius recommended three different methods for underwater concrete constructions. The first

64

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

tional Pier 5 was discovered under water at the eastern tip of the submerged breakwater, ca. 55 m east of Pier 4 (Map 6). No remains of any connecting piers were found. Airlifting around Piers 4 and 5 indicated that they rested upon an extensive foundation of boulders. Curving in a semicircle for a distance of ca. 100 m, a series of submerged, unconnected rock piles labeled A, B, C, D, E, and F was discovered (Map 6). To determine whether these rock piles were connected to the breakwater and to each other in antiquity, airlifting was done for a total working time of thirty-five hours between them and also between A and the eastern end of the breakwater. No rock connections were found. We have interpreted these rock piles as separated extensions of the main breakwater; they protected the anchorage area on the east and southeast while allowing the offshore currents traveling up the coast in a northwesterly direction to circulate through the harbor area. With the dominant wave action from the southwest, a scouring action would have been achieved.38 The two main tools used in the underwater excavations were a water jet prober (Fig. 111-52) and an airlift (Figs. 111-20, 53).39 Jet probes reaching to a depth of 3.50 m below the present ocean floor were made throughout the harbor area but particularly between the breakwater and its visible extensions to discover if they might rest upon a bedrock foundation and whether there were any connections between them not visible above the sand today. The main breakwater is now submerged in places to a depth of about 3 m, but it must have been above sea method, which must have been employed for the Cosa harbor piers, used a prefabricated form made of single wooden uprights held together by tie beams. The form was then floated into place and anchored to the bottom by driving the stakes into the sea floor. After the lower surface within was prepared, hydraulic concrete (made with the essential component, pozzolana) was poured into the frame. To strengthen and prevent buckling while the concrete was being poured, crossbeams were laid within. Spaces for such crossbeams are still visible in Piers 2 and 3 at Cosa. See Gazda's description in chapter five. This single-walled cofferdam is suitable for concrete construction in protected and shallow water with sandy bottom conditions, as at Cosa. Compare this method with that used at Herod's harbor at Caesarea Maritima (Sebastos), built into the open sea between 22 and 10/9 B.C., where recent underwater excavations have revealed evidence that a double-walled, prefabricated cofferdam was used. Once positioned underwater, the uprights were not pounded into the sea bottom, but mortised into the upper surfaces of heavy sleeper beams. Mortar was then poured between the section of the hollow frames and rubble piled around the outside of the block to steady it while the hydraulic concrete was dumped into the bottomless form. J. P. Oleson describes in detail the concrete technology used at Caesarea and its specific relationship to Vitruvius' third method of construction (5.12.5-6) in his recent paper, StHarbArch I, forthcoming. Oleson suggests that this mortar-

level in antiquity to have functioned as a protection against the prevailing winds from the south and southwest. Although the results obtained from the jet probing survey and by airlifting below the present sandy ocean floor are not conclusive, the evidence offers some support to geologist Jelle de Boer's suggestion that the main breakwater and at least some of its extensions may rest upon a submerged natural platform of rock, an extension of the present limestone promontory. 40 If this is the case, the design of the main breakwater and its extensions would in part have been conditioned by the bedrock foundation that once offered more extended protection along the southern and eastern sides of the barrier lagoon complex. A series of probes from the visible eastern end of the main breakwater directly east toward extension A hit a consistent rock level for a distance of about 28 m at depths of between 1.50 m and 2.75 m below the present ocean floor (Fig. 111-14). Likewise, probes running at 10-m intervals and angled out from the eastern tip of the main breakwater along its northeastern exposure record hard obstructions for a distance of about 22 m at similar depths. In an additional series of probes between breakwater extensions A and B, B and C, and C and F rock was recorded at depths between 0.20 m and 2.50 m below the present sand. These obstructions may be scattered rocks rather than bedrock, but the consistency of the rock at lower levels suggests bedrock. On the other hand, probes between the breakwater extensions D and F and D and E toward the shore did not encounter any hard rock obstructions. 41 packed, double-walled form was used at Caesarea because of the open exposure of the site and bottom conditions which prevented the driving of pilings into the sea floor. Since hydraulic concrete was available, probably made with imported pozzolana, the cofferdams did not have to be pumped dry, as described by Vitruvius. The concrete constructions found at Caesarea thus reveal a more developed technology designed to provide greater strength and flexibility of construction than the single-walled cofferdam used probably at least a century earlier at Cosa. 38 See McCann, chapter seven, reconstruction of the port. 39 See Lewis's discussion in the Appendix to this chapter. See also Lewis 1972, pp. 169-170. 40 Verbal communication, based on de Boer's geological survey of the Ager Cosanus in the summers of 1974-1980 with the Wesleyan University excavations. Compare discussion by Bourgeois, chapter two. 41 The probes are recorded in: J. D. Lewis, "Underwater Notebook, Cosa 1969," pp. 49, 50, 56, 57, 62, 67-69; R. L. Hohlfelder, "Outer Harbor Underwater Notebook, Cosa 1969," pp. 54, 55, 66-78, 82-87, 137-152. Probing by Hohlfelder in 1970 within the visible sand pockets of the main breakwater also indicated rock obstruction below 0.50 m; Hohlfelder, "Underwater Notebook, Populonia and Cosa, 1970," pp. 126-129. Compare Lewis 1973, p. 241.

Ill: MODERN SCHOLARSHIP AND EXCAVATIONS

65

Airlifting revealed further possible evidence for the existence of a submerged natural platform beneath parts of the breakwater area. Open trenches were dug to depths of about 1.50 m below the sea floor at the eastern tip of the main breakwater, at the western end of extension A, on the eastern and western sides of extension C, and on the north side of B. Beneath the large cut boulders in each trench, a layer of smaller stones of golf-ball size was revealed.42 These may be the remains of a worn upper surface of a natural platform upon which the larger rocks were placed. In the trenches at C and B fragments of concrete work were found, probably isolated tumble; but it is also possible that the fragments belonged to architectural structures, such as beacon lights, on these rock piles. In an attempt to establish the relationship between the concrete piers and the breakwater, further airlifting was done around Pier 5 (Fig. IV-15). Unfortunately, its foundation could not be established, since the airlift, placed on the pier's western side, could not penetrate the surrounding rock below a depth of 1.60 m. One can only assume, therefore, that the concrete piers were originally placed on top of the main breakwater at some time following its construction. On the southern side of Pier 5, pieces of tuff and concrete extend for a distance of about 1.50 m. Whether these are broken fragments from the pier or, more probably, an extention of its main core also could not be firmly established due to the difficulties of penetrating the rock mass with the equipment available.43 Breakwater extension D, the largest of the isolated islands, was selected for further excavation to determine its original height, to discover whether it rested upon any foundation, and to obtain datable evidence from the ancient harbor level (Fig. IV-5; Map 6). Two open trenches, D l (Fig. HI-15) and D2 (Figs. Ill—16, 17), were dug with the airlift on the western and eastern sides of D. Five rough courses of boulders were found to rest directly upon a sandy bottom with a total preserved height of 3.30 m (Fig. Ill—18). Two of these courses are visible above the sand for a height of 1.40 m. Information on the depth of the ancient harbor floor and the date of its period of use was also obtained (described below in detail). Three additional caisson trenches, C l , C2, and

C3, provided further data on the ancient harbor. Two steel cylinders, each 1 m in height and 1.30 m in diameter, were joined together, providing adequate space for a diver to excavate inside with the airlift (Figs. 111-19, 20, and 21). As the sand was removed, the caisson sunk into the sandy sea floor. This system of controlled excavation provided the chief stratigraphic information for the Portus Cosanus. The ancient harbor level was thus established at a depth of between Is m and 2.10 m below the present ocean floor, or at 5.10-6.20 m below the present sea level.

42 Recorded by J. D. Lewis, "Outer Harbor Notebook, Cosa, 1968," pp. 42-52. 43 Ibid., pp. 53-54. For the suggestion that Pier 5 originally carried a beacon light and for its probable association with a votive model of a tower from Vulci see McCann, chapter seven, recon-

struction of the port. 44 The amphoras from Level III in Trench Cl included: four fragments of Will Type 4a (Cats. A115-117, A144) and one fragment of 4b (Cat. A211).

CAISSON TRENCH Cl

Caisson Trench C l was located 26 m north of the eastern end of breakwater extension C in water 4.10 m deep (Map 6). Three stratified levels were distinguished (Figs. 111-22; IV-I). Level I, extending from the present ocean floor to a depth of 0.75 m, consisted of sterile sand. Level II, extending from 0.75 m to 1.15 m, was composed of rotted vegetation, mud, and gravel. The ancient archaeological layer began at Level III. In its upper section, 1.15 m to 1.25 m, the first ancient pottery fragments were found in a layer of gravel. These sherds are fragments of Sestius amphoras (Will Types 4a and 4b) dating from the last quarter of the second century B.C. through the first century B.C.44 Mixed with the amphora sherds were pieces of slate. From 1.25 m to 1.40 m the gravel changed to larger packed stones with some amphora sherds of the same type. From 1.40 m to 1.50 m a layer of small pebbles with some more amphora sherds was found. From 1.50 m to 2.10 m, or Level III, many amphora sherds appeared with fragments of tuff, a fragment of roof tile, and a piece of window glass (Cat. G13). In this level our oldest datable material from the harbor site was found. A hollow toe from the earliest identified type of Greco-Italic amphora (Type la, Cat. Al), dating from the late fourth century or first quarter of the third century B.C., is one of two of this form found in the excavations. In the same level two further amphora toes of Type Ic (Cats. A4 and A5), dating about 200 B.C., were recovered. These pieces provide archaeological evidence in support of the ancient historical sources for the use of the Portus Cosanus in the third century B.C. and as early as the founding of the colony in 273 B.C. Below 2.10 m in

66

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

Trench C l sherds and stones were absent, and only blackish sand occurred. The ancient harbor level was thus established here between 1.15 m and 2.10 m below the present ocean floor, or from 5.25 m to 6.20 m below the present sea level and from 4.25 m to 5.20 m below the estimated ancient sea level. CAISSON TRENCH C2

Caisson Trench C2 was set up in water 4.20 m deep 20 m to the southwest of the southern tip of breakwater extension D, approximately midway in the presumed harbor channel entrance (Map 6). Three levels of stratification were distinguished. Level I, extending to a depth of 1 m below the present sea bottom, was composed again of sterile sand. Light-colored sand was found to extend down to 0.80 m, where it then changed to a blackish color. In Level II, from 1 m to 1.15 m, a few amphora sherds of Types 4a and 4b were found in a layer of small stones and rocks. Below this level no further ancient pottery was found in a layer of blackish, firmly packed sand. Airlifting continued, however, to a depth of 1.50 m below the sand, and probes went down to 2 m. No further pottery or changes in stratification were discovered. The lack of pottery in this caisson, compared with C l , led us to conclude that Trench C2 was within the open entrance channel, where a concentration of sherds would be less likely. Jet probes made between extensions D and F later bore out this assumption. CAISSON TRENCH C3

The third caisson Trench C3 was located 29 m southwest of the northernmost tip of extension D (Map 6). Unfortunately, after airlifting to a depth of 0.40 m below the sterile sand, a heavy storm carried away the cylinders, and no further information was obtained. TRENCH Dl

Trench Dl was located on the western side of breakwater extension D, 6 m to the north of its southernmost tip and 4.10 m b.s.l. (Map 6). This open trench measured 2 m north-south by 6.30 m east-west, with a section 4.50 m deep cut into the breakwater itself (Figs. Ill—15; IV-I). The boulders 45 The amphora fragments from D include: one fragment of Will Type Ic dating about 200 B.C. (Cat. A3) and one fragment of Will Type Id dating about 180-150 B.C. (Cat. A13); one fragment of Will Type 4b dating from about 75-0 B.C. (Cat. A207); four fragments of Will Type 12b dating in the first or second century A.D. (Cats. A261, A262, A264, A265); one fragment of

from this section, weighing 200 to 300 kg, were removed by means of large crowbars, chained to inflated plastic bags, and then lifted away. Because so many ancient amphora sherds and tiles were wedged and encrusted between the rocks, the larger boulders were held in a tight bond, making them extremely difficult to dislodge. When the rocks were cleared from the trench to the level of the present sand, the airlift was used for excavating. Airlifting proceeded to 1.90 m below the present ocean floor, or 6 m below the present sea level, where the boulders stop. Jet probes for an additional depth of 4 m struck no further rock. The foundations of the ancient breakwater extension D were thus laid directly upon the sandy sea bottom. Five rough courses of boulders were revealed in Trench D l for a preserved height of 3.30 m. Two of these courses are visible above the sand for a height of 1.40 m, or 2.70 m b.s.l. Since no timber was found—which might suggest the use of sunken barges for the transport of rock in ancient times—the heavy boulders must have been brought separately by barge or float and dropped on the predetermined location. Within Trench Dl three levels of stratification were distinguished. Level I, extending down for a depth of 0.90 m below the present ocean floor, consisted of sterile sand. This modern layer was followed by a thin layer of rotted vegetation 0.10 m thick (Level II). The ancient archaeological material began in Level III at a depth of 1 m and continued to 2 m below the present harbor floor. Amphora sherds ranging in date from about 200 B.C. through the second century A.D. were recovered in this gravel layer.45 The greatest concentration of sherds occurred between 1 m and 1.60 m at 5.10 m and 5.70 m below the present sea level, or between 4.10 m and 4.70 m below the estimated ancient sea level. The nature of the sea environment made it impossible to distinguish individual layers within the general archaeological level. TRENCH D2

Trench D2 was located on the eastern side of breakwater extension D, 12 m from its southern end and 2.50 m to the east of it (Map 6; Fig. IV-I). The trench was cut in a semicircle with a radius of ca. Will Type 18b, dating in the second century A.D. (Cat. A283); one fragment of a jar related to Dressel's Form 29 and dated by Will in the second or first half of the third century A.D. (Cat. A300); and a fragment of an unidentified type like Dressel Form 28 or 29 (Cat. A301).

Ill: MODERN SCHOLARSHIP AND EXCAVATIONS

3.50 m under and out from the rocks (Fig. Ill—16). At a depth of 1.30 m to 1.70 m a level of ancient pottery was found (Fig. 111-17) which included amphora fragments of the Sestius types (Type 4a [Cat. Al 13] and 4b [Cats. A208 and A210]) dating from the late second through the first century B.C. and one amphora fragment from the fifth century A.D. (Type 21c [Cat. A294]). The latter is the latest ancient fragment from the harbor and appears to be an isolated one. Besides the amphora material in this ancient Level II, we found the rim of an undecorated Late Italian sigillata plate, dating from the second to the fourth quarters of the first century A.D. (Cat. RG6), and two pieces of utilitarian ware (Cats. K29 and C l 5). The Olesons date the fragment of a skillet (Cat. K29) between 125 B.C. and 25 B.C. and the fragment of a jug (Cat. C15) about the mid-first century A.D. Directly below this layer of sherds and gravel a shallow, uneven layer of rotted vegetation occurred (Fig. IH-18). Below 1.70 m only occasional pottery fragments were found, and at a depth of 1.90 m the sand was sterile in Level III. Jet probes made at slanting angles underneath the breakwater indicated that the rocks extended down to the bottom of the ancient archaeological level. The evidence of Trench D l was thus confirmed: breakwater extension D was originally built up directly from the sandy floor and does not rest upon a bedrock foundation.

The Harbor Entrance Channel Using the water jet prober, we located the ship channel entrance 33 m wide between breakwater extensions D and F (Map 6). A plastic line was laid out and marked at every meter on the sea floor between the southernmost visible end of extension D and the northernmost tip of extension F. This area appeared to be the most likely place for a ship channel, since it is the largest gap in the series of breakwater extensions. Along this line forty-three probes were made, from which a profile of the area was constructed. At depths of between 1.30 m and 1.80 m below the sand sherd material was sounded, but no rock barriers were encountered. Readings also revealed a clear, deeper channel in the center of this area, with an estimated ancient water depth of about 6 m. Along the sides of this presumed entrance, rock has tumbled toward the center of the channel from the breakwater extensions. Nevertheless, an uninterrupted width of about 33 m was established—ample room for medium-sized ships to enter and depart the

67

harbor safely. A point of particular interest within this area was found at approximately 15.50 m south of extension D, where a soft spot was felt with the prober. From the hole made by the prober, dark organic material was thrown up which appeared to be remains of wood. Further investigation of this area is planned for the future.

THE BEACH WALLS AND CLIFF CHANNELS

In the exploration of the shore area in 1968, a concrete embankment, Wall M, was uncovered in line with Pier 1 and traced for a distance of 7.43 m to the north across the present parking lot (Map 6). Trench N (Figs. 111-23; IV-I) was cut about 13 m to the north of the exposed section of Wall M, but no further evidence for its extension could be found. Presumably this retaining wall was destroyed in the building of the modern parking lot, if not earlier. Along the beach 18 m to the east of Wall M, two facing limestone polygonal walls, Walls A and B, were uncovered. They bridge a channel 7 m wide at their northern extension and 4 m wide on their southern one. Cuttings for sluice gates also appear on their inward faces. Just to the east, parallel rubble and concrete walls, Walls C, D, and E (Fig. IV-18), were identified as later constructions belonging to the period of the villa. About one kilometer east of the Portus Cosanus a pair of additional rubble concrete walls was found extending into the sea (Figs. IV-26, 27), and their length was traced for a distance of about 56 m. These walls embank a channel about 10 m wide. Their mouth in the sea, which is divided by two small piers and another larger pier to the south, was mapped. The channels in the limestone cliffs that protect the harbor on the west received further attention, particularly the northernmost section of the Tagliata leading north from the quarry basin into the ancient lagoon area. Cuttings for sluice gates along the face of the limestone cliffs were noted (Fig. IV-43). The ancient eastern side of this water channel, formed by a concrete wall resting upon bedrock, was traced for a distance of ca. 52 m along the base of the presentday concrete wall (Fig. 111-24). A profile of the channel (Sl) revealed a rock bottom at an average depth of 1.36 m b.s.l., with a width varying from 3 m to 5 m (Text Fig. II—5). Today a modern drainage canal that cuts north and east through the now silted lagoon follows the course of this ancient waterway. From its shallow depth and narrow width, we con-

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

68

eluded that this waterway was not an ancient ship channel, as had been previously suggested.

villa walls perhaps indicates a second building phase for the maritime villa.46 Architectural fragments associated with the period of the Imperial villa included many pieces of colored marble mosaics found in the beach area, two columns of marble (left under water) and one of gray granite (Inv. no. PC 69-133), a piece of window glass (Cat. Gl 3; Fig. XII-20), and many fragments of bricks and tiles. An abundant number of dolia fragments were also recorded, as well as slate and a millstone. The remains of the wooden hull and prow of a ship found submerged in the shallow water east of Piers 1 and 2 and probably dating in the late sixteenth century A.D. will be discussed by Lewis in an Appendix to chapter eight.

THE HARBOR FINDS

Although the finds from the harbor are included in the discussion of the chronology of the site as a whole in chapter eighteen, it is useful to summarize them as a separate group, for they reinforce the larger picture. The underwater excavations turned up only Roman material, ranging in date from the late fourth or early third century B.C. to the third century A.D., with one fifth-century A.D. find, an amphora fragment from North Africa (Type 21c [Cat. A294]). No material was found to indicate the port's use in Etruscan times. The earliest catalogued find is a fragment of a Greco-Italic amphora of Type la (Cat. Al; Fig. IX-5), dated in the late fourth or early third century B.C. Documenting the use of the port in the second century B.C. are three catalogued fragments of amphoras of Type Ic, dated ca. 200 B.C., and five catalogued fragments of amphoras of Type Id, dated from about 180 to 150 B.C. One piece of black glaze ware of Campana A dating in the third quarter of the second century B. c. was catalogued (Cat. BG2). But the clear floruit of the port's trading life, as that of the whole site, is established for the late second century through the first century B.c. by the dominance of the Sestius amphoras of Types 4a (most numerous) and 4b. From the harbor alone the catalogued pieces of these two major forms numbered fifty-eight, and many uncatalogued pieces were also recorded. Within the concrete itself, in the upper repaired sections of Pier 1 and Wall M, two early Type 4b rosette stamps were noted (Cat. A195; Fig. IX-299). After the end of the first century B.C., a distinct gap is evident in the material until the second half of the first century A.D., when the amphora finds begin again and continue into the third century A.D. Eleven amphora fragments of Imperial types were catalogued from this time span. Pieces of commonware recovered range in date from ca. 125 to 25 B.C. (Cat. K29), with one catalogued fragment after about 50 A.D. (Cat. C15). Further documenting the Imperial period of the harbor's use are a Late Italian sigillata plate (Cat. RG6; Figs. X-82, 83) from the later first century A.D. and a lamp fragment, dated after ca. 50 A.D. (Cat. L5; Fig. XII-28). A coin of Faustina the Younger (A.D. 161-176, Cat. C O l ; Figs. XII-33, 34) found in the

THE LAGOON EXCAVATIONS

Simultaneously with the excavations in the seaport, digging also began in 1968 in the silted lagoon, which we believed at that time to be an inner harbor (Maps 7, 8, and 10; Fig. 111-25). Two trenches were dug along the western embankment of the lagoon where sections of ancient limestone walls were exposed. Trench IH (Inner Harbor) was located where a stretch of rubble repairs was observed in the polygonal embankment wall, identified as Wall P (Maps 8 and 10; Figs. 111-26, 27). Placed perpendicular to the wall, the original Trench IH measured east-west 7.5 m, with a north-south width of 2 m. It was later extended for an additional 9 m terraced to the west up the slope (Fig. 111-28). The foundations for Wall P at this point were found to rest upon bedrock. The wall was buttressed by a retaining wall 1.20 m thick. An additional parallel polygonal wall, Wall PW, was traced about 3.50 m to the west of Wall P. To the east of Wall P a cobblestone pavement 5.84 m wide came to light 0.20 m below the surface at present sea level (Fig. 111-29). A horseshoe was found embedded in the pavement. This postRoman road must have served the lagoon during some period of its later use, probably during the sixteenth century, when the Fontana della Regina was used and the port was guarded by the defense tower. A second trench, IC (Inner Harbor, C), was placed about 36 m to the south of Trench IH to the east of the southernmost pair of projecting concrete rubble walls, later identified as Wall a of the Spring House (Map 8). These walls lay directly below a cistern about 32 m up the slope to the west (Fig. Ill— 30). The original trench projected to the east for a

See Gazda's discussion of the villa walls in chapters four and seven.

Ill: MODERN SCHOLARSHIP AND EXCAVATIONS

distance of 7 m, with a north-south width of 2 m. The cobblestone pavement found in Trench IH was also picked up in Trench IC. Below it, at a depth of 1.50 m and 0.28 m a.s.l., a portion of a tuff-andconcrete platform 3.55 m long and 1.55 m wide was discovered, surrounded by the excellently preserved wooden planking of its form work (Figs. Ill—31, 32). A concrete pedestal, pier 2, was uncovered upon it. A section of Wall P was revealed abutting the platform to the south. Above Trench IC, the cistern was surveyed. On the eastern side of the modern drainage canal 35 m to the east of Wall P, a parallel limestone polygonal wall, PE, was traced for a distance of about 78 m (Figs. IV-77, 78; Maps 7, 8). Its northern tip fades into a low rubble wall (Fig. IV-79), and its southern end is lost under the modern road. Three courses of Wall PE visible above the modern canal were cleared, but its depth below the water level could not be defined with the equipment available. Later rubble repairs along Wall PE were also identified. Excavations were continued in the silted lagoon in 1969 (Maps 5 and 10). The line of the polygonal embankment Wall P was traced for 131 m south of Trench IC in a series of three probes (Pl, P2, and P3; (Figs. 111-33, 34) and three trenches (ID, IF, and IG; Figs. 111-35 to 39). In Trench ID four courses of the polygonal blocks of Wall P and a backing of smaller stones were revealed. In Trench IF a layer of packed lime, pebbles, and ash was found in Level III below the surface but above Wall P (Figs. 111-37, 38). This layer probably represents the remains of a limekiln used in the refurbishing of the lagoon and spring during the sixteenth century, also the probable period of the cobblestone pavement found in Trench IH in 1968. A horseshoe was also found in this layer of Trench IF. The southern terminus of Wall P was uncovered in Trench IG where it joins the natural bedrock of the northern end of the cliffs protecting the harbor (Figs. 111-39, 40). In 1969 work also continued in Trench IC, which was extended eastward to reveal the complete tuffand-concrete platform of the Spring House and its wooden formwork (Figs. V-4 to 7). The rectangular shape with two projecting wings supporting two concrete pedestals or piers measured 4.58 m at its southern end and 6.90 m along its western side. The platform, bisected by an open terracotta drain, was found to form the eastern wall of a basin in which fresh water gushed from an underground spring calculated to flow at about 400 gallons per minute. During excavation continuous pumping was neces-

69

sary, and a modern cofferdam of wooden planking had to be erected around the trench to keep its walls from collapsing (Fig. 111-41). Below a lower ledge of the southern wing of the platform, polygonal Wall P was revealed, indicating that it must have been designed and built at the same time. To the east, the tops of two further concrete pedestals, piers 3 and 4, came to light. In conjunction with the excavations in 1969, a cesium magnetometer survey and a resistivity survey were carried out over the excavated area of the silted lagoon by E. Ralph of the University of Pennsylvania (Fig. Ill—42). Unfortunately, the results of this survey were largely negative, owing to the highly magnetic sediments present in the silted pasture land. In 1970 a more successful magnetic survey with a differential proton magnetometer was made by R. E. Linington of the Lerici Foundation in Rome (Map 9). He also carried out a drilling survey over the western section of the ancient lagoon in 1971 in which fifty-three cores were made to an average depth of 6 m b.s.l. (maximum depth, 11 m b.s.l.). The material from these cores was further analyzed by soil specialists I. Barshad and B. Viani of the University of California at Berkeley (chapter fifteen) and also used by geologist J. Bourgeois for her sedimentological study of the lagoon (chapter two). Because of the enormous difficulties involved in digging in the silted lagoon, where water and mud occur within a meter of the surface, it was decided that mechanical help would be needed to complete the desired excavation plan of this western boundary of the ancient Cosa lagoon. In 1972 over 2 m of sterile mud covering about two and one-half acres was removed by means of back-hoe and bulldozer with continuous pumping (Color Fig. 2; Fig. 111-43). This season was devoted to the excavation of the silted lagoon, and a plan of its western and northern sections was achieved (Maps 5, 7, 8). Western embankment Wall P was revealed for its completed length of 250 m, extending from the limestone cliffs on the south to its juncture with concrete Wall X to the north where one road probably entered between the low hillocks from the town site above. Wall X forms the northern boundary of the lagoon and was traced to the east for a length of 43.75 m. At this point it turns south to form Wall U for a length of 54 m. At the base of Wall U, along its eastern side, a series of almost complete amphoras of Type 4b was found in neat rows (Cats. A224-A236, A239A242; Fig. 111-44). They must have been used for fill to support a packed earth embankment behind the wall.

70

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

In 1972 the plan of the Spring House was also completed (Map 14). Behind the previously uncovered platform and water basin two attached rooms were revealed to form a rectangular complex 13 m long (east-west) by 6.25 m wide (north-south). Of particular interest was the discovery in Room 2 of the remains of a wooden water-lifting device—a bucket chain driven by an angled gear (known as a saqiya in the Arabic world today). Remains of seven of the buckets were found with other wooden parts (chapter thirteen, Cats. W1-W16). For lack of means to preserve them at that time, the wooden pieces had to be reburied. They were reexcavated in 1975 by J. P. Oleson, who experimentally preserved some of the wooden parts by the acetone/rosin impregnation process. A modern reconstruction of one of the buckets was later made at the University of Victoria, British Columbia, for display in the Cosa Museum (Figs. V-58, 59, 60). The connection of the Spring House and its water source during its different periods of use with both the eastern side of the lagoon and with the cistern above to the west was also established. An Aqueduct composed of at least six piers (nos. 3-8; another, pier 9, was located east of Wall PE) was found extending eastward from a collecting tank on the south side of the Spring House, a structure associated with the spring's earliest period of use (Fig. Ill—45). A later phase of the spring's use was indicated by another water channel, carried on a series of six other masonry piers supporting rubble columns, found leading westward from the Spring House to the cistern. This unique complex was excavated by J. P. Oleson and is discussed in detail in chapter five. The remains of two long rectangular structures found in the center of the ancient lagoon have since been identified as fish tanks (Map 8 and see below, chapter seven). Together they enclose an area of about one hectare. Remains of some of their original wooden forms were well preserved around them. The northern tank, defined by narrow concrete Walls W and V, was traced for a length of 96 m until it was broken off by the Spring House complex. Its east-west width was 32.75 m. An additional spring was released out of the bedrock into the tank's northeast corner, where a small channel was placed (Trench U5, Fig. 11-15). A second fish tank to the south of the Spring House was separated from the northern one by the Aqueduct. The north-south length of the southern fish tank defined by concrete Wall N was traced for a length of 91 m until the modern road prevented further excavation. Its eastwest width was traced for 11 m, but probes indi-

cated that this wall continued at least to the modern drainage canal for an additional 6 m. The height of Wall N was recorded at about 1.40 m at its northern end and at about 2.75 m at its southern end, with probes indicating that it continued downward (Fig. 11-18). Today the height of the rubble fish tank walls in Lago di Burano above the sandy bottom of the lagoon is about 1.70 m. Unfortunately, the bottom of the ancient lagoon behind the Portus Cosanus could not be reached by excavation with the equipment available. Nevertheless, probes were made to the east of pier 3, where a hard-packed layer of sand was encountered at a depth of about 2.50 m b.s.l. In this lowest dark sand layer also revealed within the southern end of the fish tank formed by Wall N, small bones were found, along with bits of tiles— perhaps our one indication of the ancient fish population. Unfortunately, the Lerici Foundation did not practice flotation at that time, so further identification of this fine bone material is lacking. The drilling survey suggested a bottom for the ancient lagoon along Wall W south of area YZ in Cores 44, 45, 46, and 47, where bedrock appears to have been struck at a depth of about 5 m b.s.l. An additional structure was found to interrupt Wall P some 50 m to the north of its junction with the northern section of the Spring House platform in the area called YZ (Maps 10 and 12). Two rubble piers constructed in a later building technique like that of the piers leading from the Spring House to the cistern were found projecting from Wall P (Fig. Ill—46). Fragments of two additional piers were recovered to the east with rubble and concrete crosswalls that enclosed a small rectangular structure 5.80 m (east-west at its southern end) by 7 m (northsouth along its eastern side; Fig. 111-47). Another freshwater spring was released, rising from the bedrock under Wall P, and a terracotta drain was found leading from the enclosure at its southeastern corner (Fig. 111-48). This structure has thus been identified as another spring house or small nymphaeam used by the villa inhabitants during Imperial times. Remains of animal bones and a wooden trough were recovered from the area immediately to the east of the fountain building, indicating that this area just west of the northern fish tank was filled in at that time. The latest Roman material from the lagoon, dating from the second and third centuries, is concentrated in this YZ zone. In the area east of the modern drainage canal and in line with concrete piers 3 to 8 an additional pier 9 of different construction was uncovered (Fig. 111-49). A corner of a structure defined by Wall S was also

Ill: MODERN SCHOLARSHIP AND EXCAVATIONS

revealed 3.6 m to the south of the southern face of pier 9 (Fig. 111-50). These ancient remains indicate that the area to the east of embankment Wall PE and the modern drainage canal was land in antiquity and attached to the freshwater source on the western bank of the lagoon for at least some period during the Spring House's use. Excavations were also carried out in 1972 in the modern parking lot along the beach. Seven probes were made in search of the extensions of polygonal Walls A and B and the presumed eastern side of a channel formed by Wall M (Map 6). In Probe V, between Walls A and B, a faced brick and limestone concrete wall (O) was found above Wall A and attached to it (Figs. IV-I, 21). This rubble wall is documented by a coin of Faustina the Younger (A. D. 161-176).47 This later Wall O must have belonged to the Imperial villa, and it indicates that the channel connecting the lagoon to the sea (bridged by Walls 47

See McCann, chapter twelve, Cat. COl.

71

A and B) was not functioning by that time. Wall B was traced further inland for a length of ca. 25 m, where it disappeared under the parking lot. Additional probes to the east and parallel to Wall M brought no further walls to light. The presumed western side of this water course must have been destroyed in the building of the modern parking lot, if not before. Scientific studies were carried out during the excavations and continued along with the study of the archaeological material. The reports of the various scientists are presented here in Part IV. This book represents the collective efforts of a team of over thirty scholars, scientists, architects, and engineers. Previous reports are superseded by the present publication, in which the results of all earlier investigations have been coordinated with the later archaeological and scientific research.

Appendix. The Tools and Techniques of the Underwater Excavations JOHN DAVID LEWIS The marine archaeologists and engineers who excavate ancient harbors today along Italy's sandy shores face extraordinary challenges. Owing to silt from river deposits, many of the ancient ports along the Tyrrhenian seacoast, such as Ostia and Luni, are now actually miles from the sea. Other ancient harbors not located at river mouths but in partially sheltered coves, such as Populonia, Pyrgi, and Cosa, have been partially covered by sand. The ancient remains are there for the digging, but access to the ancient harbor floor and to the intermediate layers has required techniques for exploration and excavation different from those now commonly used in uncovering ancient shipwrecks. Since the port excavations at Cosa are among the very first in harbor research since the development of SCUBA, it seems worthwhile to summarize and illustrate here some of the tools adapted and used for the recovery of the underwater data presented in this book. The port of Cosa also presents a typical sand-filled harbor situation, and it is hoped that some of our solutions may provide insight for later excavations of similar ancient harbor sites. The ancient man-made breakwater at Cosa both protected the harbor from the predominantly southwesterly winds and also acted as a barrier for the natural currents traveling up the seacoast in a northwesterly direction with deposits of silt and sand (Map 6; Figs. IV-2 to 5). Thus the ancient Cosa breakwater and its extensions have been partially covered by sand over the centuries (Fig. 111-14), a process that probably began in antiquity 1 despite the ingenious solution of its Roman engineer, who calculated in his design both the wave direction and the need for a continuous flow of water through the harbor to reduce silting.2 The placement of the separated islands of rock off the main breakwater must have been part of a carefully thought out scheme, the product of a knowledge of the particular coastal problems, and a sophisticated appreciation of previous ancient harbor engineering inherited from the 1

See geology discussion by Bourgeois, chapter two. See Lewis 1973, pp. 239-240, and McCann, chapter seven, reconstruction of the port. 2

Greeks and probably also the Etruscans. The breakwater extensions, as they have been so named, may be considered forerunners of the later Imperial Roman harbor works, with their elaborate moles and piers connected by arches such as may be observed underwater today at Puteoli, Baia, and Anzio. In studying the visible ancient remains still above the sand in the port area of Cosa in 1965, it was obvious that we needed to penetrate below the present sandy sea floor to answer our questions about the ancient harbor plan. First, what were the original dimensions of the breakwater, and what was its foundation? In 1965 only the main breakwater was visible above the present sea bottom, and it was only in the winter storms of 1968 and 1969, when over 0.50 m of sand was washed away, that the extensions suddenly became visible. Second, where was the most likely entrance for ships coming into the harbor, and what was the depth and stratigraphy of the ancient seabed? Finally, while remains of ancient amphora sherds were everywhere on the surface of the present harbor floor, we hoped to recover artifacts that could be dated from the context of the ancient harbor floor itself. We obviously needed special tools that could penetrate the sand in a controlled digging environment. The most common pieces of equipment used in underwater excavations today are the airlift and the dredge, both of which have the disadvantage of creating enormous cones when they are operated in an open sandy environment such as Cosa. These cones continually fill up with sand, preventing accurate measurements and photography of structures found. Two pieces of equipment were thus specially designed and adapted for our particular needs.

THE CAISSON

One means of controlling the flowing sand is to dig within a confined and protected space such as

Ill, APPENDIX: TOOLS AND T E C H N I Q U E S

provided by a caisson or vertical cylinder. The cais­ son technique has long been used for the construc­ tion of river tunnels and bridges, and we adapted it on a smaller scale to our needs at Cosa. The caisson 3 used at Cosa was constructed of two pieces of 20gauge sheet steel (1.00 m χ 4.30 m), each sheet weighing about 25 kg. The sheets were rolled into two cylinders, and then each lap was welded every 0.50 m. They were then bolted together to give a maximum working depth under the sand of over 2 m (Fig. 111-51). Additional cylinders could have been added to increase the operating depth. The interior space was just large enough to accommodate a diver working the nozzle of the airlift and provided the needed protection against backfilling sand (Fig. III20). As sand was excavated and removed from within the caisson, it sank into the sea floor (Fig. Ill— 21). In this way the stratigraphy of the harbor floor could be accurately measured and photographed without haste (Fig. 111-22). Three to four hours was the time required to excavate down through 2 m of sand and record measurements. The use of light steel cylinders also had the advantage of mobility. The caisson could be moved easily by inflating air bags and floating it under water to another location (Fig. Ill—19). Three caisson trenches were dug in the har­ bor in 1969, and the results are recorded here in the above section in chapter three on the underwater ex­ cavations.

THE WATER JET PROBER

A simple and inexpensive piece of equipment, the water jet prober proved invaluable both for locating and defining structures hidden beneath the sand and for making soundings throughout the harbor area to obtain data on depth and stratigraphy. This tool proved a viable alternative to more expensive acous­ tic, magnetic, or coring surveys. 4 The jet prober uses water pressure directed through a nozzle; the force of the water digs the nozzle and accompanying pipe into the sand (Fig. 111-52). The prober consists of two main parts: a portable gasoline-driven water pump that delivers 40 psi pressure mounted on the surface of the water in an inflatable Zodiac rubber boat; and attached to it, a lVa-inch diameter firehose 3

See Lewis 1972, pp. 169-170.

73

connected to a 3/4-inch pipe marked every 0.50 m and fitted with a modified firehose nozzle. The depth at which the nozzle strikes any hard object can therefore be accurately recorded. The maximum depth that this jet prober can penetrate is 3.5 m; a more powerful water pump would be required for deeper probing. One diver operated the prober un­ derwater while another recorded readings. The flow of water up and around the pipe prevented it from becoming wedged in the sand. In operation, a diver can soon learn to "feel" both changes in the sedimental layers of the harbor and hard material. Thus, the prober proved useful in obtaining stratigraphic information as well as in defining the outlines of the main breakwater on its northern and southern sides, its various extensions, and the profile of the ancient ship channel entrance. The results of our probing surveys are recorded in the above section on the underwater excavations.

T H E AIRLIFT

The airlift, a type of suction hose acting as an un­ derwater vacuum cleaner, has been in common use since its development in 1952 by Jacques Cousteau and his team in the excavations of the wreck at the Grand Congloue. A number of open trenches were dug with the airlift at Cosa, including ones at the easternmost tip of the main breakwater and exten­ sion A (Fig. 111-53) and on both the western and the eastern sides of extension D (Figs. 111-16, 17). With the airlift, a full profile of these breakwater elements was revealed. Since the underwater excavations at Cosa had the advantages of being near shore and in relatively shallow water (under 30 feet) the air for the airlift could be provided from a low-pressure air compressor placed on the beach. Simple garden hoses anchored underwater brought the air to the airlift nozzle, which was made from an irrigation pipe about 5 inches in diameter fitted with anchors and a float. For its use within the caisson, a flexible type of exhaust tubing was fitted onto the end of the irrigation pipe. The airlift proved effective for re­ vealing major architectural elements. For obtaining stratigraphic information, the airlift was used within the caisson. 4

McCann and Lewis 1970, pp. 204-206, fig. on p. 205.

Chapter IV. The Port and Fishery: Description of the Extant Remains and Sequence of Construction ELAINE K. GAZDA

The purpose of this chapter is to describe fully all the pertinent remains thus far found of the ancient harbor and fishery complex at Cosa and to present a sequence of construction based largely upon the visual evidence. Excluded from this discussion is the one complete architectural unit discovered, the Spring House, which is presented in chapter five by its excavator, J. P. Oleson. However, the material character, structural history, and chronology of the Spring House are integrally related to those of the fishery and port in at least two main periods of their development. The artifacts found within the Spring House, along with the amphora finds from the whole site, will largely provide the dates for the phases of architectural development discussed by McCann in chapter eighteen. Although the form and function of most of the structures are readily identifiable, some remains pose problems for which the present study cannot offer definitive solutions. These problems are due in part to the fortunes of preservation. Other uncertainties remain because modern construction inhibited excavation in certain critical areas, including the western end of the harbor basin, where it joins the quarry in the cliffs, and the juncture of the northernmost end of the Tagliata channel with the ancient fishing lagoon. Also, the modern drainage canal that cuts through the middle of the lagoon area has destroyed some of the remains of the ancient fishery. In addition, the Torre della Tagliata with the adjacent modern villa, coast guard station, tourist bar, and restaurant-bathing establishment called La Strega are all built upon areas rich in Roman material but beyond the limits of the excavational permit and plan. Since further excavation in these areas might modify the current interpretations, an account as objective as possible of the extant structures in the main sections of the excavated site—the port, the cliffs, and the lagoon—will be given before an interpretation of

them is offered. A summary of the evidence for the sequence of building construction in each of these areas will also be included after each descriptive section to prepare the way for the reconstruction and chronological discussions of the site as a whole in chapters seven and eighteen.

THE PORT AREA AND RELATED COASTAL STRUCTURES

Within the port area of Cosa there still remain impressive ancient structures of varied size and function. The largest among them is the rough limestone breakwater, now completely submerged, upon which two of five large concrete piers, identified here as Piers 1 through 5, were built. Three embankment walls protruding from the beach—one of concrete (Wall M) and a pair of polygonal construction (Walls A and B)—formed part of a system of channels connecting the ancient fishing lagoon to the sea. In antiquity a pair of concrete walls extending into the sea one kilometer to the east of the harbor and a submerged polygonal wall off the Torre di Burano some three kilometers further to the east provided additional links between the long, narrow barrier lagoon and the sea. Finally, two series of parallel walls (O, C, D, E, F), which extend toward the harbor basin on either side of the Torre della Tagliata, and a vaulted cistern nearby attest to the presence of a large, Imperial seaside villa, constructed after at least some of the channel walls (A and B) had gone out of use. BREAKWATER

Protecting the cove against the prevailing winds from the south and southwest is a broad breakwater constructed of large boulders quarried from the ad-

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N 1

jacent limestone cliffs. This breakwater, now com­ pletely under water, extends for ca. 110 m almost directly eastward from the promontory (Map 6; Figs. IV-I to 5). All along the southern (seaward) face of the breakwater the rocks are laid in a straight line, and on the northeastern (inner) face a similar line can be traced for about 70 m from the eastern tip inward toward the shore. This inner face angles sharply toward the beach at about 25° west of north (Fig. IV-4). Beyond the point where a clear line can be traced, rocks spread out in an area to the north of the breakwater, forming an irregular curving con­ tour that ends approximately 23 m to the east of Pier 3. This curved line does not represent the original contour of the breakwater; rather, it appears to be the result of wave action, which spread rocks from the top of the breakwater toward the north and the northeast, and of sand erosion from beneath the boulders. Some dislodging may also have been caused by earthquake activity. Nevertheless, the sharply pointed breakwater was very broad, meas­ uring up to 70 m across in some places. Today the breakwater lies at 1.5-3.0 m b.s.l. (ca. 0.5-2.0 m below the ancient sea level), and much of its structure is obscured by sand. The boulders are visible at heights varying between 0.5 m and 2.0 m above the sea floor, but they continue downward for at least another 1.5 m. 2 It is not surprising that the largest rocks—some of them measuring ca. 2.0 m χ 1.5 m χ 1.0 m and weighing up to 2,000 kg—are found along the seaward face of the breakwater. The smallest rocks (ca. 0.5 m χ 0.5 m χ 0.4 m) are found along the innerside of the breakwater, probably be­ cause less protection was needed there and also be­ cause waves tumbled the lighter boulders in that di­ rection (Fig. Ill—14). Beginning about 15 m to the east of the tip of the breakwater is a series of discontinuous submerged rock mounds (A-F; Fig. IV-5), the visible remains of which differ greatly in dimensions (measured eastwest by north-south): (A) 16 m χ 13 m; (B) 10 m χ 10 m; (C) 8.5 m χ 17.0 m; (D) 15 m χ 36 m; (E) 6 m χ 13 m; and (F) 6 m χ 23 m. The mounds, lying at depths that vary between 2 m and 4 m b.s.l., ex1 Much of the information presented in this description of the breakwater and harbor remains is derived from yearly excavational reports by McCann and the 1968 and 1969 notebooks of Lewis; McCann and Lewis 1970; Lewis 1973; and McCann 1979. Additional observations and interpretations are based on the pres­ ent author's first-hand examination of all accessible structures in 1972, 1973, and 1974. Further information relevant to the inter­ pretation of the structures is presented by McCann in chapters three and seven. 2 For the problem of the present depth of the breakwater, see Lewis 1973, pp. 241-242.

75

tend eastward in a semicircle for a distance of about 100 m beyond the breakwater. The first four mounds in the series—A, B, C, and F—form a dia­ mond-shaped cluster. A (Fig. IV-6) and B (Fig. IV-7) roughly continue the line of the inner, northeastern face of the breakwater for about 65 m toward the southeast; F (Fig. IV-8), located 23 m northeast of B, begins the return toward shore, which is contin­ ued by D (Figs. 111-15 to 18; IV-9), situated ca. 43 m farther to the northeast, and by E (Fig. IV-10), which lies ca. 22 m to the northwest of D. In the 85-m space between E and the present shoreline, no other rock mounds were found. The breakwater and associated rock mounds A-F must have been constructed as coordinated parts of the same system of protection for the harbor of Cosa. Today they define a basin of some 25,000 sq. m; measured from the present shoreline to the northern side of extension C, the basin has a diam­ eter of ca. 190 m. Allowing for the rise in sea level of approximately 1 m, the ancient shoreline must have been somewhat farther to the south than the modern beach; thus an ancient harbor basin corre­ spondingly smaller in area must be envisioned. The entrance to the harbor has been established between mounds D and F, where there is an open space of 33 m (Map 6). The depth of the channel in antiquity was approximately 6 m b.s.l. Construction of the breakwater of the port of Cosa, must have been rapid and economical. The rough boulders of the main breakwater were prob­ ably quarried and hauled directly from the nearby cliffs, using a completed section of the broad plat­ form for extending the breakwater eastward. Rocks for the isolated mounds might have been carried on ropes slung between two barges and then dropped directly upon the designated areas. Some of the smaller stones may have been brought in as ballast by ships entering the port. 3 Other architectural elements closely associated with the breakwater system are the remains of five large concrete piers (1-5), a smaller pier (1.5), and a segment of a concrete wall (M) (Map 6; Fig. IV-3). The badly eroded remains of Piers 4 and 5 appear to 3 Excavation beneath the rock mounds revealed no traces of timber, which might have indicated that the mounds had been constructed by sinking boats filled with stone, a method used at the harbor of Ostia for the construction of the concrete island and for certain sections of the left mole. A similar mefhod was used at Centumcellae. For Ostia, see Pliny NH 16.202; Testaguzza, pp. 105-120; O. Testaguzza, "The Port of Rome," Archaeology 17 (1964), pp. 173-179; V. Scrinari, "Strutture portuali relative al 'porto di Claudio' messo in luce durante i lavori dell'Aeroporto Intercontinentale di Fiumicino (Roma)," Rassegna dei lavoripubblici 7 (1960), pp. 173-190; Blackman 1982-2, p. 198.

76

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

rest on top of the breakwater. 4 Pier 4 is located 40 m to the east of the cliffs, nearly centered on the breakwater, while Pier 5 is situated about 15 m from the tip of the breakwater approximately equidistant from its inner and outer faces (Fig. IV-5). The materials used in the construction of these two piers are tuff (from the Volsinian volcanic complex ca. 60-80 km northeast of Cosa) and pozzolana (from the region of Pozzuoli). 5 The same materials are found in Piers 1, 1.5, 2, and 3 and in Wall M just to the north of Pier 1. Since the same north-northwest/southsoutheast axis is repeated in Wall M and Piers 1 through 3, with Pier 4 placed on a parallel axis slightly to the east, it appears that all these concrete structures were designed and built at one time or that they at least formed part of an interrelated plan executed over a period of time. PIERS 1-5

Piers 1 through 3 are relatively well preserved, no doubt owing to their more sheltered position close to the cliffs and shore (Color Fig. 1; Figs. Ill—12, 13; IV-Il, 12, and 13). The widths (east-west) of the piers (not including the small Pier 1.5, Fig. IV-13) are 6.3 m, 6.8 m, and 7.5 m, respectively. Their preserved lengths (north-south) exhibit somewhat greater variation: 12.6 m, 10.5 m, and 6.4 m. The last measurement given, that of Pier 3, does not represent a finished dimension, for that pier is partially immured in a modern concrete drainage structure. Originally, Pier 3 was comparable in size to Piers 1 and 2. The present dimensions of Pier 4—7.5 m (east-west) by ca. 10 m (north-south)—are closest to those of Pier 2, but the badly eroded condition of Pier 4 indicates that it must have been considerably larger in antiquity (Fig. IV-14). The visible portion of the severely damaged Pier 5 is now only ca. 4.8 m (east-west) by ca. 4.3 m (north-south), with an additional section 1.5 m long extending under the sand to the south (Fig. IV-15). Of all the piers, the best preserved is Pier 1.5, located between Piers 1 4 See McCann, chapters three and seven for her suggestion that Pier 5 may have served as the base for a lighthouse. 5 On the identification of the tuff aggregate from Pier 3, see Trigila, chapter sixteen. Professor Mario Fornaseri of the Institute of Geochemistry at the University of Rome examined samples of tuff from the other piers and concluded that all are probably from the Volsinian complex. A comparative analysis of pozzolana mortar from Pier 1 at Cosa and of a pozzolana sample (possibly ancient) taken from the modern pier at Pozzuoli showed the two to be very similar. See D. M. Roy and C. A. Langton, Longevity of Borehole and Shafl Sealing Materials: 2. Characterization of CementBased Ancient Building Materials. A Topical Report for the Office of Nuclear Waste Isolation Report No. ONWI-202, Materials Research Laboratory, The Pennsylvania State University, Univer-

and 2 (Fig. IV-3). It measures 2.5 m (east-west) by 5.8 m (north-south) and appears to stand to its full original height at present mean sea level. The full heights of the other piers are not preserved: Piers 1, 2, and 3 rise to 1.75 m, 2.80 m, and 2.72 m a.s.l., respectively; the top of Pier 4 just breaks the water line, and that of Pier 5 is ca. 2.5 m b.s.l. The bottoms of the piers were not located, even though probes made along the southwest side of Pier 1, the southeast side of Pier 1.5, and the east side of Pier 2 reached depths of 1.9 m, 2.05 m, and 2.05 m, respectively. Probing around Piers 4 and 5 suggested that these rest on the rocks of the breakwater, but the precise depth of their foundations could not be determined because of heavy rocks surrounding them. 6 The remains of the piers span an overall distance of ca. 150 m along a staggered line, with considerable variation in the spacing between them. The intervals range from ca. 5 m between Wall M and Pier 1 and between Piers 1 and 2, to 6.5 m between Piers 2 and 3, ca. 36 m between Piers 3 and 4, and ca. 55 m between Piers 4 and 5. The small Pier 1.5 blocks the space between Piers 1 and 2, but no evidence was found during the excavation of the harbor area to indicate that other intermediate piers once filled the open expanses between the last three piers. It may be that a plan to construct a continuous line of piers was never completed or that additional piers did exist but were destroyed by the same forces of nature that damaged the breakwater. It seems more likely, however, that a continuous line was never intended. 7 The lower portions of Piers 1, 2, and 3 show several grooves, holes, and impressions that clearly were left by the formwork used during construction. In excavating along the western side of Pier 1 to a depth of 0.44 m b.s.l., shallow vertical impressions, varying in width from 0.10 m to 0.15 m and in depth from 0.15 m to 0.20 m, were observed (Fig. IH-13). These were made by rather thin, narrow planks. The intervals between the impressions are of sity Park, October 31, 1980. In oral communication Langton offered the opinion that both samples derive from the same lava flow but are probably from different levels or from different quarries within it. Given the proximity of pozzolana quarries to the harbor at Pozzuoli and the fame of these quarries in antiquity (see, for example, Vitruvius 2.6.1-4 and 5.12.2), it may be assumed that the pozzolana in both samples is from this region. I am most grateful to Nicholas Hartmann, Research Fellow at MASCA, The University Museum of The University of Pennsylvania, for arranging to have these samples of pozzolana cement analyzed. 6 McCann, chapter three. 7 It seems very unlikely that other piers would have disappeared without a trace, as presumed by Brown 1980, p. 60 and fig. 77.

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

77

approximately the same width as the impressions themselves (0.11-0.15 m), indicating that the planks overlapped one another. These vertical planks must have been held together on the exterior of the form by some type of horizontal bracing such as that 8 found at Pyrgi; at Cosa, however, none survives. Yet evidence that there were two horizontal braces on the interior of the forms does remain in Piers 2 and 3. Just above the present sea level two squared holes ca. 4 m apart pass through each of the concrete masses in an east-west direction, each at about 3 m from the northern and southern ends of the piers, marking the positions of large beams that have since rotted away (Figs. IV-Il, 12). The dimensions of the holes (ca. 0.26 m χ 0.25 m) indicate that the beams were slightly less than a Roman foot square (0.296 sq. m) in cross-section. A third hole is now nearly destroyed by the large crack that passes diagonally through Pier 2, dividing the upper portion of the pier in two. Originally the top of the hole was ca. 0.25 m a.s. 1., as one small preserved portion of its upper surface attests, but no traces of the lower and side surfaces survive. Today the depth of the crack varies between 0.56 m and 0.68 m b.s.l. and the width between 0.55 m and 0.58 m. It seems likely that this diagonal hole once held a third horizontal brace for the interior of the formwork of Pier 2. At the southern end of Pier 1 a shallow diagonal impression, ca. 0.6 m wide, may mark the position of a similar internal brace (Fig. IV-13, lower center). In Pier 2, immediately above the level of the beam holes, there are several rounded hollows that have diameters of 0.12-0.14 m (Figs. IV-Il, 12). The ran­ dom arrangement of these hollows and their posi­ tions just above the bracing beams suggest that they were left by timbers used either for the reinforce­ ment of the concrete 9 or for scaffolding or other equipment that collapsed or was tossed into the con­ crete during construction. Such planks, beams, and other construction equipment recall Vitruvius' de­ scription of the type of single walled cofferdam to be built when erecting piers in the sea using concrete containing pozzolana from the region of Pozzuoli. Such a cofferdam should be built "with its sides formed of oaken stakes with ties between them . . . to be driven down into the water and firmly

propped there; then, the lower surface inside, under the water, must be levelled off and dredged, work­ ing from beams laid across; and finally, concrete from the mortar trough . . . must be heaped up until the empty space which was within the cofferdam is 10 filled up by the wall." In the piers at Cosa broken pieces of aggregate rock, varying in length from 0.10 m to 0.30 m, were carefully layered in a large amount of pozzolana mortar, resulting in a very compact mass that has resisted disintegration re­ markably well for more than two thousand years. A distinctly different type of concrete was used to construct the upper parts of Piers 1, 2, and 3; con­ sequently, the seams between the upper and lower portions of these piers are clearly visible (Hg. IV11). The yellow-brown of the tuff in the lower sec­ tions contrasts with the steely gray of the local lime­ stone and the orange of the amphora fragments that constitute the main aggregate materials of the con­ crete of the upper levels. Mixed in with the broken limestone and amphoras are small quantities of sand­ stone and an occasional piece of yellow tuff. In Pier 2 four distinct layers, or pours, are visible: the first, or lowest, measures 0.48 m thick, the second 0.53 m, the third ca. 0.42 m, and the fourth is preserved to a maximum thickness of 0.65 m (Fig. IV-12). In each layer the aggregate rock and sherds are laid in relatively neat rows. Body sherds normally curve downward, and other amphora parts, such as rims, handles, and toes, are placed horizontally. The first pour contains fewer terracotta fragments than the other three, but its mortar is comparable to that of the second and includes small bits of yellow tuff along with lime and the local dark sand. By contrast, the mortar of the third and fourth pours contains only sand and lime. The bits of tuff in the mortar of the first two pours probably derive from pozzolana, from the region around Pozzuoli as indicated from the information reported above in footnote 5. These upper portions of the piers preserve no traces of formwork. It may be that walls of cut lime­ stone were built around the outer sides to act both as formwork for the concrete cores and as protective facings.11 The high proportion of terracotta frag­ ments in this concrete is very likey due both to the availability of large quantities of amphora sherds and

8 J . P. Oleson, "Underwater Survey and Excavation in the Port of Pyrgi (Santa Severa), 1974," JFA 4 (1977), pp. 305-307, fig. 5. 9 See Lewis, "Outer Harbour Notebook, 1968," p. 41. 10 Vitruvius 5.12.3, trans. M. H. Morgan (Cambridge, Mass., 1914). For discussions of the Vitruvian methods of building formwork for underwater construction, see C. Dubois, "Observations sur un passage de Vitruve," MelRom 22 (1902) pp. 439-467, and

Schlager, "Die Texte Vitruvs" (cited in chapter three, n. 37) pp. 150-161. Cf. Oleson, StHarbArch I, and McCann, chapter three, n. 37, above. 11 The use of cut stone facings for concrete structures was com­ mon by the end of the second century B.C., according to Blake 1947, pp. 1-2.

78

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

to the belief that terracotta was well suited for hydraulic construction. 12 Immured in the upper portion of Pier 1 is a Sestius-type rim fragment with a rosette or medallion stamp (an early example of Type 4b; Will, Cat. Al95), and the upper section of Pier 2 contains toes of amphora Types 4a and 4b (uncatalogued). The question naturally arises: Do the two types of construction indicate two distinct phases in the history of these concrete structures, or do they represent two stages in the same building program? The evidence provided by the piers themselves is not conclusive but an argument can be made in favor of the first alternative. Although the seams between the two types of concrete in Piers 1 and 2 are at a comparable height (ca. 1.5 m a.s.l.), in Pier 3 that seam is higher and very irregular. At the northwest corner of Pier 3 the tuff-and-pozzolana concrete ends at 2.1 m a.s.l., but toward the center of the pier it continues upward another 0.42 m. This irregularity suggests that the original part of Pier 3, made of imported materials, was damaged and later repaired with concrete composed of local materials.13 In addition, the manner of laying the amphora fragments in rows within the concrete core recalls a construction technique used elsewhere on the site in a structure of a demonstrably later date than those made of tuff-and-pozzolana concrete.14 The small quantities of pozzolana and tuff in the concrete of the upper layers of the piers may well have been reused from damaged portions of the original structures. WALL M

Wall M, 5.12 m to the north of Pier 1, is aligned on an axis parallel to those of Piers 1-3 and, like the lower portions of the piers, is made of concrete whose aggregate is composed of yellow tuff (Fig. IV-16). Impressions of vertical planks on the southern end suggest that the wall was built within wooden forms similar to those used for the piers (Fig. IV-17). Wall M was followed for a distance of 7.43 m into the embankment of the parking lot, but in Trench N, ca. 10 m to the north, no trace of it was found (Figs. 111-23; IV-I). Presumably its northern end was destroyed by the modern construction that is very much in evidence in the area. For ex12

For example, crushed terracotta was commonly mixed with pozzolana and lime to make waterproof linings for cisterns, such as that of the Spring House cistern at Cosa (Oleson, chapter five). Vitruvius (2.5.1) recommends that ground-up bricks be added to mortars containing river or sea sand and lime in order to obtain a better composition. 13 It should also be noted that some areas of the uppermost preserved surface of Pier 3 are modern. For example, a two-holed

ample, at a point ca. 5.7 m from its southern end, Wall M is intersected by a loosely packed rubble wall or fill of broken limestone. In Probe 4, about 10 m to the northeast of Trench M (Fig. IV-I), a short section of very rough polygonal wall backed by a similar loose limestone fill was located. Although this polygonal wall is parallel to Wall M and to the long axes of the harbor piers, its construction does not appear to be ancient. Moreover, the discovery just to the east of the wall of a modern stone fishing marker inscribed "LIMITE TRA LA PESCA MARITTIMA E FLUVIALE" (Fig. III-2) supports the attribution of the wall to recent times. 15 The top of Wall M lies at ca. 1 m a.s.l. The bottom was not located, but a probe made alongside the wall indicated that it continues downward for at least another 2.06 m. The upper part of the eastern face of the wall has a smooth surface that slopes at an angle of ca. 45°, and just below this angled surface the wall is 2.44 m thick. On top of the tuffconcrete portion of the wall, ca. 0.38 m from its southern end, is a narrow segment of wall made of limestone rubble and sherds set in a sand and lime mortar (Fig. IV-16). This narrow construction, preserved to a height of 0.26 m, was traced for a distance of 1.3 m. The different building materials and method of construction, and the fact that the original upper surface of Wall M was not prepared for a superstructure, lead to the conclusion that this narrow wall is a later addition. Immured in it was another rim fragment of an amphora of an early Type 4b (uncatalogued) impressed with a rosette or medallion stamp. Subsequently, at an undetermined date, the western side of Wall M, including this later upper portion, was chopped away (Fig. IV-17). IfWaIl M did originally extend northward to the cliffs, as is likely, it would have intersected the end of a concrete embankment wall on the eastern side of the Tagliata Canal. Such a connection to the canal would imply that Wall M and harbor Piers 1-3, in at least one phase of their use, functioned in connection with the structures in the lagoon. WALLS A A N D B

Two polygonal walls that define another channel are located to the east of Wall M: Wall A at a disbrick of recent manufacture and two iron clamps are immured, and there are small patches of concrete that differ in appearance from the surrounding limestone and sherd construction. Also see Lewis 1973, p. 235, who suggests that the upper sections of these piers represent later repairs to the original structures. 14 One may compare Wall F of the harborside villa. 15 See McCann, chapter three, for discussion of the post-Roman use of the site.

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

tance of ca. 30 m and Wall B an additional 5 m to the east (Map 6). These walls extend from the pres­ ent beach into the water, where the large stones of their upper courses have tumbled (Fig. IV-18). It is possible nevertheless to trace the lines of the walls from those blocks that still lie in situ in the sand and from those tracts that were excavated farther to the north. Wall A is nearly parallel to Wall M, whereas Wall B angles slightly more toward the north, with the result that the southern ends of the walls con­ verge to form a funnel-shaped channel 4.55 m wide that is narrowest at its mouth to control the influx of seawater. Beginning at the shoreline, Walls A and B were followed northward for distances of ca. 26.5 m and ca. 19 m respectively. The three courses of Wall A that were uncovered in Probe 5 stand to a height of 1.7 m a.s.1., but the bottom of the wall was not reached (Fig. IV-19). Approximately 17 m from the seaward end of Wall A, on its eastern face, there are two square cuts in the second course of boulders, one ca. 0.20 m wide χ 0.24 m high and the other 0.24 m wide χ 0.32 m high. The fact that the top of Wall B has two slots 0.21 m long χ 0.05 m wide at about the same point along the wall as the recesses in Wall A suggests that these cuttings held some sort of apparatus that spanned the constricted channel. Both walls are made exclusively of roughly po­ lygonal limestone boulders, on the average ca. 0.57 m in height and between 0.80 m and 1 m in length. These compare very closely with the boulders of the polygonal walls in the lagoon (Walls P, PW, PE; Figs. IV-55, 64, 65, 78). Like those boulders, the boulders of Walls A and B are dressed only on their exposed surfaces and are laid horizontally without mortar, their long ends tapering into the sandy em­ bankments. These similarities, and the fact that the line of Wall B does not diverge greatly from that of Wall PE in the lagoon, suggest that at least Wall B continued northward, bending its course slightly, to join Wall PE. The northern continuation of Wall A is more difficult to envision, for if it connected with Wall P on the western side of the lagoon, it would have cut off the Tagliata Canal (with which it was contemporary, as will be argued elsewhere). Further excavation would be necessary to locate the northern course and terminus of Wall A. VILLA WALLS

Just to the east of polygonal Wall A and partly obstructing the passageway between it and Wall B is 16

79

the seaward end of a brick and limestone faced con­ crete wall (O) of a Roman villa built later along the shore (Map 6). Additional walls of similar materials and construction probably belonging to the same villa are normally visible at various intervals along the beach for some 125 m to the east (Fig. IV-20). This villa was partially explored in 1923 and 1924, and those portions of its walls that remained exposed have been recorded by scholars at various times since 16 then. In the excavations of 1972 an additional sec­ tion of Wall O and another wall (C) were uncovered in Beach Probes V, VI, and VII (Fig. IV-I), but no attempt could be made to explore or study the villa systematically since this area lay outside the limits of the excavation permit. Nevertheless, the materials and methods employed in the construction of the villa walls provide some useful evidence for distin­ guishing the phases of architectural development at the harbor and lagoon sites. The observations re­ corded below are limited to those parts of the villa that were visible during 1972, either excavated or exposed by the elements. These include two groups of parallel walls on the beach, one to the west and the other to the east of the Torre della Tagliata, and a cistern immediately to the west of that tower. The five long concrete walls to the west of the Torre della Tagliata run parallel to one another and, on the whole, are similar in materials and techniques of construction (Map 6; Figs. IV-18, 20). As noted above, the westernmost of these walls (Wall O) lies between polygonal Walls A and B, blocking the channel that they once defined (Fig. IV-21). Portions of Wall O were exposed in two probes (nos. V and VII) in 1972. In Probe 7, the more northern of the two, the wall is ca. 0.50 m thick, but its breadth increases to 0.65 m toward its southern end, which protrudes from the sandy bank of the parking lot onto the beach. The northern end of this wall was not located, but the exposed portions span a total distance of ca. 23 m from Probe 7 to the southern extremity where the wall has been eroded by the sea. Here a squared pilaster is visible on the western side of the wall. A similar pilaster was revealed in Probe 5 (Fig. IV-21). The latter has a facing of five or six bricks surmounted by several courses of small quadratic limestone blocks; farther to the south the facing of the wall itself consists of about six courses of trian­ gular bricks (0.27 m long, 0.03-0.035 m thick) alter­ nating with several courses of limestone tesserae (0.11-0.12 m on a side). These tesserae were evi­ dently intended to be reticulate in form but, owing

McCann, chapter three. See also P. Raveggi's report on the finds made in 1923 and 1924 in NSC 3 (1927) pp. 204-210.

80

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

to the hardness of the limestone, are rather irregular. The full height of this roughly reticulate limestone course is not preserved. Although the next three walls in this series (Walls C, D, and E) were not excavated in 1972, they were exposed for several meters from the sand embank­ ment at the edge of the modern parking lot down to the sea (Fig. IV-18). Near the sand embankment the heavier foundations of some of the walls support su­ perstructures with brick and limestone facings, some preserved to a height of ca. 1 m above the tops of the foundations. The facing of the upper portion of Wall E at its northern end is very similar to that of Wall O. It preserves six courses of roughly reticulate limestone tesserae and five courses of brick. The concrete cores of Walls O, C, D, and E are com­ posed of closely packed, small pieces of broken lime­ stone (0.10 m in length on the average) and a small quantity of broken terracotta set in a sandy mortar that varies in composition (Fig. IV-22). Wall F is dis­ tinctive in that it has several clearly discernible courses of brick and amphora fragments on top of the limestone aggregate. This method of laying the aggregate bears a general resemblance to that fol­ lowed in the construction of the upper portions of the harbor piers and suggests that those piers were refurbished in connection with the villa, perhaps as part of a renovation of the port area for use by the occupants of the sumptuous seaside residence. Another structure, probably to be associated with the western group of villa walls, is a large concrete cistern located immediately to the west of the Torre della Tagliata (Fig. IV-23). The aggregate of the cis­ tern vault is composed of slabs of porous sandstone mixed with other local stones and broken bricks or tiles (ca. 0.4 m long χ 0.04 m thick). This aggregate is set in a sandy mortar containing a small amount of finely crushed terracotta. Except for one small portion of a concrete vault within the Tagliata, the construction of the cistern is unique among the ex­ posed structures at the harbor and lagoon sites. The group of walls to the east of the Torre della Tagliata differs from that to the west in size and ori­ entation as well as in materials and methods of con­ struction. In contrast to the faced concrete of the western walls, the eastern walls appear to be of a type of pseudo-concrete construction, comparable to much of the rubblework masonry found in this re­ gion of Italy. Limestone fragments and a few am­ phora sherds set in a coarse, sandy mortar form the 17 In addition to the author's own observations made in 1972 and 1974 and those recorded in Lewis's notebook of 1969, infor­ mation has been drawn from the 1972 notebook of V. J. Bruno,

cores of the walls, and these are faced with irregular pieces of limestone that are ca. 0.2-0.3 m in length, roughly dressed on their outer surfaces (Fig. IV-24). In one area of the structure, however, just to the west of the restaurant La Strega, the facing of two walls resembles that of the western group of villa walls (Fig. IV-25). The irregularly cut reticulate limestone facing of the lower level of the walls is crowned by an offset course composed of roughly quadrilateral limestone blocks that measure on the average 0.20 m in length and 0.10 m in thickness. Above the offset the walls are faced with a type of opus mixtum. Single quadrilateral limestone blocks alternate with double courses of brick set in thick layers of sandy mortar. As will be seen below, the villa walls provide a gauge for attributing numerous structures in the port and lagoon areas to the Impe­ rial period. WALLS ONE KILOMETER EAST OF THE PROMONTORY OF COSA

Approximately 1 km to the east of the promon­ tory of Cosa two long parallel walls traverse the beach from the dunes to the sea (Fig. IV-26). These may be observed on land for a distance of more than 25 m and in the water for an additional 31.5 m. Be­ cause these walls have not been excavated, the fol­ lowing observations are based on the portions that were visible above the sand in 1969, 1972, and 1974.17 The portions of the walls that traverse the beach are ca. 9 m apart. Their -heavy foundations are ca. 1.1 m a.s.l. and are 2.2 m wide at the seaward end, but they become narrower toward the dunes. For example, at 6 m inland from the sea, they are only 1.75 m wide. These foundations support narrower upper walls (0.40 m wide), which are preserved be­ ginning at a point ca. 9.2 m from the seaward ends of the walls. The foundations are of concrete con­ taining an aggregate of limestone and other local rocks set in a mortar of lime and sand, some of which is coarse and gravelly. The upper walls con­ tain amphora fragments in addition to the stone ag­ gregate and have a sandy mortar of a fine texture. These beach walls are continued on line into the water by walls constructed of tuff-and-pozzolana concrete identical to that found in a number of struc­ tures in the harbor and lagoon (Fig. IV-27). All of the walls in the sea are completely submerged. Both the eastern and western walls are damaged, but they who made a special study of both these walls and the polygonal wall at the outlet of the Lago di Burano.

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

have an average preserved width toward the south of 2.8 m. Between the southern ends of the walls are two low rectangular piers composed of the same type of tuff-and-pozzolana concrete. The pier on the east is ca. 1.5 m sq.; the one on the west measures ca. 2.0 m (east-west) χ 1.5 m (north-south). About 10 m farther to the south, opposite the western wall, is a larger rectangular concrete mass that measures 6.3 m (north-south) χ 5.3 m (east-west). Running along the interior face of the western wall is a chan­ nel 0.50 m wide and 0.45 m deep, which is pre­ served for a distance of 9.9 m. The internal wall of the channel is 0.32 m thick. A number of breaks through the western wall may be eroded remains of east-west channels. The eastern wall has a series of five such channels that run across it in an east-west direction at varying intervals. POLYGONAL WALL AT THE OUTLET OF LAGO DI BURANO

Just to the southeast of the Torre di Burano at the present outlet of the vestigal lagoon of Cosa (Fig. IV-28), a submerged wall lies near a modern pump that is used to bring water from the sea into the la­ goon whenever the natural outlet is blocked by sand (Fig. IV-29). 18 The course bearing of the wall is ca. 165° to 170°, an angle similar to that of the channel walls a kilometer east of the Tagliata. The wall is composed of large, roughly shaped limestone blocks that measure ca. 1.00 m χ 0.80 m, similar in form to polygonal Walls PW, P, and PE in the lagoon and to Walls A and B in the port area. It is not certain whether the interstices of the submerged wall are filled with mortar or with a natural concretion of sand and lime like that found in the breakwater. Some amphora fragments are scattered among the stones, but these do not appear to be mortared into the wall fabric. Some loose blocks of tuff were ob­ served along with some bricks (0.30 m long χ 0.05 m thick). This wall is preserved to a width of ca. 2 m and a length of 10 m to 15 m.

Sequence of Construction of the Port and Coastal Structures It is clear from the extant remains described above that there were several stages in the construction of 18 Since this wall was never visible above the sand during my explorations of the site, the measurements and description given here have been drawn from Bruno's notebook. 19 Compare McCann's discussion of the chronology of the site in chapter eighteen. She suggests that the harbor piers probably preceded the construction of Wall M and the concrete fishery in-

81

the harbor at Cosa. The evidence of stratigraphy and construction show that the breakwater and its rockmound extensions (A-F) were built at the same time, while probes around the bases of Piers 4 and 5 in­ dicate that in all likelihood these piers were built subsequently, their foundations resting on the rocks of the breakwater (Fig. IV-14). Two factors suggest that this subsequent stage included all of the harbor piers and probably Wall M as well. 19 Piers 4 and 5 are identical in fabric to the lower tuff-and-pozzo­ lana concrete portions of Piers 1-3 and Wall M, and the axis of orientation of Pier 4 is parallel to those of the latter structures. The identical tuff-and-poz­ zolana construction of the seaward ends of the walls a kilometer to the east of the promontory suggests that these walls, too, belong to the same building phase. 20 The actual length of time that separated this phase of construction from that of the breakwater and the duration of time involved in building all these extensive concrete structures can be surmised only on the basis of the other architectural evidence and the finds that will be discussed in chapters seven and eighteen. Wall M and Piers 1-3 were probably refurbished in at least one subsequent phase of building. Not only do the materials and techniques used for the upper portions of these structures differ from those of the lower portions, but also the irregular juncture between the upper and lower sections of Pier 3 very likely indicates that the original structure had been damaged before the upper part was added. The fact that these additions resemble some of the villa walls (especially Wall F) in materials and techniques of construction suggests that the building of the villa occasioned the refurbishment of the existing port structures. By that time, presumably, the channel formed by Walls A and B was no longer functional, for it was blocked by one of the villa walls (O). These sequences, in turn, may indicate the following stages of construction: 1. The breakwater and its extensions (A-F). 2. The lower tuff-and-pozzolana concrete por­ tions of the harbor piers, Wall M, and the sea­ ward ends of the walls located a kilometer to the east of the promontory. 3. The upper portions of the piers and of Wall M, the western group of villa walls, the villa cisstallations in the sequence of construction within this second building phase of the site. 20 For the further use of tuff-and-pozzolana concrete at Cosa, see Gazda and McCann, chapter seven, and the second building phase of both the port and fishery.

82

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

tern, and probably also the eastern group of walls belonging to the villa. The place in this sequence of poygonal Walls A and B and of the submerged polygonal wall off Torre di Burano cannot be determined solely on the basis of materials and techniques of construction. It is clear only that Walls A and B preceded those of the villa. Evidence of comparable polygonal struc­ tures in the lagoonal area and their place in the ar­ chitectural development of the site will be discussed elsewhere. 21

THE CLIFF CHANNELS AND QUARRY

Perhaps the most striking feature of the landscape around the harbor today is a picturesque cleft through the edge of the promontory that constitutes the western boundary of the port (Map 6; Fig. 1-4). This cleft, known as the Tagliata, forms the seaward end of a narrow rock-cut sluiceway that runs from the lagoon southward along the base of the cliffs. Along the course of the sluiceway there is a deep trapezoidal recess in the face of the cliff, from which rock for many of the ancient structures in the vicin­ ity must have been quarried. In the southwestern corner of the quarry is the inconspicuous entrance to the Spacco della Regina, a long, dramatic crevice through the promontory that ends at the sea. A third cleft, known as the Piccolo Spacco, extends from the northern portion of the Tagliata channel to the sea, running between and nearly parallel to the Spacco della Regina and the southern seaward section of the Tagliata. AU three of these clefts are natural fractures related to Neocene-age rotation and uplift of the Ital­ ian Peninsula. 22 They were subsequently widened by the action of water dissolving the limestone, and parts of them were reshaped by the hand of man. The extant remains of the man-made alterations are described below.

The Tagliata The Tagliata proper consists of two high-walled, open-air passages (here referred to as the Tagliata South and North) that are connected by a low tunnel (Fig. IV-30). A second, narrower tunnel, known as the Tagliata Piccola, joins the Tagliata South toward its seaward end, then branches diagonally off from it and curves toward the southwest. Each of these 21 See Gazda and McCann, chapter seven, reconstruction of the fishery.

segments was hand scarped, following the lines of fractures in the rock. The Tagliata South is 60 m long and varies between 4 and 5 m in width (Figs. IV-31, 32, and 33). Its eastern wall is preserved to an average height of ca. 5 m a.s.l.; the western one rises to heights that vary with the contours of the cliff from ca. 6 m to 10 m a.s.l. The connecting tun­ nel measures 17 m in length and is ca. 2.5 m wide, and its ceiling is ca. 1.8 m a.s.l. The Tagliata North is approximately the same width as the tunnel, ca. 2.6 m. Approximately 20 m to the north of the tun­ nel the high walls of the Tagliata North (on the northeast ca. 7 m a.s.l.; on the southwest ca. 10 m a.s.l.) drop away sharply where an ancient quarry cuts deeply into the cliff. From this point northward to the lagoon the narrow sluiceway is completely artificial on its eastern side and thus will be distin­ guished from the other sections of the Tagliata by referring to it as the Tagliata Canal. Today this canal carries the overflow from the Lago di Burano to the sea. Each section of the Tagliata waterway preserves numerous details of construction that will be de­ scribed in sequence from south to north. TAGLIATA SOUTH AND THE TAGLIATA PICCOLA

The southern end of the main channel, which must have originally been closed to the sea at its northeast end, contains traces of rock-cut as well as concrete construction (Fig. IV-31). Although most of the features are badly eroded, a considerable amount of evidence of their original architectural character still remains. The seaward mouth of the Tagliata is cut through a wall of rock that forms a barrier between the channel and the open sea (Figs. IV-32, 33, 34 and 35). The opening is in the form of a short tunnel with a domical ceiling. On the side facing the sea, the tunnel entrance is 3 m wide and very low, having a maximum height of only 0.50 m a.s.l. Whether a vertical slot (0.40 m wide χ 0.75 m high) on the western side of this seaward face was caused by erosion or whether it served a specific purpose is uncertain. Conceivably, it represents an abortive effort to reshape the seaward opening. Within the outlet tunnel the domical space is ca. 5 m in diameter and rises to a height of ca. 2 m a.s.l. From its ceiling, an air shaft 5 m high, measuring 1.8 m (east-west) by 0.9 m (north-south), extends upward through the rock (Figs. IV-31, 33). The floor inside the tunnel is only ca. 1.6 m b.s.l., but that of the sea immediately to the south drops to 22

See Bourgeois, chapter two.

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

more than 3.0 m (Fig. IV-32). On the floor of the tunnel is a large boulder that obstructs much of the opening. Today, as undoubtedly also in antiquity, this boulder serves to break the force of waves crash­ ing through the domed space. Rapidly entering water splashes first against the boulder, then against the north wall of the domical ceiling and up the air shaft, which functions as a blow hole. Three of the interior walls of the dome—on the south, east, and west—are lined with concrete containing broken limestone, sherds, and some yellow tuff set in pozzolana mortar. The north wall, which absorbs most of the impact of the waves, is made of a type of concrete composed of thin slabs of local sandstone cemented in courses (Fig. IV-36). This portion of the domical structure projects ca. 1 m northward into the Tagliata South, where it merges with a large mass of concrete that adheres to the three interior channel walls. (Fig. IV-35) This concrete mass ap­ parently formed a vault that reinforced the northern wall of the domical tunnel. The remains of the vault, preserved to a maximum height of 2.5 m a.s.l. on the southern wall of the channel, taper downward along the western and eastern sides of the channel as they continue toward the north. Erosion obscures the shape of the vault, but cut recesses in the cliff walls, which appear to have acted as brackets for the structure, provide some notion of the original form. On the south wall the cutting follows a fairly straight line just above the irregularly preserved con­ crete mass, but on the western wall the line of the cut recess arcs downward for a distance of almost 3 m toward the north, outlining a broad low vault. The vault was made of a concrete whose aggre­ gate consists of fragments of limestone, tiles, and amphoras along with scattered pieces of tuff. In the lower portion of the vault, close to the present water line, the aggregate contains a higher percentage of terracotta materials than that of the rest of the struc­ ture. In this lower level of concrete, near the outlet, two pairs of rectangular tile-lined slots appear, one pair on either side of the channel (Fig. IV-37). AU four slots are 0.16 m high, but they vary in width from 0.31 to 0.61 m, and two of them have holes inside measuring 0.10 m sq. The function of these slots is not clear, but their spacing and staggered alignment across the channel suggest that they held planks, perhaps for scaffolding or formwork used in constructing the upper levels of the vault. The lower levels of the vault merge with the con­ crete of what now appear to be ledges along the sides of the channel, the one on the east preserved to a maximum width of 1 m and that on the west to

83

0.6 m. Above these ledges, cut into each of the rock walls of the channel, is a series of recesses. On the western wall three of the recesses are in the form of shallow rectangles (Fig. IV-38). The respective heights, widths, and depths of these recesses, meas­ uring from south to north, are: 0.38 m χ 0.32 m χ 0.15 m; 0.36 m χ 0.29 m χ 0.15 m; and 0.41 m χ 0.29 m χ 0.18 m. The two shallower recesses toward the south are at approximately the same level (1.8 m a.s.l.) and are placed ca. 0.45 m apart, while the third recess is about 0.28 m higher and 0.26 m far­ ther to the north. The latter is situated adjacent to a long vertical slot that extends from a height of 2.6 m a.s.l. to an undetermined point below the water line. The upper part of this slot is narrow and shal­ low (0.12 m wide and deep), but a meter below the top, the slot widens to 0.39 m and deepens to 0.24 m. Directly across the channel from these cuttings are additional recesses that correspond in position to those on the western side but differ in form. The two cuts on the south are in the form of shallow notches in the top of a ledge 0.45 m wide that runs along the eastern wall at 1.95 m a.s.l. These notches penetrate 0.35 m into the rock, but their vertical sides are only 0.20 m high, and they are open at the top. The width of the southernmost notch is 0.25 m and of the other 0.42 m. Like the corresponding re­ cesses on the opposite wall, they are 0.45 m apart. Just to the north of these two notches on the east wall is a T-shaped cut 0.75 m wide at the top of a long slot (0.20 m wide χ 0.18 m deep) that continues into the concrete below. This T-shaped cut corre­ sponds in position to the long slot and adjacent re­ cess in the west wall. The form of the slot and of its mate across the channel suggests that provision had been made for a sliding gate or other mechanism at this point along the channel. The two pairs of rec­ tangular recesses immediately to the south evidently held beams, each about a Roman foot square (0.296 sq. m), that spanned the waterway. Immediately to the north of the long vertical slots are two large concave recesses (ca. 1.4 m wide) to which portions of an arched concrete bridge still ad­ here (Fig. IV-38). The tops of these recesses are at ca. 1.70-1.80 m a.s.l., approximately the same level as the tops of the beam holes on the western wall. Presumably, this was also the height of the bridge, which evidently provided a platform across the channel from which to raise and lower the mecha­ nism held in the adjacent slots. The concrete core of this bridge is composed solely of pozzolana mortar and yellow tuff aggregate, some pieces of which are quasi-reticulate in shape, but the voussoirs and some

84

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

of the facing blocks were cut from local limestone. Concrete continuing from the ledges of the vault to the south of the bridge partially lines the rock-cut slots, then passes beneath the bridge and extends northward for 0.34 m on the east side and 1.10 m on the west. Clearly, these ledges were constructed some time after the bridge, perhaps as reinforcement for the latter. Approximately 1.5 m to the north of the bridge, on the western side of the channel, the small tun­ neled outlet of the Tagliata Piccola meets the main channel (Figs. IV-31, 35). Roughly 0.20-0.30 m in­ side this tunnel are two cuttings in the southern side wall, one below the other, apparently intended to hold some sort of apparatus at the mouth. From the main channel the tunnel runs westward for ca. 3 m then turns toward the south for a distance of ca. 6 m to the sea (Fig. IV-39). The seaward mouth of the Tagliata Piccola is thus angled away from the main wave direction coming from the southwest and empties into a protected cove. Its oblique angle to the waves contrasts with the seaward mouth of the Tagliata proper, which faces into the waves. At the bend of the tunnel of the Tagliata Piccola are remains overhead of a concrete arch made of the yellow tuff, limestone, and sherds set in pozzolana mortar (Fig. IV-40). The tunnel itself is 1.8 m wide, and its ceil­ ing is ca. 1.8 m a.s.l., but the depth varies consid­ erably. Most of the tunnel floor along the eastern side is a rock-cut shelf that lies at 1.4 m b.s.l., but at the seaward end, along the western side, a narrow section of the floor descends to 2.6 m b.s.l., approx­ imately one meter lower than the seaward mouth of the Tagliata South. This change in level in the Ta­ gliata Piccola allows for an outward flow of water from the main channel of the Tagliata South. At the seaward opening of the Tagliata Piccola the tunnel has collapsed for a distance of some 3 m, but a num­ ber of details of its original form may still be dis­ cerned from the remaining rock. Above the opening a rectangular air shaft nearly identical in dimensions to the one above the domical opening in the mouth of the Tagliata South was cut through 7 m of rock (Fig. IV-41). Presumably it, like its companion ca. 3 m to the east, helped to break the force of entering waves. Within the Tagliata South, 8.4 m to the north of its juncture with the Tagliata Piccola, there are more rock-cut recesses and a pair of submerged rectangu­ lar platforms made of concrete (Fig. IV-31). The up­ per surfaces of the platforms now lie at sea level, ca. 0.7-0.9 m above the present floor of the channel. The platform on the east side is ca. 1.2 m wide χ 2.5

m long, while that on the west is only ca. 1 m wide χ ca. 2 m long. These platforms are located imme­ diately to the north of several cuttings in the channel walls, which may have held some device related in function to the sliding gatelike apparatus adjacent to the concrete bridge. The form of the device next to the platforms is difficult to envision owing to the poor state of preservation of the cuttings and to the presence of naturally formed pits in the brecciated limestone wall. However, one cutting near the plat­ form on the east wall is quite clear. It is 0.22 m wide, 0.53 m long, and 0.21 m deep, and its top is situated at 0.95 m a.s.l. Although this location is about one meter lower than the slots next to the bridge, it is still high enough to have held a gate or other mech­ anism that could be raised above the ancient water level. Thus one can imagine an enclosed tank for fish within the Tagliata South that was provided with continuous water circulation. Under normal condi­ tions water flowing in would have come mainly from the mouth of the Tagliata South and exited through the Tagliata Piccola (Fig. IV-31). The eastern wall of the Tagliata South, to either side of the submerged concrete platforms, is pierced by three diagonal openings (Figs. IV-31, 33). The first lies just to the south of the platforms; the sec­ ond, which is visible only under water, is located 10 m to the north of the first; and the third is 25 m north of the second. These openings may have served to increase the circulation of sea water in the channel north of the concrete bridge. A fourth open­ ing in the eastern wall, at the extreme northern end of this southern section of the Tagliata, appears to have resulted from a later collapse of the natural cliff wall. All along the top of the eastern side of the Ta­ gliata a low limestone rubblework wall exists, and portions of narrow, irregular ledges are preserved at various intervals and levels cut into the interior face of the rock wall below. The rubble wall has no close counterparts at the harbor and lagoon site and may be of a relatively recent date. The rock-cut ledges, however, probably gave access to the various ancient structures within the channels. TAGLIATA NORTH

The southern open-air section of the channel (the Tagliata South) is joined to the northern open-air section (the Tagliata North) by a low tunnel that leads northwestward through the cliff (Fig. IV-30). On the eastern wall of this connecting tunnel, near its northern end, some cuts indicate that another tunnel running toward the northeast had been planned but was abandoned. Beyond the tunnel, in

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

the northern open-air channel (Fig. HI—5), several de­ tails of cuts in the rock are preserved. Along the eastern side at a height of ca. 4.5 m a.s.l. is a roughly cut ledge ca. 0.5 m wide. On the southwestern side of the channel, near the water line and 2.7 m from the tunnel entrance, is the beginning of the Piccolo Spacco cleft, which will be described below. Ap­ proximately 8.4 m to the north of this point are two additional cuttings in the rock of the Tagliata walls. These recesses, which face each other across the channel, may have held another gate or barrier of some kind (Fig. IV-42). The tops of the recesses are at the present water line, and both vary in width from 0.12 m to 0.17 m. The depth of the eastern recess is 0.06 m, compared with the 0.09 m depth of the one on the west; the former continues below the water for 0.4 m, while its western mate reaches only 0.3 m below the water. At 7.5 m to the north of this set of recesses the high walls of the Tagliata North end abruptly at the quarry. From this point north­ ward the high-walled channel takes the form of a shallow artificial canal leading northward into the ancient lagoon (Fig. III-6). TAGLIATA CANAL

The jagged course of the Tagliata Canal between the clefts to the south and the lagoon to the north can be followed along the cliffs for a distance of more than 140 m (Map 6; Fig. IV-30). Because this ancient waterway has been reused as part of the modern drainage system, the precise location of its original northern terminus is obscure. The average width of the canal is 3.35 m, and the average depth of its floor (cut out of the bedrock) is 1.3 m b.s.l. A section drawing of the canal made at about 25 m north of the quarry at point Sl on Map 6 shows a typical profile (Text Fig. II-5). The eastern embank­ ment has three ledges: the lowest, located ca. 0.7 m above the canal bottom, is 0.7 m wide; the second, 0.5 m above the first, is 0.5 m wide; and the third, or uppermost, 0.4 m above the second ledge, is also 0.5 m wide. For most of the observable length, the ledges of the eastern embankment are cut out of bed­ rock, but at a point ca. 36 m north of the quarry, the bedrock apparently drops away. From there northward for a distance of 14 m, the upper ledge is constructed of concrete composed of tuff aggregate set in a mortar of pozzolana, sand, and lime (Fig. 111-24). At ca. 30 m to the north of the quarry are two recesses cut into the cliff wall: one of these (0.12 m wide χ 0.11 m high χ 0.07 m deep) is located at 2.5 m a.s.l.; the other (0.07 m wide χ 0.09 m high χ 0.06 m deep) is at 1.3 m a.s.l. Farther to the north,

85

ca. 50 m from the quarry, near the point where the concrete canal wall appears to end, there are two more recesses: one (0.08 m sq. χ 0.06 m deep) is at 1.5 m a.s.l., and the other (0.4 m wide χ 0.2 m high χ 0.1 m deep) is at 0.3 m a.s.l. (Fig. IV-43). Like the cuttings in the walls of the Tagliata South and North, these too must have been intended for some sort of mechanism. Beyond this point, it is difficult to determine where the ancient canal ends and the modern system begins. The present waterway bends slightly eastward and then toward the north, follow­ ing a somewhat northeastward course as it passes along the eastern embankment wall of the ancient lagoon (Wall PE). At grid point 20N/820E (Maps 6 and 7) the modern waterway angles sharply east­ ward and continues on a straight line for about three kilometers to the Lago di Burano, the only remain­ ing vestige of the old Cosa lagoon. The Piccolo Spacco The northern terminus of the Piccolo Spacco is cut into the southwestern wall of the Tagliata North de­ scribed above (Fig. IV-30). Here the cliff face was hollowed out for a depth of 1.7 m in the shape of a rounded arch (1.65 m wide and ca. 1.80 m a.s.l.) and then was abandoned. The arched contour of the cut indicates that a tunnel was intended (Fig. IV-44). Two ledges, each ca. 0.55 m a.s.l., flank the lower portion of the unfinished tunnel, separated by a cut 0.7 m wide that extends below the present water line (Fig. IV-45). The function of these ledges is uncer­ tain; perhaps they were used as platforms for the workmen as they cut through the rock. The natural fissure of the Piccolo Spacco (best observed in aerial photographs [Fig. 1-4]) runs southwestward from this point for a distance of ca. 140 m to the sea. At its seaward end is a natural tunnel that can be traced for ca. 20 m from the sea into the cliff, but there it is blocked by fallen rock beneath a modern villa. The Spacco della Regina The Spacco della Regina, the third and largest of the natural fractures to be considered, runs across the southeastern point of the promontory for ca. 260 m along a northeast-southwest line from the quarry to the sea (Map 6; Figs. 1-4; IV-46). The Spacco varies in width from 1 m to 6 m, excluding the intersecting wider caves, and has a maximum depth of 30 m to 35 m. The sides of this natural fracture have been scarped along various sections to remove splintering and overhanging limestone (Fig. IV-47). About 50

86

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

m from its seaward end the sea has breached its wall. Slightly farther to the southwest its course is contin­ ued by a natural tunnel for about another 18 m. This tunnel joins at right angles another natural cleft run­ ning northeast-southwest across the southernmost point of the Cosa promontory (Figs. IV-48). At their intersection a rectangular skylight measuring 2.10 m χ 1.10 m was cut into the rock. This northeastsouthwest channel, open to the sea at both ends, is about 45 m long and between 2 m and 3 m wide (Fig. IV-49). Approximately 7 m to the east of the skylight a third natural fracture, ca. 2.5 m wide, leads southwest for about 17 m and exits into the sea (Figs. IV-50, 51). Although the seaward mouth of this third channel is a natural one, the tunnel con­ necting it with the long northeast-southwest channel appears to have been artificially cut. The seaward entrances of the long northeast-southwest channel both show signs of man-made cuttings underwater. The southwestern entrance is in about 5 m of water, and the sea floor around it is covered with boulders and sand for a distance of about 50 m from the shore (Fig. IV-52). The entrance itself is not cut com­ pletely down to the seabed but ends on a cut ledge some 2 m above the sea floor. The northeastern en­ trance is not as deep. It rises only about 3 m off the bottom, but it has a similar ledge cut into the rock. Above this ledge is an additional hand-cut rock shelf, the top of which is now at sea level (Fig. IV53). This raised shelf across the entrance probably reduced the flow of sand into the channels and also obstructed the reverse flow of water out of the chan­ nel, thus acting as a natural sluice gate. In this way, a calm basin of water could be maintained within the long channel while also allowing for circulation. The system of triple entrances at this seaward mouth of the Spacco further provided for the circulation of sea water into the long channel from whatever direction the swells and winds were running. 2 3 At the northern end of the Spacco the configura­ tion of the fracture is altogether different. From the quarry (Fig. IV-46), one enters on dry ground into a narrow passageway illuminated by a hand-cut sky­ light (Fig. III-8). About 25 m down the passageway is a large cave, ca. 20 m long, 8 m wide, and 30-35 m high. The cave is lit by a skylight at its southern end (Fig. III-7), and along the eastern side is an arte­ sian freshwater spring situated ca. 1 m a.s.l. Today the spring's waters are piped through the skylight to 23 The observations recorded above are those of McCann and Lewis. For further discussion of the Spacco della Regina, see Gazda and McCann, chapter seven, and the first building phase of the fishery.

a villa on top of the cliffs. A further examination of the Spacco by geologist J. de Boer yielded sufficient evidence to suggest that the waters of this spring may have been lifted to the top of the cliff in antiq­ uity as well. 24 De Boer noted that the upper part of the cave had been opened up and widened and that ca. 3 m from the top there were remains of a rockcut platform, ca. 1 m wide, as well as cuttings for a staircase leading down from the skylight. He further observed that the spring, which is very powerful, could have filled the floor of the cave to a depth of at least one meter. This spring empties naturally along the course of the Spacco in the direction of the sea, but de Boer found no marine borings or bar­ nacles on the walls of the Spacco to indicate that sea water had penetrated into the fracture as far as the cave in antiquity. Rather, fresh water from the nat­ ural cistern at the northern end must have flowed outward toward the sea into the complex of tunnels at the seaward end of the Spacco described above.

The Quarry The quarry, located at the very northern end of the Spacco della Regina and to the west of the Tagliata Canal, covers an area of ca. 390 sq. m. (Map 6; Fig. IV-30; Color Fig. 1). Its sides, which form an irregular trapezoid, measure roughly 35 m on the northwest, 12 m on the southeast, 26 m on the southwest, and 40 m on the west. This quarry un­ doubtedly supplied most of the rock for the break­ water and other limestone structures in the imme­ diate vicinity of the port. Except for a small portion of the bedrock wall shared by the Tagliata Canal and the quarry, the juncture of the two units is obscured by a recent concrete wall associated with the current drainage system. However, it is clear from the pres­ ent height of the quarry floor, roughly 1.5-2.0 m a.s.l., and the evidence of a continuing line of bed­ rock beneath the modern concrete wall that the canal and quarry were distinct and separate in antiquity and that the quarry floor was dry. At the north, where the quarry meets the high scarped wall of the cliff along the western side of the Tagliata Canal, there are incised outlines of several polygonal blocks (Fig. IV-54) resembling those used in the fortification wall of the town begun shortly after the foundation of the colony in 273 B.C. and in a number of other structures dating between that 24 I am grateful to de Boer for permitting me to cite his unpub­ lished observations. See his comments on sources of water at the site in chapter sixteen.

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

time and the middle of the following century. 25 The fact that these outlines were cut into the neatly dressed canal wall suggests that at least this portion of the Tagliata was already finished when blocks of this type were being quarried, conceivably as early as the second quarter of third century. From his ex­ tensive study of the area de Boer concluded that a number of the blocks used in the fortification wall came from this harborside quarry, where the quality of rock is finer than that available from the hill sources along the perimeter of the wall itself.26 But there is no way to prove that the blocks that remain in situ along the Tagliata Canal were among those intended for the town wall. In fact, they are rather smaller in comparison with most blocks of the fortification. They could just as well have been in­ tended for a later structure—for example, the po­ dium of the port temple erected ca. 170 B.C., to name the only building in the immediate vicinity of the harbor in which the use of such polygonal blocks is attested. 27 At most, this evidence provides a ter­ minus ante quern of the second quarter to mid-second century B.C., for after that time such finely cut po­ lygonal blocks apparently ceased to be used. Sequence of Construction of the Cliff Channels The stages of the cutting of the cliff channels may be surmised from a few points along the Tagliata where sequences of construction are evident. The southern section of the main Tagliata channel (the Tagliata South) appears to have been completed first, together with the small tunnel of the Tagliata Piccola, which functioned as a unit with it. Like­ wise, it is clear that the high walls of the northern segment of the Tagliata channel (the Tagliata North) were scarped before the northern end of the intended Piccolo Spacco tunnel was cut into its southwestern wall. The place in this sequence of the isolated, in­ tersecting tunnels at the seaward end of the Spacco della Regina is more difficult to determine. The tun­ nels themselves appear to be natural for the most part, yet they share certain dimensions with the Ta­ gliata Piccola and the unfinished northern end of the Piccolo Spacco (all are ca. 1.8-2.0 m wide), and the lateral dimensions of the hand-cut skylight above the intersecting tunnels of the Spacco della Regina (2.1 25 On the date of the town wall, see Brown 1951, p. 57, and Brown 1980, p. 21. Other structures in which such polygonal blocks were used include the terrace of the Capitolium and the podia of Temples B, C, and D, the terrace wall, the horrea, and the retaining walls of three large reservoirs.

87

m x 1.1 m) are roughly comparable to those of the air shafts at the outlets of the Tagliata Piccola and the Tagliata South (1.8 m χ 0.9 m). The rock-cut ledges at the ends of the intersecting Spacco tunnels may perhaps be compared in function with the boul­ der at the outlet of the Tagliata South. But this ar­ chitectural evidence alone does not prove that all of the rock-cut channels were contemporary. The whole of the Tagliata waterway could have preceded the other cliff channels by a considerable amount of time. It is equally difficult to determine when in the se­ quence the cutting of the platform and stairs ob­ served by de Boer in the Spacco cave took place. Originally, access to the cave must have been along the fracture line that now forms the northwestern wall of the quarry. As quarriers penetrated more deeply into the cliff, removing part of the fracture, the Spacco cave would have become more accessi­ ble. It seems very likely that the maximum use of the artesian spring within the cave would have been made after the cave had become easier to reach. If this were the case, then these rock-cut features within the cave would most logically be attributed to some time after the quarry was opened. They could even belong in the medieval period, when the cave was apparently used as a hiding place. Such a sequence, in turn, would imply that the cuttings in the Spacco cave were later than the Tagliata since at least that portion of the Tagliata which passed along the eastern edge of the quarry was probably finished before the quarry was opened, for outlines of several polygonal blocks were incised sometime after the canal wall had been cut smooth (Fig. IV-54). If these unquarried blocks can be taken as indication that the canal was already completed when quarrying began, then the Tagliata emerges as the earliest of the rockcut structures. The amount of time that may have elapsed between all these stages will be considered in a later chapter. Parts of the Tagliata underwent modification at least twice after the initial cutting of the entire waterway. In the Tagliata South two types of con­ crete construction can be identified, and more con­ crete construction is found along the eastern side of the canal portion of that waterway to the north of the quarry. As has already been noted, most of the concrete construction within the Tagliata South may 26 De Boer, personal communication. Brown 1951, p. 35, and Brown 1980, p. 20, on the hilltop location of the quarries for this wall. 27 See McCann, chapter six, for discussion of the lagoon tem­ ple.

88

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

be reconstructed as belonging to a small bridge and a heavy vault with low ledges. That the concrete of the bridge was put in place before the vault to the south of it is also clear, for concrete of the same mass as that of the vault overlaps the concrete of the bridge. This sequence is made all the more obvious by the fact that a different type of concrete was used in each of the stages. The concrete of the bridge is composed of tuff and pozzolana, whereas that of the vault and its low ledges contains mainly broken limestone and terracotta with only a few scattered pieces of tuff. The sandstone aggregate of the portion of the vault above the seaward outlet clearly belongs to the same phase of construction as the vault itself, for it bonds with the limestone and sherd concrete of the vault to either side of it. The concrete overhead arch located at the bend of the Tagliata Piccola is of the same composition as that of the vault in the mouth of the Tagliata South and thus probably belongs to the same building phase, while the two submerged platforms to the north of the bridge in the main Tagliata channel appear similar in construction to the tuff-and-pozzolana concrete bridge. The 14-m-long tuff-and-pozzolana segment of the eastern embankment wall of the Tagliata Canal located ca. 40 m to the north of the quarry must also belong to the same stage as the Tagliata bridge and submerged platforms. In summary, the sequence of construction discernible within the cliff channels might reasonably be reconstructed as follows: 1. The cutting of the entire course of the Tagliata waterway with the Tagliata Piccola from the lagoon to the sea, followed (after undetermined intervals of time) by the opening of the quarry, the northern end of the Piccolo Spacco, the intersecting tunnels at the seaward end of the Spacco della Regina, and the cutting of the platform and stairs within the Spacco cave. 2. The construction of the bridge of tuff-and-pozzolana concrete in the Tagliata South and of the 14-m-long segment of the upper ledge of the embankment on the eastern side of the Tagliata Canal to the north of the quarry; probably also the construction of the two submerged platforms north of the concrete bridge in the Tagliata South. 3. The construction of the vault in the seaward mouth of the Tagliata South, along with the adjacent walls of the domical tunnel of the seaward outlet and the concrete arch at the bend of the Tagliata Piccola.

Clearly, several substages may have occurred within stage 1, and these will be taken up in chapter seven, where the function of the Tagliata in different periods of the site's use is considered. In this connection, it should also be acknowledged that many of the cuttings in the walls of the channels are difficult to place within the sequence. Those within the Tagliata South that acted as brackets for the later concrete bridge and vault must belong to the respective periods of those structures. There is no internal evidence, however, for determining the stage at which the beam holes and slots to the south of the bridge were made, and the same must be said of several other cuttings along the course of the Tagliata. Some of these cuttings may have preceded the concrete structures by a considerable amount of time, as will be shown in the discussion of their probable functions.

ARCHITECTURAL REMAINS IN THE LAGOON

Within the extensive western area of the now silted ancient lagoon, various structures of great interest were discovered during the excavations of 1968, 1969, and 1972 (Maps 7 to 13). These include the Spring House complex on the western embankment, with its Aqueduct and cistern, and an elaborate series of low walls that originally lay within the water of the lagoon itself. AU the structures, with the exception of the Spring House, will be described here: the embankment walls of the lagoon (P, PW, PE, X, and U); segments of walls that appear to belong to two elongated, trapezoidal fish tanks (Walls V and X to the north of the Spring House Aqueduct; Walls L and N to the south of the same Aqueduct); a small rectangular enclosure (YZ) to the north of the Spring House; and the Aqueduct, which leads from the Spring House eastward across the lagoon.

The Western and Northern Embankment Walls WALL PW

The steep hillside to the west of the lagoon was stabilized by two retaining walls of rough polygonal construction (Maps 7, 8). The upper wall (PW) was exposed in four segments. Immediately to the south of the Spring House the line of Wall PW coincides with that of Wall d against which it was built (Fig. IV-55). At this point the polygonal wall (PW) is pre-

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

served to a height of two courses (0.75 m high; 4.26 m a.s.l.) and is 1 m thick, including its backing of small limestone rubble (possibly mortared). Farther to the north, in Trench IH, Wall PW is 1.25 m thick but only one course high (0.4 m) (Fig. 111-26). The four segments span a distance of ca. 107 m, but the wall was probably much longer in antiquity, per­ haps comparable in length to Wall P, which lies be­ low it 3-4 m to the east. WALL P

Wall P runs in a northeasterly direction along the base of the promontory for more than 250 m from the south (Maps 7, 8), at the point where the high cliffs of the promontory recede (Trench IG; Figs. Ill— 39, 40) to an undetermined point beyond its inter­ section with Wall X, which bounds the lagoonal area on the north (Maps 10, 11). The line of the polyg­ onal wall was engineered to accommodate the nat­ ural contour of the hillside, which becomes steeper and projects farther eastward toward the northern end of the lagoon. At a point ca. 110 m from its southernmost end Wall P changes course and angles more sharply toward the northeast for the rest of its exposed length (140 m). At its highest preserved point (Trench IL, Fig. 11-11) the wall consists of six courses of boulders that rise to 1.82 m a.s.l. The limestone boulders of Wall P vary considerably in size, the largest measuring roughly 0.8 by 0.6 m and the smallest 0.3 m by 0.2 m (Fig. IV-56). The lengths of the larger rocks vary from 0.8 m to 1 m. These stones were laid dry with their long ends ta­ pering into the earth of the hill. Only their outer faces were dressed to a relatively smooth surface. In some areas loose limestone rubble, fragments of am­ phoras (Will Type 4), and broken tiles were ob­ served in the fill behind the wall. Beginning at approximately 14 m to the north of the Spring House, at the point where bedrock pro­ trudes from beneath Wall P, and continuing for ca. 40 m to the end of the bedrock outcrop, the original polygonal masonry of Wall P was replaced with a mortared rubblework wall (Figs. 111-27; IV-57, 58) Apparently the bedrock did not provide a firm foot­ ing for the heavy polygonal boulders. In Trench IH the repaired wall is preserved to a height of 2.65 m above the bedrock (Figs. 111-26, 27, and 28). It is composed of limestone rubble and fragments of Will Type 4 amphoras, bricks, and tiles set in a mortar of lime and fine gray sand. The roughly shaped stones, dressed only on their outer surfaces, measure on an average 0.3 m χ 0.2 m. Some of the large boulders from the original wall were reused in the facing. The

89

liberal use of amphora parts, particularly handles and toes for snecking, imparts a lively speckled appear­ ance to the wall. This construction closely resembles that of the cistern to the west of the Spring House. A very small repair of Wall P is located to the north of Pier Z, just above the foundation (Map 12; Fig. 111-46). The most distinctive feature of this re­ pair is a l-m-long area composed of one course of amphora handles bordered above and below by a course of downturned amphora belly fragments (Fig. IV-59). This decorative arrangement of am­ phora fragments is similar to the sherd-and-tile construction of the column drums of the cistern conduit described by J. P. Oleson in chapter five (Fig. IV-60). The area surrounding this small repair of Wall P was also reworked with mortared sherds and smaller limestone fragments. The heavy polygonal boulders of the original por­ tions of Wall P rest on a base of several courses of relatively small, flat stones. The whole wall leans slightly backward toward the hill, the uppermost courses lying 0.05-0.10 m to the west of the base of the wall. Below the base of small stones the nature of the wall foundation varies. In areas that had been above sea level in antiquity the wall rests either on a foundation of rock and sand or directly on the earth or bedrock of the natural terrain (Trenches IF, ID [Map 10; Figs. 111-34 to 38]). As has already been noted, the 40-m-long repaired segment of the wall rests on an outcrop of bedrock. At four points along the course of Wall P ledges of concrete were con­ structed to serve as foundations and/or reinforce­ ments. Two of these are adjacent to the Spring House (Figs. H-19; IV-61, 62; Map 14), a third is visible in Trench YZ (Map 12; Fig. IV-63), and a fourth was revealed at the intersection of Walls P and X (Map 11; Fig. IV-64). The relationship of these concrete ledges to Wall P and adjoining structures is crucial for reconstructing the plan of the entire la­ goon area and thus must be analyzed in some detail. The structural relationship of Wall P to the Spring House is discussed thoroughly in chapter five, but some essential points must be reiterated here. The ledges to the south and north of the Spring House were constructed within wooden forms that con­ tinue those used for the Spring House basin plat­ form. On the southern side of this platform is a se­ ries of three irregular ledges at varying levels, and it is clear that Wall P was built on top of them. The southwestern corner of the platform itself provides a foundation for Wall P where it meets the buttress of Wall a (Fig. IV-61). AU of these ledges are of the same yellow tuff-and-pozzolana concrete from

90

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

which the platform and some of the upper walls of the Spring House were constructed. The portions of this concrete that were submerged in lagoon water have weathered to a blue-green color. 28 The ledge to the north of the Spring House runs for a distance of ca. 14 m to the above-mentioned outcrop of bedrock (Fig. IV-62). This ledge has a lower section of tuff-and-pozzolana concrete that shares formwork with the Spring House platform; on top is a thinner layer of another type of concrete that was laid against Wall P, partially obscuring the small stones at its base. The lower portion, made of tuff, most likely serves as the foundation of Wall P at this juncture with the Spring House, while the upper layer acts as reinforcement or sealing, proba­ bly to prevent moisture from eroding the base of the wall. This reinforcing concrete consists mainly of an aggregate of limestone rubble and amphora frag­ ments laid in a generous amount of lime mortar con­ taining the local gray sand and small bits of fibrous material, probably derived from pozzolana such as that used in the Spring House walls. Here and there amphora body fragments form a finished surface. This ledge, which is ca. 0.5 m wide, slopes along its course from 0.6 m a.s.l. at the south to 0.42 m a.s.l. at the northern end. A third concrete ledge, exposed in Trench YZ for a distance of ca. 9.5 m, is lower and slightly nar­ rower than the ledge adjacent to the Spring House (Maps 10, 12; Fig. IV-63). Its width varies from 0.5 m to 0.4 m, and its upper surface varies between 0.23 m and 0.29 m b.s.l. At its southern end a break in the ledge affords a view of the footing of Wall P (Fig. 111-46). Here the polygonal blocks appear to rest directly on the concrete. The construction, which includes an area paved with amphora sherds to the north of Pier Z, is virtually identical to that of the upper portion of the ledge north of the Spring House. O n the southern side of the intersection of Walls P and X, exposed in Trench IK, a fourth concrete ledge, ca. 0.5 m wide, was uncovered for a distance of 8.2 m along Wall P (Maps 10, 11; Fig. IV-64). The upper portion of this ledge, which is of limestone-and-sherd concrete similar to that of the two ledges farther to the south, runs along and overlaps the lower courses of Wall P; its short end abuts the foundation of Wall X (Fig. IV-65). The lower por­ tion of the ledge is obscured by a layer of tuff-andpozzolana concrete that appears to be identical in

composition to that of the Spring House platform and adjacent ledges. 29 This concrete is visible all along the base of the ledge of Wall P and along the southern face of the foundation of Wall X, ca. 0.7 m below the level of the foundation offset. It is not certain that the tuff-and-pozzolana concrete forms a foundation for Wall P. However, it is clear that along Wall X it acts only as reinforcement, as may be observed at one point where the concrete has fallen away from the surface of the limestone foun­ dation of Wall X to which it adheres. This hydraulic concrete evidently was added as extra reinforcement at the unbonded intersection, perhaps to protect against the erosive force of the spring water that is­ sues from beneath Wall P in this area (Trench IK). Along Wall X the concrete reinforcement narrows from a thickness of 0.45 m at the western end to 0.36 m at the eastern end of the trench. N o traces of it were found in the eastern extension of Trench IK (IKE). WALL X

Wall X, which defines the northern end of a series of interconnecting walls, served as an embankment wall, judging from the irregularity of its northern face (Maps 8, 10, 11; Fig. IV-66). The upper section of the wall at its western end is 0.53 m thick and rises 0.7 m from a foundation 0.9 m wide. The en­ tire wall is made of broken limestone set in a mortar of fine gray sand and lime. A few amphora frag­ ments are scattered in the core. The facing stones vary in size from larger rocks of 0.35 m χ 0.20 m to smaller ones measuring 0.10 m χ 0.12 m. The mor­ tar of the foundation forms a flush surface with the dressed faces of the limestone blocks, suggesting that the concrete was laid within a wooden form. This foundation wall continues eastward for 43.75 m, where it merges with Wall U (Fig. IV-67). The upper part of Wall X, however, continues for only 15 m to the east of Wall P, and the foundation of the wall from that point up to its juncture with Wall U shows no traces of having carried a superstructure (Fig. IV-68). It would appear that the upper part of Wall X was left unfinished. The Northern Fish Tank WALL υ

The northern end of Wall U, exposed in Trench IKE, is identical to Wall X in dimensions and con-

28 29

See Trigila on the weathering of the tuff, chapter sixteen. N o samples of this tuff were taken, but its blue-green color

appeared identical to that of the Spring House basin platform.

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

struction (Maps 8, 10, 13; Figs. IV-67). The two ap­ pear to have been laid within continuous formwork. To the south, in Probes 1 and 2, Wall U becomes thicker (1.0-1.1 m) but its construction remains es­ sentially the same. Irregular compaction of lagoonal sediments probably caused the variation in the level of Wall U from 0.58 m b.s.l. at the northern end to 0.50 m and 0.71 m b.s.l. toward the south. The bot­ tom of Wall U was not found, but a probe at the northern end indicated that the structure is at least 2 m high. The jagged eastern surface of the southern section of Wall U abutted an artificial earth embank­ ment that was built up on a layer of crushed stone, followed by miscellaneous logs and planks, and fi­ nally a row of nearly intact amphoras (Will Type 4b) laid side by side with their broken rims turned to­ ward the wall (Figs. 11-14; 111-44; IV-69).30 The am­ phoras were located ca. 1.5 m below the top of Wall U (see Will Cats. A224-A236, A239-A242). At its northern end Wall U runs parallel and ad­ jacent to a double concrete wall that also abutted the earth embankment to the east (Map 13). The west­ ernmost of the two walls is 0.64 m wide, while the one on the east measures 0.40 m. Between them is a channel 0.44 m wide (Fig. 11-12). These walls are constructed of limestone concrete, the mortar of which lines the channel walls. Several meters to the north, this channeled wall meets a deep V-shaped channel (1.2 m wide at the top; 0.80 m below) cut into bedrock, evidently at the source of a spring that has since gone dry (Fig. 111-15). The constriction of the channel from 0.80 m to 0.44 m must have caused the spring water to flow swiftly toward the south. At a point 7 m south of its juncture with Wall U, the channel turns and goes under Wall U, emptying into the enclosure formed by Walls U, V, and W, identified here as a fish tank (Fig. IV-70). Here the channel is lined with tiles, probably to guard against erosion by the strong current. WALL ν

Wall V is the northern boundary of the large trap­ ezoidal fish tank into which the channel empties (Maps 10, 13; Fig. IV-70). It lies parallel to Wall X ca. 4.8 m to the south and is 32.75 m long. Wall V has a foundation ca. 0.6 m wide and a narrower up­ per wall that measures ca. 0.3 m across. Its upper surface is at approximately the same level as that of Wall U and the foundation of Wall X, varying be­ tween 0.57 m b.s.l. on the east and 0.53 m b.s.l. on the west. The bottom of the wall could not be

91

reached owing to the limits of the machinery avail­ able for excavating through the mud. Probes indi­ cated that it probably continues downward, for 2 m or more, as does Wall U. At 0.15 m from its western end there is a narrow channel (or break) in the top of the wall. The construction of Wall V is very sim­ ilar to that of the reinforcing ledges along the base of Wall P. It is an unfaced concrete structure com­ posed principally of broken limestone and amphora fragments set in a large amount of mortar containing coarse, pebbly sand and lime. Some broken tiles ap­ pear on the upper surface along with many Type 4 amphora fragments, which form areas of paved sur­ face. An identical type of construction is found in Walls W, L, and N and in the reinforcing offset of the foundations of Wall P. WALL

w

Wall W intersects Wall V at an angle of ca. 80° to form the western side of the trapezoidal enclosure (Maps 8, 10, 12; Fig. IV-71). This wall, which runs parallel to Wall P at a distance of 11 m to the east, was traced for 49.5 m in three trenches (IK southern extension, probe east of IJ, and YZ) and in five probes to the south of Trench YZ. The latter probes revealed a heavier, lower wall continuing along the same line to a point 6.5 m to the northeast of Aq­ ueduct pier 5. If all of these segments belong to the same wall, as seems likely, the total preserved length of Wall W would be 106.5 m. The uppermost portion of Wall W, which lies at 0.54 m b.s.l., is 0.38 m wide and rests on a founda­ tion 0.62 m wide and 0.27 m below its upper surface (0.81 m b.s.l.). What appears to be the continuation of this foundation wall to the south varies slightly in width (0.65 m to 0.68 m) and is 1 m b.s.l. In Trench YZ (Map 12), the eastern side of Wall W was exca­ vated to a depth of 1.5 m, but the bottom of the wall was not reached. On the western side, about 0.3 m below the top of the wall (at approximately the level of the foundation offset), a rough paved surface of lime and small stones adjoins the wall (Figs. 11-16; 111-47). This surface is ca. 0.24 m thick and continues to the west for an undetermined dis­ tance. An undecorated late Italian sigillata sherd was found resting on this pavement (Cat. RG5). In Trench YZ, Wall W is interrupted for ca. 2 m where it appears to have been damaged by debris that fell from the small rectangular structure YZ on the west (Map 12; Figs. 111-47). Wall W is compa­ rable in construction to Wall V except that it has a

For a similar use of amphoras to stabilize marshy ground in the ancient port of Marseille, see Euzennat, p. 134, pi. 22, fig. 3.

92

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

few pieces of tuff in its aggregate as do Walls N and L, the reinforcing ledges of Wall P, and the southern continuation of the foundation of Wall W. At the southernmost end of the foundation of Wall W, a horizontal plank from the formwork was found on the western side 0.78 m below the preserved top of the wall. Impressions of other planks are preserved in the layer of mortar that coats both surfaces of the wall. This southern extension of the foundation of Wall W shows no traces of an upper wall, and it ends abruptly on the south ca. 2 m from pier 5 of the Aqueduct, suggesting that it too may never have been completed or was broken off. The Aqueduct PIERS 3-8

Six piers, numbered 3 through 8, traverse the la­ goon between the polygonal embankment Walls P and PE immediately to the south of the trapezoidal enclosure just described (Map 8; Fig. 111-45). The westernmost, pier 3, is located 1.45 m east of Wall P and 3.9 m south of the Spring House platform (Fig. IV-72). It rests upon a 1.95 m-wide ruff-andpozzolana concrete platform that abuts the founda­ tion of Wall P at 0.57 m b.s.l. Whether this platform continues across the lagoon as a foundation for all the piers is not certain. The remaining four piers, placed at intervals of ca. 3 m, are in line with Walls d and e on an east-west axis parallel to that of the Spring House (Fig. IV-73). It is likely that two or three additional piers continued the line to Wall PE, but no traces of them were found. They, along with the eastern end of Wall L, may have been destroyed when the modern drainage canal was dug. Further piers to carry the water of the Aqueduct to the em­ porium or commercial center of the port probably existed east of Wall PE, leading south in this still unexcavated area. Of the six nearly identical piers, pier 3 is the best preserved (Fig. IV-72). Like the others, it has a base of unfaced tuff-and-pozzolana concrete that was poured into wooden forms. Impressions of horizon­ tal planks are preserved in the mortar on the surface of the concrete. The superstructures of the piers are built of mortared rectangular limestone blocks sur­ rounding a small core containing broken limestone and amphora fragments. The lengths of the blocks vary, but the heights are aligned in relatively even courses snecked, where necessary, with fragments of tiles. None of the piers stands to its full original height. Pier 3 is preserved to a height of 2.16 m measured from the concrete foundation platform on

the west (or 1.6 m a.s.l.). A probe made beneath the pier on the south side determined that the bottom of the structure lies at 1.40 m b.s.l. (Fig. 11-13). The concrete base of the pier rises 0.90 m above the foundation platform and measures 1.15 m (eastwest) χ 1.0 m (north-south). The limestone super­ structure is preserved to a height of 1.35 m above the concrete base and measures approximately 0.85 m on each side. The upper and lower parts of the piers meet at a point ca. 0.3 m a.s.l. PIER 9

Pier 9 is situated on higher ground to the east of the modern drainage canal and the polygonal em­ bankment Wall PE (Figs. 111-49, 50). It lies 19.4 m to the east of pier 8 of the Aqueduct approximately in line with piers 3-8. However, the dimensions and construction of pier 9 differ considerably from those of the others; thus it may date from a later period. Its base is 1.50 m high, 1.2 m long (north-south), and 2 m wide (east-west). The partially preserved superstructure rises an additional 0.4 m above the base and measures ca. 1.5 m (east-west) by 0.4 m (north-south). The total height of pier 9 is 2.10 m (or 2.09 m a.s.l.), about half a meter higher than the other piers. In contrast to piers 3-8, pier 9 was built on dry land. Limestone and amphora fragments, rather than tuff-and-pozzolana concrete, were used for the foun­ dation as well as the superstructure (Fig. IV-74). The foundation appears to be of unfaced limestone con­ crete constructed within wooden forms. In certain places, particularly on the western face, there appear to be some impressions of horizontal planks. The upper portion of the pier is of sherd and limestone rubblework set in a mortar of fine gray sand and lime. The facing stones at the corners are roughly rectangular in shape, while those of the intervening surfaces are irregular and are snecked with amphora fragments. An amphora toe is clearly visible on the western face, along with some belly fragments. Limestone and sherds are also present in the core. This construction technique compares best with that of several other structures in the lagoonal area, in­ cluding the Spring House cistern and Pier f of the cistern conduit colonnade.

The Southern Fish Tank: WALLS L AND N

A second trapezoidal structure lies just to the south of the Aqueduct (Map 8). Two long walls, L and N, intersect at an angle of ca. 80° approximately

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

5 m to the southeast of the end of Wall W to form the northern and western boundaries of the enclosure identified here as a fish tank. The northern Wall L, which exactly parallels the Aqueduct piers (Fig. IV-73), was followed for 1 1 m , but owing to the impossibility of excavating in the modern drainage canal, the eastern end of the wall was not located. Wall N (Fig. IV-75), which is roughly parallel to the course of Wall P at its southern extremity, was exposed in three trenches (IB, IE, IN; Figs. 11-10, 17, and 18) for a total distance of 91 m, but here again the end of the wall was not found. The wall may continue beneath the modern road that runs between the lagoonal area and the beach. Walls L and N vary in width between 0.7 m and 0.6 m, roughly comparable in dimension to the foundation of Wall W of the northern trapezoidal structure. Like Wall W and other submerged walls in the lagoon area, N and L were constructed within wooden forms. On the northern face of Wall L, between Aqueduct piers 7 and 8, a long horizontal plank was found in situ ca. 0.6 m below the top of the wall. Farther to the west, another plank was found near the top of the wall, and several vertical planks were observed. At the intersection of Walls L and N a wooden post remains in place, and many impressions of horizontal planking are preserved in the mortar that coats the sides of both walls. The formwork appears to be comparable to that of other submerged walls in the lagoon, including that of the Spring House basin platform which, however, also preserves an outer wall of upright planks on its southern side (Fig. V-7). The aggregate and mortar of Walls N and L are identical to those of Walls W and V and the ledges of Wall P. The partial paving of tile and amphora fragments on top of Wall L is also comparable. Two stamped rims of Sestius amphoras (Type 4a, Cats. A99, AlOl) were found immured in the paved surface (Fig. IV-76). Although the two walls of this southern enclosure compare closely with the foundation portion of Wall W in design, dimensions, and construction, they do not seem to have served as foundations for thinner upper walls. Wall L has a finished upper surface, and both L and N are nearly 1.3 m higher than the southern end of Wall W. Also, the top of Wall L is at approximately the same level as the socles of the Aqueduct piers. It is possible that these relationships have been somewhat distorted by the compaction or shifting of sediments, but it is unlikely that any shift would account fully for a difference of 1.3 m in 31 See Gazda and McCann, chapter seven, reconstruction of the fishery.

93

height. Rather, it would seem that the northern and southern enclosures were intended to serve different purposes, a possibility that will be discussed elsewhere. 31

The Eastern Embankment: WALL PE

Polygonal Wall PE lies ca. 35 m to the east of Wall P (Maps 7, 8; Figs. IV-77, 78, and 79). It served as an embankment wall for the higher ground at this point in the lagoonal area, and it may also have formed the eastern side of the trapezoidal enclosure defined on the north and west by Walls L and N. Today, Wall PE functions as a retaining wall for the modern drainage canal. The northern end of the wall is located 60 m to the southwest of the southernmost exposed segment of Wall U. At its northern end the polygonal boulders of Wall PE are adjoined by a limestone rubble wall that angles to the east (Fig. IV-79). Whether this rubble construction marks the point where the original polygonal construction of Wall PE ended in antiquity is not certain, but it seems probable, since no other polygonal blocks were found to the north. Furthermore, the pattern of sedimentation in this area, in particular the spread of gray sand to the north and east of the end of Wall PE, indicates that Wall PE probably never extended northward beyond this point (Text Figs. II—1, 2, and 4). 32 The polygonal wall is preserved in several segments for an overall distance of 77.5 m, but in antiquity it may have been connected to the easternmost of the similar polygonal walls (A and B) on the beach (Map 6; Fig. IV-19). The polygonal sections of PE alternate with stretches of limestone rubblework that appear to be later repairs or alterations of the original wall (Fig. IV-77). The first rubble segment on the north extends for 7.70 m. It is followed by a 1.53-m-wide passageway with walls leading to the east for a distance of ca. 1.45 m. The corners of PE and the passageway walls are quoined with rectangular limestone blocks (Fig. IV-79). To the south of this passageway another stretch of rubble wall extends for 4.1 m until it meets a segment of polygonal construction 7.2 m long (Fig. IV-78) which, in turn joins another section of rubble wall that continues for 6.5 m t o the south. Another polygonal segment begins at that point and continues southward for ca. 45 m. Approximately four polygonal courses, which See Bourgeois, chapter two, for discussion of sedimentation.

94

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

are preserved to a height of 1.57 m in Trench E l , are visible above the water of the modern canal. The limestone boulders of the polygonal segments of Wall PE compare in size with those of Walls P, PW, A, and B and with those of the submerged wall off the Torre di Burano. In Trench El (Map 8) the wall is 0.82 m thick, including its backing of smaller limestone fragments that are embedded in earth. Be­ cause the wall extends downward into the water of the modern drainage canal, it was not possible to determine whether it rests upon concrete foundation ledges similar to those of Wall P. The rubble repair sections of Wall PE and the short walls of the pas­ sageway are of limestone rubblework construction of a type commonly found at Cosa. The repairs are ca. 0.5 m thick. WaIlS Two low, narrow walls that intersect at a point 3.61 m south of pier 9 appear to be aligned with the sides of that pier (Map 8; Fig. 111-50). The materials and methods of construction were not studied in de­ tail because the wall was found on the final day of the excavations in 1972. The walls, which are ca. 0.48 m wide, are faced with irregular limestone rub­ ble, and their cores contain small limestone and am­ phora fragments. Structure YZ Structure YZ is a small rectangular enclosure measuring ca. 5.8 m (east-west) by ca. 7 m (northsouth), situated along the embankment Wall P ca. 50 m north of the Spring House and 41 m south of Wall X (Maps 8, 12; Fig. 111-48). This structure, appar­ ently related to the repaired section of Wall P (de­ scribed above), is composed of three low walls (Y, Z, and Cross wall YZ) and heavy rectangular piers (Figs. 111-46, 47). Pier Y on the south is built against the repaired section of Wall P, while Pier Z (Fig. IV80) on the north abuts the original polygonal ma­ sonry. The lower courses of two additional piers are in place on Crosswall YZ, one at the northeast cor­ ner of the structure and the other slightly to the south of the center of the crosswall. The upper parts of these two piers, which had fallen eastward, were found in Trench YZ (Fig. 111-47), and there are traces of a fifth pier that stood at the southeast cor­ ner of the structure. Piers Y and Z are ca. 0.7 m square in section, while the pier that fell from the northeast corner is slightly larger, measuring 0.78 m on a side. Pier Y is preserved to a height of 1.1 m

and Pier Z to 1.3 m, measured from the top of the foundations of Wall P. The fallen pier has a pre­ served length of 2.0 m. The second fallen pier was not completely uncovered, but the portion of it that remains on the crosswall measures 1.0 m χ 0.7 m. The pier on the southwest corner of the structure is ca. 0.7 m square. Other irregularities are apparent in the walls. Wall Y measures 5.8 m (east-west), while the parallel Wall Z is only 5.3 m (east-west). Both are ca. 0.7 m wide. The crosswall that constitutes the eastern side of the enclosure varies in width from the northern end, where it is 1.5 m wide, to the southern end, where it is only 1 m wide. The levels of the three walls also vary. Walls Y and Z are 0.27 m a.s.l., while the crosswall is only 0.09 m a.s.l. Some of these irregularities are due to varying states of pres­ ervation, but others may have resulted from hasty workmanship. At the southeast corner of the structure, where Wall Y and Crosswall YZ intersect, a narrow chan­ nel passes through the wall (Fig. IV-81). Its sides are lined with bricks and its bottom with tiles. A cover tile, 0.2m χ 0.4m χ 0.8 m, originally roofed the channel. A fragment of a small marble half-column' was found at the south side of the structure lying over the channel. (See Cat. AD2 and Fig. 111-48). Other fragmentary architectural elements were found strewn in Trench YZ along with the fallen piers. These included a section of a brick arch and several limestone blocks (Fig. 111-47). The arch frag­ ment has an extrados width of 0.51 m and an intrados width of 0.32 m. The individual bricks are 0.20 m long. A threshold block was also recovered. Its preserved dimensions are 0.46 m χ 0.16 m, with a projection of 0.17 m and an indentation of 0.10 m. Three ashlar limestone blocks were found in this area. One measured 2 m long (preserved) χ 0.95 m wide; another was 1.5 m long χ 0.7 m wide. This fallen debris suggests that the construction of the YZ enclosure was quite substantial and that the piers car­ ried a heavy, arched superstructure. The concrete of Walls Y, Z, and Crosswall YZ has an aggregate of broken limestone, bricks (0.05 m thick), tiles, sherds, and reused pieces of opus signinum set in a lime mortar consisting of fine sand, a small amount of gravel, small fragments of yellow tuff, and some bits of red material, probably crushed brick. Walls Y and Z were built on top of the foun­ dation ledges of Wall P (Map 12; Figs. 111-46; IV63). Bricks were stacked on top of the ledges to build up their level to that of the walls and also to provide solid foundations for the piers. The walls

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

themselves were constructed within wooden forms, parts of which remain in place. A horizontal plank was found near the top of the northern face of Wall Y, and on the southern side of the same wall more horizontal planks were found ca. 0.4 m below the top. Several rounded vertical posts and a vertical plank remain in place against the horizontal planks. Walls Y and Z are unfaced except for their upper surfaces, which are paved with broken tiles, some sherds, and possibly some brick fragments. The surface of the crosswall is not well preserved. The piers of the YZ structure are of very sturdy construction (Fig. IV-80). Single courses of roughly rectangular limestone blocks whose heights vary from 0.08 m to 0.15 m or 0.20 m alternate with single courses of tiles. The tiles, which are buff or red-orange in color, are 0.25 m thick and more than 0.55 m long. The small cores of the piers contain small pieces of broken limestone, some sandstone, sherds, and bricks (0.45 m thick) along with some reused opus signinum (Pier Y), all set in a mortar of coarse sand, a bit of gravel, crushed terracotta, and lime. This mortar differs from that of Walls Y and Z only in that it does not contain fragments of yellow tuff. Sequence of Construction in the Lagoon At least two distinct phases of building activity can be identified among the structures of the lagoon, stages that are closely tied to the two periods of construction of the Spring House distinguished by Oleson in chapter five. As Oleson demonstrates, the concrete foundations of Wall P to the south and north of the Spring House are contemporary with those of the Spring House itself; indeed, both were constructed within continuous wooden forms. The segment of Wall PW to the south of the Spring House appears to be contemporary with the Spring House complex because of its relationship in construction and plan to Wall P below. Given the similarity of the polygonal embankment walls (P and PW) on the western side of the lagoon to Wall PE on the eastern side, there seems little reason to doubt that Wall PE also belongs to this phase. As we proceed northward along Wall P, another section of its concrete foundation appears to be identical in construction to the one that lies immediately to the north of the Spring House. At the extreme northern end of the site Wall P meets Wall X at an intersection that must also be contemporary with the construction of the foundations of the Spring House and Wall P. Proof lies in the materials and techniques

95

of construction. This intersection of Walls P and X is reinforced with the same type of tuff-and-pozzolana concrete that was used in the Spring House, and the upper level of the reinforcing offset along the base of Wall P at this intersection is identical in construction to the two exposed segments of foundation that lie between the Spring House and Wall X. The other walls of the northern enclosure must also belong to this stage of construction. At its eastern end Wall X merges with Wall U, which in turn bonds with Wall V, and the latter bonds with Wall W at its western end. The channeled wall that runs along the western side of Wall U at its northern end turns beneath Wall U just to the south of its intersection with V. These bonded connections thus attest that all of the walls at the northern end of the lagoon were built as part of the same project and that the project also included Wall P and the Spring House. To the south of area YZ the line of Wall W is continued by a thicker, lower wall that corresponds in width to the foundation of Wall W excavated in Trench YZ. Because it is made of the same type of concrete as Walls V and W to the north, there is little reason to doubt that it was built at the same time. Likewise, Walls N and L, which lie to the south of the Aqueduct, must also be contemporary. These walls are of a dimension comparable to that of the lower portion of Wall W, and they are identical in materials and methods of construction to Walls W, V, and the three reinforcing offsets of the foundations of Wall P to the north of the Spring House. Further evidence for the unity of all these walls with the first stage of construction at the Spring House lies in the overall plan. The long axis of the Spring House, which is paralleled by that of the Aqueduct piers, is repeated by the lines of Walls X, V, and L. Wall W is parallel to the northern portion of Wall P, while Wall N follows the slightly more southerly direction taken by Wall P from a point 22 m to the south of the Spring House toward the sea. Moreover, in many of the structures (including the Spring House) standard dimensions, which can be converted into multiples or fractions of the Roman foot (0.296 m), were used. Thus, the lagoonal complex gives the impression of a highly controlled and functional plan. The efficient execution of that plan is apparent from the deliberate choices of materials and techniques of construction and their consistent application throughout. All submerged load-bearing structures, as well as walls that had to resist the pressure of water, were constructed of hydraulic concrete made from tuff from the deposits of the Volsinian volcanic complex

96

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

and pozzolana most likely brought from the region of Pozzuoli. These structures include the basin platform of the Spring House and the walls of the room that contained the surging waters of the spring (Room 2), the foundation of Wall P to the south of the Spring House, the contiguous foundation beneath pier 3 of the Aqueduct, and the socles of all six of the extant Aqueduct piers. Exactly the same type of concrete was used to reinforce the unbonded intersection of Walls P and X that is located near another spring. For submerged walls that bore no weight of superstructures a more economical concrete was used. This second type consisted of an aggregate of local materials (limestone, amphora, and tile fragments) and a few scattered pieces of the imported tuff set in a mortar of sand, lime, and small bits of tuff, probably from pozzolana. This concrete was used for the reinforcing offsets above the foundation ledges of Wall P and for the entire fabric of Walls N, L, W, and V. The channeled portion of Wall U, which conducted water from a spring into the trapezoidal enclosure, was also made of this material. Moreover, all of the concrete walls of the lagoon area were built within wooden forms of the same type. Horizontal planks were attached to upright poles that stood outside the forms. The best-preserved example of this type of formwork is that of the Spring House platform (Figs. Ill—32; V-4 to 7), but other fragments of wood were found in situ on Walls W and L, and impressions of horizontal planks are preserved in the mortar on the vertical surfaces of virtually every concrete wall. For structures that were above the level of the water or abutted earth embankments, the builders chose other types of construction appropriate to the function and location. The heavy polygonal masonry of Walls P, PW, and PE served best to restrain natural earth embankments. Wall U, however, was built on lower ground to support an artificial fill. The thick proportions of Wall U and its concrete construction of broken limestone set in a mortar of sand and lime enabled the wall to stand independently as a brace for the fill to the east. The heavy foundation of Wall X is of a similar type of concrete construction, and, like Wall U, it stood on low ground against an artificial earth fill. Squared stone masonry is found only in the upper portions of the Aqueduct piers and at the corners and buttresses of the Spring House. This careful orchestration of building materials and methods is also characteristic of the construction of the Spring House. The rebuilding of the Spring House, together

with the construction of the cistern (Figs. V-38 to 41) and upper water conduit (Figs. V-36, 37) signals a revitalization of building activity in the lagoon. Wall P was repaired using materials and methods comparable to those employed for the construction of the cistern. A small repair to Wall P, at the base of the wall in Trench YZ (Fig. IV-59), reflects the same tendency to use amphora fragments in the decorative manner that is also apparent in the cistern conduit colonnade (Fig. IV-60). The small rectangular structure in area YZ is built up against the repaired section of Wall P, and the details of its construction materials and techniques suggest that it is contemporary with that repair, although it could be somewhat later. The low concrete walls of the YZ enclosure are similar in most details of construction to those of the earlier walls in the lagoon (including Walls V, W, L, and N, and the foundation offsets of Wall P), but they are not as well made, and their mortar, unlike the type used for the earlier structures, contains crushed terracotta. The masonry of the piers built on top of these walls alternates roughly quadratic limestone blocks with single courses of tile or brick, a technique that can be compared with parts of the maritime villa and with Imperial structures elsewhere. Other structures in the lagoon area that belong to this second phase of construction are few. Pier 9, which stands to the east of Wall PE, closely resembles in construction the Spring House cistern and the foundations of its upper conduit colonnade. Moreover, this pier does not fall precisely into line with piers 3-8 of the Aqueduct, thus strengthening the impression that pier 9 was a later addition. Wall PE itself appears to have been repaired in several sections with rubblework construction similar to that used for the repair of Wall P, which in turn resembles the construction of the Spring House cistern, the foundations of its conduit colonnade, and pier 9. The stages of construction within the lagoon may be summarized as follows: 1. The polygonal portions of retaining Walls PE, PW, and P, together with the concrete foundations and reinforcing offsets of Wall P; the northern trapezoidal enclosure comprising Walls X, U, V, and W; and the southern trapezoidal structure comprising Walls L and N— all concurrent with stage 1 of the Spring House. 2. Repairs of Walls P and PE; structure YZ and pier 9—probably contemporary with stage 2 of the Spring House.

IV: PORT AND FISHERY REMAINS AND C O N S T R U C T I O N

Stage 1 in this sequence clearly was designed and executed as an integrated plan; stage 2, on the other hand, may have involved a number of substages. For example, structure YZ may have followed the repair of Wall P by a considerable amount of time. The place of Wall S in the development of the site cannot be determined on the basis of the evidence now available. the structural sequences observed within the port, cliffs, and lagoon provide the basis for discerning at least three distinct ancient building phases at the Portus Cosanus. The types of building materials used, the methods of construction, and the structural relationships noted provide the evidence. Generally speaking, the three stages of construction observed within the port basin and along the shoreline correspond to the three observed within the cliff channels, while the second and third stages in both these areas can be shown to be concurrent with stages 1 and 2 in the lagoon. Thus, the first building phase of the site included the breakwater, the Tagliata, the Tagliata Piccola, and possibly also the hand-scarped portions of the Spacco della Regina and the Piccolo Spacco. The second phase entailed the erection of large tuff-and-pozzolana concrete piers in the harbor; the installation of fish tanks in IN CONCLUSION,

97

the lagoon along with the construction of the Spring House and its lower Aqueduct; the refurbishment of the Tagliata with hydraulic concrete structures made of tuff and pozzolana; and the construction of several new channels connecting the lagoon to the sea at various points along the coast. The remains of a large Imperial villa on the beach provide the clearest evidence for a third and final phase in the ancient architectural development of the site, one that also involved further additions to the Tagliata—namely, the limestone and sherd concrete vault and ledges at the southernmost end—and probably also the construction of an overhead arch at the bend of the Tagliata Piccola. Piers 1, 2, and 3 and Wall M may have undergone repairs to their upper portions as well. Within the lagoon the polygonal embankment walls (P and PE) were repaired, a small fountain house (YZ) was constructed, and the original Spring House was rebuilt and provided with a cistern and upper water conduit. Construction of a new pier 9 on the eastern embankment may indicate that the original Spring House Aqueduct was also revitalized to serve the seaside villa. A detailed reconstruction of the architecture of the site and its function in each of these three building phases will be presented in chapter seven and their chronology in chapter eighteen.

Chapter V. The Spring House Complex JOHN PETER OLESON

EXCAVATION OF THE SPRING HOUSE, AND ITS RELATIONSHIP TO THE PORT

1968-1975,

During the summer of 1968 a series of trenches was excavated in the hypothetical inner harbor at Cosa, perpendicular to a rough polygonal wall that revetted the foot of the slope along the west side of the area. One of these probes, termed IC ("Inner Harbor, C"), was located in front of the southernmost of a pair of rubblework walls (later termed Wall a) that seemed to interrupt the polygonal retaining wall approximately halfway along its length and could be traced for 13.0 m up the slope to the west (Map 8; Fig. HI-30). After several days of excavation, a concrete platform consisting of tuff-andpozzolana mortar was uncovered at a depth of 1.50 m, still surrounded by the wooden forms into which the concrete had been poured and encircled by a row of upright planks (Figs. Ill—31, 32). Although the high level of the ground water at this point necessitated the use of a pump during excavation, the water protected and continued to preserve the wood by rapidly covering it up when attention was directed elsewhere and the pumps were stopped. In 1969 probes were made at various points in the harbor area, both on land and in the water, but attention was focused on the structure discovered the previous year in trench IC. After several weeks of work, it became apparent that the concrete platform, bisected by an open drain, formed the east wall of a basin with a capacity of 17,000 liters (4,500 gallons), which collected the water of a still flowing spring. Once this basin had been partially cleared, the flow of the water increased, making the continuous use of a powerful irrigation pump necessary. The flow into the basin has been calculated at 1,500 liters per minute (400 gallons per minute), but this rate slowed 1 I would like to thank the American Council of Learned Societies for a grant-in-aid that allowed me to carry out research in Rome and at Cosa in 1975, and the Faculty Research and Travel Committee at the University of Victoria, which provided funds

down somewhat once the basin became filled, and it probably varies from season to season and year to year. In the course of the excavation the area around the basin and platform was encompassed by a modern cofferdam of upright planks designed to contain the wet earth along the sides (Fig. 111-41). During the extensive excavations in the lagoon area in 1972, the rest of the Spring House complex was uncovered (Figs. V-I, 2, and 3). Two rooms west of the basin were revealed, along with a row of piers that led to a cistern located farther up the slope (Map 14; Figs. V-42, 43). In Room 2, just behind the basin, a mass of broken and burned timber appeared at water level, containing several elements of a Roman water-lifting device—a bucket-chain driven through an angle gear (Figs. V-8, 9). This device, known through archaeological remains, ancient literary testimony, and papyri, is of a type still in use in some parts of the Mediterranean world (see Figs. V-54 to 57). It frequently is called by the Arabic name sctqiya. Since no preservation facilities were available at the time, the wooden machinery fragments were reburied in Room 2 below the water level after photography and study. In 1975 some of the pieces of wood were recovered and conserved by means of the acetone/rosin impregnation method for study and possible reassembly. A full-scale reproduction of the best preserved bucket was subsequently constructed to test the mode of chain assembly and as a museum display (Figs. V-58 to 60).1 The four seasons of excavation focused on the Spring House revealed a complex of extraordinary interest. The structure itself-—well-built and specifically designed to collect and exploit the water source—was carefully related in function and location to the other structures of the lagoon and harbor areas (Map 8). The importance of a nearby freshfor the continuation of my research and for construction of the bucket reproduction. See J. P. Oleson, chapter thirteen, for a catalogue of wood from the Spring House machinery.

V: SPRING HOUSE COMPLEX

water source for ships using the harbor, for villas and more modest habitations on the lower acropolis slopes, and for the network of fish tanks in the la­ goon must have made the Spring House a focus of activity for this entire area of the city. 2 The structure was used intermittently for approximately 250 years—from late in the first quarter of the first cen­ tury B. c. to the second half of the second century— but underwent no major modifications in plan or function. From the very beginning, it was designed to accommodate a bucket-chain driven through a sdqiya gear. At first the device was set up to lift water a vertical distance of 5 m so that it could flow east across the lagoon area on an elevated channel (Fig. V-48). Both this device and the Spring House itself were abandoned in the late first century B.C. In the second half of the first century A. D. the Spring House was adapted for installation of a device to lift water 13 m, so that it could flow west through an aqueduct into a cistern constructed farther up the hill (Fig. V-53). The preservation of elements of this machine, in addition to the wooden forms around the outer basin, give the Spring House a unique im­ portance in the study of Roman technology: it pro­ vides the earliest evidence for the presence of a geared water-lifting device in the ancient world. Furthermore, the careful orchestration of the types of tuff and limestone used as aggregate and facing, along with purposeful variations in the mortar bind­ ing, reveal a sophisticated knowledge of the capabil­ ities of the material and a surprising preference for some materials not available locally. The basin plat­ form and the piers of the port and lagoon aqueduct are the earliest known datable examples of tuff-andpozzolana concrete. Ancient harbors constitute a special type of site because they exist in a double environment: dry structures rest upon foundations that lie below water level, and both elements had to be adjusted to serv­ ice ships varying considerably in design and size. Consequently, design, construction techniques, choice of materials, and economic function could not be as straightforward at the juncture of sea and land as they are at a dry site. The previous contro­ versy concerning the purposes of Tagliata and the Spacco della Regina, which have always been visible at Cosa, is evidence of the difficulties and ambigui2 Water was always a problem on the acropolis of Cosa, and there are relatively few springs even at the base of the hill. See Brown 1951, pp. 84-88; also Brown 1980, ρ 11. The Spring House source seems to have attracted attention as late as the six­ teenth century; see n. 18 below. 3 For a discussion of the special problems of harbor excavation,

99

ties that complicate a study of this type. The Spring House is in itself a microcosm of the complex prob­ lems surrounding the port of Cosa as a whole, for it extends both above and below sea level, it was care­ fully designed and constructed to serve several hy­ draulic purposes, and it was integrated into the eco­ nomic and social system of the port area. It is not surprising that the excavation of the Spring House and the evaluation of its function have had to over­ come the same difficulties confronting the overall harbor project. 3

STRUCTURAL ELEMENTS OF THE SPRING HOUSE

(Map 14; Figs. V-42, 43) The Spring House can best be analyzed as a num­ ber of distinct but intimately related units: the plat­ form and basin, with the associated piers 3-9 (in Trench IA); Room 2; Room 1; Walls d and e; and Piers f through k, along with the cistern. Because of the complexity of the structure and the unrewarding character of the stratigraphy, the interpretation of the sequence of construction and changing patterns of use must depend on careful analysis of the struc­ ture itself. Platform, Basin, and Piers The platform is a solid concrete block, 6.90 m wide at the west end and 4.58 m long along the south edge, an irregular polygon in outline, the up­ per surface of which is divided into several different steps, levels, and piers (Figs. V-4 to 7). The heavy mass, composed of green tuff aggregate in a pozzolana mortar, rests on the limestone bedrock of the sloping hillside, which is visible at several points in­ side the basin. The highest level of the platform (0.53 m a.s.l.) is a low, T-shaped step (0.25 m high) whose bar (1.03 m—IHS?—wide, 6.90 m long)* forms the west edge of the platform and is inter­ rupted in the center by a drain that bisects the per­ pendicular leg of the T (1.07 m wide, 3.20 m long) (Figs. V-4, 5). The channel of the drain (ca. 0.30 m wide, 0.30 m deep) is lined with six terracotta cover tiles laid end to end in the concrete, snecked with tuff rubble and amphora sherds. On either side of see Schafer, pp. 663-678. and Williams, pp. 73-79. * Throughout this chapter, where the correspondence with meters seems to justify it, multiples of the Roman foot (0.296 m) will be noted in Roman numerals. " S " (semis) indicates half of such a foot. For example, "IHS" signifies 3Vi Roman feet.

100

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

the drain the leg of the T is built of roughly squared blocks of tuff somewhat larger and more regular than the aggregate in the rest of the platform. West­ ward extensions of this step (2.75 m long) form the north and south sides of the basin and serve as the foundations for the east ends of Walls a and b. The drain carried by the upper step crosses an irregular platform (0.28 m a.s.l.) whose two asymmetrical eastward arms in turn frame a low, irregular step. The southern arm (1.55 m wide, 3.55 m long, from the edge of the upper step to its southeast corner) is regular in shape (Figs. V-6, 7), while the northern arm (1.60 m wide, 2.43 m long, from the edge of the upper step to the northeast corner) has a trun­ cated forward edge angled back to the northwest (Fig. V-5). An irregular lower step (0.44 m b.s.l.) extends below the drain from the northeast corner of the north arm to a point midway along the north edge of the southern arm (Fig. V-5). At the time of excavation, a terracotta pan tile was found propped up on the step directly below the drain, as if to break the force of the jetting stream (Fig. V-7). A square concrete pedestal (between 0.80 and 0.90 m—III?— on a side, preserved to a height of 0.98 m) stands in the center of each half of the platform (pier 1 on the south, pier 2 on the north), their axes approximately on line with the north and south edges of the basin. Each has a base formed by a low (0.15 m), project­ ing socle of small, roughly squared tuff blocks; the sides are smeared with a layer of cement that flares outward slightly just below the present upper sur­ faces. The upper sections of these piers are con­ structed of yellow-brown tuff, along with a few am­ phora body sherds set in a mortar consisting of sand, some crushed terracotta, and lime, but without pozzolana. Another irregular step (ca. 1.50 m sq.) at ap­ proximately the same level (0.42 m b.s.l.) as the step below the drain abuts the south edge of the platform at its intersection with the polygonal wall (Wall P) (Figs. V-10, 11). A similar platform at a level of 0.57 m b.s.l. projects 1.80 m from the face of the polyg­ onal wall, its southern edge 5.0 m south of the southwest corner of the basin platform, evidently supporting the foundation of pier 3. The lower part of this pier (1.0 χ 1.15 m sq.; 1.70 m high), built of the same materials as the basin platform, has an up­ per offset at 0.30 m a.s.l. below a crowning section 4 The preservation of wooden formwork is not uncommon in harbor situations or bridge abutments when the water is fresh (precluding destruction by the teredo worm) or the wood covered by silt. See the references in Oleson, "Underwater Survey and Excavation in the Port of Pyrgi" (Introduction, n. 62), p. 304, n. 19; R. Aris, "Agde, ancien port, ancien quai," Etudes sur Pezenas et sa region 7, no. 1 (1976) pp. 3-12 (Agde); G. de Boe and

(0.90 m sq.—III—1.08 m high) constructed of roughly rectangular limestone blocks. Six other piers of similar design and construction were set up at 3.0 m intervals in a straight line toward the east (piers 4-9) (Figs. 111-45; V-34). At the time of excavation, the entire area of the basin and basin platform was found to be covered with the badly disintegrated remains of fallen lime­ stone rubblework walls (Th ca. 0.60 m), which probably formed the upper reaches of Walls a, b, and c. The spill reached out as far as the step below the basin spout. The basin platform is still surrounded by the wooden formwork into which the concrete was poured (Figs. V-6, 7). 4 For the most part, it consists of thick spruce, oak, and pine planks (Picea abies, Quercus, Pinus) (0.03-0.05 m thick; ca. 0.34 m wide) of lengths that vary according to position. These boards are held in place by round, upright pilings (0.10-0.14 m diameter) placed at irregular intervals where needed. At least some planks were fastened to these supports by means of iron nails. The arrange­ ment of the formwork shows that the middle level of the complex platform (the level holding piers 1 and 2) was poured in a single stage (Figs. V-4, 5), then the step below the drain and the platform at the southwest corner. The edges of the upper step may have been built up without the use of formwork, since they were composed of a larger and more reg­ ular aggregate. The shelf serving as the foundation of the polyg­ onal Wall P north of Wall b, although 0.07 m higher than the upper level of the basin platform, shares the same formwork and must have been laid at the same time (Fig. V-12). In much the same manner, the lower step (at 0.42 m b.s.l.) south of the basin plat­ form serves as the foundation for the polygonal Wall P south of Wall a. This polygonal wall is built up against the formwork along the southern edge of the main platform. Two isolated wooden pilings driven into the earth 1.20 m in front of the drain (Fig. V5) suggest that the engineers originally planned to have the low front step extend from the southeast corner of the north arm to the northeast corner of the south arm, since horizontal boards laid between those two corners would have required support at precisely these points. A post immured in the conF. Hubert, Caesarodunum 12 (1977) pp. 312-320 (Pomeroeul); C. Brusm, GH scavi di Aquileia (Udine, 1934) pp. 16-26 (Aquileia); Euzennat, pp 133-140 (Marseille); A. Ferdiere, Caesarodunum 12 (1977) pp. 326-328 (Bourges); J. P. Oleson, "The Caesarea An­ cient Harbour Excavation Project," Classical Views 27, n.s. 2:2 (1983) pp. 159-167.

V: SPRING HOUSE COMPLEX

crete of the platform close to the southeast corner of the south arm probably was meant to provide the formwork with stability from the inside (Fig. V-7). A second post partially visible in the east face of the basin suggests that further supports of this type may now be hidden from view within the platform mass. A type of retaining wall composed of upright planks (0.032-0.055 m thick, 0.21-0.46 m wide, more than 2.0 m long) driven into the earth edgeto-edge skirts the south and east faces of the southern arm of the platform and the low step adjacent to it. No horizontal tie beams or braces were discovered along this outer construction, which perhaps originally surrounded the entire platform and was partially removed in antiquity (Fig. V-Il). The west edge of the platform and the westward arms serving as the foundations for Walls a and b frame a large collecting basin (3.45 m wide, 2.75 m long—XII by IX?—1.70-2.05 m deep) (Figs. V-13 to 17). The limestone bedrock appears at the west end (Fig. V-14), but on the east, probes penetrated 1.0 m into the present sand and rubble bottom without exposing bedrock. Wooden forms identical to those surrounding the outside of the platform were partially preserved on the north, east, and south sides of the basin, revealing a skin of puddled mortar above the concrete where some of the boards had fallen (Fig. V-15). One post immured in the east face is partially visible through this mortar (Fig. V-16). Since Walls a and b are thinner than this foundation, a step 0.36 m wide extends along the north and south rim of the basin. Some water enters the basin from the bedrock at the southwest corner, but the greatest part flows in through three irregular openings at the foot of Wall c, which forms the west wall of the basin and separates it from Room 2 (Figs. V-13, 14). All three openings are framed by carefully squared limestone blocks of varying dimensions. The largest opening (0.30 m wide, 0.55 m high) is located 1.60 m from the northwest corner of the basin, the second (0.21 m high, 0.15 m wide) and third (0.17 m high, 0.48 m wide) in the southwest corner. The rest of Wall c, up to the level of the basin lip, is built of irregular, gray-green tuff-and-pozzolana concrete identical to that of the platform. The tuff blocks used at the level of the drain are somewhat larger and more regular than those in the rest of the wall. There are no traces of formwork or its imprint on Wall c, and the impressions of the boards on the north and south walls impinge on it slightly, suggesting that Wall c was constructed first. The structure of the west wall of the basin is in-

101

terrupted at the level of the basin lip (0.53 m a.s.l.) by a course of terracotta roof tiles (the pan tiles laid sideways, the cover tiles at the center laid lengthwise) with spaces between to provide overflow drainage from Room 2. These passages, along with small square holes near either end of the wall, would have prevented flooding of Room 2 if the less accessible lower openings became clogged. The upper part of Wall c, which bonds with both a and b, is built of yellow-brown tuff set in a mortar containing lapilli of pozzolana and the local peppery black sand. A few scattered sherds appear in the core of the upper part of the wall, where the tuff facing has fallen away. Where this facing is preserved, toward the bottom of the wall, it frequently takes the form of opus quasi-reticulatum (Figs. V-17, 18). The center of the wall is interrupted by an opening (0.86 m wide; III?) whose edges were protected by tuff blocks larger and more carefully squared than the rest of the aggregate (Fig. V-4). The bottom of this opening originally was at the level of the tile drains but was walled up at some time for at least part of its height, causing the corner blocks of the northern edge to be disturbed and rearranged in the process (Fig. V-14). The fill survives to a height of 0.87 m (top: 1.40 m a.s.l.), possibly the original termination, since it is level with the tops of piers 1 and 2 on the basin platform. Nine voussoir blocks of yellow Volsinian tuff (0.095 m intrados width, 0.19 m extrados width, 0.34 m long), discovered in the earth filling of the basin 1.0 m below the lip, probably spanned this opening (Fig. V-19). At its intersection with Walls a and b, Wall c is preserved to a height of 2.71 m above the level of the basin lip (3.24 m a.s.l.); it was originally 1.05 m thick (HIS). Above the level of the dock the main lines of the Spring House are formed by Walls a, b, and c and Crosswalls 1 and 2, all of which bond with each other and belong to a single construction period. These walls (including the buttresses) form a long rectangle (12.81 m long, 5.96 m wide; XLIII by XX) divided into three separate units that climb up the steep slope of the hill: the basin abutments, Room 2, and Room 1 (Map 4; Figs. V-42,43). Walls a and b; Room 2 The eastern ends of Walls a and b bracket the basin, forming slanted abutments meant to buttress the rest of the structure to the west (Figs. V-2, 20). The forward edges, which slope backward at an angle of 26 degrees from the vertical, had a facing of carefully squared limestone blocks varying between

102

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

0.20 and 0.30 m in height, while the north and south surfaces were faced with irregular limestone rubble. The core is formed of miscellaneous rubble, mostly limestone and yellow-brown Volsinian tuff, along with sandstone and potsherds, held together by a coarse mortar containing local peppery sand, pozzolana lapilli, and lumps of unslaked lime. A pair of beam holes (ca. 0.18 m sq.) traverses each buttress just east of Wall c, 1.60 m above the basin rim (Fig. V-18). A second hole in Wall b above the first (3.50 m above the basin) does not have a corresponding partner in Wall a. The outer sides of these two buttresses are concealed by fill held by the polygonal walls that were subsequently built up against them on the north and south. The Spring House is not quite perpendicular to the line of the polygonal walls. Wall a extends 2.60 m (IX) east of its intersection with Wall c, and 9.0 m (XXXS) west. East of Wall c, Walls a and b are, respectively, 0.93 and 0.89 m thick (III?). West of Wall c, however, where they form the north and south sides of Room 2, they become slightly thinner (Fig. V-21). The outside lines remain unchanged and display the same facing of limestone rubble, but on the interior they contract slightly (to a thickness of 0.65-0.68 m) and are faced, like the east and west faces of Wall c, with opus quasi-retkulatum. Wall a has lost most of this surface, probably as a result of the fire that destroyed the bucket-chain of stage 2 (Fig. V-24). Crosswall 1 (4.82 m long), which forms the west end of Room 2 (4.82 m wide, 3.25 m long; XVI by XI?), is faced with the same limestone rubble as the outer faces of Walls a and b (Fig. V-22). There is a low step along the base of Wall b in Room 2 (0.25 m wide), built of the same tuff and concrete as the basin platform and at the same level but in line with the south edge of Wall b east of Wall c. The room formed by these walls has a floor of unworked limestone bedrock covered with a layer of blue sand in its eastern half, through which the spring water percolates at various points (Figs. V-23, 24). When this level was exposed, the floor was found to be packed with broken and burned wood—beams, boards, and parts of the water-lifting machinery (Figs. V-8, 9). Pieces of lead pipes, sheeting, and grills found with the wood probably belonged to the same device. Four pairs of beam holes (0.20 m sq.) appear in the north and south walls (Fig. V-24) of the room at 1.50 m and 3.48 m a.s.1., between 0.30 m and 0.40 m in from the east and west walls. There is one corresponding hole in the lower right-hand section of Crosswall 1 (Fig. V-22).

The top of Crosswall 1, which rises 3.20 m above bedrock at its northern end and 3.30 m at the south (4.25 m a.s.l. at this point), is finished off by a single layer of bricks, 0.67 m above the paving of Room 1 (Fig. V-25). Room 1 Room 1 (W 4.82, L 5.06; XVI by XVII?) lies at a considerably higher elevation than the rest of the Spring House but is nevertheless part of the original construction period (Figs. V-43, 46, and 47); it subsequently underwent some minor structural alterations. With the exception of Crosswall la on the east (Fig. V-25), all the walls bond with one another. The lower portions of Walls a and b and of Crosswall 2 on the west, (Fig. V-26), up to a level of 2.68 m above the present paving (which is 3.58-3.62 m a.s.l.), were constructed of the same materials as the east end of Walls a and b and were faced on both sides with limestone rubble. The northwest (Fig. V27) and southwest outer corners of the structure up to the same point are formed by large, carefully squared limestone blocks of various dimensions. An offset along the lower part of the outer faces of Walls a and b forms an irregular, roughly horizontal line approximately 1.20 m above the paving inside the room (Fig. V-37). Below this offset the walls are 0.65-0.68 m thick and above it 0.58-0.61 m thick. At a point 2.68 m above the present paving, the fabric of the Walls a and b and Crosswall 2 (0.58 m thick; II) changes to yellow-brown Volsinian tuff with an opus quasi-retkulatum facing on the interior and exterior, similar to the facing of Wall c. Considerable remains of these upper walls were found lying horizontally in the fill north, west, and south of the walls of the room. The fallen section south of Wall a, which rests on a layer of roof tiles and lime rubble, originally had a vertical height of ca. 2.60 m (Fig. V-28). The outer face of Crosswall 2, which was excavated to its base at one point (3.50 m a.s.l., 0.17 m above bedrock) had no horizontal offset in its limestone section (0.71 m thick) but clearly shows the impression of the trench (1.03 m deep) into which its foundation was laid. The tuff wall above Crosswall 2 is 0.56 m thick (II?) (Fig. V-31). A broken limestone threshold block (W. 0.31 m; H. 0.24 m) projecting from the south wall of Room 1 above a series of brick-topped steps and foundation offsets originally served a door (1.70 m wide) at a level 1.52 m above the present pavement (Fig. V29). During some subsequent remodeling, perhaps when the original floor of the room was lowered

V: SPRING HOUSE COMPLEX

from a higher level to its present one (which is even with the base of Crosswall 2), the block was broken and turned in the wall, and the bottom of the door opening was lowered 0.66 m. to a brick-topped step approached by a lower step that is part of the foun­ dation offset of the wall. Still later, the door was walled up except for a square hole (ca. 0.50 m sq.) directly above the block. Finally, the rough edges around the door were smoothed off with amphora sherds packed in mud, the hole filled with the same material, and the walls and floor of the room cov­ ered with a fine-grained, white lime plaster (0. ΟΙ­ Ο.02 m thick) with a smooth but unpolished surface. This coating gave the south wall of the room a lumpy but homogeneous appearance. A second door (1.06 m wide; HIS) in the northeast corner of the room had a plaster threshold 0.20 m above the level of the present floor, originally pro­ tected by two boards that have left their imprint in the plaster (Fig. V-30). To the west of this door there is a second hole (0.30 m sq.) in Wall b directly opposite the hole above the broken threshold block. The door opening was later partially obstructed by construction of Crosswall la, a rough, unplastered limestone rubble wall (0.28 m thick) built up against the plastered west face of Crosswall 1 and surviving to a height of 1.10 m (Fig. V-25). Whereas the north face of Wall b and the west face of Crosswall 2 were either unplastered or have lost their plaster coating, the south face of Wall a, be­ tween its intersection with Wall c and its west end (Fig. V-31), has retained a smooth covering (0.02 m thick) of the same white plaster used on the interior of Room 1. Its good state of preservation along this stretch of wall and the complete absence of traces elsewhere suggests that the rest of the building's ex­ terior was not treated in this fashion. The fallen sec­ tion of the upper stretch of Wall a reveals the same plaster coating. Although there has been some cracking along the lines of the door in Wall a, it had already been closed by the time the exterior plaster was applied. Some fragments of wall plaster painted a dark red were found in the surface rubble south of Wall a, just west of its intersection with Wall c, and several large pieces of a plaster crowning molding (see catalogue of architectural decoration, chapter thirteen) appeared in the rubble below the fallen wall south of Room 1. The isolated circumstances of these finds and the utilitarian character of the Spring House suggest that they derive from some other structure in the area. A row of eighteen terracotta pan tiles 0.60 m (II) south of Wall a form an eastward sloping gutter

103

joined to the wall by a paving of coarse lime cement that seems to run under the wall plaster along the line of intersection (level of west end, 4.76 m a.s.l.; east end, 3.82 m a.s.l.) (Fig. V-32). The first eight tiles from the west are revetted along their outer edges with small chunks of limestone rubble and have had their flanges broken off neatly, perhaps to remove a possible hazard to traffic using the nearby door (assuming it was still open at the time the pave­ ment was laid). The gutter and paving terminate on a line even with the east face of Wall c (total length 9.50 m), directing their overflow over the short, steep slope held by the polygonal retaining wall. Near the west end of Wall a, 0.80 m from the south­ west corner of the Spring House, the paving and exterior plaster terminate in an irregular line where the wall originally disappeared into the brown, sandy fill surrounding the back of the building. A section of this former embankment, perpendicular to Wall a, clearly reveals the presence of a gully (1.00 m wide, 0.70 m deep) formed by run-off from the eaves on line with the tile gutter; it was filled with rubble from Wall a (Fig. V-33). Walls d and e Two isolated walls perpendicular to Wall a stand 3.30 m south of the central portion of the Spring House (Figs. V-34, 35). They are constructed of limestone rubble and a small amount of sandstone and a few amphora body sherds set in a lime and black sand mortar that lacks pozzolana. Wall d (3.03 m long, 0.56 m thick—X by II?—1.30 m high) to the east has a damaged upper surface (5.09 m a.s.l.), while Wall e (L 3.02 m; H 2.45 m; Th 0.68 m ap­ parently preserves its original level crown (5.95 m a.s.l.). There is a horizontal offset (0.05 m wide) along the west face of Wall e at the same level as the top of Wall d. The remains of a rough polygonal Wall PW (1.10 m thick, two or three courses high) parallel to the lower polygonal wall can be traced for 30.0 m to the south from the south end of Wall d (Fig. IV-55). Two further stretches of this upper wall run north of the Spring House, above Trench YZ. Walls d and e are aligned with each other and with piers 3-9 at the level of the basin platform (Fig. V-34). Halfway between the north ends of Walls d and e and the south wall of the Spring House, a large piece of a fallen pier (?) was found lying on a thin layer of rubble just above the ancient ground level (Fig. V35). The preservation of three corners shows that the original dimensions were 0.44 m by 1.28 m (IS by

104

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

IVS?), but the height cannot be determined. The pier was built of concrete with tuff aggregate, faced at the corners with regular limestone blocks (L ca. 0.33-0.36 m; W 0.19 m; Th 0.15 m) alternating with courses of bricks (one well-preserved example meas­ ured L 0.38; W 0.26; Th 0.06 m). The surface was covered with a thin layer of white plaster. The di­ mensions and material of construction set the wall apart from the rest of the Spring House structure, but it seems too large to have rolled or been dragged from farther up the hill. A smaller fragment, includ­ ing the corner of a pier, was found in the rubble over the north end of the basin platform. The method of construction resembles some Neronian piers added to the forum basilica on the acropolis, so the ma­ sonry may belong to alterations carried out in stage 2 of the Spring House development (see below).

Piers f through k and the Cistern Beginning with Pier f, located 1.40 m outside of the door in the northeast corner of Room 1, a series of six piers placed at 4-m intervals passes obliquely up the slope to the northeast corner of the cistern (Fig. V-36). The latter was built at an angle with the Spring House in order to take advantage of the to­ pography (Figs. V-37, 43). Constructed of limestone rubble mixed with amphora body sherds and toes, the piers are held together by a fine mortar contain­ ing peppery black sand and bits of unslaked lime. The original upper surfaces, which have been pre­ served only on Piers i and k, were finished with am­ phora body sherds smeared with mortar. Except for Pier f, which is rectangular (1.78 m wide, 1.93 m long; VI and VIS), these supports are all roughly square in plan, between 1.05 m and 1.53 m on a side (HIS-V). On its east face, Pier f has an offset (0.35 m wide) at a level (5.81 m a.s.l.) close to that of the offset above the limestone section of Crosswall 2 (6.06 m a.s.l.). Pier g has a similar offset (0.23 m wide) at about the same level as the foundation offset on the north face of Wall b outside Room 1 (Fig. V37). Despite the steep incline, some attempt was made to keep the upper surfaces of the piers at roughly the same level (between 6.73 m and 8.22 m a.s.l.), apparently because they were meant to sup­ port an arcade of circular columns built of alternate courses of radially laid amphora toes and body sherds (Fig. IV-60). Although traces of these col­ umns (ca. 0.61 m diameter; U?) are preserved only on Piers i and k (Fig. V-36), numerous fragments were found along the row of piers and in the rubble fill north of the basin platform. Since no traces of

plaster were discovered, it is possible that the lively decorative effect of the materials employed was meant to be appreciated. The cistern (outside dimensions, 11.06 m χ 8.16 m), located on a gentle slope 18.0 m northwest of the Spring House, has high walls (ca. 3.50 m high on the east, 0.94 m thick; III?) built of the same ma­ terials as Piers f through k (Fig. V-38). Two but­ tresses resting on bedrock (ca. 8.48 m a.s.l.) support the eastern corners of the cistern but do not bond with the walls. The edges of these buttresses were revetted with roughly squared limestone blocks. Two further buttresses built up against the central part of the east wall have now fallen away, leaving only traces on the wall and the bedrock. Five beam holes open in the east wall at more or less regular intervals along its outer face 1.70 m below the upper edge but do not appear on the inside. The upper surface of the wall, level except for one upward pro­ jection (0.60 m high; II) toward the center, seems to survive to its original height (12.47 m a.s.l.), ap­ proximately 2.20 m (VIIS) above the floor of the cistern (10.27-10.57 m a.s.l.). Traces of lime plaster are visible at several points on the exterior. There is no evidence that the cistern was roofed, although it is at least possible that there was a wooden covering supported by beams. A square drain hole (0.20 m sq.) was built into the east wall of the cistern 1.70 m north of the southeast buttress, slightly below the floor level of the tank (Fig. V-39). Unfortunately, the limestone facing and part of the core have fallen away around this hole, destroying all evidence of its arrangement on the ex­ terior (the hole was reconstructed as a simple square opening soon after excavation in order to prevent further erosion of the fabric). A second opening (0.20 m sq.) was built into the adjacent south wall 1.20 m west of the southeast buttress. A rough con­ crete and rubble basin (ca. 1.50 m sq., 0.42 m deep) was built up against the corner, just below the spout, itself having a drain through its south wall formed by two cover tiles placed edge to edge to form an oval opening (Fig. V-40). The interior of the cistern (9.18 m χ 6.28 m; XXXI by XXI), which has a cracked but level floor, is lined up to the top of the walls with a layer of waterproof cocciopesto concrete (pozzolana mortar with ground-up potsherds) (Fig. V-41). The walls carry a single layer of this waterproofing material (0.02 m thick); the floor consists of two separate lay­ ers, the upper 0.21 m thick and the lower 0.17 m thick, possibly indicating the repair of leaks caused by the settling of the structure. In the southeast cor-

V: SPRING HOUSE COMPLEX

ner by the drains irregular holes have been cut into the lower layer of the floor to the level of the outlets. The vertical and horizontal seams along the cistern interior were buffered by a thick, rounded seal of the same cement that forms the rest of the lining (Fig. V-41). The cistern has a capacity of approximately 126,830 liters (33,506 gal.). A long trench was dug between Crosswall 2 and the southeast buttress of the cistern in order to check the intervening stratigraphy. No signs of further structures were found in this trench or in a second one excavated for 15.0 m west of Wall e, on line with Walls d and e (Fig. V-2). There is no surviving evidence for any channels leading from the drains in the southeast corner of the cistern.

SEQUENCE OF CONSTRUCTION, CHRONOLOGY, FUNCTION

In general, the topography of the Spring House site and the reuse of the structure during the Empire have precluded the formation of closed occupation deposits. The site slopes steeply down from west to east, attracting debris from further up the hill and subjecting the Spring House area to erosion (Figs. V-42, 43). The presence of flowing water in Room 2 and in the basin means that these two parts of the building could not have been allowed to accumulate debris. Conversely, the sandy mud fill around the basin platform seems to have been a marsh that swallowed up and protected all ceramic material that fell into it in an unstratified deposit providing dates from 125 B.C. to the second century A.D. Room 1, which was cleared out after the destruction of the Spring House by fire in the later second century A.D., and Room 2 contained homogeneous fill that yielded ceramic material dating from 200 B.C. to the third century A.D. Even the destruction level in Room 2, consisting of the charred remains of the bucket-chain and its drive machinery, yielded only two datable sherds: a Type 4a stamped Sestius amphora rim (Cat. A106) of 125-50 B.C. and a kitchenware sherd (Cat. K3) that possibly belongs in the first three centuries A.D. Most of the fill has been thoroughly mixed and redeposited from elsewhere: a joining fragment of the amphora rim found in the wood level of Room 2, for example, was recovered from the sandy fill below the wall rubble north of Wall b. Furthermore, three joining fragments of a Type 4b amphora'(Cat. A218) datable to ca. 75 to 0 B.C. were found in three different contexts: the fill in the basin at the south end of the cistern, the fill

105

below the wall rubble north of Wall b, and the fill above the floor in Room 1. A large proportion of the datable finds in every stratum in and around the Spring House (approximately 78 percent) are amphora fragments that can be dated from the late second through the end of the first century B. c. This apparently was the period of the most intensive use of the Spring House and associated fishery, and large dumps of discarded pottery from it must have remained around the site during the later activity in the Empire. These fragments were even used as building material for Piers f through k and the cistern and for a repair in the polygonal Wall P. Fortunately, despite this later mixing, the overall chronological distribution of the datable finds and the preservation of a few closely dated pieces in crucial strata yield a convincing picture of the construction and development of the Spring House. Furthermore, a close analysis of the structures themselves, including the formwork, the aggregate and mortar of the concrete, and the alterations in design, yields convincing evidence for the sequence of construction. This sequence can in turn be tied to absolute chronology by consideration of the datable material, comparable buildings on the Cosa acropolis, and information supplied by literary and historical sources. Two main stages in the development and functioning of the Spring House can be documented through an examination of the architecture. Both stages saw the use of a gear-driven bucketchain dipping into Room 2. During the first stage water was lifted 5 m to an aqueduct and carried eastward across the present channel on Walls d and e and piers 3-9 (Fig. V-48). During the second stage the water was lifted 13 m to a trough carried back to the upper lip of the cistern on Piers f through k (Fig. V53). Apparently no special arrangement was made for the overflow from the basin channel during either period, and it probably drained off into the lagoon area to the east, possibly serving to dilute the brackish water of the fish tanks. Perhaps, as Gazda suggests in chapter seven, there was a wooden trough, since destroyed, connecting the spring to the northern fish tank.

Stage 1 SEQUENCE OP CONSTRUCTION

During the initial period of construction the main lines of the Spring House were laid out in their entirety: the basin platform, along with piers 3-9, the polygonal walls and their foundations, Walls b and

106

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

c, Crosswalls 1 and 2, and Walls d and e. Calculations of the building dimensions in multiples of the Roman foot yield reasonable figures, although exact regularity and symmetry in the overall plan or its details are not marked features of the structure (Map 14). The basin platform in particular seems to have been laid out in irrational dimensions, perhaps because of the difficulty of working in the waterlogged soil. The outside dimensions of the building, apart from the platform, seem to have been 20 by 43 Roman feet. Room 1 has inside dimensions of 16 by 17 Roman feet; Room 2, 16 by 11; and the basin, 12 by 9. The walls vary in thickness from two to three and a half feet, and the doors may have been three feet wide. The piers on the basin platform are three feet square, like those in the lagoon (nos. 3-9), while Piers f through k are irregular. The interior dimensions of the cistern are 21 by 31 Roman feet, and the walls are 3 feet thick. The unity of the platform, piers, and polygonal walls is proven by the arrangement of the wooden forms, the homogeneity of building materials, and the coincidence of important levels in the structure. The use of nonpozzolana mortar in piers 1 and 2 and the correspondence in level between their upper surfaces and the top of the fill built into the opening in Wall c suggest that they may belong to a later stage. The contemporaneity of the rest of the Spring House is proven by the bonding of the walls, the similarity of building materials, and the unified character of the plan. It is probable that the original floor in Room 1 was higher than at present, closer to the threshold of the door in the south wall, perhaps level with the top of Crosswall 1 (4.24 m a.s.l.) or even with the foundation offset behind Crosswall 2 (4.55 m a.s.l.) (Fig. V-46). Even at this level some sort of step would have been necessary to bridge the 0.59 m (II) difference between the floor and threshold. The narrower door in the north wall consequently must be considered a later addition. The earth to the south of Wall a probably was close to the level of the present paving (4.75 m a.s.l.), and the ground level to the north of Wall b may have been the same, since the offset in this wall is at about the same height (4.82 m a.s.l.). The foundation of Crosswall 2 is slightly lower (4.55 m a.s.l.), but there is an undisturbed rubbish level just above, containing tiles, amphora sherds, and miscellaneous rubble. The pavement to the south of Wall a was not part of the original design, as evidenced by a thin, dark layer of burned earth or pozzolana sand mixed with fineware sherds that forms an unbroken, sloping level from beneath the pavement up to the north end of Wall e. In con-

sequence, Wall e, which formed part of the original design, was built before this level was laid down and thus preceded the pavement (Fig. V-35). Furthermore, the pavement and gutter evidently were designed to catch runoff (from the roof?), which had formed a ditch in the brown, sandy fill west of Crosswall 2 that was deposited above the construction and occupation levels at the base of this wall (Fig. V-33). Walls d and e, on the other hand, seem to have been part of the first stage of construction: they are composed of materials closely similar to those of the limestone sections of Walls a and b; they line up with piers 3 through 9 and the top of Wall d (5.09 m a.s.l.), along with the offset on the west side of Wall e; and they are at approximately the same level as the threshold in Wall a (5.14 m a.s.l.). Finally, the upper polygonal wall was built up against the south end of Wall d in much the same manner as the lower polygonal wall abuts Walls a and b. CHRONOLOGY

The majority (ca. 78 percent) of the amphora sherds that carpeted the Spring House site, the rest of the lagoon, and the port area belong to Types 4a (dated from ca. 125 to 50 B.C.) and 4b (dated from ca. 75 to 0 B.C.). Type 4a predominates over 4b by 4:1. Approximately 15 percent of the amphoras found date before the middle of the second century B.C., and 7 percent are Imperial types. These statistics suggest a period of intense trade and prosperity beginning in the late second or early first century, as discussed in chapters nine and eighteen. This would have been an appropriate moment for the inauguration of large-scale landscaping and construction projects in the lagoon area, since it coincides with the well-documented flowering of the settlement on the acropolis. The amphora fragments used as building material in the original portions of the Spring House all appear to be from Type 4a shapes, reaffirming the appropriate terminus post quern: Cats. A161-A163 in Wall e, Cat. A182 in Wall a. The spread of Type 4a amphora from the late second through as late as the mid-first century B.C. is narrowed by the presence of a ring handle decorated with a satyr head from a thin walled cup of the Sovana class (Cat. TW23) found in the fill behind the polygonal wall just south of its intersection with Wall a. It cannot be later than the construction of the wall, which is contemporary with the rest of the Spring House proper. A reasonable date would be about the end of the first quarter of the first century B.C., a date reinforced by the rest of the thin walled ware found around the site in mixed contexts. The

V: SPRING HOUSE COMPLEX

type of early situla very popular on the hill-top settlement of Cosa in the second century B.C. does not appear, and 70 percent of the material discovered dates from about 75 B.C. to the early Augustan period. The remaining 30 percent is Imperial. Finally, radiocarbon analysis was also made on several fragments of wood from the cofferdam around the basin platform, as well as on fragments of the wooden water-lifting device in Room 2. One date for the cofferdam is untenable—one must assume some contamination of the material—but two others fall into the time span under discussion. Of particular significance is the one MASCA-corrected date of 70 B.C., which may be used as corroborative scientific evidence for the dating of the original construction of the platform and the beginning of stage 1 in the history of the Spring House. 5 The utilitarian ware found around the Spring House is more difficult to date with any exactness and consequently less useful, but the statistics resemble those provided by the amphoras. Of this material, 66 percent dates between the third and the end of the first century B.C., and 42 percent between 125 and 25 B.C. Only 24 percent seems to date to the Empire. Three catalogued pieces of black glaze pottery were found around the Spring House (Cats. BGl, BG4, BG5), and they reinforce the same pattern: all the material belongs to the second and first centuries B.C. The presence of opus quasi-reticulatum facing in Wall c and Room 1 also suggests that the structure was built in the first half of the first century B.C. Close comparisons can be made with the facing of the lupanar in the Roman forum, which Lugli dates to the first half of the first century B.C., and with that of the Casa dei Grifi, which can be dated ca. 100 B.C.6 Blake makes a case for connecting opus quasireticulatum with Sullan construction in Pompeii and elsewhere in central Italy.7 Reticulate and quasi-reticulate work does not appear on the acropolis of Cosa, both because most of the buildings predate its development and because the limestone rubble used as a facing material was too intractable to allow production of the proper blocks. The presence of such tuff facing in the Spring House points out once again the careful selection of building materials and mor5 B. Lawn, Radiocarbon 15, no. 2 (1973) p. 368, no. P-1722 from the formwork of the Spring House main platform (70 B.C.); Lawn, Radiocarbon 16, no. 2 (1974) p. 220, no. P-1933 from the formwork of the platform in Trench IA, south of pier 3 (17101520 B.C.); sample 1-3968 from the formwork of the mam platform (ca. A.D. 25), by letter from J. Buckley, Isotopes, Inc., Westwood Laboratories, Westwood, N J . , Jan. 17, 1969. For reports on dates obtained for the water-lifting device, see n. 21

107

tars in this sophisticated structure. Furthermore, the limestone rubblework with reinforced corners forming the rest of the structure is similar to the fabric of the substructures of the Temple of Jupiter Anxur at Tarracina, which belongs to the late second or early first century B.C., and to the contemporary structures in the town of Cosa. 8 There is thus ample evidence for placing the construction of the Spring House and the polygonal Wall P late in the first quarter of the first century B.C. Furthermore, the drastic reduction in finds belonging to the period from about 10 B.C to A.D. 50 and the total absence of some closely dated wares suggest that the structure went out of use at the end of the same century. Only one amphora of Type 14 (Cat. A270), which may bridge the first centuries B.C. and A.D. (although it continues into the third quarter of the first century A.D.), was found, along with one fragment of Type l i e (Cat. A259) from the first half of the first century A.D. A piece of a glass bowl (Cat. Gl) may be dated from the late first century B.C. to the first half of the first century A.D. Most significant perhaps is the clear hiatus in the thin walled ware after about 10 B.C. Black glaze, which is very sparse in any case on the whole site, ceased about 30 B.C. One piece of early Arretine ware from the Spring House can be dated about 20 B.C. (Cat. RGl), and one fragment of undecorated sigillata (Cat. RG2) probably dates before the end of the first century B.C. Then the red-gloss wares disappear from the Spring House finds until the undecorated Late Italian sigillata pieces dated in the second half of the first century A.D. (Cats. RG3-RG5). The chronology of Imperial occupation of the Spring House will be discussed below in the section on stage 2. RECONSTRUCTION A N D FUNCTION

(Figs. V-42 to 48) Any reconstruction of the Spring House must account for all the structural elements outlined above, the landscaping of the lagoon area by means of the two polygonal walls, and the fish tanks that were built at the same time a short distance to the east. A number of striking peculiarities set this Spring House apart from the other preserved Greek and Robelow. 6 Lugli, II, pi. 133, no. 1 (lupanar), no. 2 (Casa dei Grifi). 7 Blake 1947, pp. 229-232. Also see below, Gazda, chapter seven, n. 82, and McCann, chapter eighteen, n. 35. 8 Brown 1951, pp. 109-111. A. Boethius dates the sanctuary at Tarracina to the Sullan period; Boethius and Ward-Perkins, p. 146.

108

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

man fountain structures. 9 The design of the platform is unique, and the preservation of the wooden forms around it is surprising. The water is not siphoned off in a manner convenient to household use: the spout provides too strong a flow for domestic utensils, and the basin has no parapet to facilitate dipping. No apparent provision was made for access to the spring from the west, and even access from the east at present involves the negotiation of pilings and formwork or of the step below the spout, which is not only 0.70 m below the upper level but also constantly awash with water.10 Furthermore, the extensive deposits of commonware pottery to be expected at an important spring were not discovered, and the platform shows little sign of wear, if use were focused on the outer basin, the existence of Room 2— a second, apparently inaccessible collecting basin—is puzzling, since Wall c then becomes a superfluous barrier, and the addition of an upper room (Room 1) is unparalleled. Finally, the alignment of Walls d and e with piers 3-9 must somehow be related to the overall Spring House function. The most satisfactory way of accounting for all these elements together is to postulate that the Spring House was designed from the start to accommodate a gear-driven bucket-chain in Room 2: a machine designed to lift water to the level of a small reservoir on Walls d and e, where it was collected for transmission across the lagoon area on an aqueduct supported by piers 3-9 (Figs. V-45, 48). This supply of fresh water probably was carried to some point in the harbor area, perhaps near the site presently occupied by the Guardia di Finanza (customs office) building, whose construction exposed a large number of Sestius amphora sherds. It has been sug9 A complete list of comparative material would include nearly all the excavated sites of the Greek and Roman world, but the most important discussions with specific examples are B. Dunkley, "Greek Fountain Buildings before 300 B.C.," BSA 36 (19351936) pp. 142-204, pi. 21-25; O. Elia, "Fontana," Encidopedia Italiana 15 (1932) pp. 633-638; F. Glaser, "Antike Brunnenbauten (krenat) in Griechenland" (diss., Univ. of Vienna, 1976); J. A. HiId, "Fons," DarSag II.2 (1896) pp. 1227-1237; B. H. HiU, Corinth 1.6: The Springs: Peirene, Sacred Spring, Glauke (Princeton, 1964); S. Meschini, "Ninfei e fontaine," EAA 5 (1963) pp. 505512; N. Neuerberg, L'architettura delle fontane e dei ninfei nell'Italia antica, Accademia di archeologia, lettere e belle arti, Naples, Memorie, n.s. 5 (1965); A. K. Orlandos, "Parastaseis krenon epi angeion," ArchEph (1916) pp. 94-107; T. Wiegand, "Zur Entwicklung der antiken Brunnenarchitektur," m Milet 1.5: Das Nymphaeum, ed. J. Hulsen (Berlin and Leipzig, 1919) pp. 73-88; R. E. Wycherly, How the Greeks Built Cities, 2d ed. (London, 1962) pp. 198-209. See also A. Burns, "Ancient Greek Water Supply and City Planning: A Study of Syracuse and Acragas," TechCult 15 (1974) pp. 389-412; R. Cagnat and V. Chapot, Manuel d'archiologie romaine (Paris, 1916-1920) pp. 102-107; A. Gre-

gested by Will that the Sestius pottery was located on this spot. The location may also have been used by the fishery. In view of the considerable distance between these piers, the absence of arches to span them, and the lack of any evidence for tall pillars to hold the channel at the level of the tank on Walls d and e, it seems likely that this water was carried to the harbor through an inverted siphon formed of lead pipes supported between the piers by wooden planks. The height of the siphon entrance is just sufficient to carry water there, and the piers would have been necessary to carry and protect the conduit over the marshy ground. The whole arrangement corresponds to Vitruvius' instructions concerning inverted siphons." A narrow walkway may have been added as well. According to this interpretation of the Spring House, the basin platform was essentially a barrier to catch the spring water, to raise it to a sufficient height for the functioning of the bucket-chain in Room 2, and to keep it segregated from the surrounding brackish ground water. The absence of identifiable elements from the water-lifting device used in the initial stage of the Spring House development makes it impossible to restore any of the machinery from that period. Nevertheless, the possibility of any marked change in the tradition of applied technology at Cosa during the period of Spring House use is remote, and the essential similarity in floor plan between stages 1 (Fig. V-46) and 2 (Fig. V-51) of the structure suggests that there were no new developments in machinery design or the method of propulsion. The geared driving machinery of stage 1 probably differed from that of stage 2 (see below) only in the height of its lift (ca. 5.0 m as mer, Manuel d'archiologie gallo-romaine, IV: Les monuments des eaux (Paris, 1960); M. Lang, Waterworks in the Athenian Agora, Excavations in the Athenian Agora, Picture Book no. 10 (Princeton, 1968); H. Mygind, "Die Wasserversorgung Pompejis," Janus 22 (1917) pp. 315-316; J. R. Smith, Springs and Wells in Greek and Roman Literature (New York and London, 1922); H. A. Thompson and R. E. Wycherly, The Agora of Athens. The Athenian Agora XIV (Princeton, 1972) pp. 197-203; A. W. Van Buren, "Bewasserungsanlagen," RE, supp. 8 (1956) pp. 9-16. ,0 It is interesting to compare this arrangement with Wycherly's lucid summation of the purposes of a conventional fountain building: "to catch and conserve the water, to make it easily accessible for public use, and to keep it cool and clean." Wycherly, Cities (cited in n. 9) p. 200. The Cosa Spring House fulfills the first and last requirements but fails in the second—-at least in the absence of its machinery. 11 Vitruvius 8.6.5; cf. Faventinus 6, Palladius Ag. 9.11.2. There is a good discussion of siphon technology in N. A. Smith, "Attitudes to Roman Engineering and the Question of the Inverted Siphon," History of Technology 1 (1976) pp. 45-71.

V: SPRING HOUSE COMPLEX

109

opposed to 13.0 m) and in the direction in which the water was channeled. The bucket-chain wheel would have been housed above Room 2, probably on beams and floors supported by the four pairs of beam holes in Walls a and b, while the harness bar and pair of angle gears would have been located in Room 1 (Fig. V-44). The drive system in stage 1 might have used a draft animal without too much difficulty, but in stage 2 the problem of access caused by the elevation of the drive area well above ground level would have required human labor to turn the machine. The opening in Wall c, originally arched with tuff voussoirs, probably served as a door providing access to the basin platform from the slopes above by means of stairs supported by the beams and floors in Room 2 (Fig. V-47). In this case, the basin would have been covered over, probably with a simple board platform resting on the ledges running across the north and south sides a few centimeters above water level. A few wooden steps within the thickness of Wall c could have bridged the 1.0 m difference in level between the lower floor in Room 1 and the basin cover. Aside from the Spring House, there is no break in the 180 m stretch of the lower polygonal Wall P that could have been used to facilitate access from the lower slopes of the acropolis to the lagoon area during stage 1. This reconstruction still leaves partially unexplained the shape of the basin platform and the presence on it of piers 1 and 2. After the 1969 season, before completion of the excavations in the Spring House itself or those in the lagoon to the east of it, the researchers suggested that this platform had served as a watering dock. According to this interpretation, small boats could have tied up to the platform—their maneuvers or mooring facilitated by its peculiar design—filled barrels or short amphoras with fresh water from the spout, and then conveyed the containers to ships moored either in a hypothetical inner harbor or the outer harbor. The upright row of planks along the south side of the platform was interpreted as the remains of a double cofferdam, of a type described by Vitruvius (5.12), which could be pumped out to provide a dry working area below sea level.12

Further study has now revealed that the Spring House was rather part of a large-scale fish-farming project and that the water pouring off the basin probably was intended for fish ponds in the lagoon and for an emporium somewhere to the east of Wall PE. From this commercial area, ships could also have been serviced with fresh water. It is, in any case, highly unlikely that the immediate area of the platform was accessible by boat either at the time of its construction or afterward. In the first place, the abrasion of even small boats against the wooden formwork would soon have resulted in considerable wear, especially to the round poles. Many of these, however, not only are undamaged but even retain their bark. Second, the upright poles in front of the spout would have restricted access to the lower step, which would have been the natural place for loading. In addition, the platform has none of the stone mooring rings typical of Roman docks, and piers 1 and 2 are not suitably designed for such a function. Finally, the wall of planks on the south end of the platform cannot have formed the outer shell of a double, water-tight cofferdam, because there are no horizontal braces (the catenae of Vitruvius?) to lend the necessary support, and there is no sign of any interior packing, such as the baskets of clay that Vitruvius instructs should be rammed into the gap to waterproof the cofferdam (5.12.5). In addition, Vitruvius specifically proposes the use of such a waterproof cofferdam for situations in which hydraulic cement—such as that used in the platform—was not available. It is also possible to object that the presence of piers 3-9 would have hampered access to the platform by water from the direction of the outer harbor. It is more likely that at least the immediate area of the basin platform was surrounded by mud or waterlogged earth rather than by a natural or manmade lagoon. In this situation the original discovery of the spring would have been relatively simple, since a marshy pool or more verdant surface vegetation would have marked the spot—a divining technique mentioned by Vitruvius and other Roman writers. 13 A few impasto sherds discovered in the fill around the base of Crosswall 2 suggest that the spring may have been in use as early as the eighth

12 See McCann and Lewis 1970, pp. 200-211. On the passage in Vitruvius, see A. Schramm, "Der Molenbau in Vitruvs Architectural Philohgiscke Wochenschrifi 49/50 (1936) pp. 1404-1408; E. jungst and P. Thielscher, "Vitruv iiber Baugrube, Baugrund, und Grundbau," RomMitt 51 (1936) pp. 145-180; A. Schramm, "Der Molenbau in Vitruvs Architectura V. 12, 3-7," RomMitt 53 (1938) pp. 46-49; and Oleson, StHarbArch I, forthcoming.

13 Vitruvius 8.1.3; Palladius Ag. 9.8.4. Faventinus 3 and Pliny NH 31.27.44 are suspicious of dependence on surface vegetation alone. L. Callebat, "Le vocabulaire de l'hydraulique dans Ie livre VHI du 'De architectura' de Vitruve," RevPhil 48 (1974) pp. 313329, asserts that Vitruvius' careful choice of terminology in this section justifies confidence in his experience and technical competence. Isogonus Nicaeensis mentions a marvelous spring near

110

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

or seventh century B.C. There is one Type la am­ phora (Cat. A2) from the area, dating to the time of Cosa's founding (300-275 B.C.), and the twenty-one Type Ic and Id amphoras possibly indicate a sudden interest in the spring in the first half of the second century B.C. The Fonte de la Reina, which attracted the attention of Saracen pirates in the sixteenth cen­ tury, long after the destruction of the Spring House structure and the obstruction of the basin, probably bubbled up somewhere in this area. The choice of this particular water source over the others around the lagoon can be explained not only by its copious flow and location but also possibly by the originally high quality of its water. The site cer­ tainly fulfills several criteria proposed by Vitruvius: "sub radicibus autem montium et in saxis silicibus uberiores [copiae] et affluentiores; eaeque frigidiores sunt et salubriores [but (springs) at the feet of moun­ tains and in flinty rocks are more copious and abun­ dant, and they are cooler and more salubrious]." 14 The similarity to the sort of spring Hippocrates re­ gards as best is even more remarkable, since the Spring House does face very nearly due east:

The major problem facing the engineer who wished to harness the spring at Cosa was the con­ struction of a basin that, fed from the bottom, would keep the water clean and provide sufficient depth for the functioning of the bucket-chain. Since the spring water could not be stopped or diverted during construction, it must soon have proven nec­ essary to construct a wall of planks to retain the soft sides of the excavation made for the foundations of the platform, a wall probably very similar in pur­ pose to the retaining wall built by the modern ex­ cavators. This support may have been necessary only on the south, but it seems more likely that the original excavation was completely surrounded by a timber wall later removed on the north and east sides to permit freer escape of the overflow from the basin and possibly also to allow access to the plat­ form on foot from these directions. The reason for the survival of the southern section is obscure, but its presence suggests at least that access from that direction on foot was not necessary.

The best are those that flow from high places and earthy hills. By themselves they are sweet and clear. . . . In Winter they are warm, in Summer cold. They would naturally be so, coming from very deep springs. I commend especially those whose flow breaks forth toward the rising—by preference the Summer rising—of the sun. For 15 they must be brighter, sweet-smelling, and light.

A simple barrier wall closing the east side of the basin probably would have dammed the spring water with the same efficiency as the massive plat­ form, but a broader construction may have been necessary to counteract shifting of the foundation and to guard against polluting and erosive leaks— inward and outward—from beneath the basin wall. Furthermore, in view of the understandable abhor­ rence of reeds, vegetation, and mud near a spring expressed by Roman authors on water supply, 16 it may also have seemed profitable to the engineers to carry the overflow a short distance from the source and to provide a firm, dry area between. 17 Natu­ rally, the spout could have been used to fill con­ tainers to be carried elsewhere in the harbor area, but the jet is too strong for small containers and its ele­ vation too low to allow barrels or amphoras to be placed beneath it. More likely, the step below the spout was meant primarily to receive the force of the

Cosa that turned sweet wine sour {FHG, IV, p. 437, no. 14). Experiments with the Spring House water failed to duplicate this feat, and Cardarelli 1925, p. 22, notes that there are no postclassical records of such a spring in jhe area. 14 Vitruvius 8.1.2, ed. F. Krohn (Leipzig, 1912) p. 172, Cf. Faventinus 3; Palladius Ag. 9.8.3; Pliny NH 31.28.48. 15 Hippocrates, Περί Ά ε ρ ώ ν 'Υδάτων Τόπων 7, 58-65, ed. W.H.S. Jones (Cambridge, Mass., 1923) I, pp. 86-87; see also Hippocrates 5.11-14. The judgment is noted in Athenaeus Deip. 2.46 c-d. 16 Vitruvius 8.4.2; Faventinus 5. Cf. Pliny NH 31.22.36-7; Vegetius Epitoma ret militaris 3.2; and Frontinus Aq. 2.90. 17 As they now stand, the Spring House basin and Room 2 are commendably close to Pliny's prescription for the arrangement of a wholesome water source: "ex quonam ergo genere maxime

probabilis continget? puteis nimirum, ut in oppidis video constare, sed iis, quibus et exercitatioms ratio crebro haustu continget et ilia tenuitas colante terra, salubritati haec satis sunt; frigori et opacitas necessana utque caelum videant. super omnia una observatio—eadem et ad perenmtatem pertinet—,ut ilia e vado exiliat vena, non e lateribus" (JVH 31.23.38-39, ed L. Ian, C. Mayhoff [Leipzig, 1897] V, pp. 14-15). ["From which type of source, then is the most commendable water obtained. From wells, of course, as I see them arranged in towns, but from those which through frequent dipping have the recommendation of circulation and that clarity produced by filtration through the earth. This is enough for wholesomeness. For coolness, both shade is necessary and that the well be open to the sky. Above all, one observation—and the same is relevant to a continual flow: the spring should issue from the bottom of the well, not its sides" (trans. J. P. Oleson).]

"Αριστα δε όκόσα εκ μετεώρων χωρίων ρεΐ και λόφων γεηρών. Αυτά τε γάρ έστι γλυκέα και λευκά. . . . Τοΰ δε χειμώνος θερμά γίνεται, τοΰ δε θέρεος ψυχρά. Οΰτω γάρ αν εϊη εκ βαθύτατων πηγέων. Μάλιστα δε έπαινέω ών τα ρεύματα προς τάς ανατολάς τοΰ ηλίου έρρώγασι και μάλλον προς τάς θερινάς. 'Ανάγκη γάρ λαμπρότερα είναι καΐ εύώδεα και κοϋφα.

V: SPRING HOUSE COMPLEX

water and prevent the formation of a deep muddy hole. There is no evidence that any control of the overflow was attempted once it ran off the step, but it may well have been directed toward one of the nearby fish tanks. The wood around the rest of the platform must have soon become waterlogged and then covered over with mud and dirt, since it shows few signs of weathering and decay. The preservation of the wood places the ambient ground level in antiquity at approximately the present sea level. According to this interpretation, the arms projecting from the north and south ends of the platform could have served as stabilizing buttresses on the unsteady foundation. There is, however, no apparent reason for the asymmetrical design of the northern arm. Since slight irregularities in the rectangular southern arm suggest that only approximate accuracy was necessary in the execution of the platform design, the same casual attitude may have been compounded at the north end by the lack of a wooden retaining wall at that point. It is also possible that buried obstructions or irregularities in the bedrock hindered complete execution of the northern arm by affecting placement of the pilings supporting the formwork. It is possible that the broad platform was furnished specifically to provide a firm foundation for piers 1 and 2, but their purpose unfortunately remains obscure. They do not line up with pier 3 or with Walls a and b, and the presence of sloping buttresses on either side of the basin makes the restoration of a porch whose forward supports rested on the piers awkward. The utilitarian character of the Spring House makes their interpretation as statue bases equally unlikely, but the slightly flared upper surfaces suggest that they served as supports of some sort. Perhaps they supported a table where fish could be gutted and prepared for pickling and garum. Slight differences in the composition of the mortar in the basin platform as compared to that in the piers (which lacks pozzolana) suggest that piers 1 and 2 may belong to a second construction stage. It is possible that they were added to the platform at the same time the opening in Wall c was filled in, since the upper level of this fill (1.37 m a.s.l.) is close to that of the upper surfaces of the piers (1.26 m a.s.l.). During the excavation of Room 2 this coincidence of level facilitated the construction of a bridge of planks leading from the dry ground on the east to the opening in Wall c, supported by beams laid on piers 1 and 2 and in the opening itself (Fig. V-2). It is at least possible that the piers were constructed during stage 2 as supports for a platform higher than

111

the board floor restored for stage 1, which rested directly on the basin lip.

Stage 2 SEQUENCE OF CONSTRUCTION

The evidence for a second stage of construction involves Room 2, Piers f through k, and the cistern. Although it is obvious that a major shift has taken place in the areas served by the Spring House, there is no evidence of any substantial structural damage. A thin layer of gray earth below the pozzolana layer south of Wall a may consist of ash. It extends to the south of Wall a as far as Wall e and was laid down before the construction of paving at that point (Fig. V-32). It consequently precedes stage 2, and, significantly, it contains no material that can be dated after the end of the first century B.C. It certainly does not have the appearance of a destruction level, however, and there is no evidence of fire elsewhere on the site before the end of stage 2. The most striking developments concern the door in the south wall of Room 1. Its heavy limestone threshold (5.14 m a.s.l.) was broken and turned, then left projecting from the wall beside a new door with a lower threshold (4.48 m a.s.l.) approached by a brick-topped step built on the protruding foundation of Wall a at .this point (Fig. V-29). These changes probably were connected with a lowering of the floor level inside the room, which until then may have been even with the top of Crosswall 1 (4.25 m a.s.l.) or with the foundation offset behind Crosswall 2 (4.55 m a.s.l.). Since the ground to the south of the structure consequently became somewhat higher than the floor of the room, it is probable that the pavement and gutter south of the wall were laid out at this time or soon afterwards to landscape the area by controlling the earth and the surface water (Fig. V-32). The floor of Room 1 may have been paved with cement at the same time, and the foundation offset on the east side of Crosswall 2, which originally flared outward on the east side as it still does on the west, was chiseled down to a rounded protuberance. The alteration in floor level (Fig. V-51) seems to have been carried out in order to allow the opening of a narrow door in the northwest corner of the room with a threshold (3.80 m a.s.l.) presumably even with the ground level outside at that point, since no steps were found and a thin rubble paving was laid out between the door and the southeast corner of Pier f (Fig. V-30). The arrangement of

112

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

the door in the south wall evidently became inconvenient or superfluous, for it was walled up with yellow tuff set in cement (reused materials?), leaving only a square hole (0.50 m sq.) through the thickness of the fill above the threshold block. After all these arrangements had been completed, the interior of the entire room was plastered. The rough edges around the door filling were first chinked with amphora body sherds set in mud, and the hole over the threshold block was packed with the same material. The plaster coating along the south side of the building probably was applied at the same time, certainly after the door had been closed, since the layer is not disturbed at this point (Fig. V-31). Although hard and tenacious, this plaster is not of the waterproof variety used in cisterns. It is possible that the lower part of the door in Wall c was walled up simultaneously, perhaps in connection with the construction of piers 1 and 2 (see above). Piers f through k and the cistern, which were constructed specifically for this renovation of the Spring House, constitute the major structural remains of the second stage (Fig. V-36). Since no destruction debris was found inside Room 1, it is apparent that the room was cleared after the fire that destroyed the second water-lifting apparatus, probably to adapt the building to serve some other purpose—possibly storage rather than habitation, in view of the good condition of the stucco. Crosswall la, a crumbly rubble wall unlike the other portions of the Spring House, was built up against the west face of Crosswall 1 at this time, partially obstructing the opening of the new door in Wall b (Fig. V-25). This clumsy barrier has the appearance of a partition wall meant to close off the broad opening between Rooms 1 and 2 above the level of Crosswall 1 in order to give Room 1 an eastern wall. It is surprising that the portions of the wall and floor of Room 1 that were uncovered show no signs of damage from the fire that destroyed the Spring House or from falling timbers, but the absence of damage can possibly be explained by the concentration of the fire in Room 2 and by the durability of the plaster. In any case, since the interior plaster is identical to that on the outside of the structure, it is unlikely that any hypothetical renovation and reuse of Room 1 after the fire would have re18 Brown 1951, p. 90, n. 74. He refers to Venerosi-Pesciolini, pp. 15-28. The spring is mentioned on p. 18: "il Monte della Tagliata della citta di Ansedoma dove sono molte belle vestigia di detta citta et al piede di detto monte vicino al mare uno tiro di schioppo et alia detta Tagliata sta una fonte viva dove Ie galeotte Turchesche vanno a fare acqua, detta la Fonte de la Reina. . . ."

quired the plastering of the south face of Wall a. When the Spring House finally went out of use, Rooms 1 and 2 gradually filled up with the mixed debris covering the whole hillside area: the bulk of the datable material belongs to the stage 1 period of use, but several sherds of the first through third centuries A. D. do appear. CHRONOLOGY: DATE OF DESTRUCTION

In the historical and topographical study of Cosa published in 1951, F. E. Brown suggested that the cistern and Piers h through k—the only portion of the Spring House visible at that time—might be Medieval in date, perhaps related to the Fonte de la Reina mentioned in a sixteenth-century document. 18 Now that the whole complex has been excavated, the mode of construction, the materials employed, and the general plan prove that the cistern is contemporary with the piers and that the piers are clearly related to the Spring House, which contained no material later than the third or fourth century A.D. Furthermore, the rubble level resulting from the fall of the upper part of the north wall of Room 1 forms an unbroken line between Piers f and g, indicating that they were built before the ruin of the Spring House. In addition, only two Medieval sherds were found in this area (Inv. no. PC 72-176), mixed in with the surface fill near the cistern. Since there is no evidence of an upper spring that could have filled the cistern, it must have remained dry after the destruction of the bucket-chain of stage 2. The preservation of the concrete basin and platform and of the wooden forms around them show that they cannot have remained visible to the sixteenth century. Furthermore, only one postclassical sherd was found in this area, in the surface rubble over the basin. The location of the spring, however, fits that of the Fonte de la Reina mentioned in the sixteenth-century document (a "gunshot distant from the Tagliata"), and the stubs of the Spring House walls, piers, and cistern visible in the hillside could easily have spawned the name. Consequently, the spring most likely found escape at this time in a natural pool outside the Spring House proper. The subsequent draining of the lagoon area would have caused this surface flow to disappear by lowering the local water level. Unfortunately, it has not been possible to deter["the Tagliata Hill of the City of Ansedonia, where there are many beautiful traces of that city, and at the foot of that hill, a gunshot distant from the sea, by the so-called Tagliata stands a flowing spring, where the Turkish galleys go to take on water, called the Spring of the Queen. . . ."] Also see the discussion by McCann, chapter three, and by Lewis, chapter seven, Appendix.

V: SPRING HOUSE COMPLEX

113

mine from the cultural remains within the Spring House the exact date at which the second stage of reconstruction was carried out. However, the probable association of this reuse with a nearby villa suggests the early Empire, probably the second half of the first century A.D.19 The use of alternating courses of stone and brick in the fallen piece of wall found between Wall a and Walls d and e, which may belong to the remodeling of the Spring House in stage 2, resembles the Neronian (A.D. 54-68) renovations in the basilica in the forum at Cosa and supports this date (Fig. VII-8).20 The remarkable construction technique of the columns above Piers f through k cannot itself be dated, but the incorporation of large numbers of amphora toes suggests the creative reuse of occupation debris accumulated over a long period of time. The same technique, using amphora handles as well as toes, appears in a repair of the lower polygonal wall north of the Spring House (in Trench YZ). The dilapidation of this substantial wall (P) again indicates—although in an indefinite way—the passage of time. Finally, the area directly served by the cistern and Spring House in stage 2 is no longer the flat land around the harbor but the lowest slopes of the acropolis just to the west of the polygonal Wall PW, a situation that reflects the slower pace of commercial harbor use during the Empire. Ceramic material that can be dated accurately to the first century A.D. is present, though scantily, in the Spring House. Totally lacking is the fine decorated Arretine pottery found in abundance on the hill site of Cosa. The thin walled ware, so frequent at the Spring House in the first century B.C., virtually ceases by the end of this century. The metallic wares of the mature Augustan period and the orangecoated vessels so popular in the hill-top city in the Tiberian period are also completely absent. Three pieces of gray ware may be as early as the Tiberian period (Cats. TW31-TW33), but the Claudian-Neronian period is surely represented by one piece of coated ware (Cat. TW34). A fragment of a blown glass bowl (Cat. G3) may be dated after the third quarter of the first century A.D., and one piece of undecorated Late Italian sigillata is placed in the second half of the first century A. D. (Cat. RG3) and another (Cat. RG4) in the fourth quarter of the first century or early second century A.D. The remaining material—glass, lamps, amphoras, and utilitarian wares—have sliding dates that can be later than the

first century A.D. Although it must be remembered that none of this material comes from sealed deposits, it is clear that interest in the immediate area of the Spring House was renewed sometime during the second half of the first century A.D. and most probably in the last quarter of the century, after a period of neglect of over fifty years. The date of the destruction of the Spring House by fire and of its final abandonment likewise is uncertain, but once again the chronological emphasis of the dated finds is clear. The actual destruction level in Room 2, containing the burned wood of the stage 2 bucket-chain, produced only one Type 4a amphora sherd (Cat. A106) dating ca. 125-50 B.C. and a utilitarian ware sherd (Cat. K3) datable to the first three centuries A.D. The amphora fragments from the white clay layer immediately above belong to the late second or early first century B.C., while the rest of the fill produced a large number of Type 4a amphora fragments along with a single Type la sherd (Cat. A2), one of the earliest pieces from the whole site. It dates to the late fourth century B.C. or first quarter of the third century B.C. Utilitarian ware dating from the third century B.C. to A.D. 50 was also found in this layer. Along with the usual Republican pottery, the fill in Room 1 produced one Type 21a amphora sherd (Cat. A290) of the third century A.D. with a floruit of ca. 250-275. The sandy mud fill over and around the basin platform, where finer vessels were in use and were preserved after being broken, produced the most coherent evidence for stage 2 chronology. Level IV, around and below the platform, yielded fragments of two decorated Late Italian sigillata bowls: Cat. RGIl, dated in the first three decades of the second century, and Cat. RG12, the latest and most complete example, assigned to the end of the production of decorated Late Italian sigillata in the second quarter of the second century A.D. A lamp from the same context dates between A.D. 90 and 140 (Cat. Ll). Along with some early types, the sandy mud fill on the basin platform and below also produced amphora sherds of Type 12b (Cat. A263) of the first and second centuries A.D., Type 18b (Cat. A284) from the second and third centuries A.D., and Type 21c (Cat. A294) of the third century A.D. There were also two pieces of utilitarian ware dated between the first and third centuries A.D. (Cats. K2, K4). One fragment of African Red Slip Ware (Cat. RG16) was

19 See Gazda and McCann, chapter seven, and McCann, chapter thirteen, on the date of the harbor villas. Also see P. Raveggi, NSc 3 (1927) pp. 204-210; Brown 1951, p. 20; Dyson JFA, p. 261. Manacorda 1978, pp. 130-131, notes that the harbor villa

seems to be a holiday rather than a working villa. 20 Brown 1951, pp. 78 (fig. 73), 112. Cf. Blake 1959, pis. 3.4, 14.3.

114

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

found, dating between A.D. 100 and 160, but the later forms of this ware popular in the late second and third centuries were absent. A fragment of a glass prismatic bottle (Cat. G7) is dated from the Flavian period through the second century A.D. Within the Spring House water basin itself, one lamp was recovered dating again between A. D. 80 and 150 (Cat. L2) and two more pieces of African Red Slip ware ranging in date from A.D. 100 to 150 (Cats. RG13-RG14). The results of the radiocarbon dating of two fragments of the wooden buckets from the water-lifting device unfortunately did not prove useful in dating the structure. One bucket yielded a MASCA-corrected date of 840-740 B.C. (lab. no. P-1931); another tested 100-10 B.C. (lab. no. P-2435).21 The latest materials from the Spring House site are two amphora fragments from the third century A.D.: Type 21a (Cat. A290) from Room 1 and Type 21c (Cat. A294) from the basin platform. Further documentation for some activity in the third century is a lamp (Cat. L8) from the rubble level north of Wall b, a type dated in either the third or fourth century. Once again, this material is not from a unified, closed deposit, but it clearly clusters in date around an upper limit of the midsecond century. In conclusion, it is possible to suggest that the basin cannot have filled up with mud and earth and the upper walls of Room 1 cannot have fallen before the mid-second century A.D. and probably not long after. Both incidents, which would have made restoration of the bucket-chain impossible, probably followed close upon the destruction of Room 2 by fire. The upper chronological limit provided by the two third-century amphora fragments (Cats. A290, A294) and the lamp from the third or fourth century A.D. (Cat. L8) suggests that some limited activity continued in the area after this disaster, as corroborated by the finds from the adjacent area YZ, but that the machinery in the Spring House at least was not in use. The whole port area was in decline by the third century. 22 Furthermore, if the geared water-lifting device had been in operation at such a late date, it would seem likely that the cumbersome wooden boxes of the bucket-chain might have been replaced by the newly invented earthenware pots especially designed for such machines (discussed below in the section on the history of water-lifting technology). The presence of a significant number 21 Radiocarbon 16, no. 2 (1971) p. 220, no. P-1931. No. P-2435 (100-10 B.C.) by letter from E. Ralph, June 16, 1976. 22 See Brown 1951, p. 21, and Dyson JFA, p. 260. 23 The three bucket-chains published in Pemp had a lift of ca.

of unripened pine cones throughout the wood level suggests that the final destruction may have taken place in the spring. RECONSTRUCTION AND FUNCTION

(Figs. V-42, 43, 49 to 53) Although the remarkable arrangement of the platform, basin, and Room 2 remained unchanged in stage 2 (with the possible exceptions of the minor alteration in the level of the opening in Wall c and construction of the basin platform piers), the focus of Spring House activity obviously shifted to the new piers and cistern on the north and west. The obstruction of the only door in the south wall reinforces this impression, suggesting that after some sort of destruction or period of disuse, the Spring House was redesigned to send water west to the cistern on Piers f through k, rather than east on Walls d and e and piers 3-9 (Fig. V-53). The changes in the door in Wall c, if the door actually was walled up at this time, indicate the same change in orientation. The preservation of the water-level arrangements of the structure, which were meant to function as before, along with the discovery of the actual remains of the bucket-chain in Room 2, prove that the Spring House once again was designed to employ a water-lifting device. The lack of changes in the platform—with the possible exception of the two piers—despite such a radical switch of orientation reinforces the preliminary impression that, in spite of its complicated design, it was basically of peripheral importance to the functioning of the Spring House. In contrast, the extensive alterations carried out in Room 1 must reflect removal and relocation of the driving mechanism (Figs. V-49, 50), since the water now had to be lifted ca. 7.0 m higher than in stage 1, to at least 12.47 m a.s.l. (the level of the cistern lip) rather than ca. 5.50 m (the midpoint between the tops of Walls d and e). Although 13 m is a considerable height, evidence for ancient and modern bucket-chain installations that lifted water significantly higher proves that the reconstruction is feasible. The Cosa saqiya, however, is unique in that the entire differential in height was created by a man-made structure—in contrast, for example, to a ground-level machine lifting subterranean water from a well.23 The reconstruction of this tall building is difficult, since such a small proportion of its elevation has survived, but certain con25 m. For comparable material, see the section on the place of the Cosa Spring House in the history of ancient water-lifting technology.

V: SPRING HOUSE COMPLEX

115

jectures are justified. The existing stone walls of stage 1 evidently required little alteration: the beam holes in Room 2 would have continued to hold beams and support floors, while the upper stories— two over Room 2 and one or one and one-half over Room 1—were built of the sturdy timbers whose charred fragments were found in the destruction level (Figs. V-51, 52). The relative scarcity of rubble fall in relation to the height of the structure and the danger involved in piling extra loads on an unreinforced foundation suggest a half-timbered superstructure, probably finished with wattle and daub (Fig. V-53).24 Vitruvius (2.8.20) comments on the lightness and ease of assembly as the chief benefits of this type of construction but warns—significantly, in view of the ultimate fate of the Spring House—that it constituted a fire hazard. Wooden stairs or ladders would have provided access to the upper portions of the building for the individuals responsible for the driving or maintenance of the apparatus. The existence of a gutter and paving only on the south side of the building suggests that the roof over Rooms 1 and 2 had a single pitch in that direction (Fig. V-52). In view of the large interaxial distances between Piers f through k and the total lack of remains of arches to span them, the trough that carried the water back to the cistern probably was built of planks, possibly waterproofed in the same manner as the bucket-chain containers.25 Piers f through k, which provided a firm foundation for this aqueduct, probably were buried in the ground for most of their height, leaving the major visual impact to the six round, gaily patterned columns they supported (Figs. V-50; IV-60). These columns, which are thin (0.61 m) in comparison with their restored height (between 5.50 m and 4.0 m), most likely carried a simple abacus fashioned of bricks to support the trough and form an appropriate architectural transition. At the point where this channel must have intersected the cistern wall (ca. 2.0 m south of the outer northeast corner) there is presently a rough notch (ca. 0.60 m wide and deep), which marks either the joint where the trough was inserted into the cistern wall or the effect of dripping water on the unprotected rubble work (Fig. V-38). As an alternative, it is possible that the columns were low and served only to even out the course of an inverted

siphon formed by lead pipes carried on a bed of planks between the collecting tank and the cistern. Since considerable work is entailed in lifting substantial amounts of water a vertical distance of 12.50 m, there must have been important reasons for locating the cistern in its present position. Although the bedrock that supports the cistern at this point presently is covered with a thin layer of earth, the soil is only slightly deeper eight to ten meters closer to the Spring House, and even at this point much of it appears to have been deposited after the final abandonment of the site (Fig. V-43). Since these alternative sites farther down the slope were not chosen, the location of the cistern at such an elevated position must have served a specific purpose. It is possible that the cistern was meant to provide a convenient water supply to inhabitants living on the lower slopes of the acropolis above the port, since they could have dipped water out of it from the elevated ground along the back wall. Although the present road at this point is modern, the contour of the hill behind the cistern would have provided an excellent location for an ancient road circling the lagoon area. The deposits of potsherds, however, in the earth outside the cistern are too skimpy to substantiate the traffic necessary to justify such a new arrangement, and in any case the cistern is not designed for convenient dipping. In addition, it is difficult to imagine that individuals living on the slopes would have sacrificed the freshness of their water merely to avoid the scramble down the slopes to the flowing basin.26 In consequence, it is justifiable to assume that the cistern was placed in its present position in order to furnish a head of water to a nearby site, a pressurized water supply carried in pipes originating from the hole at the south end of the east wall. Although the erosion of the outer surface of the wall at this point (as at the entry to the cistern) has obscured the original system, the only alternative—the opening in the south wall—can be eliminated, since the preserved square exit hole shows no evidence of having housed a pipe. It probably functioned instead as a simple maintenance drain that would have been blocked with tiles and cement and only opened up when the cistern was to be emptied and drained. The low basin below the spout would have prevented erosion of the soil at this point. Since the water system was

24 Pemp, p. 24, notes that limestone and tuff were used for the lower story of the Stabian baths but that the upper stories were built of "Crumamauerwerk" (half-timbered rubble?). 25 Wooden troughs and pipes for conducting water (tubi vel canales lignei) are mentioned by Faventinus 6 and Pliny NH

16.81.224; see also Palladius Ag. 9.11. 26 Pliny considered cistern water far less healthy than running water; NH 31.21.34. Note also Athenaeus Deip. 2.42 c. cf. 2.46 d; Columella 1.5.1-4. Palladius Ag. 1.16-17, however, finds no harm in cisterns.

116

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

pressurized, a closed conduit of terracotta or lead pipes must have been used. Because a channel of ter­ racotta pipes can be expected to have left traces in the form of sherds or pipe sections, whereas a lead pipe would have been salvaged after the destruction of the Spring House, it is significant that the only conduit found was a short piece of lead pipe (Cat. M i l ) recovered from the loose surface fill a few me­ ters east of the cistern. This piece is identical to the longer stretches of pipe found with the remains of the bucket-chain, so it should be attributed to the same hydraulic complex. The fragment may repre­ sent a lost piece of salvage from the Spring House machinery, but in view of the total absence of ter­ racotta piping, there is also reason to attribute it to the plumbing of the cistern outlet. The parallel of the famous high-pressure piping system built by Eumenes II to supply water to the acropolis of Pergamon is instructive. The original metal pipes have disappeared from that site as well, leaving only an empty trench and line of perforated stone support slabs up the side of the hill. 27 Roman writers express a preference for terracotta piping, even in pressur­ ized systems, on the grounds of purity, health, and ease of repair. But because terracotta required more care during construction, was vulnerable to break­ age, and was more susceptible to leakage, the use of lead piping remained common. 2 8 It is highly unlikely that the pressurized system created by the cistern was meant to employ piers 39, which carried the discharge of the stage 1 machin­ ery across the lagoon toward the harbor area. The highest elevation east of the present channel and south of the Spring House is only 5 m a.s.l., a height that could easily have been served by a cistern built^ over Walls d and e, the hypothetical location of the 27 For the siphon at Pergamon, which carried a pressure of 18 atmospheres at its base, see E. Hansen, The Attalids of Pergamon, 2d ed. (Ithaca, N.Y., 1971) p. 247; A. Neuburger, The Technical Arts and Sciences of the Ancients (New York, 1930) pp. 422-424; G. Eggers, "Wasserversorgungstechnik im Altertum," TechnikGeschichte 25 (1936) pp. 5-6; R. J. Forbes, A History of Technology, ed. C. Singer (Oxford, 1956) II, pp. 668-669; R. J. Forbes, Studies in Ancient Technology, 2d ed. (Leiden, 1964) I, pp. 164-165; G. Garbrecht, "FragenderWasserwirtschaftPergamons,"Pergdmenische Forschungen 1 (1972) pp. 43-49; Garbrecht, "Die MadradagWasserleitung von Pergamon," AWeIt 9:4 (1978) pp. 40-49. The water was carried from the source at Madra Dag (ca. 45 km north of Pergamon) to the beginning of the inverted siphon m thick terracotta pipes for which—in contrast to the metal pipes—there is copious physical evidence (see Garbrecht). Terracotta piping systems always leave traces in the form of sherds and unbroken sections. There are further examples at Kourion and Sidon: J. S. Last, "Kounon: The Ancient Water Supply," ProAmerPhilSoc 119 (1975) pp. 39-76; C.E.N. Bromehead, "The Early History of Water Supply," Geographical Journal 99 (1942) pp. 183-184, 186187. On the inverted siphon in the Hellenistic and Roman periods and ancient water channels in general, see A. W. Van Buren, "Wasserleitungen," RE VIIIA (1955) cols. 453-485 (pressure sys-

stage 1 tank. This arrangement would have avoided the extra trouble of lifting the water higher up the hill. As an alternative, it is possible that the water was conducted to a dwelling on the acropolis slopes west of the harbor, not too distant from the cistern and at a slightly lower elevation. No suitable re­ mains are presently visible in this area, but some frescoed walls discovered nearby and destroyed dur­ ing the construction of the restaurant Il Pescatore may have belonged to the villa served by this re­ markable water-lifting system. The fragments of terracotta molding and red wall plaster found in the rubble south of Wall a could have belonged to such a structure. The presence of a pressurized water sys­ tem in a luxurious seaside villa is no cause for sur­ prise, since both Strabo and Frontinus note the pres­ ence of plumbing even in humble dwellings in Rome. 2 9 The complexity of the water-lifting system at Cosa, however, the large capacity of the cistern, and comparison with other Roman water-lifting in­ stallations suggest that the system was intended in large part to serve a bath in the villa rather than sim­ ple household use. 3 0

T H E WATER-LIFTING MACHINERY OF STAGE 2

(Figs. XIII-I to 33, Cats. Wl to W16) Circumstances of Discovery During the excavation of the fill in the Spring House basin in 1969, several puzzling pieces of wood were discovered. They had been worked into enig­ matic shapes, pierced with round and square holes, and in one case smeared with pitch. Although no final evaluation of their function could be made from the few fragments available at that time, it was tentems are discussed in cols. 472-475); H. Plommer, Vitruvius and the Later Roman Building Manuals (Cambridge, 1973) pp. 25-31; Smith, "Attitudes" (cited in n. 11); J, Therasse, "A propos des tuyaux de poterie dans les aqueducs romains," Les etudes classiques 46 (1978) pp. 127-132; and J. B. Ward-Perkins, "The Aqueduct of Aspendos," BSR 23 (1955) pp. 115-123. 28 On the hazards of lead and the benefits of terracotta piping, see Vitruvius 8.6.8-11; Faventinus 6; Palladius Ag. 9.11.2-3; and cf. Pliny NH 31.31.57. On pressure systems and the various sizes of lead pipes, see Vitruvius 8.6.4-7; Faventinus 6-7; Palladius Ag. 9.11-12; Pliny NH 31.31.58; Frontinus Aq. 1.23-63. Note also ILS 5743. 29 Strabo (5.3.8) said that "nearly every house" (άπασαν δε οίκίαν σχεδόν) in Rome had pipes and fountains; Frontinus (Aq. 2.76) noted with indignation the presence there of illegal pipes even in garrets (cenacula) and brothels (corruptelae). For legal plumbing systems in baths and noble houses, see CIL X 4654 and 4760, 5807 ( = I, 1529). 30 Also see discussion below under parallels for the design and function of the Spring House structure. Levi, p. 478, notes the presence of baths connected with the villa adjacent to the harbor, but the cistern is too high to have served it. For other finds from Levi's survey at the port, see McCann, chapter one, n. 149.

V: SPRING HOUSE COMPLEX

tatively suggested that they might have constituted part of a water-lifting device. Insertion of a metal probe through the openings at the base of Wall c revealed the presence of more wood—two pieces of which were recovered—on the other side, but it was assumed to be formwork similar to that surrounding the basin platform. In 1972, when the fill in Room 2 was completely removed, it was discovered that the lowest level, be­ low the rubble and tiles left by the collapsing walls, was composed of a mass of charred and broken wood partially enveloped in a layer of blue-green sand (Figs. V-8, 9). Pieces of lead pipe and grills and a large fragment of lead sheeting, all showing the effects of fire, were scattered among the wood frag­ ments (Figs. XIII-34 to 61, Cats. M l to M26). Since the top of this layer was even with the basin lip, the fire consuming the wood evidently had been extin­ guished by the spring water, which then acted as a preserving agent. It is probable that the major por­ tion of the structure and its machinery was either completely destroyed by the flames or else fell into the upper layers of the fill in Room 2 and outside the building where the environment did not favor preservation. The large amount of ash in the layer and the completely carbonized condition of several thick beams testify to the violence of the conflagra­ tion. Nevertheless, it was immediately obvious from a preliminary examination of the material exposed that significant remains of a bucket-chain from a water-lifting machine had been preserved among the beams and planks forming the structural members of the Spring House. One bucket was almost intact, although structurally very weak, and partial remains of at least six others ultimately were recovered, the majority from a strip of sand along the north wall. Many of the bucket fragments had been liberally smeared with pitch, which had melted and flowed during the fire, leaving a rough, cracked surface oc­ casionally incorporating masses of the sand into which the buckets fell. A few small fragments of rope appeared as soft, amorphous masses of fibrous mush; the partial imprint left by a rope in the melt­ ing pitch along a side panel of bucket Cat. Wl re­ veals that it had a diameter of at least 0.015 m (Figs. XIII-I to 5). Recovery and Conservation Procedure After the entire layer of wood had been cleared of the overlying debris and surrounding sand, a grid 31 The details of the preservation process are described by H. Λ pp. 111-125.

117

was marked off above the wood with string, con­ sisting of 1-m squares numbered I-IV along the north wall from east to west and A-E along the east wall from north to south. Rows IV and E had to be smaller than the others in order to accommodate the grid to the dimensions of the room. After photo­ graphs had been taken and a drawing of the stratum had been completed using this frame of reference, the better preserved pieces of wood, the lead, and the few visible sherds were lifted square by square for more detailed drawing and photography. Since it was not possible to attempt conservation at the time, the wood was replaced in the northeast corner of the room in perforated plastic bags and reburied in the fill beneath modern wooden boxes. In 1975 it was possible to recover some of these bags and carry out experimental restoration of some of the bucket elements by means of the acetone/rosin impregna­ tion method. Reassembly of the strengthened frag­ ments proved to be feasible, but until a permanent restoration procedure can be developed that will avoid both distortion of the wood and destruction of the pitch coating, it is preferable to leave the best preserved pieces in situ and to depend on less vul­ nerable photographs (Figs. V-8, 9; XIII-I to 33), drawings, and a full-scale copy (Figs. V-58 to 60) for museum display.31 The fragments of the water-lifting device have been catalogued in chapter thirteen under the cate­ gories of wood and metal. The wood is subdivided into three different machinery elements (plus some miscellaneous, unidentifiable pieces) whose purposes are discussed here; the metal finds consist of lead pipes, grills, and sheeting and iron nails. Since the design of all these pieces is carefully described in the catalogue, it will not be repeated in the next section, which attempts a restoration of the machinery. Reconstruction of the Water-Lifiing Device of Stage 2 Although little of the Spring House machinery has survived, apart from the remains of seven containers from the bucket-chains, the design of the building itself, its relationship to the site, and the evidence of ancient and modern water-lifting machinery of sim­ ilar design provide supplementary information al­ lowing at least a partial restoration. This restoration, in turn, is supported by ancient literary testimony, papyri, and archaeological remains concerned with mechanical water-lifting devices. As a first step, it is useful to note the areas in Tell, E. Roger, and A. Varsanyi, Studies in Conservation 17 (1972)

118

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

which the relevant material was recovered, since the configuration of any water-lifting device is governed largely by the location and design of its containers. The fragments of the seven buckets and the five examples of machine element type 2 (Cats. W8-W12) were concentrated along the north side of Room 2 and in the basin, while the single example of machine element type 3 (Cat. W13) was found in square AI of Room 2. Three of the lead grills were recovered from the basin fill (Cats. M13, M14, M15), one from Room 2 (Cat. M12), and one in the surface fill above the basin (Cat. Ml 6). The large piece of lead sheeting (Cat. M17) was found in Room 2, along with five pieces of pipe totaling 2.24 m in length (Cats. M1-M5), while two further pieces of pipe totaling 0.35 m in length were recovered from the basin fill (Cats. M6, M7) and one from the rubble above the drain south of Wall a (Cat. MlO). The piece of pipe found near the cistern (Cat. M i l ) is assumed to be part of the cistern discharge pipe discussed above under reconstruction and function of stage 2. The concentration of bucket fragments on the north side of Room 2 strongly suggests that the bucket-chain was designed to dip into the protected spring water collected in this area and to carry it up to the level of a wheel directly above. The absence of any bronze or iron chains and the partial preservation of pieces of fibrous rope indicate that the wooden buckets were strung together along a pair of ropes designed to form long loops suspended from a vertical bucket-chain wheel. The rapidity with which fire must have consumed these ropes, allowing the buckets to drop into the water, explains the preservation of a relatively large number of bucket fragments. The long side panels of the buckets would not actually have formed part of the loop, since the strain involved in supporting the whole chain of buckets near the top of each cycle would have quickly destroyed the planks. Instead, each bucket probably was tied to the rope loops by short pieces of thinner rope fastened through the holes at either end of the side panels, placing the whole weight on the ropes and allowing flexibility during the passage over the drive wheel (Fig. V-59). Similar precautions are taken in all modern pot-garlands or bucket-chains.32 The rubbing of these bonds has smoothed the edges of the holes in most of the preserved side panels. The location of these holes toward the flat edges of the side panels shows that the 32 See Schieler, pp. 16 (fig. 5), 97 (fig. 66); the well-designed Syrian device shown on p. 23 (fig. 13) is an exception because it is constructed of steel. Cf. also the Portuguese device illustrated

curved edges, which hold the "top" panels, were oriented toward the outside of the bucket-chain. The function of a second round hole present in each side panel behind and slightly above the forward hole is more difficult to envision. It is possible that a long dowel extended across the open end of each bucket through these holes and ca. 0.10 m beyond each side panel in order to mesh with cogs in the bucket-chain drive wheel and prevent the rope from slipping (Fig. V-60). Such a restoration is unlikely, however, since no such dowels can be recognized in the wood fill of Room 2 (the stick found adhering to the back of bucket no. 1 was too thin). Furthermore, the coefficient of friction of the ropes passing over the drive wheel was probably high enough to prevent slippage by itself. Since each bucket was ca. 0.55 m long, and since modern parallels suggest that the interval between containers should at least equal the bucket dimensions, the loops could have accommodated, at the most, approximately twenty-four containers. Slight variations in size and construction among the buckets, as well as some differences in the amount of wear on the suspension holes, reveal that they were built to a standard design but without any rigorous emphasis on uniformity. Units were likely replaced individually as they wore out. Restoration of the design of the wheel that carried and motivated the bucket-chain is more difficult. It is probable, however, that machine elements types 2 (Cats. W8-W12; Figs. XIII-18 to 23) and 3 (Cat. W13; Figs. XIII-24, 25) belonged to this important component, since they were found in the same context as the bucket fragments—the north side of the basin fill—and since they too were waterproofed with pitch. The bucket-chain and the wheel over which it passed are the only elements of this type of water-lifting device that come in contact with water and require such protection. Element type 3 probably served as one of the spokes of a heavy bucketchain wheel with a radius of slightly more than 0.50 m (including hub and rim). The thick, flat tongue at the broad end of this piece would have formed a strong support when socketed into the hub, and the stout peg at the other end would have supported the rim. Two such rays of spokes joined together laterally by crossbars at their outer ends and circling a single hub would have supported and moved the buckets in the proper manner. Since all of the fragments catalogued as machine in J. Dias and F. Galhano, Aparelhos ie elevar a agua de rega (Oporto, 1953) p. 64 (fig. 12).

V: SPRING HOUSE COMPLEX

element type 2 appear to have been inserted into some other part of the mechanism and are usually broken across the joint, it is possible that they were set radially into the rim of this wheel. The square holes could then have accommodated strong bars extending between the two wheels to provide reinforcement and to support the two rope loops of the bucket-chain passing over them. These projections would also have served as guides to keep the ropes from slipping off the wheel, and their curved outer ends may have been designed to engage and disengage smoothly like cogs the dowel that possibly projected beyond the sides of each bucket. As each bucket passed over this wheel, the water it contained would have spilled out into a catch basin. The lower panel of each bucket was designed to extend farther forward than the upper panel so that the water would not begin to spill out until the bucket had traveled partially over the curve of the wheel (Fig. V-59), enhancing the efficiency of collection. Many modern bucket-chain and pot-wheel devices have a trough within the wheel to catch the water and funnel it out through one side to the collecting basin, avoiding the splashing, dripping, and excessive fall that lower the efficiency of machines emptying across the wheel (Figs. V-54, 55, 62, 76; cf. Fig. V-65). Unfortunately, the design of the catch-basin system in the Cosa device can be restored only on the basis of conjecture, but the care taken to adjust the lower bucket panel length and the presence of lead sheeting and pipes in the destruction level suggest that the importance of efficiency at this point was not overlooked. The restoration of a garland wheel with two sets of spokes prevents insertion of a catch basin within it, but modern parallels suggest that the wheel itself might have carried a series of small troughs to funnel the water to a collecting tank beside the wheel.33 The perforated grills recovered from the basin and Room 2 may have been designed to filter the water as it passed from such individual collecting troughs and prevent any vegetable material or aquatic animals from entering the channel leading to the cistern. Such a system would have been self-cleaning, since filtered debris would have fallen back into Room 2 as the wheel turned over.34 It is also possible that the water simply poured through the whole diameter of the wheel to a tank directly beneath it. The large piece of lead 33

See Schioler, pp. 16-17 (figs. 5-6) for the former arrangement, p. 23 (figs. 12-13) for the latter. 34 Three similar lead strainers were found on the Cosa acropolis (CB 1509, CF 1013, C 68-304), at least one of them in the context of a cistern (CF 1013). Large perforated lead screens filtered the water passing through the castellum aquae at Pompeii: see R. Pari-

119

sheeting found in square DII in Room 2 (Cat. Ml 7) probably lined the final collecting basin, which was connected to the aqueduct on Piers f through k by means of the pipes found in the destruction level. One of these (Cat. MlO) bears a flange marking its joint with a tank (Fig. XIII-44). The drive that propelled this bucket-chain—like that of every other ancient bucket-chain device—was located at the top of the system, an arrangement necessitated by problems of design and mechanical efficiency.35 Although a man-powered tread-wheel mounted on the same axle as the garland wheel was commonly used to drive water-lifting devices in the Roman world, such a system does not seem suitable for the Spring House at Cosa. The building was not large enough to accommodate comfortably beneath its roof a wheel of the appropriate dimensions (at least 4.5 m diameter), and the structural problems of mounting such a wheel outside the north or south face of the Spring House would have been prohibitive. The mounting of the tread-wheel at the top of the Spring House would have been further complicated by the difficulty of designing a tread-wheelpowered bucket-chain that would discharge water at a level significantly higher than the axle height (see Fig. V-65). The top of such a treading wheel would have had to project several meters above the already lofty level (at least 12.50 m a.s.l.) of the channel leading to the cistern, well above the protection of the Spring House roof. A more specific objection is that the Spring House plan simply does not resemble the configuration of other Roman structures that housed tread-wheels; at the same time, it shows significant parallels to the few surviving gear-driven bucket-chain installations discussed below in the section on parallels for the design and function of the Spring House structure. In consequence, it is more reasonable to restore a horizontal drive wheel that transmitted power to the bucket-chain by means of a right-angle gear. This system cannot have employed the short drive-shaft arrangement in which the horizontal drive wheel meshes either with a vertical cogwheel mounted adjacent to the bucket-garland wheel or with cogs on the garland wheel itself (Figs. V-61, 62), because the superstructure above Room 2 would not have provided sufficient space for its operation. Such a system, in which force is imparted to a long elevated beni, NSc (1903) pp. 27-28; A. Mau, RomMitt 19 (1904) p. 46. 35 Philo's water-wheel-driven bucket-chain, described in the Pneumatka 65, is the only ancient device to violate this principle, but it clearly is an unworkable, imaginary design. It is possible that the drive system was added by the Arabic translator. See the discussion in Schioler, pp. 65-66, and below, n. 46.

120

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

beam that is fastened at a right angle to the top of the horizontal drive wheel axle and is then pushed in a circle around the whole apparatus, would re­ quire a circular space at least 8.0 m in diameter. Fur­ thermore, this arrangement generally raises water no higher than the level at which the draft animal walks. 36 AU of these difficulties can be avoided by the res­ toration of an elevated shaft, angle gear drive. In this arrangement, a horizontal gear wheel (placed above the horizontal beam to which the motivating force is applied) transmits power to a vertical gear wheel mounted on the end of a long, elevated horizontal shaft, which serves as the hub of the bucket-chain wheel (see Figs. V-57, 70). 37 The treading circle and the gears, located above Room 1, would have driven a horizontal shaft oriented east-west (at a level slightly above 12.50 m a.s.l.) that crossed the space over Room 2 and carried the bucket-chain wheel. The structural timbers of the building would have provided the sturdy framework needed to hold the sockets in which each end of the vertical and hori­ zontal shafts turned. The amount of force required to drive the machinery would have varied according to the ratio of the two parts of the right-angle gear, but any exclusive increase in the radius of the verti­ cal gear wheel would not only have decreased the force necessary to motivate the device but also have increased the height to which water was raised. The power to work the bucket-chain in the second stage of the Spring House can have been supplied only by human labor. The maximum possible radius for a circular path around a drive wheel located over Room 1 is 2.30 m, probably insufficient for a large draft animal. In any case, even with the provision of stairs and ramps, it would have been difficult, if not impossible, to arrange convenient and safe access for draft animals to the elevated chamber in which the work had to be performed. 38 The use of human la­ bor would have allowed not only access by ladder but also the provision of a lighter-weight superstruc­ ture and machinery. There are no mechanical prob­ lems that would preclude the use of human labor in this situation, and it is possible to restore a respect­ able output for the device by hypothesizing two in36 This is the system commonly used to drive the Balearic bucket-chains; Schioler, pp. 12 (fig. 2), 16 (fig. 5), 25 (fig. 16). 37 For representations see Schioler, pp. 13 (fig. 4), 32-33 (fig. 26 a-b). Schioler's classification of the machinery generally fol­ lows that of Dias and Galhano, Aparelhos (n. 32), pp. 65-66, 7277, where this type appears. 38 A Chinese geared, mule-driven bucket-chain illustrated in J. Needham, Science and Civilization in China (Cambridge, 1965) vol. IV, pt. 2 (fig. 591), is an instructive contrast: the machine is located at the top of a tower and consequently requires a huge

dividuals working simultaneously, either side by side on a short bar or at opposite ends of a long horizontal bar extending through the drive-wheel shaft across the whole diameter of the treading cir­ cle. The latter system uses labor more efficiently. The horizontal drive shaft would have been mounted high enough to allow the worker(s) to pass under it during each revolution. Modern angle gears of similar design make exclusive use of draft animals simply because the utilization of human labor for such purposes is not feasible or efficient outside of a slave society. It is at least possible that draft animals were used to drive the angle gear that the Spring House was designed to house in its first stage, since the treading circle was located more or less at ground level, but this space still seems too confined for animals. Furthermore, the absence of any change in the plan of the Spring House between stages 1 and 2 suggests that there was no marked alteration in machinery design.

Discharge of the Water-Lifting Device of Stage 2 Although the details of the drive system of the Cosa bucket-chain can only be surmised, the data provided by Schieler's study of modern gear-driven bucket-chains can be compared with the calculated bucket volume, the height of the lift, and the as­ sumed constants of human physiology to yield at least partial confirmation of the feasibility of the re­ stored design. 39 The minimum lift necessary to al­ low water to flow back into the cistern is 12.0 m (from 0.54 m a.s.l. to 12.47 m a.s.l.), but the bucket-chain must have been at least 27.0 m long (2 χ 13.5 m) in order to allow for passage over the wheel at the top and dipping below water level at the bottom. Assuming twenty-four buckets holding ca. 6.5 liters each (allowing for spillage; average ca­ pacity of the preserved buckets is 7.2 liters), the total capacity of the chain would have been ca. 156 liters. Since each bucket probably weighed four or five kil­ ograms when new, the total weight of the bucketchain itself would have been ca. 108 kg, plus ca. 30 kg for the ropes. 4 0 Since only eleven of the buckets would have contained water at any one time (the access ramp for the animals. A right clavicle from a muscular, upper-middle-age male and a skull fragment possibly from the same individual were found in a deep probe in Trench IA, near pier 3 (J. L. Angel, chapter seventeen, Cats. H l , H2). Can these be the mortal remains of a Spring House slave? 39 Similar calculations, with different variables, are attempted by Pemp, pp. 57-60, 65-67. 40 The bucket reproduction built for the museum at Cosa weighed 2.96 kg without any pitch coating.

V: SPRING HOUSE COMPLEX

121

submerged bucket naturally is not counted), 71.5 kg should be added, for a total of 209.5 kg, or 105.0 kg for each of the two rope loops. This load should be well within the limits of hemp rope with a diameter of ca. 0.03 m. Since the weight of the ascending and descending buckets balances, the machine is required to lift only the weight of the water in the buckets at any one moment, that is, 71.5 kg, plus an allowance for friction. One complete cycle of the chain, lifting 156 liters of water 12 m high, represents 1,872 kg m of work, without taking spillage into account. It is possible to make a rough calculation of the flow that could be produced by such a machine by using Schioler's figures on the efficiency of a modern gear-driven bucket-chain at Ibiza (0.6) in conjunction with an estimate of the human power that could be exerted on such a machine.41 There are no statistics on the human capacity for work in a situation exactly comparable to that of driving a geared bucket-chain, but in view of the need for continuous effort over a significant period of time, an output of 0.1 h.p. (456.42 kg m/min.) seems a likely maximum. 42 Since the lift is 12 m, the effort of one man equals an output of 38.03 kg/min. at the top of the machine. But the hypothetical efficiency of the machine is only 0.6, so the maximum output a single human could provide would have been 22.82 kg/ min., or 22.82 liters of water per minute. Since the chain of buckets holds 156 liters, it would take 6.84 minutes for a complete revolution. Naturally, the design of the gearing system would require the unfortunate individual driving the machine to make a larger number of turns for each rotation of the chain. At a flow of 22.82 liters/min., it would take 92.64 hours to fill the cistern (capacity 126,832 liters). All of these figures are approximate, but seem reasonable. Adding a second driver would more or less double the discharge.43

Although the architectural evolution of the Spring House cannot be precisely dated, and although the purpose of some parts of its structure remains obscure, the general outlines of its development and function are clear. The initial phase, consisting of an outer basin with platform, an inner basin, and an elevated room to the west, was constructed as a unit sometime around the end of the first quarter of the first century B.C. during large-scale landscaping and construction work in the lagoon area. The peculiar design of the original structure and the discovery within it of remains of a bucket-chain belonging to stage 2 strongly suggest that it was designed from the start to house an elevated shaft, gear-driven bucket-chain lifting water to supply an aqueduct leading across the lagoon to the harbor area. This device was destroyed or abandoned at the end of the first century B.C. Sometime during the second half of the first century A.D. the structure was altered so that a new bucket-chain could elevate water high enough to flow west along a newly constructed aqueduct into a cistern. This system, which probably supplied water to a nearby villa, fell out of use before the end of the second century. A summary of the building history makes it clear that there are three interrelated elements to be considered in any evaluation of the place of the Spring House complex in the history of ancient hydraulic technology: the water-lifting machinery, the structures that housed it and carried or stored the water it lifted, and the lagoon or harbor environment it was designed to serve. The location of the spring determined the placement of the Spring House, but the initial anticipated need for water determined the

41 Scht0ler, pp. 21-22. There is a brief discussion of machine efficiency in J. Barois, Irrigation in Egypt (Washington, D. C , 1889) pp. 77-78, where a figure of 0.33 is implied. In fact, the calculation has no validity, because Barois assumes that the ox exerts maximum force on the machine, and he uses this theoretical value as a basis for comparing the discharge. What he is really calculating is the efficiency of this application of potential ox labor. 42 It is difficult to calculate human capabilities with respect to driving irrigation machinery without constructing full-scale models. Even leaving aside the question of the average body weight and physical condition of Egyptian peasants or maltreated Roman slaves, there are no calculations of the present human capacity for work in situations exactly parallel to those of working a saqiya gear or a tread compartmented wheel. Various treadmill and stair-climbing exercises have shown an adult male capable of ca. 0.20-0.22 h.p. See, for example, D. K. Matthews, Physiology of Muscular Activity and Exercise (New York, 1964) p. 163, and B. Ricci, Physiological Basis of Human Performance (Philadelphia,

1967) p. 261. Treading a compartmented wheel, however, or turning a saqiya gear quite probably was harder, and it certainly was expected to continue for longer periods of time than those measured in these modern tests. The poor physical condition of slaves and their understandable disinclination to work at top efficiency should also be taken into account. I prefer to calculate possible output at 0.1 h.p., as does J. G. Landels, Engineering in the Ancient World (London, 1978) p. 63. This figure agrees with a table in A. P. Usher, A History of Mechanical Inventions, 2d ed. (Cambridge, Mass., 1966) p. 156. 43 Pemp's figure (p. 58) for the discharge of the tread-wheelpowered bucket-chain in the Stabian Baths at Pompeii is about half the Cosa Spring House discharge (25.8 hters/min.). The correspondence seems striking, since the lift is twice as great (25 m), but the figures are not really comparable. The drive systems are different, and a single driver has been restored for the Cosa system, whereas Pemp calculates the use of two in the Stabian baths. Pemp's device would have taken ca. 22 hours to fill its 38,000liter cistern.

THE COSA SPRING HOUSE COMPLEX IN THE HISTORY OF ANCIENT WATER-LIFTING TECHNOLOGY

122

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

height and volume of lift, and thus the design of the machinery. The prospective machinery design, in turn, along with the direction of the water conduits and the topography, determined the design of the Spring House itself. When the harbor fell out of use and its industries vanished, a domestic application of the spring water was envisioned, and the structure accordingly underwent modification and addition. However, since the same class of machinery was used, the basic plan did not require alteration. In both stages of its development the design of the ma­ chinery naturally remained an expression of both Roman capacities in applied technology and the availability of cheap labor in the Roman slave econ­ omy. The Cosa Spring House and the remains of ma­ chinery found within it constitute the earliest evi­ dence yet found for the existence of the bucket-chain driven through an angle gear and are thus of con­ siderable importance for the history of ancient water-lifting technology. The Romans subsequently developed this branch of applied mechanics to a re­ markable degree of perfection and applied various machines to difficult pumping problems involved with mining, agriculture, navigation, and urban life. The Roman success was conspicuous, but every one of these machines was the product of Hellenistic Greek science, either in its final form or in its basic 44 Oleson, Water-Lifiing Devices. Shorter, partial studies of the subject—but without sufficient reference to its Hellenistic Greek origins—appear in Schieler's book and in Landels, Engineering (n. 42) pp. 58-83. R. J. Forbes has dealt with pumping devices m parts of his Studies in Ancient Technology, 2d ed., vol. 2 (Leiden, 1965) pp. 32-43, and vol. 7 (Leiden, 1966) pp. 216-23, but his accounts are uneven and not infrequently in error. Most earlier attempts at a well-rounded discussion appear in general histories of technology and are derivative in character. Among them, see T. Ewbank, A Descriptive and Historical Account of Hydraulic and Other Machines for Raising Water, Ancient and Modern (London, 1842) pp. 94-158, 259-260; A. G. Drachmann, The Mechanical Technology of Greek and Roman Antiquity (Copenhagen, 1963) pp. 150-157; Needham, Science (cited in n. 38) pp. 330-362; Usher, History (cited in n. 42) pp. 129-131. 45 One possible exception is a wooden object from Egypt in the collection of University College, London, which Petrie described as a "pawl of a saqieh wheel of sont wood"; W. M. Flinders Pet­ rie, Objects of Daily Use (London, 1927) p. 48. The only handy— though flawed and incomplete—collection of sources on saqiya pots is found in Schiuler, pp. 97-109. J. Ramin felt that a small esparto grass bucket from the Esperanza mine in Mazarron be­ longed to a bucket-chain, but the design and material correspond much better with a container for hand bailing. See Gosse, "Las minas y el arte mmero de Espana en la Antiquedad," Ampurias 4 (1942) p. 55 (pi. VI. 1); J. Ramin, La technique minihe et metallurgique des Anciens (Brussels, 1977) p. 81. J. M. Luzon, "Los sistemas de desague en minas romanas del suroeste peninsular," Ar­ chive espanol de arqueologia 41 (1968) p. 112 (fig. 12), interprets a round-bottomed bronze bowl found at Rio Tinto as a container from a bucket-chain, but the shape is not adapted for such a func­ tion, and the interpretation is highly unlikely; the same must be

principles. A detailed study of the origins of this de­ vice, however, or of the other devices the Roman engineers at Cosa might have used in its place is be­ yond the scope of the present excavation report and has been presented elsewhere by this author. 44 Here the aim is only to elucidate the design of the Cosa Spring House and its machinery by comparison with some related Greek and Roman devices and instal­ lations and to supplement the direct archaeological evidence. Parallels for the Design and Function of the Spring House Machinery In archaeological terms, the remains of the Cosa bucket-chain and the attendant machinery are unique. Large numbers of Roman saqiya pots (uni­ form terracotta pots, mass produced specifically as bucket-chain containers) have been found in Egypt, but no convincing metal or wood containers or frag­ ments of gear-driven machinery have yet appeared there or elsewhere. 45 Philo of Byzantium, in his Pneumatica (chap. 65), specifies either wooden or metal buckets for his bucket-chain (Figs. V-63, 64). In his description of a related device, Vitruvius (10.4.4) mentions only bronze buckets with a capac­ ity of one congius (3.3 liters) (Fig. V-65). 46 Reil's con­ jecture that the εκχύσεις of P. Lond. 1177. 237 are said of a bucket found m the Sotiel Coronada mine in 1883-1884. See A. Engle, "Godet de noria provenant des mines de Coronada (Huelva)," Bulletin Hispanique 1 (1899) pp. 127-130; Gosse, "Las minas," pp. 56-57. 46 Unfortunately, the text of Philo's Pneumatica, composed in the late third century B.C., is preserved only in an Arabic trans­ lation or paraphrase (a Latin translation of the first few chapters exists as well). However, because of its relevance to the Cosa device, it is worth quoting here. For discussion of the Arabic version and its relation to the Greek original, see Oleson, WaterLifting Devices, chapter two; F. D. Prager, Philo of Byzantium: Pneumatica (Wiesbaden, 1974) pp. 66-67, 241; and A. G. Drach­ mann, Ktesibios, Philon and Heron (Copenhagen, 1948) pp. 50-68. It is quite possible that the propulsion by means of a primitive type of water wheel is an addition by the Arabic editor. I have prepared the following English translation from B. Carra de Vaux's French translation Academie des Inscriptions et des Belles Lettres: Notice et Extraits des mss. de la bibliotheque nationale 38 [Paris, 1903], pp. 185-188), assisted at several points by direct reference to the Arabic. The Greek letters refer to Carra de Vaux's drawing (here, Fig. V-63), which is, however, in error at several points. A drawing published by T. Beck, Beitrage zur Geschichte der Technik und Industrie 2 (1911) pp 74-75 (here, Fig. V-64), gives a more accurate impression of the device. "Construction of another ele­ gant device. Let us build yet another device which might be used for many other functions. With it water can be lifted from rivers or other places in order to deliver it to elevated places to water gardens and farms. This water can also be lifted to flow into for­ tresses and elevated hidden places. The river which is to be used for irrigation with this device must have a strong current flowing downhill, copious enough in relation to the water which this de­ vice lifts.

V: SPRING HOUSE COMPLEX

123

wooden containers for a bucket-chain is unlikely, since in other contexts the words seems to mean "trough." The κεραμεΐδες ("pots") mentioned elsewhere in the document (lines 158-63, 185, etc.) might be saqiya pots, but the significance of the ter­ racotta and bronze κάδοι ("jars") is uncertain. 47 Bronze, of course, is salvageable, so the absence of archaeological evidence for metal buckets is not sur­ prising, but the metal was also expensive, and many farmers must have substituted locally available lum­ ber instead. The more complicated force pump underwent the same reworking in wood during the Empire, with surprisingly good results. 48 Excava­ tion of the four known bucket-chain wells in Pom-

peii might provide evidence for bronze containers, if the installations indeed remained in use from the time of the earthquake in 62 up to the final destruc­ tion. 4 9 A leather pouch found at Saqqara by J. E. Quibell was termed a "leather bucket from a waterwheel." 5 0 Leather would seem to be an excellent ma­ terial for water-lifting containers, but this specific discovery can no longer be verified. Since no pots were found at the first- or second-century geardriven bucket-chain installation at Hermoupolis, Schioler may be right in proposing the use of a chain of leather containers there. 5 1 There is no evidence for the use of the saqiya pot, which allowed the rapid, inexpensive production of

"Let us construct a rectangular building similar to a tower. Its proportions are such that it is not weakened by its height, and it is removed a certain distance from the river so that the mass of the river's water might not enter the space from which the water is drawn. Let the front and back part be spaced so that this con­ struction is restricted to the place where the water is drawn. A wooden floor is placed on these foundations, resting on masonry, and the water directed over it. A trench is cut from the river up to this building. This trench is one and a half fathoms deep below the level line, that is below the level of the water which is in the trench. The sides of the trench are solidly built, and its bottom made of lime and plaster, carefully worked, up to where it emp­ ties into the basin. This basin has two walls six cubits long, and its width is such that the device which discharges the water for irrigation is fixed within it. "This device is set up on a very solid cross-shaft, an axle which carries pully wheels two cubits [ca. 0.92 m?] m diameter. Each end of the axle is clad, and fit into a square bearing member presenting a socket in which it can turn easily. The entire device is solidly fitted, because the movement is strong. Another solid axle is placed in the upper part of the tower, similar to that which we have described in its lower part. The irrigation wheel is in the middle of this axle. Its diameter is four cubits. Let there be a triangular device made of copper: its sides have such length as results when they are tangent to the radius of the water wheel, and its width is one cubit. At each end of the axle are identical pulley wheels, like the pulley wheels which we described in the lower part. The water wheel placed in the middle, on which is the triangular device, and the pulley wheels are fixed to the axle. "The lower water wheel is labeled αβ, its axle γ, the pulleys δε. The upper wheel is labelled ζ, the triangle device η, the [up­ per] axle Θ, and the [upper] pulley wheels ικ. "It is also necessary to prepare an iron structure similar to a vertebral column, fastened to the pulley wheels, whose length is such that, being placed on the triangular structure, it comes to within one cubit of the basin floor. The length of each of its seg­ ments is also one cubit. It is hinged by means of iron nails. This structure is labeled O, and the nails λ. Then prepare some rectan­ gular copper or wooden buckets, pinned to this device, and jointed on their inner sides. They are labeled μ. Let the iron struc­ ture be placed around the triangular device, as we have said. The pinned buckets are labeled μ. If the axle is made to turn by force, the triangular device turns and the buckets rise up, full of water. The portion of the construction which holds the buckets neces­ sarily falls on the corners of the triangle, so that when the wheel turns and the buckets have filled themselves, they empty. They empty above point v. Beneath the spot where the buckets empty, place a container to receive the water and make it flow toward the channel placed on masonry pillars, as we have described. "It is left to explain how the axle moves without anyone ap-

proaching it, and lifts the water by means of the buckets. The device must dip in the water we have mentioned, coming from the trench. Ducts are contrived which empty on the containers of the water wheel. Let them be thick and strong. These ducts are so arranged that when the containers are full, the lower axle moves with great force. When this lower axle moves with force and vibrates, the upper one moves as well, because of the chains on which the buckets are mounted. Four containers fill on each section of the water-wheel device, and each has a capacity of two kouz. The discharge depends on the abundance or scarcity of the water. "It should be noted that, of the water wheels, the largest are the water wheels with the triangular device. If the apparatus has enough force to lift twenty buckets, it should have a height of sixty cubits [ca. 27.42 m?], and the ascension of the buckets will be easy. Enough water must be left in the basin for the buckets to be immersed and fill up. The water which is m excess of this amount ought to be drained off by another trench going down­ hill. This apparatus is built as we have said. Here is the figure." Vitruvius' passage (10.4.4) is less detailed because of the ab­ sence of a complicated drive system. Instead of employing water power, it is driven by a man-powered tread-wheel over whose axle the bucket-chain is hung (see Fig. V-65): "sin autem magis altis locis erit praebendum, in eiusdem rotae axe involuta duplex ferrea catena demissaque ad imum libramentum conlocabitur, habens situlos pendentes aereos congiales. ita versatio rotae catenam in axem involvendo efferet situlos in summum, qui < c u m > super axem pervehuntur, cogentur inverti et infundere in castellum aquae quod extulerint" (ed. F. Krohn, 1912, p. 236). ["But if a supply is required at still greater heights, a double iron chain will be set up, wound around the axle of the same sort of wheel and allowed to hang down to the lowest level, with bronze buckets the capacity of a congius [3.3 liters] suspended from it. Thus the turning of the wheel, by winding the chain over the axle, will carry the buckets to the top, and as they are borne over the wheel they will necessarily turn over and pour out into a reservoir what they have raised" (trans. J. P. Oleson).] 47 T. Reil, Beitrage zur Ketmtnis des Gewerbes im hellenistischen Agypten (Borna and Leipzig, 1913) p. 83; M. Schnebel, Die Landwirtschafl im hellenistischen Agypten (Munich, 1925) p. 78. 48 There is a lengthy discussion of this development in my book on pumping technology, Oleson, Water-Lifling Devices, chapter five. See also T. Schioler, History of Technology 5 (1980) pp. 1738, for the bronze force pumps, and A. Neyses, Trierer Zeitschrifl 35 (1972) pp. 109-121, for the wood-block pumps. 49 See Pemp, pp. 9-10. 50 J . E. Quibell, Excavations at Saqqara, 1908-10, Monastery of Apajeremias (Cairo, 1912) p. 29, no. 2068. 51 Schioler, p. 148.

124

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

uniform series of water containers in a standard, ef­ fective shape, until the late third or fourth century A.D. These pots, generally ca. 0.30-0.40 m tall, with a knob foot and short, concave neck to facilitate lashing, a wide mouth for rapid filling and empty­ ing, and an average volume of ca. 5-8 liters, were ideal for the treeless conditions of Egypt (Fig. V-66). Under the name qddus they are still in use in many parts of North Africa and the Near East today (Figs. V-54, 55). When such pots are carefully mounted on a well-built machine and properly cushioned by the garland ropes, the rate of breakage is not as high as might be expected, and the Roman designs continue in use today nearly unchanged. At present, I am aware of possibly Roman sdqiya pots at only one site outside Egypt (Kefar-Manda in Galilee), but others 52 may lie hidden in excavation reports and museums. A pair of grooves worn in the side of a brick compartmented wheel installation in the Terme della Trinacria at Ostia may have been caused by a potgarland, but no sdqiya pot sherds were found there. 5 3 The intensive use of such pots by the Islamic cul­ tures of the Near East, North Africa, and Spain in the Medieval and modern periods makes the subject of chronology a difficult one. At any rate, the sdqiya pot is a type artifact of the Roman sites in Egypt from the fourth century on, and it seems likely that it was developed at the beginning of that century to meet the need for a rapid expansion in the use of the gear-driven bucket-chain for agricultural irrigation. This expansion, in turn, probably was stimulated by Diocletian through reforms in the tax system that fostered or enforced the reclamation of uninundated lands. By allowing the mass production of bucketchain containers and thus facilitating rapid construc­ tion and repair, the use of sdqiya pots made the cost of irrigation less burdensome. 5 4 Both Philo and Vitruvius specify that iron chains should form the loops on which the buckets of their water-lifting devices are suspended. Here again the archaeological record is silent, since no chains of this type have been preserved in the proper context, but the record may have been distorted by accidents of preservation and destruction. Pemp restores iron chains on the bucket-chains at Pompeii because of 52

Schioler, pp. 100-101, 107 (fig. 76). Schi0ler, pp. 140-141. 54 This hypothesis is expressed by D. Bonneau, "L'administration de !'irrigation dans les grands domaines en Egypte," Proceed­ ings of the Twelfth International Congress of Papyrology (Toronto, 1970), p. 49. 55 Pemp, pp. 9-10. 56 For example, see P.Harr. 79. 23-25 (HI); P.Lond. 131R. 609610 (78-79); P.Lond. 131*. 1, 12 (78); P.Lond. 1177. 164-65, 185 (113); P.Michael. 19. 7,11 (III). 53

the depth of the wells they served (Fig. V-76), but he rightly rejects a chain found at Herculaneum, be­ cause it carried only a single bucket. 55 Vegetable fi­ ber rope, like that used for the Cosa bucket-chain, must have been the most common material, and ropes (τριχίαι, σχοινία are occasionally men­ tioned in the papyri in the context of gear-driven irrigation machines. 56 Sdqiya pots would break as they passed over the pot-garland wheel if mounted on chains instead of ropes. The papyri also provide lists of parts used for fitting out gear-driven bucket-chains or compartmented wheels and of the personnel needed for their repair and operation. Some of these parts can be at least tentatively identified, and many describe items that have been found or would have ap­ peared on the Cosa machinery: άξων ("shaft"), άψείδεις ("rims"), βάλανοι ("pegs"), έκχύσις ("trough"?), έργάται ("cogwheels"), ζευκτήριαι ("bindings," "yokes"?), ζυγοτράχηλα ("yokes"?), ήλοι ("nails"), κεραμείδες ("pots"?, "bricks"?), κνήμαι ("spokes"), κόμματα ("strips"), κυκλευτήριον ("wheel"?), κυλλή κύκλας ("pot-garland wheel"?), κύφωνα ("curved beams"?), μηχανάριος ("mechanic," "engineer"?), ξυλική καρταρεία ("wooden fittings"), πίσση ("pitch"), ραιστήρ ("pawl-block"?), σιδήρωσις ("ironwork"), σκυτάλαι ("cogs"?), τέκτονες ("carpenters"), τροχός ("wheel"), τύμπανον ("wheel," "cogwheel"), χελώνια ("sockets"?). 57 Vitruvius' description of the bucket-chain mentions a rota ("tread-wheel"), axis ("axle"), ferrea catena ("iron chain"), aerei situli ("bronze buckets"), and castellum ("water tank"); the cogwheels of his mill are called tympana dentata and the cogs denies. The lack of loose iron nails in the debris of the Cosa device may indicate that most of the structure was pegged together, but the buckets, at least, were fastened by nails. The iron tenons needed to support the pivoting vertical and horizontal axles are also lacking, but most of the drive machinery—located at the top of the building—must have been con­ sumed by the fire or exposed to salvage or decay as it lay in the upper part of the cellar fill. The debris above water level may even have been cleared out 57 In the present context it would be impossible, and probably superfluous, to cite all the occurrences of these terms in the pa­ pyri. There are some excellent lists and explanations of vocabu­ lary in Reil, Beitrage, pp. 82-84, and Schnebel, Landwirtschaft, pp. 73-84 (both cited in n. 47), and some discussion in Schioler, pp. 110-128. A thorough treatment appears in Oleson, Water-Lifting Devices, chapters three, five, six. The most informative papyri in regard to the terminology of parts are P. Bad. 95 (VI/VIl), P.Flor, 16 (239), P.Lond. 776 (552), P.Lond. 1177 (113), P.Mil. 64 (441), P.Ross.Georg. 11.19 (141), SB 5300 (VVVU).

V: SPRING HOUSE COMPLEX

later on. Pitch appears in abundance as a protective coating on the wooden buckets at Cosa; Vitruvius' bronze bucket-chain containers naturally did not require such protection, but he specifies the use of pitch and wax to finish his compartmented wheels (10.4.2-3). One essential feature of the Cosa machinery sets it apart from all other ancient geared water-lifting devices so far recognized in archaeological remains, literary testimony, and the papyri: at least in its second stage, and probably in its first stage as well, it was designed to use human labor! The use of manpower to drive a geared machine is surprising because—as L. A. Moritz has pointed out in the case of the rotary grain mill—the basic advantage to this sort of drive system is that it allows the use of animal labor with minimal supervision.58 An animal working an oscillating device like the cerd—a large, self-emptying water bag lifted by a rope over a pulley wheel—must be led along the track and held as the water container fills or empties. 59 Human labor is more adaptable and can be applied to treadwheels, water-screw barrels, and the dipping action of the shaduj—z bucket suspended from one end of a counterweighted pole—without the need for special gearing. The only real advantage of animals is their strength and endurance, but in Egyptian agriculture, at least, where the rewards of large-scale, long-term irrigation were very rich, this must have been a powerful attraction. Despite the need for heavy capital outlay for animals and machinery, geared, animal-driven compartmented wheels and bucket-chains spread rapidly there from the first century A.D. Egyptian conditions, however, must have been special, for there is no evidence for the use of animal-powered pumping devices outside Egypt (and Nubia) during the Republic or Empire, 58

L. A. Moritz, Grain-Mills and Flour in Classical Antiquity (Oxford, 1958) p. 102. 59 The cerd is illustrated and explained by Schioler, pp. 85-86. 60 The passage, written ca. A.D. 210, appears in a chapter making fun of the doctrine of the reincarnation of human souls into the bodies of animals that are appropriate to their previous lives, good or bad. The pistrina are the ass-driven Pompeian mills for grinding grain, while the aquilegae rotae are compartmented water-lifting wheels most likely powered through an angle gear. Tertullian may well be thinking of an Egyptian machine, although his connections with Carthage may indicate a machine located there or elsewhere in North Africa. "Nam et qui laboribus atque servitiis puniendi in asinos utique et mulos recorporabuntur, quantum sibi de pistrims et aquilegis rotis gratulabuntur, si metallorum et ergastulorum et operum publicorum ipsorumque carcerum, licet otiosorum, recordentur!" (ed. J. H. Waszink [Amsterdam, 1947] p. 48). ["For even those who will be reincarnated as asses and mules to be punished by toil and servitude, how will they congratulate themselves on the mild labor of the mill and the water wheel, when they recollect the mines, and the convict

125

with the exception of a reference in Tertullian to a compartmented wheel possibly located elsewhere in North Africa (De anitna 33.7).60 Since the efficient use of human labor does not require the use of a circular treading path, human-powered, gear-driven pumping devices consequently do not appear either —except at Cosa. The need for a constant, relatively high discharge over long periods of time undoubtedly led to the adoption of the ox-powered saqiya gear in Egypt, both for bucket-chains and for compartmented wheels, and the expense would have been offset by the high yield of properly handled plots of the rich, river valley soil. It is probable that the nature of the regional labor pool had some effect on the use of animals as well, since Egypt did not receive the flood of slaves that had such a profound effect on Roman society elsewhere from the second century B.C. Irrigation workers, according to the Egyptian papyri, usually are salaried, whereas Roman literary texts that specify the status of such workers nearly always indicate the use of slaves, criminals, or prisoners.61 Although there is occasional mention in Arabic texts of prisoners set to work on geared irrigation devices, the need for perennial irrigation in the eastern Mediterranean and the decline of slavery in the post-Roman West meant that Medieval and modern geared irrigation devices were, almost without exception, animal-driven (Figs. V-54 to 57, 61, 62).62 The atypical character of the Cosa bucket-chain drive in the second stage of the Spring House development is easily explained: the great height of the structure made the application of draft animals impossible and rendered the use of a tread-wheel impractical. The installation of an elevated shaft, human-powered saqiya gear avoided both problems. It is less easy to visualize the need for this type of degangs, and the public works, and the very prisons, although there was more leisure there!" (trans. J. P. Oleson).] 61 The status of irrigation laborers as described in the papyri is discussed at length by D. Bonneau, "Esclavage et irrigation d'apres la documentation papyrologique," Actes du Colloque 1973 sur I'esclavage (Pans, 1976) pp. 314-331. Examples of servile or convict status among water-lifting personnel are: Artemidorus Onirocritica 1.48; Manetho Apotelesmatica 1.83-88; Scriptores Historiae Augustae Heliogab. 24.5; Strabo 17.1.30; Suetonius Tiberius 51. A partial exception to this low status is found among the pumpers (sifonarii/siponarif) attached to the quasi-military vigiles in Rome, who operated fire engines (ClL VI. 1057-58, 2194, 31075), and their citizen counterparts in the eastern provinces; see Pionius Vita Sancti Polycarpi 28 and Pliny Epistulae 10.33.2. 62 B. Laufer, "The Noria or Persian Wheel," in Oriental Studies in Honour of Cursetji Erachji Pavry, ed. V. D. C. Pavry (London, 1933) p. 240. See also E. Wiedemann, "Zur Technik bei den Arabern," Sitzungberichte der Physikal. Medizin. Soz. Erlangen 38 (1906) p. 332.

126

HISTORY, GEOGRAPHY, A R C H I T E C T U R A L R E M A I N S

vice in the initial stage of Spring House develop­ ment. The structure is too carefully planned and constructed to allow credence for any claims of local inexperience with such devices, and a tread-wheeloperated bucket-chain may already have been in use for over fifty years in the acropolis baths at Cosa (Fig. V-67). 63 The steeply sloping site—so different from the situation of Egyptian pumping installations (see below)—may be a factor, since it necessitated the construction of an extensive superstructure di­ rectly over the well in order to provide a working platform for the drive machinery at the level of dis­ charge. The peculiar, massive design of the Spring House basin, along with the presence of buttresses against the ends of Walls a and b, show that the ar­ chitect had justified or imagined fears concerning the stability of his foundation, perhaps leading him to avoid any unnecessary vertical expansion. In such a case the use of a geared drive with elevated shaft would have had a distinct advantage over a treadwheel-driven bucket-chain, which could have lifted water only a short distance above the level of its axle. Although Room 1 presumably could have been made large enough to accommodate the treading path of a draft animal, and access would have been easy enough in the first stage, the installation appar­ ently was planned from the start for human propul­ sion: animal drive must have been uneconomical either in terms of the local labor market or in view of the volume of discharge desired.

Parallels for the Design and Function of the Spring House Structure

identical to those of Medieval and modern installa­ tions in the Islamic world: a deep circular well, an adjacent circular ox walk, and a cistern (Figs. V-68 to 72). 64 The device represented in a fresco of the Augustan period in Alexandria seems to be essen­ tially the same (Fig. V-73). Here, however, the lift­ ing element is a compartmented wheel. 65 In the basic design a bedding stone for the lower end of the ver­ tical axle of the horizontal cogwheel stands at the center of the treading circle, and pillars at either side support the crossbeam holding the upper socket. The elevated shaft arrangements at Abu Mena would have required a supporting framework above the well to hold the drive shaft and pot-garland wheel; the sunken drive shaft at Hermoupolis was mounted in a trench below the ox walk. The Abu Mena installations were enclosed in rooms, probably because of their busy, secular environments, whereas the Hermoupolis ox walk was exposed on a slightly elevated brick platform in the center of the temple precinct. In the papyri such an installation seems to have been called μηχανοστάσιον, the ox walk βοοστάσιον or possibly κυκλευτήριον, the well φρέαρ, and the storage cistern λάκκος or κάστελλος. The irrigation installation as a whole might be referred to as a μηχανή αντλούσα ("irrigating machine"), or possibly ύδρεύματα. 6 6 The last term, however, might also refer to the distribution chan­ nels for the water lifted, like· the aqueducts and pipes restored for the Spring House device. Roman rep­ resentations of Egyptian irrigation machines at work, along with modern parallels, show that an at­ tempt was made to roof or to train vines over such installations to provide shade for man and beast. 67

The three well-preserved gear-driven bucketchain installations so far excavated in Egypt—at Hermoupolis West of the first and second centuries A.D. and at Abu Mena in the "Grosse Doppelbad" and "Palast" of the fifth and sixth centuries A.D.— all show essentially the same features, more or less

The bucket-chains driven by tread-wheels in­ stalled at three public baths and a tannery in Pompeii required a different type of arrangement, but one equally uniform and characteristic: the square well shaft is flanked by a cistern and a long narrow cham­ ber meant to accommodate the large tread-wheel

63 This bath structure, which has not yet been excavated, was built ca. 150-125 B.C. (oral communication, F. E. Brown, 1981). The upper portion is badly damaged, but the chronology and basic structural elements are clear. A rectangular, rubble masonry shaft (ca. 1.50 χ 1.25 m) connects two large subterranean cisterns with the upper level of the bath structure, about 4.0 m higher, where the remains of a cistern can still be seen. No oblong treadwheel room can be seen at present, but the relatively low lift involved and the small dimensions of the bath structure suggest that a tread-wheel drive of the type described by Vitruvius (10.4.4) would have been sufficient. 64 Hermoupolis West: A. Badawy, "Au grand temple d'Hermoupolis-Ouest: !'installation hydraulique," RA 48 (1956) pp. 140-154; Schioler, pp. 141-148. Abu Mena: "Der grosse Doppel­ bad," M. Muller-Wiener, "Abu Mena. 3 Vorlaufiger Bericht," MittKairo 20 (1965) pp. 131-132; Muller-Wiener, "Abu Mena. 4

Vorlaufiger Bericht," MittKairo 21 (1966) pp. 175-178; MullerWiener, "Abu Mena. 5 Vorlaufiger Bericht," MittKairo 22 (1967) pp. 207, 209; Schieler, pp. 131-136; W. Muller-Wiener, "Der Pa­ last," MittKairo 22 (1967) pp 206-224; W. Muller-Wiener, "Abu Mena. 6 Vorlaufiger Bericht," AA (1967) pp. 458-459; Schioler, pp. 135-136. 65 H. Riad, "Tomb Paintings from the Necropolis of Alexan­ dria," Archaeology 17 (1964) pp. 169-172; Schualer, pp. 152-153. 66 See the references cited in n. 57. 67 See, for example, the fresco in Alexandria mentioned in n. 65. A fresco representation of a water-screw treader at Pompeii is shown in A. Maiun, NSc (1927) pp. 59-60 (pi IX). On the use of shaded ox walks in modern machines, see Barois, Irrigation (cited in n. 41) p. 77; Schnebel, Landwirtschafi (cited in n. 47) p. 83.

V: SPRING HOUSE COMPLEX

(Figs. V-74, 75). 68 Similarly, when the gear drive in the "Grosse Doppelbad" at Abu Mena was replaced at one point by a tread-wheel drive, the latter had to be accommodated in a narrow trench dug across one of the earlier ox walks (Fig. V-71). 69 It is likely that the bucket-chain supplying the forum baths at Cosa was driven by a tread-wheel mounted in a similar room (Figs. V-67) and Vitruvius probably had the same layout in mind for the installation he describes (10.4.4). The Cosa Spring House is unique among ancient geared water-lifting installations in provid­ ing an artificial elevation for a ground-level spring rather than ground-level discharge of a subterranean water source. There are, however, some modern parallels, in particular a well-preserved but now dis­ used gear-driven bucket-chain at Formia in Cam­ 70 pania. Both the use of an elevated shaft saqiya gear and the positioning of the device in a structure built up against a sloping hillside above the water source resemble the Cosa arrangement (Figs. V-56, 57). But despite its atypical design, the Cosa Spring House still manifests the characteristic elements of a geared installation: a deep well, an adjacent working platform, and a cistern. The long, narrow room for the tread-wheel is absent. The circular walk, bed­ ding stones or blocks, and supporting frames of the stage 1 machine were lost as a result of the lowering of the floor of Room 1, and the machinery of stage 2 perished in the destruction of the upper story where it was installed. The stage 1 cistern apparently consisted of a wooden tank on Walls d and e. All the ancient water-lifting installations differ radically from conventional spring house design in antiquity in not attempting to provide convenient access to the water at its original level. Both the probable industrial function of stage 1 of the Cosa Spring House and the hypothetical domes­ tic function of stage 2 can be paralleled in all three classes of evidence concerning Roman bucket-chain devices: archaeological remains, literary texts, and papyri. Since it is quite possible that at least part of the water provided by the first stage of the Spring House was intended for the hypothetical Sestius am­ phora factory possibly located in the port area, it is interesting that two Egyptian papyri hint at the use of pumping installations to serve a pottery and a brick factory. The earlier document, P. Flor. 50 68

See Pemp, passim, for the Pompeian devices. Schioler, p. 134. 70 Schioler, pp. 32-33. 71 Reil, Beitrage (cited in n. 47) p. 85; A. Calderini, "Appunti di terminologia secondo ι documenti dei papin," Aegyptus 1 (1920) p. 310. On evidence for a pottery in the port area, see Manacorda 1978, pp. 124-125. 69

127

(A.D. 268), is a legal record of the division of prop­ erty, including a number of ύδρεύματα ("water works"? "cisterns"?). In lines 12, 47-48, and 89 there appear όργανα δύο ("two devices") set up at the spring of the μέγα ϋδρευμα ("big cistern," "big water-work"?), and further isolated mention of ύδρεύματα is made in lines 35, 59, 79, and 106. In line 107 they are to provide water for some cattle and a κεράμιον ("brick factory"?). There is an άναβατικόν ΰδρευμα ("raising water work") in line 54 that is either an elevated cistern or a groundlevel cistern to which water is lifted from below; in either case, pumping machinery must be present. The first four letters of κεράμιον are unclear but probable. The other document, P. Lond. 994 (A.D. 517), consists of the record of a lease of a share in a pottery (κεράμιον; line 10). At the factory there is a λάκκος ("cistern," "well") and a ξυλικόν όργανον ("wooden device"; lines 10-12) completely fitted out. The editor of P. Hamb. 23 interprets this device as a "Topferapparat," but, in view of the connection with a λάκκος, a water-lifting machine also is possi­ ble. Reil uses the ambiguous term "Ziehbrunnen," but Calderini considers it a pumping machine of some sort. 71 The vast majority of the Egyptian papyri that seem to mention mechanical pumping devices are concerned with irrigation, but P. Lond. 1177 (A.D. 113) has preserved the report of a commission over­ seeing a municipal water supply system served by shadufs, water-screws, tread compartmented wheels, and gear-driven devices, either wheels or bucketchains. 72 The machinery, reservoirs, and distribution system (lead pipes) were publicly owned (like the Cosa Spring House?) and supplied water to a bath, public (and private?) fountains, a brewery, and a synagogue. Revenues for labor costs and for main­ tenance of the large complex were obtained from town officials and by daily or monthly charges on water usage. Like most public corporations, it seems to have lost money. In his Pneutnatica, chap. 65, Philo specifies that his bucket-chain device can lift water either for irriga­ tion or for household use. Vitruvius (10.4.4) does not mention any specific application for his related machine. The Hermoupolis installation was de­ signed to provide water to a basin frequented by the 72 This document is discussed by Oleson, Water-Lifting Devices, chapters three and five; Reil, Beitrage, pp. 81-83; Schnebel, Landwirtschafl, pp. 74-77; and Schioler, pp. 123-124, 158. A partial translation containing many inaccuracies appears in A. C. John­ son, ESAR. II. Roman Egypt to the Reign of Diocletian (1936) pp. 685-691.

128

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

sacred animals of Thot, whereas the Abu Mena devices served the needs of a bath building and "palace." The tread-wheel-driven bucket-chain on the Cosa acropolis served a bath, as did three similar devices at Pompeii and the single possible example at Ostia. The fourth bucket-chain at Pompeii (in the Casa del Cambio) served a tannery. The small, crank-driven bucket-chain bilge pump restored on the Nemi ships is a complete fabrication, and the evidence proposed for mine-draining bucket-chains

at Mazarron, Rio Tinto, and Sotiel Coronada is weak. 73 There may have been bucket-chains to serve the "ritual baths" at Cyrene and the private baths in the Palazzo delle Colonne at Ptolemais.74 Since the provision of water for baths is the function of most bucket-chains in the Roman West (along with a number of the compartmented wheels), the supposition that this may have been one of the functions of the Cosa Spring House machinery as well, in its second stage, is not farfetched.

73 The Nemi "bucket-chain" is presented in G. Ucelli, Le navi di Nemi, 2d ed. (Rome, 1950) pp. 181-184, 428-429. There is no basis for the restoration, which was a fabrication meant to serve Fascist propaganda, and it is rejected by L. White, Jr., Medieval Technology and Social Change (Oxford, 1962) pp. 105-106, and by Schioler, p. 157. For Mazzaron, Rio Tinto, and Sotiel Coronada, see n. 45.

74 The Cyrene device is discussed by Schioler, p. 156; to his bibliography add G.R.H. Wright, "Cyrene: A Survey of Certain Rock-Cut Features," JHS 77 (1957) pp. 299-310. Schioler is the only scholar to propose the presence of such devices here and at Ptolemais (p. 156).

Chapter VI. The Lagoon Temple ANNA MARGUERITE M C C A N N

Just to the east of where the modern hotel Le Rocce sits, on the crest of the rocky hill overlooking the northern end of the ancient fishing lagoon (Fig. VI-I), the foundations of a small temple and some of its terracotta decoration were found in 1950. Renewed quarrying of these cliffs, which had already cut away most of the southern side of the building with its faqade, made documentation imperative. 1 The excavators F. E. Brown and L. Richardson uncovered in a one-day summary excavation a small square cella, 8.88 m on a side, made of coursed rubblework, and remains of a pronaos with part of one anta. The latter was unbonded to the cella wall and thus of a later period when also the floor of both the cella and the pronaos were resurfaced. The building rose directly from the natural rock without a podium. It faced south-southeast and looked down upon the fishing lagoon with its freshwater springs and Spring House. No remains of columns were found, but Brown restores four along the fagade in both its phases (Text Figs. VI-1,2), following the plan of Temple D on the arx of Cosa above which appears to be a slightly larger architectural twin (Fig. 1-10). Many fragments of decorative terracottas were recovered—namely 207 fragments of revetment plaques decorated with a looped palmette design with only 8 fragments of other revetments. 2 Also found in 1950 was the upper torso and head of a half-life-size nude and helmeted warrior, in terracotta evidently from the pediment (Fig. VI-2). 3 The figure was meant to be seen from the side for

the face is a simple oval without features and turned toward the left shoulder. Short curls hang down the back and below the cheek pieces. E. H. Richardson compared the type of helmet and hair style of the warrior with the opponent of Artemis on the great Hellenistic frieze on the Altar of Zeus at Pergamon 4 and suggested an attempt by the artist of the Cosa pediment to emulate the style of the Pergamene monument, so famous throughout antiquity. She further associated the subject matter with that of the pedimental sculpture from the neighboring temple at Telamon, which is now dated on the basis of its style into the second quarter of the second century B.C.5 She apparently did not know, however, of the sideand back-view warriors from this pediment, now restudied, restored, and fully illustrated by O. W. Von Vacano.6 In the left wing of the pediment is a headless nude warrior turned to the right, very like the Cosa figure.7 The subject of the Telamon pediment has long been recognized as the defeat of the Seven against Thebes. 8 The most recent reconstruction of the pedimental group has been offered by Von Vacano.9 On the right, Amphiaraus is seen disappearing with his chariot into the cleft in the earth opened by Zeus. On the left, Adrastus, the only one of the seven to survive, leads off in his chariot drawn by the magical horse Arion, offspring of Poseidon and Demeter, who provides his escape. In the center Von Vacano would restore the figures of Oedipus with his two dead sons, Eteocles and Polynices, previously thought to belong to another pediment or

1 Brown et al., pp. 142-145, with plan and restoration drawings of its two phases, figs. 108-109. 2 Ibid., pp. 204-205. 3 Ibid., pp. 330-332, figs. 20, 21 (E. H. Richardson). 4 Kahler, pis. 7, 30. For this type of helmet also compare the Gallic iron helmet found at Port bei Nidau in the Landesmuseum, Zurich, dated in the middle of the first century B.C. and illustrated in H. R. Robinson, The Armour of Imperial Rome (London, 1975) p. 43, fig. 95. It is pointed out that this type of Gallic helmet influenced the design of Roman iron helmets. 5 "Telamon," PECS, p. 891, "ca. 170-150 B.C." (O. W. Von Vacano); Brown et al., p. 307, n. 20 (E. H. Richardson).

6 Von Vacano 1961 and 1969. Also M. Martelli, "Le manifestaziom artistiche," in GIi etruschi, pp. 278-281, fig. 273, who dates the pediment between 180 B.C. and 160 B.C. It is now on display m the Museo Archeologico in Florence. 7 Von Vacano, 1969, foldout opposite p. 142, panel IX. Also compare the warrior seen in back view under the horse from the left side, panel VI. 8 First recognized by L. A. Milam, Il museo archeologico di Firenze (Florence, 1912) pp. 66-69, 257-260, pis. 114-116; Andren, vol. I, pp. 227-238; II, pis. 82, 83. 9 Von Vacano 1969, foldout opposite p. 142.

Text Fig. VI-2. Elevation of lagoon temple, phase 1 and phase 2. B r o w n et al., fig. 109.

VI: LAGOON TEMPLE

131

frieze. Intermingled with the heroes and warriors are winged female Etruscan daemons. Behind the design of this elaborate and pictorial composition may be a lost Greek painting, but the insertion of the typically Etruscan winged figures indicates the inventiveness and immediate heritage of the Telamon artist.10 If the pediment from Telamon may be dated in the second quarter of the second century B.C., as is now accepted, it would be contemporary with the building of Temple D and the lagoon temple. While no pottery or other dated material appears to have been found during the excavation of the lagoon temple, the two phases of Temple D are well documented by both the archaeological and stylistic evidence.11 Temple D preceded the building of the original Capitoleum on the arx constructed about 150 B.C. and was built shortly after the construction of Temple B in the forum, completed about 175 B.C. Its second building phase occurred in the first quarter of the first century B . C , dated by both numismatic evidence and a comparison with the new decoration of the Capitoleum. The pronaos of Temple D was now extended by creating long antae and a new fagade of columns. Because the plans of Temple D and the lagoon temple are similar in their two phases as well as the use of identical revetment plaques (Figs. VI-3,4), these dates proposed by Brown and Richardson are accepted here as also the most probable for the two building periods of the lagoon temple. Furthermore, this chronology correlates closely with the activity and building phases of the fishery and port below, now dated by our excavations. The earliest large body of amphora finds from the lagoon and port are Will's Type Id, dated in the third and second quarters of the second

century B.C.12 Many of these were found in the area of the major spring along the western lagoon embankment not yet contained in its permanent structure. At the end of the first quarter or beginning of the second quarter of the first century B.C., the Spring House with the large fishery in the lagoon and the main concrete installations in the port were constructed. It is logical that at this time renovations were made in the lagoon temple fagade to render it a more prominent guardian of the new and extensive building complex below. No traces of the little temple over the lagoon could be found in our surface survey of the site in 1968. Apparently it had been obliterated by the modern quarry. A few more pieces from the original temple decoration, however, were picked up on the surface and are catalogued below. These include five fragments of terracotta revetment plaques of the looped palmette pattern and two fragments of sculpture probably belonging to the pediment. The larger of the sculptural pieces and the one of particular significance in identifying the subject of the pediment is a draped female torso, about one-quarter life-size, running toward the right. This figure closely resembles one of the winged and draped Etruscan daemons from the Telamon pediment. 13 The soft rendering of the drapery, which clings and reveals the active body, may also be related to a Pergamene stylistic tradition. 14 This new addition to the small pedimental group is further evidence of a strong link with the iconography of the Telamon pediment. Why the myth of the Seven against Thebes is so appropriate for the Cosa lagoon temple deity will be suggested below.15 To what deity then might our little temple over-

10 Von Vacano 1961, p. 63. Compare Martelli in GIi etruschi (cited in n. 6), who suggests artists from Telamon also worked at Cosa. 11 Brown et al., pp. 43-44, 111-118. 12 See below Will, chaper nine. 13 Von Vacano 1969, foldout opposite p. 142, panel XVIII from right side of pediment. 14 See Cat. LT7. 15 In a recent article cited in the abbreviations under Zancani Montuoro, Dr. Paola Zancani Montuoro has proposed an identification with Dionysus and Aphrodite for the pedimental figures of the Cosa lagoon temple. She apparently did not know of our female figure. Her suggestion is based upon a small votive model of a temple found in a deposit at Vulci dated in the first half of the first century B.C., which shows these deities (or Dionysus and Ariadne?) in the pediment. It was found with two other little building models of similar fabric and style—one a colonnade or stoa and the other probably a unique representation of the lighthouse or beacon tower of the Portus Cosanus. The association of the tower model with a beacon light at Cosa seems most probable to me—and is also supported by the archaeological evidence presented in chapter seven (McCann, reconstruction of the port)—

but the identification of the temple model with the lagoon temple at Cosa seems less likely. This connection is largely based upon the fact that the temple model and the Cosa lagoon temple both lack podiums. The votive model, however, shows a pseudoperipteral temple, whereas the lagoon temple was clearly of the Latin type with columns only across the front. Furthermore, besides the new running female figure, the half-life-sized warrior must be fitted into any pedimental group. Neither figure belongs in the Dionysiac composition. It also does not seem necessary to me to presume a close topographical relationship between the three Vulci building models. The nature of a thank offering implies a religious rather than a topographical association on the part of the dedicator, and several important but separated monuments from the region might be so united. Votive models of other lighthouses, particularly the famous Pharos at Alexandria, are common enough and are discussed in detail by Dr. Zancani Montuoro; also see references cited in chapter seven, reconstruction of the port. A resident of nearby Vulci returning safely home through the port of Cosa might well have dedicated such a model along with one of the temple of his faith more likely in his home town. There also does not appear to be enough space on the sharp cliffs above the Cosa

132

HISTORY, GEOGRAPHY, A R C H I T E C T U R A L R E M A I N S

looking the fishing lagoon be dedicated? In the absence of any inscriptional evidence, the answer to this intriguing question must remain speculative. Brown has proposed an association with Portunus or Palaemon, son of Mater Matuta, the Italian goddess of childbirth and nursing who also became associated with the Greek goddess of the sea, Leucothea.16 Portunus is first known in Rome as the guardian of the ferry crossing of the Tiber, and apparently only later did he become the god of ports and gates.17 His particular attributes are the anchor and the key. The lagoon temple's association with Portunus is based primarily upon the identification of its twin on the arx, Temple D, with Mater Matuta. Brown assumes that a similarity in architecture implies a kinship of cult. Although this may be true, the evidence for identifying Temple D with Mater Matuta appears inconclusive. Among the evidence cited, first are fourteen fragments of a two-thirdssized terracotta statue found near the temple and identified as the remains of a female cult image. 18 None of the drapery fragments, however, reveal a specifically female form, and while the hair style on the head fragment is most commonly feminine, it is also sometimes used for representations of male divinities, namely Dionysus or Apollo. 19 Moreover, the other separate fragments of irregularly massed curls20 do not appear to belong to this hair style in its usual form. Rather, the tumbled mass of isolated curls looks more like a beard. Perhaps we have the remains of a male divinity—or indeed the remains of two cult images, one masculine and one feminine.

Leucothea, but it is not unique to this goddess and marine motifs were a very common architectural decoration in Etruria at this time, as E. H. Richardson has pointed out. 22 In fact, on the temple at Telamon, dolphins are used for acroteria.23 Finally, Brown cites several fragmentary inscriptions from small statuary bases found in the Capitoleum and forum that record dedications by women, a collegium or matronae lead by magistrae, which he associates with the worship of Mater Matuta. 24 In the recent study of the Cosa inscriptions, however, E. Bace questions the association of one of these with a female goddess, based on the use of the masculine adjective "Cosano." 25 Bace further notes that dedications by matronae to male divinities (Apollo and Liber) also occur. If the identification of Temple D with Mater Matuta is then not assured— as seems the case—an identification of the temple in the lagoon with her son Portunus may also be questioned. Another suggestion for this god's identity seems more in keeping with the site to which it belonged as well as with the archaeological evidence, both old and new.

Specifically suggesting a sea divinity for the god or gods of Temple D is the little frieze of dolphins and sea monsters that once formed part of the temple's decoration.21 Such a frieze would be appropriate for Mater Matuta in her role as the sea goddess,

First, on topographical grounds, one might ask why the ancient Cosans placed a temple to Portunus so far behind the port, overlooking the basin of a fishing lagoon, when the limestone promontory directly above the port's western exposure would be the most appropriate site for any deity protecting the port. The temple's location is actually some 300 m behind the harbor and sits directly above one of several freshwater springs we uncovered, which was channeled into the northern fish tank. This location over a fishing lagoon gushing with sweet water rather suggests the Latin god associated with fresh water and fishermen as well as the sea—namely Neptunus. 26 His attributes are the pine or palm

lagoon for a surrounding stoa. It thus seems more likely that both the temple and the stoa relate to buildings in Vulci as R. Bartoccini has already suggested in "Scoperte recenti negli scavi di Vulci," StRom 6 (1958) pp. 125-129. For further discussion of the whole votive deposit and its date see: Helbig4, HI, pp. 508-510, with earlier bibliography and especially R. A. Staccioli, Modelli di edifici etrusco-italici. I modelli votivi (Rome, 1968) pp. 24-28, 69-70, 75, pis. 12-18. 16 Brown et al., pp. 45-47; Brown 1980, p. 49. For the worship of Mater Matuta see Q. F. Maule and R R . W . Smith, Votive Religion at Caere: Prolegomena, (Berkeley and Los Angeles, 1959) pp. 74-87; G. Radke, Die Goiter Altitaliens (Minister, 1965) pp. 206209; R. Bloch, "Recherches sur les religions de l'ltalia antique," Centre de recherches d'histoire et de philologie, Ecole pratique des Hautes Etudes 3, 7 (1976) pp. 1-9. 17 L. A. Holland, Janus and the Bridge, PAAR 21 (1961) pp. 141178; "Portunus," RE 22, 1 (1953) pp. 400-402 (H. Papenhoff). 18 Brown et al., pp 324-328, figs. 13-17. 19 For example, see G. E. Rizzo, Prassitele (Rome, 1932-1940) pis. 115-117, 128; J. Charbonneaux, Hellenistic Art (330-50 B.C.) (New York, 1973) pis. 217, 262. Also see bearded male terracotta

head from Orvieto, Banti, The Etruscan Cities (cited in chapter one, n. 117) pi. 59a. Note particularly the head of Dionysus from the Tempietto di Ponte Rotto Vulci, in the Archaeological Museum, Florence, published by Martelli in GIi etruschi (n. 6) p. 280, fig. 275. 20 Brown et al., p. 326, fig. 15 a, b. 21 Ibid., pp. 328-330, figs. 18-19. 22 For example, see Andren, II, pi. 14, fig. 47; pi. 19, fig. 60; pi. 45, fig. 147; pi. 74, figs. 252-253; pi. 77, fig. 264; pi. 78, figs. 269-270. 23 G. Libertini, "Le decorazioni fittih di un tempio italo-etrusco sul colle di Telamonaccio," AttiPontAcc 15 (1921) pp. 135-178, figs. 17, 19. 24 Brown et al., p. 46; Brown 1980, p. 49, n. 6. Cosa Inv. CB 580, CB 963, CB 1739 and CE 108, Bace, pp. 90-94, nos. IHAl and IIIA2. 25 Bace, p. 93, no. IHAl and n. 87. 26 Catullus 31.1-4. I am grateful to H. Comfort for this reference. Also see G. Wissowa, Religion und Kultus der Rbmer (Munich, 1902) pp. 250-253; OCD, "Neptunus," pp. 728-729; OCD, "Poseidon," pp. 866-867. For deities especially worshiped by fish-

VI: LAGOON TEMPLE

133

bough used in his festival, the dolphin, the horse, tunny fish, hook, and the trident. The last symbolizes his special power to shatter rocks, bring forth fresh water, and shake the earth. He became thoroughly assimilated with the Greek god of flowing fresh water and the sea, Poseidon. Although this god lost the contest with Athena for the land of Attica, his gift to mankind must have been particularly valued by the ancient Cosans, whose hill town was without a spring but whose lagoon and port below were so blessed. The largest spring in the lagoon area, unleashed again in our excavations of the Spring House, gushed forth sweet water at over four hundred gallons per minute. This spring is probably the famous Fonte de la Reina known in the Middle Ages from the documents of the port site27 and was the chief freshwater source for both the ancient port with its ships as well as the fishery. It apparently had a long use and also a sacred association. Another important aspect of the Poseidon/Neptune myth is his association with horses. He is credited with creating the first horse for mankind, and horse and chariot races were held in his honor. He is also said to have fathered Pegasus and the magical horse Arion, who saved the life of Adrastus, one of the Seven against Thebes. Poseidon/Neptune was thus also worshiped as the god of equestrian accomplishments and the horse is one of his symbols. Furthermore, inscriptional evidence in the Roman world indicates that Neptune, rather than Portunus, 28 was the popular deity of fishermen and sailors and his statue appears on coins as well as on a relief representing the Claudian harbor at Ostia.29 There is apparently no evidence of a Portunus cult at Ostia, where one would have been most appropriate. Neptune may also have had other honorings at Cosa and was a god honored elsewhere in Etruria, whereas Portunus does not appear to have been popular during the Republic, apart from Mater Matuta. 30

These images bring us back to the early bronze coinage of Cosa, the half-litra with Mars on the obverse and the protome of a horse's head mounted on a dolphin on the reverse (Figs. 1-13 to 15).31 Although Buttrey interprets the reverse image as symbolizing Cosa's function as a port, which indeed may be implied, the particular combination of the horse and the dolphin belongs only to Neptune. Could it be that the god of the little lagoon temple built during the early history of the port site was Neptune himself, chief god of the sea and fresh water, whose festival, the Neptunalia, was one of the oldest celebrated in the Roman world? The Roman numen of springing water is Amphitrite, wife of Neptune and also goddess of the sea. In Roman art, Neptune and Amphitrite are often represented together with a sea thiasos, as in their marriage representation on the well-known Republican reliefs in Munich from the so-called Altar of Ahenobarbus. These controversial panels have been widely interpreted and dated. Inez Ryberg offers a convincing identification with Gnaeus Domitius Ahenobarbus, consul in 32 B.C. and famous as a naval commander in the civil wars following the death of Caesar. She believes the reliefs commemorate his naval victory of 42 B.C. and were placed in the Temple of Neptune in the Roman Campus Martius, which he either built or restored.32 If the temple above the Cosa lagoon also belonged to Neptune, its twin within the town walls may be associated with Amphitrite rather than Mater Matuta. It may also be significant that this special honoring of freshwater and sea deities is so closely associated with the Ahenobarbi, one of the wealthiest families of the late Republic and known to have owned vast property in the Cosa area.33 That they used the port we know from Caesar, who tells how the younger Lucius Domitius Ahenobarbus embarked troops, including some gathered from among his own slaves and

ermen see Radcliffe, pp. 125-129; G. Lafaye, "Piscatio et Piscatus," DarSag 4, pp. 489-494. 27 See above, J. P. Oleson, chapter five. 28 For inscriptional evidence of honoring of Neptune at the site of a Roman fish factory at Douarnenez, Brittany, see R. Sanquer, "Bretagne: Douarnenez," Gallia 33 (1975) p. 349; Sanquer, Gallia 35 (1977) pp. 335-360. See here chapter one, n. 185. Also see N. Lambogha, "L'ara a Nettuno dei pescatori di Pedona," RStLig 18 (1952) pp. 19-31. 29 For statue on coin of Nero see Testaguzza, p. 97. For illustration of relief, Casson, SSAW, fig. 144, dating ca. A.D. 200, here, Fig. VIII-4. For mosaic with statute of Neptune from Ostia, see G. Becatti, Stavi di Ostia. IV. Mosaici e pavimenti marmorei (Rome, 1962) no. 45, pp. 26-27, pi. 161. 30 For inscription on statue base in the forum of Cosa which may be restored as a dedication to Neptunus see Cosa Inv. C67.440, Bace, p. 71, no. ICl, fig. 6. The remaining letters fit a restoration of either Portunus or Neptunus and Bace favors the

latter. The inscription appears to date sometime before the 60s B.C.

For cult of Neptune in Etruria see L. R. Taylor, pp. 35, 173, 244. Portunus is absent from her study of the cults of Etruria. 31 See chapter one, the historical background; Buttrey, pp. 1727, pi. 2. 32 I. S. Ryberg, .Rites of the State Religion in Roman Art, MAAR 22 (1955) pp. 27-34, with earlier bibliography. For a date of the reliefs between 115 and 70 B.C., see R. Bianchi Bandmelli, Rome, the Center of Power 500 B.C. to A.D. 200 (New York, 1970) pp. 52-53, and fig. 52 for the Neptune and Amphitrite reliefs in Munich. For the popular use of these sea motifs on Roman sarcophagi see McCann, Roman Sarcophagi, vp. 76-78, and A. Rumpf, Die Meerwesen auf den antiken Sarkophagreliefs, Die antiken Sarkophagrehefs, V, 1 (Berlin, 1939). 33 Caesar BCiu. 1.34 (Lucius Domitius Ahenobarbus); I. Shatzman, Senatorial Wealth and Roman Politics, Collection Latomus,

134

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

freedmen to aid Marseille (Massilia) against Caesar in 49 B.C.34 While no inscriptional evidence has been uncovered directly linking any of the extensive building projects in the port with the Ahenobarbi, Will has suggested here that amphora Type 24a and the stamp SEX D O M may be associated with this family. One can imagine that the leading wealthy family of the region may well have been involved in such a large-scale project, which also must have brought in commercial profit. Nero, the natural son of Gnaeus Domitius Ahenobarbus, we know remodeled the basilica in the forum of Cosa into an odeon (Fig. VII-8), indicating a specific association of this family with Cosa, at least in the first century A.D. 3 5

The Sestii were another wealthy plebeian noble family known from the ancient writers to have owned property at Cosa.36 Their involvement in the life and commerce of the Portus Cosanus is documented by our excavations, in which great quantities of Sestius-type amphoras were found. The caduceus stamp, representing trade, may be explained as a reference to the successful enterprises of the Sestii during the prime years of the port's life. An identification of the temple in the lagoon with Neptune might also explain the otherwise puzzling frequent use of sea devices on the Sestius amphora stamps. These include several of the specific attributes of Neptune, namely, the trident (Color Fig. 3), hook, fish spine, anchor, palm, and pine bough. 37 Such an identification might also account for a condensed form of the myth of the Seven against Thebes in the small lagoon temple pediment, inspired by the larger pediment at Telamon previously discussed. Adrastus escapes death on his fabulous horse Arion, offspring of Poseidon and Demeter and once the horse of the god himself. 142 (Brussels, 1975) p. 339, no. 139 Evidence of the vast land wealth of this family is Lucius Domitius Ahenobarbus' promise of land grants to each of his soldiers, estimated at some ten to thirteen thousand (Caesar BCiv. 7.17.3). However, as P. A. Brunt has pointed out, probably not all of the Ahenobarbi landholdings were in Etruria, but included other parts of Italy as well as outside Italy ("Two Great Roman Landowners," Latomus 34 [1975] pp. 619-635). For location of their property in the Ager Cosanus see E. Galli, "Antiche vestigia nel dominio cosano dei Domizi Ahenobarbi," Historia 1 (1927) pp. 15-16; Dyson, JFA, p. 260, who cites inscriptional evidence for the possible location of one of their villas near Le Grotte m the Albenga valley. For a discussion of this and the other inscriptional evidence of the Ahenobarbi in the area see Manacorda, Athenaeum, pp. 80-91; Manacorda 1980, pp. 180-182; and Bace, pp. 27, 45, and 50, with notes. J. Balsdon, in Life and Leisure in Ancient Rome (London, 1969) p. 199, cites a local tradition that the name, Monte Argentano, derives from the Domitii Ahenobarbi who are believed to have owned property on the peninsula and were also bankers or argentarii. Cf. Santangelo, p. 76. The remains of the large seaside Roman villa at Santa Liberata

Finally, a further relationship between Cosa and Poseidon/Neptune may be recalled. Cosa's sister colony, Paestum, had been founded also in 273 B.C. upon the earlier Greek colony of Poseidonia, city of Poseidon, linked to the god by its name and coinage.38 Would it not be appropriate for Cosa to honor the patron god of their sister colony, an additional link to their counterpart in the south and a reaffirmation of their joint maritime purposes as Latin colonies guarding the Tyrrhenian coast and promoting its commerce? For all these reasons, coupled with the fact that Neptune, not Portunus, was the major god of the sea and fresh water, it seems most likely that the temple, erected in the colony's early period after the Second Punic War, was dedicated to Neptune. Only this god could have guarded and controlled both the fishing complex, with its springs, and the ships and sailors embarking and debarking in the port of Cosa below.

Catalogue LTl. (PC 68-71). (Fig. VI-5). Lagoon, Temple of Neptune (?). Surface, cliffs above quarry immediately east of Le Rocce Hotel. Fragment of architectural terracotta revetment plaque. Discolored and worn; original paint no longer preserved. Light buff-colored clay with coarse grog of red pozzolana. MPL 0.06 m; MPH 0.06 m; Th ca. 0.02 m. Fragment of three strigils from upper border of revetment plaque probably belonging to the original decoration of the small temple fagade. The concave strigils have rolled edges articulated by a central channel. They project slightly at their tops and are about 0.008 m wide. The fragment must have been associated by some with the Ahenobarbi: Santangelo, p. 86; Manacorda 1980, p. 174.The Maritime Itineraries do document a Domitiana positio in the Cosa area, but its exact location remains uncertain (It. Ant. Marit., 499.8). On the Domitius Ahenobarbus family as a whole see RE V, cols. 1315ff. (Miinzer); also Broughton 1951-1952, II, p. 332. 34 Caesar BCiV. 1.34. For recent evidence of the Domitii Ahenobarbi's involvement in trade, note the inscription of their name on a lead anchor in the Palermo National Museum; P. A. Gianfrotta, "Ancore 'romane.' Nuovi matenali per Io studio dei traffici manttimi," MAAR 36 (1980) p. I l l and fig. 26. 35 See chapter one, n. 152, above. For a newly identified portrait of Gnaeus Domitius Ahenobarbus on the famous Ravenna relief see J. Pollini, "Gnaeus Domitius Ahenobarbus and the Ravenna Relief," RomMitt 88 (1981) pp. 117-140. 36 Cicero Alt. 15.27.1; Will 1979, p. 349. 37 See Will's discussion and illustrations of stamps in chapter nine. 38 For bibliography on Paestum see "Paestum," PECS, pp. 663665 (W.D.E. Coulson).

VI: L A G O O N T E M P L E

belong to plaques decorated with looped palmettes similar to those found in the temple excavations in 1950. They are presumed to have decorated both the raking and horizontal roof timbers. 39 These plaques, with their looped palmette design crowned with a border of concave strigilation, exactly duplicate plaques of fabric C, Type B, that belong to the original decoration of Temple D on the arx in the town of Cosa above, as well as replacement plaques for Temple B from the forum. 40 No traces of color are left, but the strigils with looped palmette designs of Type A from Temple D were painted in repeated patterns of red, yellow, and blue with white borders. The intervening channels were painted black.41 Date: Second quarter of the second century B.C. LT2. (PC 68-69). (Fig. VI-6). Lagoon, Temple of Neptune (?). Surface, cliffs above quarry immediately east of Le Rocce Hotel. Fragment of architectural terracotta revetment plaque. Discolored and worn; original paint no longer preserved. Light buff clay with coarse grog of red pozzolana. MPL 0.11 m; MPW 0.053 m; Th 0.02 m. Fragment of looped palmette design from revetment plaque belonging to the original temple decoration. One complete rounded leaf of the seven-petaled palmette is preserved with sections of two other leaves and part of the encircling, concave serpentine loop. This particular revetment design, duplicated in plaques from Temples D and B in the hill town, became common in central Italy only later in the second century B.C. L. Richardson suggests that the Cosa plaques are thus the earliest of their type and represent an early, experimental stage in the development of the looped palmette design.42 No plaques of this type were found at Talamone, although other designs used for plaques from Temple D are duplicated at Talamone indicating the close relationship of these temples and the Cosa lagoon temple—all three probably of contemporary date. No trace of the original paint is preserved. The leaves from the palmettes of this type from Temple D were painted alternately red and blue, with the serpentine ribbon and ground within the loop painted white. 43 Date: Second quarter of the second century B.C. LT3. (PC 68-73). (Fig. VI-7). Lagoon, Temple of Neptune (?). Surface, cliffs above quarry imme39

Brown et al., pp. 204-205. Ibid., pp. 192-193, pis. 26, 1 and 26, 2 for reconstruction of revetment plaque from lagoon temple. « Ibid., p. 191. 40

135

diately east of Le Rocce Hotel. Fragment of architectural terracotta revetment plaque. Discolored and worn; no trace of original paint.. Light buff clay with coarse grog of red pozzolana. MPL 0.09 m; MPW 0.048 m; Th 0.02 m. Fragment of looped palmette design from revetment plaque belonging to the original temple decoration. One complete rounded leaf of a palmette is preserved with outlines of two others and the edge of the encircling serpentine loop. Compare Cat. LT2. Date: Second quarter of the second century B.C. LT4. (PC 68-70). (Fig. VI-8). Lagoon, Temple of Neptune (?). Surface, cliffs above quarry immediately east of Le Rocce Hotel. Fragment of architectural terracotta revetment plaque. Discolored and worn; original paint no longer preserved. Light buff clay, with coarse grog of red pozzolana. MPL 0.09 m; MPW 0.065 m; Th 0.02 m. Fragment of looped palmette design from revetment plaque belonging to the original temple decoration. Preserved are two sections of opposing, slightly concave loops, which encircled the palmette design. The loops are 0.02 m wide. Compare Cat. LT2. Date: Second quarter of the second century B.C. LT5. (PC 68-74). (Fig. VI-9). Lagoon, Temple of Neptune (?). Surface, cliffs above quarry immediately east of Le Rocce Hotel. Fragment of architectural terracotta revetment plaque (?). Discolored and worn; no trace of the original paint. Pink clay with fine pozzolana grog. MPL 0.026 m; MPW 0.25 m; Th ca. 0.015 m. Probably a fragment of the architectural decoration from the original temple. The clay appears to be that of Type B used in the decoration of Temple D on the hill of Cosa.44 The fragment has a rounded surface with a portion of a raised edge preserved. While the fragment is too small for precise identification, it may be a section of a leaf of a palmette from one of the decorative revetment plaques similar to Cat. LT2. Date: Second quarter of the second century B.C. LT6. (PC 68-72). (Fig. VI-10). Lagoon,Temple of Neptune (?). Surface, cliffs above quarry immediately east of Le Rocce Hotel. Fragment of architectural sculpture (?). Pink clay with fine pozzolana grog. Surface worn with no traces of original paint preserved. MPL 0.075 m; MPW 0.05 m· Th 0.02 m. «Ibid., pp. 193, 195. «Ibid., p. 193. «Ibid., pp. 184-185.

136

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

The small fragment shows a rounded, oblong form which is concave on the interior suggesting that it originally belonged to the sculptural rather than the architectural decoration of the temple. The clay appears to be that of Type B used for the scultpural decoration of Temple D on the hill.45 One side of the raised form is outlined with a section of a flat form that resembles a feather or scale pattern. Perhaps the fragment belonged to the armor of a warrior from the pedimental scultpure. See the discussion of the pediment design above. Date: Second quarter of the second century B.C. LT7. (PC 68-67). (Fig. VI-Il). Lagoon, Temple of Neptune (?). Surface, cliffs above quarry immediately east of Le Rocce Hotel. Fragment of a terracotta female figure from the pediment (?). Discolored and worn; no traces of paint preserved. Light orange clay with coarse pozzolana grog. MPH 0.14 m; MPW 0.12 m; Th ca. 0.03 m. The fragment shows the upper torso of a draped female figure of about one-quarter life-size running toward the right. Her clinging chiton is high-belted with one flowing ribbon of the sash preserved. On the interior is a knob for the figure's attachment, probably to the pediment of the temple. Only one other figure has been preserved from the original pedimental sculpture of the small temple—a nude and helmeted warrior found in the 1950 excavations. He is made of a similar fabric and is about half life-size. Both are an appropriate size for a small pediment estimated to be about 7 m in length with a maximum height of about 1 m.46 Adding now to the original lagoon temple sculpture a striding or running female figure in slightly smaller scale suggests a further connection with the contemporary temple at nearby TeIamon, already cited as a source for the architectural decoration of Temple D on the hill of Cosa—the twin in form and size of the lagoon temple. The subject of the Telamon pediment has long been identified with the Theban cycle in Greek mythology—the defeat of the Seven against Thebes. 47 45

Ibid. Ibid., pp. 330-333, figs. 20-21. For temple reconstruction see Brown et al., fig. 109. 47 Milani, Museo archeologico (cited in n. 8) pp. 66-69, 257-260, pis. 114-116; Andren, I, pp. 228-238; II, pi. 82, no. 285; Von Vacano 1961, pp. 9-63; Von Vacano 1969, pp. 141-161; "Telamon," PECS, p. 891 (O. W. Von Vacano); Martelh, in GIi etmschi (cited in n. 6) pp. 278-281, fig. 273. 46

Besides the heroes and their chariots, the pediment also preserves fragments of other figures including warriors and smaller winged female daemons. Although in the background of this pictorial pediment design may well be a lost Greek painting, the insertion at Telamon of typical Etruscan winged furies appears to be a creative contribution of the Etruscan artist himself. They may be interpreted as defenders of the city perhaps also present in the original prototype in a different form.48 For a proposed reconstruction of the Telamon pediment see Von Vacano 1969, foldout opposite p. 142. For our female figure, compare particularly the hovering winged fury dressed in a long, high-belted chiton with one breast bare and holding a down-turned torch in her right hand from panel xvm. She is very like the Cosa torso but in reverse. Although not enough of the Cosa figure is preserved to tell whether one breast was bare— a typical feature of Etruscan daemons—its association with the warrior, which has been specifically connected with a figure from the Telamon pediment, 49 strongly argues for an identification with such a winged fury. The appropriateness of this particular Greek myth for the decoration of a temple dedicated to Neptune is discussed in the introduction above. The style of the figure's softly clinging drapery, however, differs from that of the sculptures on the pediment at Telamon. Rather, the fluid style of the Cosa figure is closer to Hellenistic Greek models; some female figures on the reliefs of the great Altar of Zeus at Pergamon may be cited. As one of the seven wonders of the ancient world, this monument strongly influenced both Etruscan and Roman art and has already been associated with the style and iconography of the Cosa warrior. 50 This stylistic connection suggests a different artist for the Cosa lagoon temple pediment—a Roman artist who drew upon both the regional Etruscan as well as the Hellenistic Greek artistic traditions. Date: Second quarter of the second century B.C. 48

Von Vacano 1961, p. 63. See discussion above in introduction to this chapter. Compare particularly the drapery style of the figure of Athena from the Altar of Zeus, Kahler, pi. 3. For comparison with the warrior of the lagoon pediment see Brown et al., p. 332 (E. H. Richardson). For influence of the Pergamene school on Roman art in the second century B.C., see recent meaningful discussion by Martelli in GIi etmschi, p. 279. 49 50

Chapter VII. Reconstruction and Function: Port, Fishery, and Villa ELAINE K. GAZDA AND ANNA MARGUERITE M C C A N N

RECONSTRUCTION OF THE PORT 1

The underwater excavations in the Portus Cosanus established the harbor plan, with its protecting massive rock breakwater on the south and its exten­ sions in the form of isolated rock islands curving in a semicircle on the east (Map 6; Figs. IV-I to 10).2 They protected a harbor basin encompassing some 25,000 sq. m. While small in comparison to later Im­ perial harbors, that of Cosa is roughly comparable in size to the Roman port of Anzio o r Kenchreai, the eastern port of Corinth. 3 Although the main break­ water and its extensions lie submerged today in places to a depth of ca. 3 m, they presumably rose above water in antiquity. Lewis has suggested that the breakwater originally was laid on a sand foun­ dation which has washed away, causing the blocks gradually to sink down to the bedrock surface, which we were in places able to reach with the jet probing device. Thus, 1-2 m of sand have eroded away, owing mainly to wave action and possibly to marine borers. With a postulated sea level change of about 1 m probably caused by eustacy rather than by tectonic subsidence, we arrive at an approximate total change in depth of 3 m. 4 Lewis's extensive study of the breakwater system 1

In this chapter, McCann is largely responsible for the sections on the port, and Gazda for those on the fishery and villa. McCann has also contributed the material on the fishery at Javea, as well as the references on modern fishing practices. For the suggested reconstruction of the port, McCann is especially indebted to Lew­ is's study for the engineerii.-g and to Casson's suggestions con­ cerning ancient ships and navigation. Casson has also kindly ed­ ited the section on the port reconstruction. See Lewis 1973, pp. 233-259, and Casson, SSAW, pp. 361-370. 2 See chapter four for full description and measurements of the breakwater system, and chapter three for detailed discussion of the underwater excavations. 3 Both Kenchreai and Anzio are estimated to have enclosed ca. 30,000 sq. m. The Trajamc port of Portus, by comparison, is estimated at ca. 333,000 sq. m; the outer harbor at Caesarea Maritima, ca. 200,000 sq. m; Leptis Magna, ca. 102,000 sq. m; and

has revealed that its plan was based upon considered engineering solutions designed for both the safety of ships within the harbor basin and for the reduction of silting effects caused by the placement of artificial structures within a natural environment. 5 Although Cosa was a relatively small port, its design and the materials used indicate a sophisticated knowledge of previous harbor works as well as familiarity with local navigational conditions and needs. The main platform of the breakwater extended directly east­ ward from the cliffs for a distance of ca. 110 m with a width of about 70 m at the point estimated to be its broadest. Its length would have allowed extended berthing space on its inner, or harbor, side, while its breadth permitted the force of the waves to be bro­ ken gradually over and through its surface, allowing needed currents to enter into the anchorage area. The rough, angular rocks quarried from the lime­ stone cliffs flanking the harbor's western side were also more effective in absorbing the force of the waves than ashlar blocks with plane surfaces—so loved in the harbors of the Greek world—which are not only more difficult and costly to build, but can be easily displaced.6 Over the years, a rubble break­ water also becomes stronger as the bond between the rocks is increased by concretions and debris. The Centumcellae (Civitavecchia) ca. 100,000 sq. m. For these and other measurements of ancient harbors, see Scranton, Kenchreai I, p. 14, n. 2 (J. W. Shaw); for Portus, see Testaguzza, p. 58; for Caesarea, Raban and Hohlfelder, "The Ancient Harbors of Cae­ sarea Maritima" (cited in Introduction, η 5, above) p. 59. 4 See Lewis 1973, pp. 241-242, and personal letter, November 1981; Bourgeois, chapter two, for discussion of eustatic change in sea level at Cosa. For examples of other ancient breakwaters which have also subsided under their own weight, helped by ero­ sion, see Blackman 1982-2, pp. 196-197. 5 Lewis 1973, pp. 237-242. 6 Note, for example, the famous mole at Samos, built by Polycrates ca. 530 B.C. (Herodotus 3.60). For discussion of Samos and other Greek harbors see chapter nineteen. For a recent study of ancient breakwaters with further bibliography see Blackman 1982-2, pp. 196-199.

138

HISTORY, GEOGRAPHY,

RCHITECTURAL REMAINS

placement of Cosa's breakwater units likewise reveals that the harbor engineer was familiar with the local wave and current directions. He purposely angled the main breakwater 45° away from the oncoming waves from the southwest so that only a portion of the platform was under the force of a wave at one time. The offshore currents that travel up the coast in a northwesterly direction would have been directed around the breakwater extensions and also allowed to circulate through the loosely packed rocks of the main platform. In this way, a scouring action would have been achieved, a necessity to prevent sedimentation in a single-entrance harbor like Cosa's.7 The Romans apparently knew about dredging, but the methods used must have been limited and never sufficient to cure the problem. 8 Natural solutions were thus imperative and their success at Cosa may help explain why the port there remained active for over six hundred years despite its sandy coastline environment and hence ever-present danger of sedimentation. While the lowest and largest blocks of the main platform rest today on a sandy sea floor, probes through this floor suggest that the main breakwater as well as extensions A, B, C, and F may lie upon a natural bedrock platform now below 1.50 m under the present ocean floor.9 Such a preexisting bedrock support for the cut rock might explain in part why the harbor did not silt up more rapidly, as well as account for the trapezoidal design of extensions A, B, C, and F. Further, this bedrock ledge would have acted as a baffle against sand deposits within the port by directing offshore currents up the coast. In any case, the type of broad rubble breakwater found at Cosa may have its immediate heritage in the Italic world. The recent underwater excavations of the Etruscan port of Populonia have brought to light a similar type of broad platform still functioning as the protection for the modern anchorage.10

The entrance channel for ships into the protected harbor basin of the Portus Cosanus was established between breakwater extensions D and F where an unobstructed distance of 33 m was recorded. In the center of this gap a deeper channel was found extending at least 3.50 m below the present sea floor. The estimated ancient water depth of this central channel is 6 m, ample clearance for any Roman merchant vessel of size seeking protection and anchorage in the Portus Cosanus. The ship carrying the Sestius jars of Type 4a wrecked off the Grand Congloue is estimated to have measured between 15 m and 20 m in length, and, with a cargo of some 1,200 jars, about 40 to 60 tons burden. 11 Through this channel, ships could have entered and departed without excessive maneuvering, sailing ships under reduced sail if the wind was favorable, galleys under their oars. Ships could also have been warped into the harbor either by their own ship's boat or by oared tugs attached to the harbor. The orientation of the harbor entrance, east-southeast by west-northwest, would have provided ships with clearance from the limestone cliffs on the west and also comfortable distance from the beach to the north. The prevailing southwesterly summer winds would have allowed a ship to enter closehauled on a port tack and depart on a starboard broad reach. These are almost ideal conditions for any sailor, ancient or modern. Inside the harbor basin, immediate shelter would have been provided by the breakwater extensions A, B, C, and F as well as the long arm of the main breakwater itself. Larger ships could have anchored within the protected basin or been moored along its outer Piers 4 and 5 as well as directly to the larger boulders of the inner side of the breakwater itself, a common practice in harbors today (Figs. VII-I, 2). Some floating dock units for the mooring of small boats might have extended out from the inner, northern side of the breakwater, piers, or beach.12 For ships

7 For other ancient solutions to the problem of silting in singleentrance harbors, note, for example, the use of rock-cut channels with flushing basins carved into the breakwater at Sidon (Introduction, n. 20 above, for references on Sidon). In the recent excavations at Caesarea, Israel, channels have been found cut through the large Herodian breakwater for desilting devices (Introduction, n. 67 above, for references on Caesarea). 8 Blackman 1982-2, p. 199, for reference to dredging at harbor of Ephesus. 9 See evidence of underwater excavations presented in chapter three, and Bourgeois, chapter two, n. 25; cf. de Boer in chapter sixteen. 10 McCann 1971, and McCann, Bourgeois, and Will. 11 See Long, chapter eight, Appendix, "The Grand Congloue Site: A Reassessment." 12 It is clear from sources m literature and art that the Romans could engineer elaborate floating docks and bridges For example, note Xerxes's famous bridge built across the Hellespont by joining

warships together (Herodotus 7.36) or Caesar's wooden bridge across the Rhine (BGaIl 4.17). Caligula built a bridge of merchant ships extending from the mole at Puteoli to Baiae, a distance of over 3½ miles (Suetonius CaMg. 19). On the method used in building such bridges, see Arrian Anab. 5.7. For illustrations of pontoon bridges, see, for example, bridges on Trajan's column, Holland, Janus and the Bridge (chapter six, n. 17 above) pp. 298-299, fig. 14; a bridge of ships perhaps over the Nile on a Roman mosaic from Ostia, Becatti, Scavi di Ostia. IV (chapter six, n. 29, above) pp. 74-76, no. 108, pi. 84. Cf. H. F. Cornick, Dock and Harbor Engineering (London, 1962) vol. 4. Concerning the mooring of ships at Cosa, no remains of bollards or stone mooring rings such as have been found in many later Roman harbors were uncovered in our excavations at Cosa (Blackman 1982-2, pp. 203-204). Likewise, at the Roman ports of Kenchreai and Caesarea no traces of such standard equipment have been found (Scranton, Kenchreai I, p. 15 and n. 5 Q. W. Shaw]). It must be assumed, as suggested for Kenchreai, that such

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

139

anchored in the basin, loading and unloading would have been carried out by the use of barges or lighters working from the shore. 13 Or those that tied up to the piers could have been unloaded by cranes placed on the piers.14 And small craft anchored in shallow water could have been loaded by men wading out from the beach.15 Craft able to pass between Piers 3 and 4—a width of about 36 m—could have anchored or pulled up on the beach behind Piers 1, 2, and 3 within the sheltered "inner harbor" basin to the west. Now partially filled by a modern concrete platform, in antiquity this enclosed area between the cliffs and the piers must have been larger, probably encompassing an area of at least 30 square meters (Map 6). Today, the piers still provide a calm body of water where small dories draw up on the beach. Perhaps this area served as a private harbor for the Sestii within the larger commercial one. Such division of harbor space was not uncommon. 16 During the floruit of the fishery in the last three quarters of the first century B.C., Piers 1, 2, and 3 also served as a protective barrier with Wall M for the fish entering the channel to the lagoon (Fig. IV-Il to 17). Besides protection, all these piers could also have served to hold mooring lines. Since there is little tidal change in the Mediterranean, it is estimated that there would have been sufficient depth for medium-sized craft even in this shallower section of the harbor.17 That cargoes were also unloaded in this area seems clear from the low, rectangular platform, labeled Pier 1.5, located between Piers 1 and 2, whose surface was just above sea level in antiquity (Fig. IV-13). So placed, it would have served as an ideal support for a gangplank level with the gunwale of a ship as pictured in the famous early third-cen-

tury-A.D. Roman relief from Ostia of the harbor of Portus (Fig. VIII-4).18 During the most active period of the lagoon fishery in the last three quarters of the first century B.C., at the seasonal times of the ascent or descent of the fish to be trapped within Channel M, probably most of the loading of boats with cargo took place in the eastern area of the harbor away from the entrances to the fish channels leading into the lagoon. 19 In any case, this eastern area of the harbor would have been the closest to the commercial center or emporium of the port (Fig. VII-10). The size and design of the ancient harbor at Cosa with its single-entrance channel thus reveals a facility that in all its periods of use satisfied the requirements of medium-sized merchant- or warships for entering, loading and unloading, and departing. Large ships could have anchored in deep water outside the harbor basin where in a storm they could quickly maneuver away from the cliffs and breakwater (Text Fig. 1-1). One other element—essential to any working harbor, ancient or modern—is a lighthouse or beacon tower. The earliest example of a lighthouse in the ancient world is the famous Pharos of Alexandria, built in the early third century B.C. However, smaller beacon towers may well have been used at harbor entrances before this time. 20 In any case, the addition of the lighthouse to standard harbor works is a particularly Roman development. They were normally placed at the end of a breakwater or upon a separate island within the harbor entrance, as we can see from the numerous illustrations of them in Roman art. The bases of a very few of these ancient lighthouses are also known from archaeological remains.21 It is thus natural to assume that from its

devices were made of wood rather than stone or iron and have thus disappeared. At Cosa, as suggested in the text, ships could also have cast lines around the boulders of the main breakwater. 13 L. Casson, "Harbor and River Boats of Ancient Rome," JRS 55 (1965) pp. 31-39, and here, Casson, chapter eight. This was common Roman practice in earlier harbors, and lighters were used exclusively at Ostia until the Claudian port was built m the first century A. D. 14 Literary evidence for the use of cranes in loading and unloading Roman ships is in Vitruvius 10.2.10. The Romans called them "storks" rather than "cranes"; cf. Casson, SSAW, p. 370, n. 42. See Blackman 1982-2, n. 103, for possible base of crane found at York and graffito at Sabratha showing a floating crane. 15 For a representation of a ship being unloaded by men wading directly through the water see Roman mosaic from Sousse, Casson, SSAW, fig. 191. Also illustrated in L. Foucher, Inventaire des mosaiques. Feuille no. 51 de I'atlas archeologique, Sousse (Tunis, 1960) pi. 41. 16 For example, note the private harbor of the Ptolemies within the eastern harbor at Alexandria and the "Secret Harbor" at HaIicarnassus, Blackman 1982-2, pp. 188-189. For the separation of military and commercial harbors there is much evidence, first presented in Lehmann-Hartleben's basic

study of ancient ports. The most famous harbor of this type was that of Athens at Piraeus. For other examples also see Blackman 1982-2, p. 189. There is no evidence, however, such as shipsheds or slips, that this "inner harbor" at Cosa ever served military needs, and the area, in any case, is too small for warships 17 It is estimated that in this area around Piers 1, 2, and 3—it is now in part covered with scattered rock and hence an exact estimate is difficult-—there was in antiquity up to 2 to 3 m of water, which would have been adequate for medium-sized merchantmen, certainly those of shallow draft. On shallow depth of ancient harbors, see Blackman 1982-2, p. 193. 18 See illustration in Casson, SSAW, fig. 144. 19 See discussion of Wall M by Gazda in chapter four. 20 See suggestion of Blackman 1982-2, p. 207. For illustration of such a beacon tower in a Pompeiian painting from Stabiae of the first century A.D., see Blackman 1982-1, fig. 5. Numismatic evidence of terracotta lanterns on the roofs of the quays in the Roman harbor Pompeiopolis is in Boyce, p. 69. Also see evidence of light tower at Gorsium, T. G. Radan, "Angaben zur Frage der sogenannten Leuchttiirme," Gorsium Forschungen 1 (1974) pp. 149-157. 21 See Zancam Montuoro, pp. 5-21, for study of ancient lighthouses, with bibliography, and most recently, Blackman 1982-2,

140

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

earliest period of use the port of ancient Cosa had at least some kind of simple beacon. Archaeological evidence for such a light during this early period is lacking. But from the second architectural phase of the site's development we may have some important evidence. We have suggested that Pier 5 may have served as the foundations for such a small lighthouse or simple beacon tower, since it is so appropriately placed, isolated from the other piers at the end of the main breakwater, and 55 m east of Pier 4 (Map 6; Fig. IV-15). Here is the ideal place for a marker to warn ships of the long artificial barrier and the channel entrance to the east. Now Dr. Zancani Montuoro's new, fascinating, and convincing identification of a small votive terracotta tower model from Vulci as the Portus Cosanus lighthouse would seem to support our archaeological evidence (Fig. VII-3).22 The votive model may be dated by the other material from the deposit and its stratigraphy to the first half of the first century B.C., and Zancani Montuoro feels it might even go back to the late second century B.C.23 The reconstructed height of this simple, square model is ca. 0.30 m, with the sides each 0.075 m. Zancani Montuoro projects an original height for the Cosa lighthouse, on the basis of her studies of such ancient structures, as about 30 m with a base of about 7.50 m on a side. Pier 5, now submerged to a depth of ca. 2.50 m b.s.l., is badly eroded and underwater probes around it indicate that it extends under the sand for at least 1.50 m to the south. Its visible preserved core now measures ca. 4.80 (eastwest) by 4.30 m (north-south). Thus, Pier 5 appears to have been originally a good deal larger and of an appropriate size to carry such a beacon tower. It seems very likely that the little model from Vulci is precious archaeological evidence for the light of the Portus Cosanus, only the base of which remains today, under the sea. Built during the port's apex, it may have become, like the Pharos of Alexandria, her identifying symbol.

also be used to corroborate the probable date for the construction of the harbor's concrete piers, since it seems most likely that the inspiration for such a stamp derived from a contemporary monument. Our underwater excavations have thus added new archaeological evidence for this type of beacon tower in the Roman world, an architectural form not otherwise well documented. Through a series of caisson trenches as well as open trenches, it was established that the ancient harbor floor of the port of Cosa lay between 1.0 m and 2.10 m below the present ocean bed or at 5.10 m to 6.20 m below the present sea level, 4.10 m to 5.20 m below the estimated ancient sea level (Figs. UI-15 to 22). Ancient use of the harbor is represented by a layer of Roman sherds in compact, brackish sand and gravel. Beneath it lies the previous sea bed of dark, sterile sand about 6.20 m below the present sea level. A shallow layer of rotted vegetation occurs in Trench D l on the western side of extension D from 5.0 m to 5.10 m below the present sea level and just above the archaeological layer (Fig. Ill—15). Trench D2 on the eastern side of extension D also revealed a shallow, disturbed vegetation layer but at the base of the archaeological layer, 5.80 m to 5.90 m below the present sea level (Figs. Ill-16 to 18). In the stratigraphy of Caisson 1 (Cl), the rotted vegetation layer is thicker and occurred just above the ancient level at 4.85 m to 5.25 m below the present sea level (Fig. 111-22). The thin vegetation layers in Dl and D2 suggest heavy storm disturbance, apparently both occurring before the ancient use of the harbor and also just following. The thicker vegetation layer in Caisson Cl may derive from a postRoman period when the harbor was idle and seaweed accumulated within the protected basin.

The Sestii may have used the light of Cosa for one of their stamps, the rare one found on amphoras of Type 4a, Cat. A79 (Figs. IX-122, 123, 124).24 If so, then this is additional archaeological evidence for the Cosa light and additional support for Will's thesis of a Sestius pottery in the harbor area. Furthermore, since Will dates the floruit of Type 4a from the late second century B.C. through the first quarter of the first century B.C.—with some material occurring as late as the mid-first century B.C.—this stamp may

Study of the archaeological material found in the port area alone, both in the underwater trenches and sporadically, indicates that the harbor came into use by the time of the colony's founding in 273 B.C. and continued in use into at least the mid-third century A. D. Owing to the nature of the underwater environment, time differentiations between the various layers of sherds in the archaeological level could not be established with any consistency. It is clear, however, from the architectural evidence presented in chapter four by Gazda, that the port underwent three main building stages. The breakwater with its extensions probably belongs to its first period and represents the earliest permanent installation, con-

pp. 207-208. Also see S. Loeschke, "Antike Laternen und Lichthauser," Bonnjbb 118 (1909) pp. 370-430; G. Stuhlfauth, "Der Leuchtturm von Ostia," RomMitt 53 (1938) pp. 139-163. 22 Zancani Montuoro, pp. 5-21. Also see McCann, chapter six,

n. 15. 23 See T. Dohrn in Helbig4, III, no. 2535, pp. 508-510, with earlier bibliography. 24 See Will, chapter nine, and McCann, chapter eighteen.

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

structed sometime during the course of the third or early second century B.C. (Fig. VII-9). It is also clear from the architectural evidence that the concrete piers followed the construction of the breakwater and are probably to be associated with the port's busiest period of trading activity from the last quarter of the second century B.C. until the end of the first century B.C., documented by the overwhelming accumulation of amphoras of Types 4a and 4b (Text Fig. 1-1; Figs. VII-IO, 11, and 12). When, within this period of time, the concrete structures of the harbor were constructed will be further discussed in the chapter on chronology. Beginning with the first century A.D., little amphora material was found, with more fragments beginning again in the later first, second, and third centuries A.D., evidence for a time of renewed life but not anywhere near on the previous scale. During this last and Imperial period of the harbor's life, one must think of a port largely for the private use of the large maritime villa now on its shores (Fig. VII13). Concrete Walls C, D, and E extending from the beach into the harbor have been associated with this third architectural phase of the site (Fig. IV-18). These narrow walls probably carried wharves or served as ship magazines for private,yachts. Piers 1, 2, and 3 appear to have been reused during the villa period, for their upper concrete sections probably show repairs to the original tuff-and-pozzolana bases, and they may have been heightened to add some superstructure (Figs. IV-Il, 12). Their staggered arrangement precludes the reconstruction of arches in any period but they could have carried wooden platforms in all phases of their use. Whether the fish channel embanked by Wall M and Piers 1-3 was in use during the villa period is not certain. But its entrance was not closed off by .the villa construction as was that of the channel to the east, embanked by polygonal Walls A and B (Map 6). The major elements of the fishery in the lagoon, however, do not appear to have been in use during this last period. The amphora finds from Imperial times are also largely imported types with the exception of Type 12, indicating that the Portus Cosanus was no longer a commercial center, no longer an export center.

RECONSTRUCTION OF THE FISHERY

From the fact that the remaining vestige of the old lagoon of Cosa at Lago di Burano functions today 25 26

De Angehs 1959, p. 3. Note theories of E. Rodenwaldt and H. Lehmann, discussed

141

as a fishery for certain species which migrate between the sea and lagoon, it may be assumed that the sheltered body of water at the base of the Cosa promontory provided a similar natural feeding ground utilized by the ancient Cosans from their earliest colonial days. According to hydrobiologist R. De Angelis: ". . . the use of brackish water lagoons has always depended on the extent and continuity of their communications with the sea and on the fertility of the sea bottom with which they are in direct relation." 25 At the Lago di Burano fishery today, it is necessary at times to use a bulldozer to open the natural channel through the sand barrier to the sea or a pump to promote the circulation and mixture of fresh and salt water. The ancient Cosans at the western end of the lagoon were favored by their limestone promontory with its natural fractures and freshwater springs (Map 6; Fig. 1-4). These cliffs allowed them to create a permanent link through the natural rock between the sea and the brackish lagoon, which was not possible elsewhere along the length of the lagoon's sandy barrier—originally stretching some 20 km eastward to the mouth of the Tafone River (Map. 2). In dry seasons, when the level of the lagoon was lower than that of the sea and the threat of stagnation was greatest, the Tagliata would have ensured the flow of seawater into the lagoon (Figs. IV-30 to 43). This inward current would have been all the stronger during storms or high tides. Conversely, in rainy seasons, owing to runoff from the surrounding hills and increased activity of the springs along their base, the water of the lagoon would have risen and flowed vigorously toward the sea. Indeed, one of the subsidiary functions of the Tagliata must have been to aid against the ever-present problem of lagoonal silts being deposited within the harbor basin.26 The lagoonal fishing operations would have been most active during periods when there was a substantial flow through the Tagliata channel, for migrating species of fish tend to swim against the prevailing current, attracted to a change in the temperature and salinity of the water. Beyond providing a stable inlet-outlet, sluice gates within the Tagliata allowed for a measure of regulation of this natural regime (Fig. IV-31). A number of rock-cut details at the southern, seaward end of the Tagliata suggest that provision had been made for controlling the flow of water and the movement offish (Fig. IV-38). This seaward end was designed primarily to break the force of high waves with the high rock wall above the mouth acting as an effecby McCann in chapter three.

142

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

tive barrier to the surf. The tall air shaft cut through this vertical mass of rock just above the tunnel may have aided in this task by diverting some of the turbulent water upward through the shaft as it moved through the tunnel toward the channel (Figs. IV-31, 32). Even so, however, during stormy weather, water gushes so forcefully through the tunnel that further means of control must have been necessary. It is highly likely that some of the rock-cut recesses in the walls of the Tagliata just to the north of the inlet-outlet date to the first period and that they held one or more devices to control the variable flow of water as well as the passage of fish. Both would have been needed for the successful operation of the fishery. At the southernmost end of the Tagliata are cuttings on either side of the channel for two horizontal beams, at roughly 2 m a.s.l. (Fig. IV-38). Just to the north of them a third beam apparently traversed the channel at a slightly higher level than the first two. Adjacent to the third beam was a sliding mechanism, the frame of which was held in long vertical slots. While it is not possible to be certain of all of the details of this mechanism, the arrangement of the cuttings suggests the following reconstruction. Two beams, spaced 0.45 m apart, spanned the channel, probably to support a wooden platform from which a sluice gate, situated in the long vertical slots, could be operated. The height of the proposed platform would have corresponded roughly to that of a rockcut ledge along the eastern wall of the channel, which could have provided access to the platform. In times of high seas, the gate could have been lowered in order to prevent excessive turbulence within the channel, or raised if it was desirable to allow seawater to flow northward toward the lagoon. On calm days, the gate might have been replaced with a screen or grate of cane or "brass with small holes," the latter especially recommended by Columella to prevent mature fish from escaping to the sea.27 Some comparisons with modern fishery installations allow us to visualize further their ancient forerunners (Fig. VII-4). De Angelis, describing a fishery in Sardinia in the S'Ena Arrubia lagoon, notes the value of having water-and fish-control devices incorporated into a single structure. Such a system "holds the gates with their working parts and at the same time supports a screen system that prevents the escape of the fish and catches them as they de27 Columella Rust. 8.17.6, trans. E. S. Forster and E. H. Heffner (London, 1954). AU translations from Columella quoted hereafter are taken from this edition. 28 De Angelis 1959, p. 14.

scend."28 The suggested wooden platform from which to operate such a device also has a parallel at the same fishery. De Angelis, who designed these installations, points out that "the piles have reinforced concrete ledges which can support a movable wooden plank on which the men controlling the sluices can move about. They also have rotating metal ledges on which planks can be laid so that the fishermen can walk on them to change the position of the screens. We avoided using fixed catwalks so as not to produce permanently shaded areas to mislead the fish that are swimming upstream, as these are the fish that represent the wealth of the lagoon." 29 Another advantage is that "In daily practice, the manager of the fisheries can quickly rearrange his installation according to the needs of the moment . . Z'30 Such a system may have been used in the Tagliata through which certain varieties of migrating fish would surely have entered the lagoon while others that favored rocky environments would have hovered about the walls of the channel of the Tagliata itself. Even today the mouth of the Tagliata is a favorite spot for local fishermen, who perch on its seaward wall and cast their lines into the ocean. That circulation of water in the main channel was a primary concern to the ancient Cosans is indicated both by the cutting of the second seaward outlet, the Tagliata Piccola, and by the presence of three diagonal passages in the eastern wall of the Tagliata South. These must have been screened off to prevent fish from escaping while allowing water to pass to and from the sea. The Roman fishery at Javea on the southern coast of Spain in Alicante,31 seen and studied by McCann, provides a clue to the primary function of the southern Tagliata and the Tagliata Piccola. In the center of the bay at Javea, near the the mouth of the Gorgos River, remains of a Roman fish factory have been found, dating from early Imperial times (Text Fig. VII-I). Cut into the promontory known as the Punta de l'Arenal, which divides the bay, is a large fish tank measuring 28 m by 7 m with a depth of 4 m. This man-made basin is connected to the sea by two cut channels, each about 40 m long. The larger mouth of the main channel, 1.60 m wide, is perpendicular to the sea, facing the wave direction from the northeast (Fig. VII-5). Fresh seawater thus continually flows into the tank. The second, narrower channel, 0.50 m in width, angles obliquely off across 29 30 31

Ibid., p. 15. Ibid. For bibliography on Javea, see McCann, chapter one, n. 230.

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

143

Text Fig. VII-I. Plan of Roman fishery and saltery at Javea, Spain. Fish tank connected to sea by two channels cut through rock in left center. Remains of fish factory with garum vats and dolia, center right; houses, right (A, B, C, D). G. Martin, "Las pesquerias romanas de la costa de Alicante," Trabajos de arqueologid dedicados a D. Pio Betran. Papeles del laboratorio de arqueologia

de Valencia, no. 10 (Valencia, 1970), fig. 2. the promontory to the northwest. Furthermore, its mouth is cut about 1 m lower than that of the larger channel, forcing water from the tank to flow out to sea. There are cuttings for sluice gates at the inner mouths of both channels. In this manner, the fish tank, which is divided into three sections by two cross walls, is provided with continuous water circulation so essential for any fishery (Text Fig. VII1). M. Ponsich and M. Tarradell suggest that the Javea fish tank was used as a holding unit before processing in the nearby factory.32 G. Martin, however, suggests that it was primarily used for raising and feeding fish.33 Both these functions appear possible and compatible. The divisions within the tank may have further served to segregate the smaller fry from the larger fish awaiting processing. The seaward mouths of the Tagliata and the Tagliata Piccola face the sea in a manner similar to the channels at Javea (Figs. IV-31 to 33). The larger mouth of the Tagliata South faces the main wave direction from the southwest. The narrower channel of the Tagliata Piccola, about 3 m to the northwest, angles away from the waves and empties into a small 32

Ponsich and Tarradell, p. 81.

cove (Figs. IV-31, 33, 39, and 40). As at Javea, its seaward mouth is 1 m lower than that of the Tagliata South, a deliberate device that forces an outward flow of water under normal sea conditions. Thus, continual circulation was provided for the enclosed fish tank contained within the walls of the Tagliata South. Regulated by sluice gates, the area between the two rock cuttings within the open section of the Tagliata South formed a small catching basin, measuring about 10 m in length with a width varying from 4 m to 5 m (Fig. IV-31). Fish could thus either be held in this area or allowed to swim up the channel to the lagoon or back out to sea. Certain species, like the wrasse, which favor a rocky environment rather than the brackish lagoon, would be allowed to circulate and feed within the rocky channels, until they were large enough to be caught. The Tagliata's main purpose thus seems clear: to attract migratory fish into the lagoon, to protect and nurture other species within its rocky walls, and to serve as a small trapping basin for both. A secondary function of the Tagliata also emerges: to aid water circulation while allowing at least some of the silt— 33

Martin, Trabajos (cited in chapter one, n. 230) p. 144.

144

HISTORY, GEOGRAPHY, A R C H I T E C T U R A L R E M A I N S

laden waters of the lagoon to pass safely through its shallow channel to the sea—well away from the harbor. During the first stage in the use of the port and lagoon, fishing on a larger scale must have taken place somewhere other than in the narrow and highwalled southern section of the Tagliata. One can imagine larger traps set up along the shore within the presumed natural inlet-outlet between the lagoon and the sea within the harbor area (Fig. VII-9). It is common today, as it must have been in antiquity, to trap the mature fish both entering the lagoon (the ascent) and exiting (the descent).34 Moreover, different species migrate at different times of the year. For example, one species of mullet, Mugil auratus, ascends in February, another, Mugil cephalus, from October to December. Eels, Anguilla anguilla L., ascend in the spring and descend in the autumn and early winter to spawn only once in deep, warm water. The lagoonal fishing season thus lasts from February to December. The various trapping devices used today in the Italian lagoons probably derive from ancient forms. 35 These devices—barrages, trelliswork, or screens forming labyrinths—are designed to ensnare the mature fish and to spare the fry, or smaller fish. We have already suggested that small traps were located at the southern end of the Tagliata along with sluice gates, and larger ones must have also been used in other trapping areas of the lagoon. The ancient authors tell us a good deal about Roman fishing practices. That the Romans of the first century A.D. not only knew about the migratory habits of certain species but also erected appropriate devices to trap them is clear from Pliny the Elder's description of how great numbers of eels were caught during their fall migration: "There is a lake called Garda [ Lacus Benacus] in the territory of Verona through which flows the river Mincio, at the overflow of which on a yearly occasion, about the month of October, when the lake is made rough ev34

It is well known that certain species of fish migrate between the sea and such coastal lagoons at specific times of year to seek richer sources of food in sheltered waters, to return to the sea to spawn. The precise reasons for this migration are not known, but according to specialists in pisciculture, they involve the level of salinity and temperature, as well as the amounts of dissolved oxygen and nitrates present within the lagoons. Also, within the lagoons is abundant food, particularly rotting organic matter. See De Angelis 1960, pp. 4, 5. On different species that migrate at various seasons, see De Angelis 1959, pp. 3-4, and 1960, pp. 3ff. On the life cycle of eels, the chief fish caught in the Orbetello lagoons today, see Moriarty, pp. 28-94. On eel farming see ibid., pp. 130-142. For discussion offish culture today in the Orbetello lagoons see Risanamento; Alessio (cited in chapter one, n. 21), "Riproduzione artificiale e piscicoltura intensiva," pp. 315-332; FAO Aquaculture Bulletin 1 (Oct. 1974-Jan. 1975) p. 3; FAO Aquaculture Bulletin (Jan. 1977) p. 20; Brasola, Kalfa, Cannas, "Esperienze positive"

idently by the autumn star, they [the eels] are massed together by the waves and rolled in such a marvelous shoal that masses of fish, a thousand in each, are found in the receptacles constructed in the river for this purpose." 36 Columella, writing ca. A.D. 60-65, also advises the prospective owner of profitable fish ponds to observe closely the nature of the seaside terrain and the types of fish that thrive in it, "for every fish cannot exist in every sea."37 Moreover, he implies that Romans from very early times were aware of the principles involved in lagoonal fishing. Although he does not specifically refer to lagoonal operations, he attributes to earlier generations the ability to raise saltwater fish in fresh water and to stock lakes with fish spawn brought from the sea.38 Also in discussing the construction of artificial fish ponds, he stresses the importance of good circulation for maintaining a healthy environment. 39 No fishing installations within the lagoon of Cosa remain from this first period. If any did exist, as seems likely, they were probably constructed of perishable materials, such as reeds, or cane, similar in form to those used in Lago di Burano today (Fig. VII-6). Probably, the large concrete tanks found in the lagoon and associated with the second and main period of the fishery were anticipated by barriers of this kind. Whether the harbor breakwater originally played any role in relation to the lagoon fishery is also uncertain, but its orientation and pointed form suggest that it may have had a second purpose, to direct seawater toward traps along the shore, encouraging massive descents from the lagoon. An additional natural sea tank for fish within the Cosa promontory may have been in use from the beginning of the port's Roman life. At the southern end of the Spacco della Regina are intersecting tunnels (Map 6; Figs. IV-48 to 53), previously interpreted as parts of a continuous waterway from a collecting basin (the quarry) to the sea; their purpose (cited in chapter one, n. 21) pp. 1-6. Both McCann and Gazda are especially grateful to V. Brasola for further information by letter on fishing in the Orbetello lagoons. For fishing today in the VaIh of the Adriatic lagoons see D'Ancona. On some other aspects of modern brackish-water fish culture see Aspects of Brackish Water Fish and Crustacean Culture in the Meditenanean, General Fisheries Council for the Mediterranean, Studies and Reviews 57 (Rome. FAO, 1980). 35 De Angelis 1959, p. 3. Compare F. L. Cabo, Fishing Gear Used in Spanish Coastal Lagoons, General Fisheries Council for the Mediterranean, Studies and Reviews 9 (Rome: FAO, 1959). For some ancient literary sources on fishing methods see Radcliffe, pp. 235-253. 36 Pliny NH38.75, trans. H. Rackham (London, 1940). For additional ancient references on eels see Radcliffe, pp. 247-253. 37 Columella Rust. 8.16.9. 38 IWd., 8.16.1-2. 39 IHd., 8.17.1-6.

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

must now be reevaluated.40 Most likely they formed another haven for fish. The artesian spring within the cave at the northern end of the Spacco della Regina drains towards the southwest along the fracture into the sea. Before part of the southern end of this fracture collapsed, the fresh water from the spring may have flowed into the sea at the point where these tunnels intersect. If this was so, the same sort of estuary environment would have existed here as at the mouth of the Tagliata. Fish that preferred a rocky haven would have hovered about the tunnels, to be caught from time to time by fishermen approaching by sea. Because the tunnels of the Spacco appear to be largely natural, they no doubt existed in some form during this early period. But it is uncertain whether the rock-cut details, such as the skylight above the main intersection and the ledges at the tunnels' ends, belong to this time. Modifications may well have been made at various times. If the proposed reconstruction of the Tagliata, Tagliata Piccola, and Spacco della Regina is accurate, and if the breakwater also played a role in fish protection, then the Portus Cosanus illustrates a sophisticated and interrelated harbor and fishery plan, hitherto unknown in the Roman world at this early date. Such a plan made effective use both of the natural cove for anchorage and of the natural resources of the adjacent fishing lagoon. When such a plan may have been executed is not clear from the architectural evidence alone, but it must have been completed no later than the middle of the second century B.C.41 A second development phase in the fishing life of the Portus Cosanus is attested by new construction in all three main areas of the site—port, cliffs, and lagoon (Text Fig. 1-1; Figs. VII-IO, 11, and 12). The building program entailed substantial modifications of the breakwater, the Tagliata, and the entire western end of the lagoon, including the hillside along its western edge. In the port, large concrete piers (15) were added to the breakwater system along with a lower wall (M), which very likely formed a connection to the Tagliata Canal, and Walls A and B, which ran northward toward the lagoon. Within the Tagliata South, the earlier water- and fish-control arrangements were replaced or supplemented by a concrete bridge and a pair of low platforms. At the same time, construction of retaining walls (P, PW, 40

See McCann, chapter two and n. 13, as well as description of the Spacco in chapter four. 41 See evidence of incised polygonal blocks along the Tagliata Canal, discussed under the quarry in chapter four. 42 See chapter four for construction sequences and illustrations of walls, and Oleson's discussion of the chronology of the Spring House in chapter five.

145

PE, X, and U), fish tanks (Walls V and W, L, and N), and the Spring House and lower Aqueduct completely altered the natural aspect of the lagoon. The pair of walls located a kilometer to the east of Cosa and the submerged polygonal wall off the Torre di Burano appear to belong to this period of development as well. The structures of this phase coincide with stage 2 of the construction sequence seen in the port and cliffs, and with stage 1 of the Spring House and other structures in the lagoon. 42 The evidence for linking these stages in each area of the site comes primarily from the materials and construction techniques described in detail in chapter four. As noted there, a consistent rationale appears to have guided the builders' choice of methods and materials for the varied structures both in the Iagoonal and in the Spring House complexes. In particular, hydraulic concrete composed of volcanic tuff and pozzolana was used selectively for those portions of the structures that had to be in constant contact with water while sustaining the weight of heavy loads or the constant rush of currents. The same rationale was applied to structures in the port and the cliffs during this second period of its architectural development. The lower portions of the harbor piers and Wall M, the small bridge and platforms in the Tagliata South, and the 14-m segment of embankment wall along the Tagliata Canal all are made of hydraulic tuff-and-pozzolana concrete in accordance with their respective locations and functions. This is also true of the seaward ends of the pair of walls a kilometer to the east of the harbor site (Fig. IV-27). The case for assigning all structures of this hydraulic concrete to the same phase of building is substantially strengthened by a comparative analysis of the tuff aggregate from two of the main representatives. Tuff samples from harbor Pier 3, the basin platform, and Wall c of the Spring House, were shown to have the same source, a large deposit of the Volsinian volcanic complex to the northeast of Cosa. 43 Moreover, the very fact that this volcanic rock was brought from a considerable distance (very likely from the same quarry) to be used at a site naturally endowed with an abundant source of building stone attests that it was regarded as superior for works of hydraulic construction.44 One is left with the distinct impression that the entire enterprise was 43

See report here of Tngila, chapter sixteen. This type of tuff-and-pozzolana concrete was commonly used by Romans for underwater construction. Vitruvius, writing in the Augustan age, describes its remarkable hydraulic properties (2.6; 5.12.2-3). Numerous examples of such construction in Roman harbors and fish tanks may be found along the western coast of Italy from Campania to Tuscany. The remains at Cosa are the 44

146

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

directed by an architect-engineer who specialized in the design and construction of maritime and other hydraulic installations. Certain architectural relationships also contribute to this conclusion. It has been shown elsewhere, for example, that the Spring House was erected at the same time as the polygonal embankments and other walls in the lagoon.45 It has also been observed that the lower portions of the harbor piers and Wall M are built of the same materials and that the axes of Wall M and Piers 1-3 join (Map 6). Whether these structures were originally conceived as part of a master plan encompassing both the port and lagoon fishery executed at once or one that evolved over time will be further evaluated with the evidence of the finds that document the use of the site in its various periods of growth. 46 In any case, that these coastal structures may be associated with the second stage of construction in the Tagliata is suggested by the fact that the line of Wall M, if projected northward, meets the northernmost end of the 14 m long segment of the concrete embankment wall of the Tagliata Canal. 47 The quasi-reticulate construction found in the Tagliata bridge and in Walls a, b, and c of the Spring House suggests that those two structures were built in the same period. 48 Polygonal Walls A and B must also belong to the same building program for the lagoon fishery. Their construction is identical to that of the embankment walls in the lagoon (Walls P, PW, and PE), and Wall B probably joined Wall PE to form a continuous embankment for the higher ground along the eastern side of the fishery complex. 49 The submerged wall off the Torre di Burano probably was also built along with these structures, judging from its similar polygonal construction (Fig. IV-29).

bare outlines of an intricate harbor/fishery plan emerge. The attempt to supply the details of that plan for which actual evidence was not recovered during the excavations runs the risk of misinterpretation, for among extant Roman fisheries, none are closely comparable. 50 Yet a certain amount of speculation, based upon the principles of modern lagoonal fishery operation and the ancient literary testimony of Roman knowledge of those principles, yields a reasonable idea of how the fishing complex functioned at its peak in the first century B.C. The architecture from this main period of the lagoonal fishery is most easily comprehended if one proceeds from the northern end of the lagoon toward the sea (Map 8). Wall X forms the northern boundary of the excavated lagoon complex, but Wall P continued beyond it to the north for an additional 18 m, and the channel of Wall U for at least 14 m. At its northernmost excavated point (Trench IL), Wall P was found to rest upon a dry, artificial landfill (Map 11). The stratigraphy of Trenches IKE and U4 indicates that Wall X acted as a retaining wall for this artificial fill which extended at least as far to the east as the intersection of Walls X and U and as far north as the hillside just beyond the excavated portion of Wall P.51

From these diverse sources of information the

Why the upper part of Wall X does not continue all the way to the intersection with Wall U is not certain. It would appear that the wall was simply left unfinished. Wall U, which is very similar in construction to Wall X, bordered a landfill to the east, but how far eastward the fill extended is not known (Fig. 11-14). However, a dike supporting a roadway and bridge here where the lagoon was narrowest, to connect the northern embankment to the southern commercial area, seems very likely.52 Such a dike,

earliest thus far known to have survived. Predating Vitruvius by about a half a century, they throw new light on the development of this remarkable medium during the late Republican period. In all likelihood, the discovery and use of this kind of concrete for underwater construction occurred around Puteoli, where nature had fortuitously carved excellent harbors from the tuff bedrock of the volcanic terrain. Overlying this bedrock were, and still are, large pits of pozzolana of the type that gained notoriety in antiquity for hydraulic and other construction. Called pulvis puteolanus after its place of origin (Pliny NH 16.202), it became one of the main export commodities of this region in Roman times. See Dubois, Pouzzoles antique (Introduction, n. 33 above) p. 119. Vitruvius implies that the tuff from the same region was considered the best variety to use as aggregate with this pozzolana, although he notes other locales where a similar kind of rock could be found (2.6). The remains at Cosa show that at least by the early first century B.C. other sources of tuff were being exploited, in this case the nearby deposits of the Volsinian volcanic complex. (See Trigila, chapter sixteen.) The expertise with which tuff-and-pozzolana concrete were used at Cosa implies that there had already been considerable experimentation with this medium for hy-

draulic construction. The development of this type of concrete will be discussed by Gazda in a forthcoming article. See also E. K. Gazda, "The Use of Tufo in Roman Underwater Construction," AJA 79 (1975) p. 151, abstracts bound separately. 45 See Gazda, chapter four, and J. P. Oleson, chapter five. 46 See McCann, chapter nineteen. 47 See chapter four, Tagliata Canal. 48 See chapter four, Tagliata South; and J. P. Oleson, chapter five. 49 See chapter four, port area, Walls A and B. 50 See McCann, chapter one. For a recent discussion of Roman fish tanks, see J. W. Shaw, in Scranton, Kenchreai I, pp. 31-33. The tanks at Kenchreai, like those at Cosa, have no known exact ancient parallels. 51 See Bourgeois, chapter two, for further details concerning stratigraphy in this area and Figs. 11-11, 12. 52 See Bourgeois, chapter two, on the topography of the lagoon site and the schematic plan of this Phase 2, (Fig. VII-10). Architect J. F. Warren pointed out the need for such a road connection at this point while studying the circulation system in preparation for the reconstruction drawings. Compare Brown 1980, p. 59,

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

dividing the fish farm on the west from the open lagoon on the east, would explain the extensive fill of amphoras of Type 4b found lying along the base of the eastern side of Wall U (Fig. IV-69). While the amphora fill is only about 1 m wide, only further excavation east of Wall U can determine the actual width of this landfill and proposed dike. In this period other connections between the lagoon to the east and the sea consisted of the pair of walls 1 km east of the port and the polygonal wall off the Torre di Burano. These suggest that additional fish trapping and farming occurred farther down the coast.53 That the enclosure formed by Walls U, V, and W contained water is attested both by the stratigraphy revealed through excavation and coring (Figs. H-16 to 18) and by certain architectural details. The sediments, which are analyzed by Bourgeois, and by Barshad and Viani elsewhere in this volume,54^ include a thick layer of gray sand that washed into this area from the sea, proving that this enclosure communicated with the sea (Figs. 11-12, 17, and 18). Fresh water entered the enclosure via the channel adjacent to Wall U, which originates in bedrock at the source of a now dry spring (Map 13; Fig. 11-15). The height of the Spring House basin platform suggests that the water level in the lagoon was roughly 1 m lower in antiquity than at present (Fig. H-19). Thus, the tops of the walls of the northern enclosure, which at present are situated at levels varying between 0.53 m and 0.50 m b.s.l., may be reconstructed as having been ca. 0.50 m above the ancient mean sea level. Although it is possible that compaction of the lagoonal sediments has caused a slight subsidence of the walls since antiquity, there is no evidence for substantial changes. The influx of water into the enclosure would have varied with the seasonal differences in rainfall and activity of the spring, but provision seems to have been made for regulating the water level, as will be seen below. Probes made alongside the enclosure walls proved that they continued downward from the top for at least 1.50 m to 2 m, but the bottoms of these walls were not reached. The depth of the water within the enclosure was, therefore, at least 1.0 m to 1.50 m. Both the stratigraphic and architectural evidence permits the northern enclosure to be restored as a fish tank. The presence of a water channel and the fig. 77, who reconstructs a dike about 8 m wide. This and numerous other details of Brown's plan of the fishery are not fully supported by the evidence. 53 See chapter four, related coastal structures. 54 See Bourgeois, chapter two, for more detail on the points that follow, and Barshad and Viani, chapter fifteen.

147

height of the walls recall the arrangement of certain artificial ponds described by Columella: " . . . the pond should be excavated to a depth of nine feet [2.66 m at 0.296 m/Roman ft.], and two feet [0.59 m] below the top streams of water should be conducted along small channels, and care must be taken that the flow is very abundant, since the quantity of water which lies below the level of the sea is only forced out by the greater violence of the fresh water rushing in." 55 Columella refers to a type of saltwater tank, built along the coast in the sea itself, in which water is circulated by the action of the waves, but the principle of directing a forceful stream of water through a channel into the tank is similar to that applied in the northern enclosure. The channel along Wall U, which narrows from 0.80 m at the source of the spring to ca. 0.44 m at the point where it empties into the tank, would have caused a swift current to enter the tank. In the same passage, Columella recommends that good circulation be maintained by providing channels on every side, "for the old water is more easily carried away if there is an outlet on the side opposite to that from which the wave forces its way in." 56 How far this principle was adapted to the lagoonal situation at Cosa is uncertain. A narrow crack at the eastern end of Wall V could possibly be the remnant of a connection to the spring that surges from beneath Wall P near its intersection with Wall X. Another spring, in area YZ, was contained within the basin of a small fountain house (structure YZ, Map 12) during the Imperial phase of the site, but there is no evidence that its water was directed into the northern enclosure during this period. More puzzling is the absence of a firm indication that the substantial runoff from the Spring House basin platform was channeled into the southern end of this fish enclosure. It is probable, however, that some connection was effected by piping or a wooden trough that does not survive, for regular flushing of the fish ponds to prevent accumulation of dead algae and to supply fresh oxygen is routine in fish farming. 57 It is also possible that a connection was never completed. In any case, the chief area drawing upon the fresh water during this second phase of the lagoon's life was clearly east of the eastern embankment Wall PE, an area directly joined to the Spring

55

Rust. 8.17.4-5. IWd., 8.17.3. 57 See J. P. Oleson, chapter five, for the archaeological evidence. On the freshwater needs offish farming, see Moriarty, pp. 132, 133. 56

148

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

House by the Aqueduct. The excavators suggest that here was a large commercial center with an amphora pottery and fish factory, to be discussed further in chapter eighteen. The development of an elaborate Spring House with its water-lifting device and Aqueduct strengthens this still unproved hypothesis (Text Figs. 1-1; Figs-V-48; VII-IO, 11, 12). There must have also been one or more outlets in the northern tank to maintain good circulation. If, as seems apparent, the main source of fresh water flowing into this tank entered through the channel in Wall U at the northeastern corner, then outlets would most likely have been located at the southern end. Unfortunately, no southern boundary wall for this tank was found during the excavation. It is conceivable that this and perhaps other east-west walls were missed, since much of the structure was not exposed. Without absolute evidence, one may imagine various possibilities. At first glance, it appears that the southern wall of the tank would have crossed the lagoon just to the north of the Aqueduct. The ragged southern end of Wall W, however, at the point where a corner should have been located, may indicate that the southern wall of the structure was not completed or that it was destroyed. Another possibility is that the southern boundary of the northern tank lay farther to the north, in the area between Trench YZ and the probe made ca. 20 m to the south. Somewhere in this unexcavated area the form of Wall W changes from a narrow wall carried on a thick foundation to one that resembles in both height and breadth the foundation alone.58 This change may indicate that either there were two separate fish-tank structures north of the Aqueduct or the upper portion of Wall W toward the southern end also remained unfinished or was destroyed. It is also possible that the southern boundary wall of the northern tank was never finished in the permanent medium of concrete, but held fish barrages made of perishable materials—cane, wood, or metal grates— for its entire length. This would have allowed a freer movement of water and fish between the tank and the open lagoon directly to the east (Fig. VII-IO). In any case, the southern wall of the tank or tanks formed by Walls U, V, and W probably was interrupted by at least some openings to provide good water circulation within the tank. Indeed, the presence of gray sea sand within the tank, as noted above, proves that the enclosure was also exposed to sand-laden water from the sea. 58

See chapter four for details of construction of Wall W. For an example of amphoras built into the side of a fish pond at Pompeii, see Jashemski, Gardens of Pompeii (cited in Introduction, n. 30, above) p. 109, fig. 178. Also note her useful discus59

Another problematical area in the reconstruction of the northern enclosure is between the southernmost excavated segment of Wall U and the northern end of Wall PE. This area must have had a passageway to the open lagoon that lay to the east, for in the lagoon there would have been the richest organic matter on which the fish could feed, much as they do today in the Lago di Burano. Unfortunately, the southern end of Wall U was not found during the excavation, and no trace of it appears along the bank of the modern drainage canal which, in fact, may have destroyed it. If Wall U ended just to the north of the modern canal, then a wide opening to the lagoon must be reconstructed between it and Wall PE, perhaps bridged by a walkway with a further fishgate control system between the fish farm and the open lagoon. It is also possible that Wall U turned westward, perhaps even joining Wall W in one of the unexcavated areas to the north and south of Trench YZ, to form a smaller northern tank than the one envisioned here. To the south of the Aqueduct, Walls L and N (Map 8; Figs. H-17, 18; IV-73, 75, and 76) of similar construction and form to Walls U, V, and W to the north, must have enclosed another fish tank. Although again no clearly identifiable features, such as channels or amphoras commonly used as nests or retreats for fish in other ancient fish tanks, were found in the excavated portions, 59 their form and scale so closely resemble the modern commercial fish enclosures in the Orbetello and Lago di Burano lagoons that their identification seems certain. Previously known ancient fish tanks are all much smaller and are associated with private use in a sea environment. The Cosa structures are the largest ancient fish tanks thus far uncovered to our knowledge and would appear also to have functioned largely for commercial purposes during this period of the later Republic, as will be discussed in chapters nine and eighteen. Whether this southern fish enclosure was bordered on the east by another concrete wall or simply by the polygonal retaining Wall PE is not certain. It was not possible to excavate Wall L to its easternmost end, owing to the presence of the modern drainage canal. Nor was the southern boundary of this structure found. It is curious that although Walls L and N are the same width as the southern end of Wall W, they are nearly 1.3 m higher (Fig. H-16). If one presumes that the levels of Walls L, N, and W have sion of fish culture as a whole at Pompeii, pp. 108-112, with illustrations. Compare the use of terracotta channels in the walls of fish tanks, for example, in the tank in the grotto of the villa it Sperlonga, McKay, fig. 51.

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

149

not shifted appreciably since antiquity, the height of these southern enclosure walls then would have been between 0.80 m and 1.30 m above the ancient mean sea level in comparison with about 0.50 m for the height of the preserved walls of the northern tank. The higher walls probably served a particular purpose; perhaps they prevented certain varieties offish, such as the mullet so common along the coast, from leaping over the sides. In modern coastal fisheries, barriers are erected to heights of 1.70 m to 2.40 m specifically for this purpose. 60 The lower walled tank or tanks to the north may in turn have been designed for other types of fish such as eels, which are carnivorous in contrast to the herbivorous mullet. Eels comprise about 50 percent of the fish caught in the Orbetello lagoons today. In modern eel farms the elvers, or small eels, are not placed in dense concentration with adult eels, who are their predators, but are spread out into different feeding grounds and caught in special traps. Indeed, the ancients well knew the habits of eels, for Columella specifically recommends that lamprey eels be segregated from other species, " . . . because, if they are seized with madness, which sometimes happens to this sort of fish just as it happens to dogs, they very often pursue their scaly companions and chew them and devour great numbers of them." 61 A complete eel farm today has four sizes of ponds—small and larger elver tanks and two for the adult eels. The adult female eel (Anguilla anguilla L.) between seven and twenty years old measures from 0.50 m to 1 m and the male from 0.30 m to 0.50 m. The production of market fish for a year in 1,000 sq. m is about 4 tons. This requires stocking 10 kg of elvers, perhaps 60,000 individuals. With these facts in mind, one can imagine appropriate cane or wooden divisions within our ancient fish tanks which, taken together, enclose an area of about 12,000 sq. m (1.2 hectares), space for 48 tons of eels (Text Fig. 1-1; Fig. VII-10). But probably one of the tanks was reserved for raising other fish previously mentioned, or fry, and even spawn. From Columella again we know that the ancient Romans " . . . not only stocked the fish-ponds which they had themselves constructed, but also filled the lakes which nature had formed, with fish-spawn brought from the sea."62 While it is not possible to determine from the excavated remains the precise form and function of the ancient fish enclosures at Cosa, it is clear from the

deposits of gray sand within the walls of the southern tank that it was directly connected to the sea as was its northern counterpart. In fact, the level of gray sea sand is virtually the same on the western and eastern sides of Wall N (Figs. 11-17, 18), indicating that the area to the west between Walls N and P was also open to the sea. In addition, because the height of Wall L corresponds closely to that of the socles of the Aqueduct piers, showing that the levels of these structures have remained fairly constant, one may infer some intentional relationship between these two units. During this second and chief commercial period of the Cosa fishery, in addition to the Tagliata already in use from earlier times, two new permanent connections between the fishing lagoon and the sea were constructed in the harbor area. Polygonal Walls A and B, the latter in line with the lagoon's eastern embankment Wall PE, and concrete Wall M were now added, probably in areas where previously some form of natural connections had served (Map 6). Within the southern end of the Tagliata a concrete bridge and platforms were constructed and a concrete embankment wall built along a portion of the eastern side of its canal leading into the lagoon. The channel contained between polygonal Walls A and B probably served as an outlet for the eastern side of the lagoon. Wall B probably joined Wall PE to form a continuous retaining wall for the eastern embankment, while Wall A may have braced another embankment that extended in the direction of the Tagliata Canal (Map 7). Between the Tagliata and Channel A-B, there appears to have been a third channel, the western side of which was bordered by Wall M. The construction of Wall M in hydraulic concrete implies that this wall stood in water, and its angled upper surface resembles those of many modern channel walls in this region. Moreover, as noted earlier, if Wall M had continued uninterrupted along its northwestward course, it would have met the Tagliata Canal at the northern end of the newly installed concrete portion of the embankment wall along its eastern side. Excavation in this critical zone was not possible because the concrete wall of the modern drainage canal rests directly upon the ancient construction. The presence of two cuttings in the scarped cliff facing this point, however, strengthens the impression that some sort of juncture existed here in antiquity (Figs. IV-30, 43).63 The

60 De Angelis 1959, pp. 14 and 16. The depth of the modern fish tanks at Lago di Burano above the sandy bottom of the lagoon is 1.70 m. See further discussion offish tanks by McCann in chapters one and three. 61 Rust. 8.17.2-3. On modern eel farming, see Moriarty, pp.

133-135. For species and sizes of eels, see Lythgoe, pp. 61-64. 62 Rust. 8.16.2. 63 See chapter four, Tagliata Canal, for a description of these cuttings.

150

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

cuttings may well have been for devices used in regulating the movement offish or water such as would have been needed at a point where two water channels converged. The location of the eastern wall of the proposed "Channel M " is not known, but the space between Wall M and Wall A could have accommodated a waterway up to 25 m in breadth. Coastal lagoons are normally open to the sea by only one channel.64 Presumably, therefore, each of the three channels at Cosa served distinct purposes. The Tagliata, whose northern course is now obscured by its modern counterpart, no doubt continued in use as a means of debouching water from the western side of the lagoonal complex, where the overflow from the springs along the base of Wall P would have accumulated. As before, the estuary conditions thus created by the mixture of fresh and salt water at the seaward end of the Tagliata would have attracted fish seeking to enter the channel, either to hover there or to find their way into the lagoon. Installation of a concrete bridge and a set of low platforms within the channel near the seaward mouth suggests that a sluice-gate/fish-trap arrangement, similar to that used earlier, was used in a somewhat revised form (Figs. IV-31, 35, and 38). The force of entering waves on stormy days evidently required a more sturdy concrete structure from which to operate the sluice gate. This structure could simultaneously provide reinforcement for the mechanism itself and a greater measure of protection for the area to the north. Another pair of concrete platforms was added 8.4 m to the north of the Tagliata Piccola probably to stabilize the enclosed fish tank and small trapping area (Fig. IV-31). The fact that some recesses are cut into the walls of the channel immediately to the south of the platforms suggests that another mechanism for water control was situated here, possibly to impede the outward flow of seawater through the Tagliata Piccola.65 To the north of this presumed barrier, fish hovering about the rocks would have found shelter during storms. It may also have been necessary to erect a barrage at this point along the channel to deter stray fish from escaping through either of the seaward outlets, but it seems more likely that those outlets, as well as the three diagonal openings in the east wall of the main channel (Fig. IV-31), were fitted with grates of the type to which Columella refers.66 As noted earlier, however, the difficulty of gaining access to the Ta64 De Angelis 1959, p. 9. At the nearby Nassa fishery at Orbetello, however, three channels are in use. See McCann, chapter one, for discussion of Orbetello and its fishery in antiquity. 65 See chapter four, Tagliata South, for a description of the preserved features.

gliata South as well as its restricted size make it an inconvenient area for trapping large quantities of fish. Additional rock-cut recesses along the channel— in the Tagliata North and along the Tagliata Canal— imply that sluice gates or barrages were placed at intervals along the course of the waterway (Fig. IV30).67 Some of them perhaps were adapted from the first period of use, but others must have been added as the system of channels connecting the lagoon and sea became more complex. For example, the cuttings ca. 11 m north of where the Piccolo Spacco meets the Tagliata North may have been located there in anticipation of completing the new tunnel to control either water or fish. Two sets of cuttings in the Tagliata Canal wall located ca. 20 m and ca. 50 m to the north of the quarry seem also to have been placed in relation to a connecting waterway, namely the postulated "Channel M " (Figs. IV-43; VII-IO). If the Tagliata provided an inlet-outlet for water along the western side of the lagoon, Channel A-B most likely was connected to the structures of the eastern side, that is, directly to the fish tanks (Fig. VII-IO). The likelihood that Wall B was once connected to Wall PE with which it is aligned along the eastern embankment of the lagoonal area supports such a reconstruction. Such a channel would not only have helped to maintain good circulation within the tanks but also a fairly constant water level primarily by providing an outlet for lagoonal overflow. The funnel shape of Channel A-B in fact suggests that this waterway served only as an outlet during normal sea conditions (Map 6; Figs. IV-18, 19). During rainy seasons, when the level of water within the tanks would have been high in relation to that of the sea, the outflowing water would have gained speed as it passed between the constricted walls of the channel's seaward mouth into the harbor basin. This rapidly moving water would not only have prevented the deposition of silt at the mouth of the channel, but would also have created conditions to which certain fish are attracted—a change in the water temperature and salinity and a strong current against which such fish prefer to swim. At these times, the cuttings on either side of the channel's mouth (Fig. IV-19) may well have held the sort of device that allows fish to enter but not to escape.68 At other times, when the channel was not in use, a 66 For the diagonal openings, see chapter four, Tagliata South; for Columella's description of grates, see Rust. 8.17.6. 67 See chapter four, Tagliata North and Tagliata Canal. 68 See chapter four, Walls A and B, for a description of the cuttings.

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

solid barrier could have been put in its place. The funnellike arrangement of Walls A and B implies, at the same time, that the flow of sea water inward to the lagoon at this point was greatly diminished and with it, any silt deposits detrimental to effective circulation. In order to attract fish toward the cooler water of the sea where they will spawn and to trap them en route, the proper conditions must be created by allowing a substantial current of sea water to enter the lagoon. In turn, in the spring when the warmer waters of the lagoon must be released to attract the fish to enter and feed, outlets to the sea are necessary. The Tagliata provided one such inlet-outlet, but since the channel is both narrow and difficult to enter, it could not have functioned as a large trapping area at any time. Rather, as in the fishery's earlier period, large quantities of fish must have been caught along the shore, where they could be hauled in with minimal difficulty and transported to the nearby processing facility presumed to have been located at the port during this time of its floruit.69 Among the shoreline structures associated with this second architectural phase of the port, Wall M and the postulated channel that it bordered provide one of the most likely locations (Map 6; Figs. IV-16, 17). One can imagine the following: water flowing northward from the Tagliata would have merged with that entering Channel M from the harbor basin; fish would have swum from the lagoon toward the sea against the currents carried by the two channels, but they would have been guided into traps situated within Channel M, their passage into the Tagliata Canal blocked by barrages. These barrages would have been held in the rock-cut recesses located at the intersection of the channels. Any fish that managed to pass through the first barrage would have been stopped ca. 30 m farther to the south by a trap held in the rock-cut recesses at that point along the Tagliata Canal. In periods of high tide or stormy weather, seawater would have moved with considerable force through the harbor basin toward the channels along the shore. At such times, Channel AB would have been closed, while Channel M, with its fish traps, would have remained open. Not only would the traps have required protection from the currents, but those currents would have had to be properly directed through the channel in order to produce the conditions conducive to a substantial descent of fish from the lagoon. By virtue of their 69 70 71

See Will, chapter nine, and McCann, chapter nineteen. Rust. 8.17.10-11. Bruno, Will, and Schwarzer, pp. 36-37, 42.

151

size and location, harbor Piers 1 through 3 could have met both requirements. Their staggered alignment and angled positions would have provided the baffles necessary to slow down rapidly moving water, stirred by storm winds from the south, southwest, or southeast, and to redirect it into the adjacent channel. Similarly, normal offshore currents flowing into the harbor basin from the east would have been deflected toward the lagoon (Map 6). The construction of protective barriers for fish tanks along the sea coast was common practice, according to Columella, who recommends them in the case of shallow tanks for sole, turbot, and other fish of a similar type. "For in this way the violence of the sea is broken by the barriers of a bank. . . . It will, however, be necessary that cuts should be made in the moles at some points, forming small but narrow passages with meandering course, so that, however fierce a winter storm is raving, they may let the seawater pass in without creating a wave." 70 At Cosa, the same result seems to have been achieved in a different form, by staggering the protective piers. Archaeological evidence from this area of the Italian coast illustrates the use of isolated piers in connection with another ancient fishery. Large Roman concrete piers are placed around the point at Santa Liberata on the Monte Argentario to protect the saltwater fish tank immediately on the south (Text Fig. VTI-2).71 These piers are, however, scattered around the point to break the oncoming waves rather than aligned as are Piers 1-3 at Cosa (compare Map 6). 72 Also, at the seaward end of the walls along the coast about one kilometer to the east of the Cosa promontory are two low piers that partially block the space between the two long walls (Fig. IV-27). A third rectangular pier stands farther out to sea, about 10 m from the end of the western wall. Presumably, these walls formed another channel between the lagoon and the sea, which was protected at the seaward mouth by large concrete baffles.73 Other evidence for elaborate man-made protection for fisheries is found in Varro, who suggests that the Romans of his era were well acquainted with structures designed to direct the flow of tides for the purpose of maintaining good circulation and cool temperatures in fish ponds. In his remarks concerning the ponds of L. Licinius Lucullus near Baiae, Varro writes that Lucullus had a mole built into the sea to cause the tides to flow into his fish tanks, 72 For other uses of the Cosa harbor piers for ships, see McCann's discussion of the port earlier in this chapter. 73 See chapter four, related coastal structures.

152

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

Text Fig. VII-2. Plan of fish tank and concrete protective piers in sea off the R o m a n villa at Santa Liberata, looking north. Scale 1:1200. Del Rosso, p. 98.

.-.

— • *

which were located some distance inland.74 Furthermore, for his fish ponds, the same Lucullus ". . . had cut through a mountain near Naples and let a stream of seawater into his ponds, so that they ebbed and flowed. . . ."75 These passages call to mind not only the breakwater and piers at the Portus Cosanus and those of the channel walls a kilometer east of the port and fishery complex, but also the polygonal wall off the Torre di Burano, which could have guided offshore currents into the lagoon at its extreme eastern end, as described in chapter four. At the very least, the angled harbor piers at Cosa would have stirred up the water at the mouth of Channel M, thereby helping to attract fish, either in ascent or descent, while also preventing silt from accumulating at the mouth of the channel. Another fish-trapping area at Cosa may also have existed within the extended channel embanked by polygonal Walls A and B (Text Fig. 1-1; Fig. VII10). While the use of its funnel-shaped seaward mouth as a desilting device has been stressed, this does not preclude the presence of fish traps within

its northern, wider passageway, particularly at its juncture with the lagoon. This long waterway would have been ideal for catching eels, as in the lagoons of Orbetello today. The southern fish tank could then have served as a holding tank for the mature fish before processing. Because of eels' burrowing habits and their long narrow shape, special traps with a funnel entrance at one end and inner sections called parlors are commonly used today for trawling and catching the immature yellow eel. Wicker eel baskets shaped like bottles with a funnel at one end and a hinged cover at the other are universal and closely correspond to those pictured in Roman mosaics.76 Such baskets set in rows on a single line must have been used in fishing for the yellow eels in the ancient open lagoon at Cosa. For catching the sexually mature eel, which has turned from a yellowbrown to a silver-gray, more complicated traps are needed since the fish have ceased feeding and do not respond to bait. But massive catching is now made easier since the habit of the silver eel is to migrate in only one direction and in large numbers. Silver-eel

74 Varro Rust. 3.17.9. Also see McCann's discussion of ancient sources on fishing in chapter one, port and fishery management. 75 Varro Rust. 3.17.9. For actual remains see R. T. Gunther, Pausilypon: The Imperial Villa Near Naples (Oxford, 1913) pp. 163 and 193. 76 For modern eel traps, see Moriarty, pp. 101-127. For a wicker jeel pot for catching yellow eels, see ibid., p. 103, fig. 13.

Compare wicker pots set in rows on a line in mosaic from Sousse, Foucher, Inventaire (cited in n. 15) pi. 31. For silver-eel traps see Moriarty, p. 126, fig. 22 and Del Rosso, pp. 471, 488, 489 with diagrams of the Nassa fishery at Orbetello, including the eel traps placed in a separate channel, and p. 465 for diagram of Comacchio fishery in the Veneto with its eel traps.

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

traps thus must be more complicated to catch masses and also to bewilder this highly sensitive animal. A common type of silver-eel trap used today in the Orbetello lagoons and in the Baltic consists of a leader net curtain which obstructs eels and guides them into the trap. One end of the curtain may be attached to the shore. These nets may extend over a distance and are usually set in a series with traps at intervals along the leader. V-shaped guiding fences of wooden stakes are often used to span narrow passages. These fences in turn lead to baskets with funnel openings, which finally trap the great silver fish as they instinctively migrate toward the sea on moonless autumn nights (Text Fig. VII-3). Simpler trapping devices for mullet and other fish must have also been in use within the various channels at Cosa. A common device of today, illustrated by De Angelis, suggests how a large trap in Channel M or elsewhere might have worked (Text Fig. VII4). 77 This traditional type of barrage, still used in some coastal lakes of Italy, is particularly appropriate for lagoons where mature fish enter from the sea along with young fry. The device is designed to catch the fish both during their ascent into the lagoons and their descent into the sea (Fig. VII-7). Moreover, in this system, the fry are always allowed to escape through the narrow openings of the cane or metal grate. According to De Angelis: The entire construction remains more or less permanently installed in the terminal canal, the screens being removed only when the canes need replacing. The manner of functioning is substantially as follows: . . . As the fish move from the sea towards the saline lagoons (called "the ascent") all the fish enter through crevices or openings, called "bochere," designated as 1 on the diagram, that remain open while those indicated as 2 become smaller and smaller until only small fry can enter. Good-sized fish are therefore caught in space C (called "colauro") while the small fry continue to move towards the lagoon. During the return from the lagoons to the sea (called "the descent"), on the other hand, all the "bochere" are completely closed except those at 3. The fish going down to the sea at high tide first enter into the "colauro" and then instinctively move towards the trap 0 (called "otella") where their capture is extremely easy. At low tide, other fish remain swimming about and try to turn back over the 77 78 79

De Angelis 1959, fig. 1. Ibid., p. 4. Pliny NH 38.75. See McCann's discussion with further ref-

153

same route but they do not succeed in reaching the lagoon because openings 2 are closed and they are therefore forced to remain in the "colauro" where they are caught with fixed nets ("bertovelli") or mobile nets ("sciabiche").78 It is clear from the passage by Pliny the Elder on catching eels, cited earlier, that at least by the first century A.D. the Romans knew how to design and build traps in lakes and rivers capable of capturing large numbers of fish. Indeed, already in the second and first centuries B.C. the Campanian villa owners caught fish in quantities in their saltwater ponds. 79 It must be assumed that the barrages at Cosa were tailored to the specific requirements of the site for, as De Angelis cautions the present-day designer of fishery installations, each lagoonal environment is unique and the habits of local fishermen must be taken into account as well.80 The architect for the large lagoonal fishery built during the boom years of the Portus Cosanus had to accommodate not only local lagoon conditions and fishing habits but also preexisting structures from the fishery's earlier period, which included both the Tagliata and breakwater. The system of trapping described above is designed to take full advantage of the instincts of fish in relation to the natural conditions of the sea and lagoon. During this second phase of the fishery at Cosa, there appear to have been additional controls over raising fish, as implied by concrete tanks of varying form within the lagoon, which probably replaced earlier reed and cane structures. As suggested above, these concrete tanks of different sizes must have served distinct purposes and been subdivided within by wooden or reed partitions to separate various species from one another and the fry from the larger fish. The arrangement within the Tagliata South suggests that shelter for rock fish was provided there. 81 Exactly how the channels on the coast were connected to the lagoonal tanks cannot be determined from the excavated evidence, nor can the location of the various necessary openings within the tank walls. The reconstruction drawing presented here offers only one of several possible solutions (Text Fig. 1-1; Fig. VII-IO). This entire building program may be generally dated on some constructional evidence. The opus quasi-reticulatum of Walls a, b, and c of the Spring House (Figs. V-13, 14, 17) and of the Tagliata bridge erences in chapter one, port and fishery management. 80 De Angelis 1959, p. 16. 81 Columella, Rust. 8.16.8, discusses the habitats of "rock-fish."

Text Fig. VH-3. Plan of trapping area (1905) in channel between lagoon and sea at Nassa fishery, Orbetello. Eels are caught in long funnel-shaped traps placed at ends of two channels leading to the sea. Scale: 1:1000. Del Rosso, p. 471. Text Fig. VII-4. Schematic plan of fish barrages in a terminal canal, after De Angelis, designed for catching mature fish with young fry both entering the lagoon and returning to the sea. De Angelis 1959, fig. 1.

SEA

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

provide a chronological range between the closing decades of the second century B.C., when this technique first appeared in Rome, to approximately the middle of the following century, when true reticulate work came into favor in the capital.82 Outside Rome, however, the use of quasi-reticulate work was probably introduced somewhat later. F. Coarelli notes, for example, that at Pompeii the earlier technique of opus incertum continued to be used down to the time of the Sullan colony. 83 Further evidence for dating this period of construction at Cosa is discussed by Oleson with reference to the Spring House and by McCann in the analysis of the chronology of the site as a whole, drawing especially upon Will's evidence for the chronology of the Type 4a and 4b amphoras. According to their conclusions, this phase of activity at the port, fishery, and Spring House is documented by finds ranging from the end of the second to the last decade of the first century B.C. It is uncertain when, within this time span, the various concrete structures in both the port and fishery were built.84 At least the fishing lagoon with its tanks connected to the sea, the embankment walls, the Spring House, and Aqueduct reveal a clear functioning unit, which was probably executed fairly rapidly, either as an adjunct to the permanent port facilities already in place or designed and constructed with them. The relationship of this extensive fishing enterprise to the whole economic history of the site and the Sestius family will be discussed in a later chapter. Another question to be considered in the chronology of the lagoon fishery in this second phase of its architectural life is whether the fish tanks were in use during the entire period of the first century B.C. Some architectural evidence already mentioned in chapter four suggests that the fish tanks were never finished. The upper part of Wall X does not continue to the eastern end of its foundation, nor does the upper part of Wall W continue along its southernmost exposed portions. Bourgeois also has noted a 82 F. Coarelli, "Public Buildings in Rome between the Second Punic War and Sulla," BSR 45 (1977) pp. 13-19, for a recent discussion of the date of the transition from opus incertum to opus retkulatum in Rome, with earlier bibliography. 83 Ibid., p. 16. M. Torelli, "Innovazioni nelle techniche edilizie romane tra il I sec. A.C. e il I sec. D . C , " Tecnologia, economia e societa nel mondo romano, Atti del convegno di Como, 27-29 settembre 1979 (Como, 1980) pp. 150-151, points out that even in Etruria, which was in direct contact with Rome, the diffusion of the new building technique (opus retkulatum) was very slow. 84 For the chronology of the Spring House, see J. P. Oleson, chapter five, and for the site as a whole, McCann, chapter eighteen. Concerning the chronology of the site, the results of the carbon-14 analysis of the wood from the form work of Wall U seem suspect to this writer (Gazda) in light of the very strong

155

thick layer of gray sea sand deposited within the area sometime after Walls P and PE were built. The fact that this layer lacks archaeological material suggests that the filling in was rapid, even as the result of a single storm. 85 Such a catastrophe might explain the unfinished portions in the fishery if it was under construction at this time. Only further excavation may clarify how long the fish tanks were active during this major commercial period of the port and fishery.

THE MARITIME VILLA

While the area of the ancient maritime villa—now occupied by the private dwelling of the Venturini family—was outside the limits of our excavation, some of its architectural elements are clearly visible within the port area (Map 6; Figs. IV-2, 3, 18). Since these remains have provided the basis for identifying a third and Imperial phase in the site's architecture, it seems important to include a description of the visible walls and structures here, even if a complete reconstruction of what must have been a large villa complex is beyond the scope of this study; the plan of this phase presented here is thus largely conjectural (Fig. VII-13). A third ancient phase in the architectural development of the port, cliffs, and lagoon was clear from the structural sequences analyzed in chapter four. Also, some of the materials and methods of constructing these walls differ markedly from those employed in the preceding two building periods. AU the architecture described here and associated with this last phase of the port and fishery probably belonged to the villa complex, although to what extent the fishing lagoon was in use during this period is still a question. The architecture thus far revealed from this Imperial period includes: the brick and limestone concrete walls along the beach (Walls O, C, D, E, F) and a vaulted cistern (Figs. IV-18 to 23); probably the upper portions of structural evidence favoring the contemporaneity of Wall U with the entire lagoonal complex. According to the MASCA-corrected date for this wood sample P-1935 (A.D. 130-60 ± 60), this wall would date at the very earliest to 1 B.C.-A.D. 1. Given the unified plan in this second architectural building phase of the site, and the common use of materials and construction methods described here, a building period lasting three-quarters of a century or more does not stand to reason to this writer. Thus, I would argue for a date for both the port and fishery concrete structures in the second quarter of the first century B.C., contemporary with the Spring House and the Type 4b amphora material, all designed in a master plan executed rapidly under the direction of a master engineer. 85 See Bourgeois, chapter two.

156

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

the harbor piers (Figs. IV-Il, 12) and the upper portion of Wall M (Figs. IV-16, 17); the vault at the southern end of the main Tagliata channel (Figs. IV32, 35, 36) and the overhead arch at the bend of the Tagliata Piccola (Fig. IV-40); the repairs of the polygonal embankment walls in the lagoon (P and PE) (Figs. IV-58, 59, 77); the small, rectangular enclosure in area YZ (Map 12; Figs. 111-46 to 48); pier 9 (Fig. IV-74) and perhaps also Wall S (Fig. 111-50); and the entire second stage of the Spring House complex, including the upper water conduit and cistern (Figs. V-36 to 41, 53).86 Structural sequences clearly show that Wall O of the villa was built after Channel A-B had gone out of use and that the concrete of the vault at the southern end of the Tagliata was poured after that of the bridge was in place. So, too, probably the upper portions of harbor Piers 1, 2, and 3 and of Wall M are later than their lower portions. The material character of these structures as well as several other walls (C-F) suggests that they were contemporary with one another. All are of concrete containing local aggregates—mainly limestone and amphora or tile fragments along with miscellaneous pieces of other native rocks such as sandstone. Only a few pieces of imported tuff are scattered in some of the construction in the Tagliata. The mortars vary in composition, even within the same structure. Some consist only of lime and local sand while others also contain volcanic particles that are probably small fragments of tuff. The clearly discernible courses of brick and amphora parts superposing the limestone aggregate in Wall F of the villa may be compared to the layering of terracotta fragments among the aggregate rock of the harbor piers' upper parts (Figs. IV-Il, 12). The limestone and sherd rubblework construction on top of Wall M appears to belong to the same building phase (Figs. IV-16, 17). It seems logical that the existing port area structures have been adapted now for use by the residents of the new villa on the beach.

ately above the opening to the sea, is composed of sandstone slabs set in thick layers of mortar (Fig. IV36). The closest parallel for this type of construction at the harbor site occurs in the villa cistern situated to the west of the Torre della Tagliata (Fig. IV-23). This coincidence would seem to place the Tagliata vault, and the contiguous concrete that lines the domed ceiling of the adjacent outlet, contemporary with the villa. The concrete of the overhead arch at the bend of the Tagliata Piccola might also be assigned to this phase on the evidence of methods and materials similar to those employed in constructing the remainder of the Tagliata vault (Fig. IV-40).87 Except for the small rectangular building in area YZ, the latest construction in the lagoon region consists almost totally of repairs to earlier structures. The most substantial of these was the rebuilding of the Spring House, to which was added a new water conduit and a large cistern. Pier 9, of very similar construction to Pier f of the new Spring House conduit, appears to have been erected at the same time (Fig. IV-74). Wall S (Fig. 111-50), which is aligned with pier 9, may have been built then as well. Wall PE of the eastern embankment was repaired in several places (Fig. IV-77), and a long stretch of Wall P on the western side was replaced (Figs. IV-58, 59, 63). In both cases the construction of mortared limestone rubblework snecked with amphora fragments appears identical to that of the new cistern on the hill above the Spring House (Figs. V-38 to 41). Against the repaired section of Wall P, the southern half of the rectangular structure in area YZ was built (Map 12; Figs. 111-46 to 48; IV-80, 81).88

The vault at the seaward mouth of the Tagliata also appears to belong to this period of activity, because its concrete construction is similar to that of the harbor piers' upper parts. It was noted in chapter four that the central portion of that vault, immedi-

The structural evidence alone is not sufficient to conclude that all these projects in the lagoon were contemporary, nor can it prove that they all belonged to the same building program as the maritime villa. In fact, several subphases may emerge upon further excavation and analysis, and there may also have been more than one villa in the area which drew upon the lagoon's resources. For now, one can say with reasonable certainty only that all the structures cited above postdate those of the second phase of the site's architectural development. Some, the Spring House complex and structure YZ included, can be dated firmly within the Imperial period on

86 See descriptions and illustrations in relevant sections of chapters four and five. 87 The use of limestone rubble, along with fragments of amphoras and tiles, as the main aggregate for concrete also occurs in the second phase of the site's architectural development. These structures from the second phase include the low fish-tank walls W, V, L, and N, and the upper courses of the foundation offsets of Wall P to the north of the Spring House. They may be distin-

guished from the examples attributed here to the third architectural phase largely on the basis of the mortar, which in the second-phase walls is generally coarser and appears to contain a higher percentage of lime. These distinctions, considered in conjunction with other evidence, such as that of structural sequences, bonding, and overall plan, sustain this proposed attribution. 88 For details on these structures, see relevant sections in chapters four and five.

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

evidence of both the construction techniques and the finds. Others, by virtue of their apparent purpose, may logically be associated with the more firmly dated structures so as to envision the character of the harbor and the fishery site in its final phase. Before approaching such a reconstruction, it will be useful to review the architectural evidence for placing all these remains within the Imperial period. A number may be dated on the basis of the construction techniques, some of which have counterparts in well-documented buildings in Rome and its environs. The wall facings found in the harborside villa provide the clearest parallels to buildings in other parts of Italy. Most distinctive are the facings on the upper parts of villa Wall O (Fig. IV-21) and Walls C through F (Figs. IV-20, 22), and on the east-west wall of the "room" adjacent to the restaurant La Strega (Fig. IV-25). A faced section of Wall O preserves six courses of roughly shaped "reticulate" limestone tesserae.89 The same sort of facing occurs on Wall E at its northern exposed end, where at least six courses of limestone tesserae and five of brick are preserved. The use of brick coursing with reticulate tesserae was not common at Rome before the Augustan period, 90 and not until the Claudian period did brick bands come into regular, well-planned use.91 The latter practice was continued in certain Trajanic structures as well as in some Hadrianic work. 92 The exposed villa walls at Cosa that show broken tiles shaped into triangles probably date in the second century A.D., when the use of broken roof tiles as triangular bricks became common. 93 In the pilasters of Wall O, exposed in Probes V (Fig. IV-21) and VII, the approximately six courses of brick alternating with seven of rectangular limestone blocks recall the construction of the large pier of the Neronian odeon which was built into the late Republican basilica in the forum of Cosa (Fig. VII-8).94 A later parallel may be found farther afield in the villa of Sette Bassi, begun in the reign of Antonius Pius.95 On some walls of this villa the facing in89 As explained in chapter four, the local limestone was not easily shaped into precisely reticulate form. These tesserae, in fact, look "quasi-reticulate" in shape. But it seems clear that the builders intended to achieve the effects of true reticulate facings. Thus, comparisons to true reticulate work for the purpose of establishing a general chronological framework seem valid. 90 Blake 1947, p. 274; Coarelli, "Public Buildings in Rome" (cited in n. 82), p. 15, cites pre-Augustan examples. 91 Blake 1959, p. 161. 92 For the Trajanic period, see the marble wharf along the Tiber, illustrated in Lugli, pi. CLV, 1. For Hadrianic structures, see Blake 1973, pp. 300-301. 93 Among the many examples, see those cited by Blake 1973, p. 301.

157

eludes six courses of brick alternating with nineteen of blocks. The facings of the east-west wall of the "room" near La Strega provide a similar chronological range from the mid-first to the mid-second century A.D. (Fig. IV-25). The upper part of the wall may be compared to the block-and-brick facings (two rows of bricks to one of blocks) used in Pompeii after the earthquake of A.D. 62.96 This type of coursed facing used in conjunction with reticulate work also has Pompeian as well as later parallels.97 Block-and-brick coursing is common in the second century A.D., but examples exhibiting a consistent alternation of two courses of brick to one of blocks are less plentiful.98 There are numerous late Imperial examples that are less relevant for dating the villa at Cosa, for a coin of Faustina II (A.D 161-176) found above the foundation of Wall O gives a terminus ante quern in the late Antonine period.99 Within the lagoon, the piers of structure YZ are built in courses that alternate roughly quadratic limestone blocks with large rectangular roof tiles (Fig. IV-80), a technqiue comparable to examples elsewhere in Italy that date no earlier than the first century A.D. 1 0 0 This structure in area YZ was built after Wall P had been repaired in a technique identical to that used in the cistern above the Spring House. As noted elsewhere, the columns of the cistern conduit are constructed of amphora fragments and tiles laid in courses that recall the decorative facings on the walls of the seaside villa (Fig. IV-60). Evidently, the builders of the colonnade adapted available materials to create the sort of lively surface pattern that appealed to contemporary taste.101 In summary, the architectural evidence for the dating of this third and last phase of both the port and lagoon areas extends from the mid-first to the third quarter of the second century A.D. Full excavation of the villa is needed to reveal its various building stages. Available evidence indicates it was begun in the mid-first century A.D., then enlarged or remodeled in the period of the Antonines. It also 94

Brown 1951, p. 78, fig. 73. Blake 1973, pp. 106-111, pi. 13, fig. 2; Lugli, pi. CXCIII, 2. Blake 1959, p. 161. 97 Lugli, pi. CXLIX, 2-3 (for Pompeii); pi. CLIII, 4 (for a midsecond century A.D. example on the Via Praenestina). 98 Ibid., pi. CXCIII, 3-4, for two late Imperial examples of variable alternations—a house near the "horrea" of the "mensores" in Ostia and the sepulcher of Valerius Romulus on the Via Appia. Additional examples are illustrated on pi. CXCIV, 1-4. 99 See McCann in chapter twelve, coin Cat. C O l . 100 For a Trajanic example at Arcinazzo, see Blake 1973, p. 237, pi. 31, 1. 101 See J. P. Oleson, chapter five. 95

96

158

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

appears that the building of this villa, and perhaps others in the area, occasioned the reconstruction of the Spring House and the erection of the smaller fountain house identified in area YZ. The rebuilding of the Spring House clearly signaled a revitalized activity in the lagoon area but on a much smaller scale, as the amphora finds also attest.102 The evidence of unfinished construction, the presence of a thick layer of sterile sea sand, and the chronological profile of the finds suggest a precipitous abandonment of the fishery site at the end of the Republic toward the close of the first century B.C. If, as it seems, the fish tanks were nearly filled up with sand at that time, they would not likely have been used again. The dredging of so large an area would have been monumental. In any case, excavation turned up little evidence to suggest largescale reuse. Apart from the repairs of both polygonal embankments, the network of walls belonging to the second phase in the development of the fishery shows no signs of alteration. How much of the lagoon was covered by water at this time is uncertain. Besides the Spring House complex, the only substantial structure added in the Empire was the small rectangular enclosure in area YZ at the site of a spring that bubbles from bedrock beneath Wall P. Its three low walls resemble those of the earlier fishery, but, as previously noted in chapter four, numerous irregularities suggest hasty or careless workmanship. The southern wall is breached by a small channel for the runoff, which apparently seeped into the ground (Fig. IV-81). A pavement on the eastern side of the structure, however, implies that the area outside the enclosure was dry, at least on the eastern side (Fig. II—16). Presumably, the pavement facilitated access to the spring from the east; the remains of animal bones (cattle and pig) as well as many amphora fragments here suggest considerable traffic of both man and animal (see chapter seventeen, Cat. FlO).

the east (in Trench L23) may have been thrown down from an upper story of the structure. The sturdy construction of the piers, and the likelihood that they held an upper story, suggest that this second fountain house was part of a larger complex, perhaps a villa, located farther up the hill. Water could have been drawn in buckets to the upper story also from the paved area below. Although smaller and less complicated than the first Spring House, it seems to have served a similar purpose. Because pier 9 on the eastern embankment is nearly aligned with the lower Spring House Aqueduct of the preceding period it is tempting to conclude that the earlier Aqueduct was reused in this last period. No traces of additional piers are visible, but neither was the area of the eastern embankment extensively excavated. The cistern of the villa on the beach must have been supplied with fresh water from a spring of the lagoon, but whether from the one at the Spring House is not yet possible to determine. Wall S, just to the south of pier 9, likewise remains an enigma that only further excavation can explain. In the area of the harbor and villa, it is clear that Channel A-B was now not functioning.104 How Wall M and the first three harbor piers were reused by the villa occupants remains uncertain, but the probable additions of concrete to the tops of the piers suggests that greater height was needed, perhaps to carry other structures. Whether Wall M still marked the course of a waterway to the lagoon is unclear from the architectural evidence.

The low enclosure walls sustained five substantial block-and-tile piers which supported, in turn, brick arches (Map 12). A fragment of one arch was found to the east of the enclosure, where it had fallen along with three of the piers and some other architectural elements (Fig. 111-47). The circumstances of the fallen debris suggest that an earthquake may have caused the final collapse of this small fountain house. 103 The quantities of pottery found farther to

The Tagliata may well have continued to provide some drainage and water circulation for the old lagoon, however shallow it may have been by this time. The heavy, broad vault built into the southern end of the main channel has every appearance of a structure designed to buffer entering waves (Figs. IV-31, 32). The northern wall of the domical tunnel diverted some of the water up the blow-hole while the vault served as an artificial extension of that tunnel, apparently to lower the level of water that surged through the seaward mouth into the channel (Figs. IV-35, 36). The whole southern end of the channel was narrowed by the concrete ledges that extended northward from the vault, passing beneath the earlier bridge and ending just south of the juncture between the main channel and the Tagliata Piccola (Fig. IV-35). These ledges must also have served to reinforce the earlier bridge. 105 The impres-

102 See Will, chapter nine, Types lie; 12b, c; 14; 16; 18a, b; 20; 21a, b, c, d; and McCann, chapter eighteen. 103 Suggestion of de Boer. 104 See chapter four, the villa.

105 See relevant sections in chapter four for further discussion of these ledges. A domical structure at Tyre may be part of a fishery. See Frost, "Recent Observations" (cited in Introduction, n. 19) pp. 103-109, fig. 1.

VII: R E C O N S T R U C T I O N OF PORT, FISHERY, VILLA

sion of a vertical slot in the concrete ledge on the eastern side of the channel suggests that a sluice gate and/or fish grate still occupied this spot (Fig. IV-37). The concrete arch above the bend in the Tagliata Piccola may have been placed there as a buttress against the waves at this point (Fig. IV-40). The whole arrangement in the Tagliata suggests a re-

159

vamping to ensure a calm environment for fish. Surely the villa's occupants would have recognized that the Tagliata could yield a constant supply for their tables. Whether the old fishery in the lagoon or Channel M was considered as a potential saltwater fish tank cannot be known without further excavation.

Chapter VIII. Ancient Shipping in the Portus Cosanus LIONEL CASSON

Our notion of what ancient ships looked like is based on miscellaneous fragments of evidence, chiefly haphazard pictures in vase paintings, wall paintings, mosaics, and reliefs that usually illustrate scenes from mythology rather than actual life.1 Some models, mostly of clay but also of wood and stone, have survived; intended as votives and never made to scale, they offer only schematic renderings. 2 The most reliable evidence, a type that has recently become available, is the remains of wrecks uncovered by underwater archaeology. Only the lowermost parts of a vessel tend to survive, usually the bottom of the hull and rarely much above the turn of the bilges, so this evidence is of limited help in reconstructing the overall aspect of ancient craft.3 For Cosa specifically, we have not even this modest material to go on. From the fifth century B.C. to the first century A.D., a span that includes the time when Cosa's port came into existence and saw its most flourishing period, we have practically no pictorial evidence from any place;4 and from no period at all, either before or after, do we have any that relates directly to Cosa. We can only assume, not unreasonably, that the ships and boats which plied the waters of the area were the same as those we know of from the evidence of other times, notably the period of the Roman Empire, and of other places such as Ostia or Pompeii.

MERCHANTMEN

The commodities that bulked largest in the ancient Mediterranean's maritime commerce were 1

Presented in extenso Casson, SSAW. A. Gottlicher, Materialien fir ein Corpus der Schiffsmodelle im Altertum (Mainz, 1978) pp. 23-38, 60-90, and, on the general nature of the models, p. 2. 3 Casson, SSAW, pp. 214-216; G. Bass, ed., A History of Seafaring Based on Underwater Archaeology (New York, 1972) pp. 502

grain, olive oil, wine, dried fish, and the fish sauce called garum. 5 Cosa's port undoubtedly handled shipments of such cargoes. They were transported in roomy freighters that were designed and rigged far more for safety than for speed. The standard rig depended for drive solely on a broad mainsail set amidships; the only other canvas regularly carried was a small squaresail on a short mast slanting over the bows, much like the bowspritsails of a later age, whose function was to aid steering (Figs. VIII-I to 3). For increased drive, larger vessels raised a triangular topsail over the main (Fig. VHI-4), while the very largest added a square mizzen, considerably smaller than the main, on the afterdeck. Such a rig, with its limited spread of canvas carried relatively low, was safe, if not fast. Moreover, sail could be shortened speedily and efficiently, thanks to a special system found only on ancient Greek and Roman vessels, a series of brailing lines, made fast to the foot, that ran up the forward surface, passed over the yard, and descended to the deck behind (Figs. VIII-I, 4); there, one or two hands could, by pulling on them, bunch the canvas up to the yard much as we raise a Venetian blind. By pulling on selected lines alone they could shorten any portion of the sail any amount (Fig. VIII-4). The procedure was always carried on from the deck; even when shortening sail all the way, no hands had to go aloft.6 The hull was generally flat-bottomed, thereby providing a maximum of room in the hold. This was first and foremost for cargo; there were no cabins below deck, although passengers might be accommodated there when the vessel sailed partly 52, 78, 140-146. 4 Casson, SSAW, pp. 173-174. 5 Cf. L. Casson, The Ancient Mariners (New York, 1959) pp. 233-234. 6 Casson, SSAW, pp. 239-243, 275-278.

VIII: ANCIENT SHIPPING IN PORTUS COSANUS

empty or in ballast. Travelers usually booked passage on deck. A cabin on the afterdeck (Figs. VIII1,4), limited in space, was reserved for the captain, the owner or his agent, and other such VIPs. Over the cabin rose the distinctive decorative element of the ancient Greek or Roman merchantman, a curving stempost ending in a gilded goose head (Figs. VIII-I, 2, 4). Most merchantmen had convex rounded bows (Figs. VIII-I, 3 left, 4), but not a few were given a concave stem that jutted forward at the waterline to end in a projecting cutwater very much like a war galley's ram (Figs. VIII-2, 3 right); no convincing explanation has so far been offered for this curious characteristic.7 Then as now, merchantmen ran the gamut of size; those that frequented Cosa's harbor must have ranged from modest coastal freighters (Figs. VIII-I, 2) to big carriers for long-distance transport (Fig. VIII-4).8 The wrecks that have been identified and investigated—many of them along the coast of Provence or Liguria and hence along Cosa's shipping lanes—reveal that the ships on the run were as a rule 20 m to 30 m in length and 6 m to 10 m in beam, with a capacity of some 100 to 150 tons of cargo. 9 Some were smaller. Off Marseille divers found the remains of a vessel that was carrying a load of wine jars, very likely taken aboard at Cosa (the Grand Congloue wreck; see appendix by Luc Long); it was no more than 14 m to 20 m long with a capacity of some 40 to 60 tons. In ancient literature the word myriophoros ("10,000-carrier") and similar expressions occur, used to indicate a freighter of very large size. For long, such terms were not taken literally. Now we know that they should be, and that the figure refers to amphoras, since at least two wrecks have been discovered whose cargo in all probability consisted of ten thousand amphoras of wine. The first, found some decades ago off Albenga on the Ligurian coast, measured about 25 m in length and some 6 m to 8 m in beam; its hold was packed with amphoras arranged in five superimposed tiers, and the best estimate puts the total at ten thousand. The second, found recently on the French coast not far from Toulon, measured about 40 m in length and 9 m or so in beam and, like the Albenga wreck, seems to have been carrying the same number of amphoras, also in five superimposed tiers.10 To garner such an imposing quantity, vessels this size very likely stopped at a number of wine-producing areas, of which Cosa could well have been one. 7

Ibid., pp. 173-182. Ibid., pp. 170-173. For merchant vessels en route to Spain passing through Cosa's waters in 217 B.C., see Livy 22.11.6. 8

161

MERCHANT GALLEYS

The ancient galley (i.e., a ship propelled by oars or by oars and sail) par excellence was the man-ofwar. But we tend to overlook that some oar-driven craft also served peaceful pursuits, the countless vessels of all sizes that plied the rivers and coastal waters of the Mediterranean, carrying miscellaneous cargo and ferrying passengers (Fig. VIII-5). For short hauls they were able to offer quicker and more certain delivery than sailing ships, since their light draft enabled them to enter shallow ports where sailing ships, with their deeper hulls, did not dare approach, and the crew of rowers enabled them to disregard the vagaries of the wind. In a place such as Cosa, which was primarily occupied with export of local products, and not, like Ostia or Puteoli, a regular port of call for large seagoing freighters, merchant galleys of all kinds must have made up a good part of the traffic. For carrying cargo, the commonest types were: lembi (lemboi in Greek), craft fast and agile enough to be conscripted in times of emergency into the navy as auxiliaries; actuariae (akatoi in Greek), craft sizable enough to require thirty to fifty rowers and also suitable for service as naval auxiliaries; cybaeae (Greek kybaiai), sizable craft which, to judge from the name, were stubby and "boxlike"; and cercuri (Greek kerkouroi), the largest merchant galleys known (some used on the Nile to haul grain downriver to Alexandria ran 45 m in length and carried 450 tons). AU these types no doubt took passengers as well, when room was available. Primarily for passengers were the celetes or celoces (Greek keletes) and phaseli (Greek phaseloi), galleys designed more for speed than capacity.11

HARBOR CRAFT

Without question, Cosa's harbor was thronged with the same small craft that are attested at Ostia or Puteoli or other harbors of Italy. There must have been any number of barges (some versions were called lenunculi, a term used of several kinds of small oared craft, others levamenta, "lighters") to service arriving freighters which, too deep in draft to moor at the piers or finding no space available there, anchored in the middle of the harbor or in the open roads. There must have been any 9

Casson, SSAW, pp. 189-190, 214-216. >° Tchernia 1978, pp. 101-107. » Casson, SSAW, pp. 157-168.

162

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

number of the equivalent of our tugboats, sturdy little craft which, in addition to a crew of oarsmen, mounted the type of fore-and-aft sail called a spritsail (Fig. VIII-6); their task was to take in tow large merchantmen, which could not use their sails in the confined waters of the harbor, as these approached Cosa and bring them to their assigned mooring. For such "tugboats" the fore-and-aft rig was more efficient than the square-rig, enabling them to sail to their tows under most wind conditions and save their rowers' muscles for the laborious return. And Cosa assuredly had its share of cumbae, scaphae, lintres—the names vary (consider our use of "skiff," "dory," "rowboat," "dinghy," and other such terms in English) but all refer to small-sized craft powered by oars, or by oars and a sail. These were used for getting about the harbor or short forays outside of it, especially for fishing. We know from Strabo (5.2.8 [225]) that shoals of tuna passed Cosa's promontory, and there were the local fish to catch by spear or hook and line, and to net (Figs. VIII-7 to 11). Those with sail might have, like the "tugboats," a fore-and-aft rig, most often the sprit but also the lateen. In shape, these little vessels were by and large what we use today, save for two types of rowboat distinguished by a curious form of prow. In one type, the prow is concave and ends in a ram-shaped projecting cutwater (Fig. VIII-12), exactly the kind of prow that, as mentioned earlier, is found on some seagoing freighters. In the other, the prow is blunt (Fig. VIII-I, the small boat), much like the transomstern of some modern rowboats; indeed, since the stern of this type is invariably curved, it rather looks like a modern rowboat being driven the wrong way. The purpose may have been to permit it to moor, head on, flush to a dock.12

grain, olive oil, wine, dried fish, fish sauce. Grain was shipped in sacks, the others in amphoras. Pictures of stevedores at work show them carrying over the shoulder either a single sack13 or a single amphora (Figs. VIII-3, 4). The sacks, then, must have been about a hundredweight, and the amphoras almost certainly were, for they averaged roughly six to seven gallons in capacity, and that much in wine or oil weighs about fifty pounds, and their weight when empty is another fifty. Since large grain carriers could stow aboard 40 tons or more, and large wine or oil carriers could be myriophoroi, Roman ports must have maintained a veritable army of stevedores. No matter how many there were, loading and unloading inevitably was slow and laborious. In loading, for example, each stevedore picked up a sack or jar at the dockside warehouse, threw it over his shoulder, trudged to the point along the pier where the loading vessel lay, stopped to be checked in by a checking clerk, climbed up a gangplank, and turned his load over to hands inside the hull who disposed it there carefully, wedged in so that it would not shift during the voyage. Amphoras, for example, were stacked upright in tiers with twigs or brush as dunnage (Fig. VIII13); the hold of a myriophoros, as we have noted, could accommodate five such tiers, and sometimes lines of amphoras were lashed on deck as well.14 At Cosa, where mooring space along the piers was limited, larger freighters must have anchored in the harbor or in the open roads. This meant that the hardworking stevedores first deposited their amphoras or sacks in barges, the barges were rowed out to the carriers, and then stevedores hauled them, one by one, up ladders to the hatchways in the deck. It is possible that time and effort was saved by rigging rope slings from the yardarms to raise cargo from the barges to the deck, but we have no concrete evidence for the practice.

THE HANDLING OF CARGO

As in all other ancient ports, traffic in the port of Cosa must have moved very slowly, requiring an extended turnaround time. This is because goods were handled in the literal sense of that verb; they were moved about by being picked up and carried by stevedores. We listed above the principal commodities that were shipped in bulk about the Mediterranean: 12 Ibid., pp. 336 (barges), pp. 336-337 (tugs), pp. 329-331 (skiffs); for further illustrations of small harbor craft, see L. Foucher, Navires et barques, Notes et Documents 15 (Tunis, 1957).

NAVAL CRAFT

Did Cosa have facilities for men-of-war as well as merchant galleys? Was it a naval base as well as a commercial harbor? We have evidence for use of the port by naval vessels on only one occasion: in 49 B.C. L. Domitius Ahenobarbus sailed from there on seven actuariae 13

Casson, Ancient Mariners (cited in n. 5) fig. 14b. Casson, SSAW, p. 177 (dunnage), figs. 140 and 148 (amphoras on deck). 14

VIII: ANCIENT SHIPPING IN PORTUS COSANUS

manned by personnel recruited from his own properties—slaves, freedmen, tenants.15 His actuariae were naval auxiliaries rather than true warcraft, and his choice of Cosa as embarkation point was unquestionably dictated by the fact that he owned enormous estates in the neighborhood. No doubt naval craft from time to time put in at Cosa's port, particularly in its earliest period of use during the Second Punic War in the late third century B.C. But 15

Caesar BCiv 1.34.

163

until new information turns up to prove the contrary, we must assume from the lack of defense installations, ship sheds, or slipways and the evidence of the amphora finds that the prime purpose of Portus Cosanus, at least in its floruit under the Sestii in the second and first centuries B.C., was purely commercial with no area reserved for the military16 (Text Fig. 1-1; Fig. VII-IO).

See chapters seven, nine, and eighteen.

Appendix 1. The Grand Congloue Site: A Reassessment Luc LONG Begun in 1952 by a team under the direction of Commandant Cousteau, the Grand Congloue exca­ vation (off Marseilles, at a depth of between 32 m and 50 m) was the first major French effort in the area of underwater archaeology. A summary of its chief results was published in 1961 by Fernand Benoit, Director of Antiquities for Provence and Cor­ 1 sica, in a work that has attracted wide attention. Even before the appearance of his report, in fact, there arose what was to become a protracted contro­ versy over the juxtaposition in a single wreck of sev­ eral hundred Sestius amphoras, a cargo of Campanian ware, Greek amphoras>t and early Greco-Italic amphoras. The fact is that, as we know today, there are two wrecks at the site. They are partially super­ imposed, and they differ in date by about a century. A detailed study of the notes and illustrations of the excavation journal (kept between 1952 and 1957), as well as a sounding made in 1980 on what is left of the site, have enabled us to verify beyond any fur­ ther question the double-wreck hypothesis. This study, the complete results of which are the subject of an article in Archaeonautica no. 5, has also made it possible to assign to each of the two wrecks the finds that were originally part of each. The first ship's cargo, doubtless taken on board in the Naples area, consisted of approximately four hundred Greco-Italic amphoras; some thirty Greek amphoras, mostly Rhodians; and a homogeneous lot of seven thousand pieces of Campanian A ware. On the same ship were also found several handleless Campanian lamps, some balsamaria, several commonware vases, and some semi-thin walled vessels of gray clay. Although some of the shapes represented in the 1

Benoit 1961. Fluted guttus, oenochoe, fish plate, plate of Lamboglia Form 36, bowl of Lamboglia Form 27. 3 Oppida of Enserune, Pech-Maho, Montlaures, Teste-Negre, Nages. 4 L. F. Gantes, "Note sur les ceramiques a vernis noir trouvees sur Γ 'oppidum' de la Teste-Negre aux Pennes (Bouches du Rhone)," Revue de la Federation Archeologique de VHirault 1 (1978) pp. 97-103. 5 M. Py, "Apparition et developpement des importations de ceramique campanienne A sur l'oppidum des Castels (Nages, 2

2

lot of black glaze vases are attested in Gaul as early 3 as the end of the third century B.C., most of the other forms do not seem to have been exported be­ fore 190-180 B.C. It is, for example, at that time that 4 they appear in the oppidum of Teste-Negre, near 5 Marseilles, and in that of Nages, in the area of Nimes. The huge increase, moreover, in the expor­ tation of wine and ceramics at the beginning of the second century was probably a direct result of Rome's victory over Carthage in 201 and the Roman expansion into Campania, an area in which the man­ ufacture of ceramics had up to that time not been actively pursued. For the first wreck, then, we propose a date in the first quarter of the second century, more precisely between 190 and 180. That date, moreover, accords chronologically with the dates given the Rhodian amphora stamps (210-175 B.C.)6 and would seem to be corroborated by tests made on fragments of wood from the wreck. 7 The soundings made in 1980 by the Department of Underwater Archaeological Research (DRASM) have permitted us to locate the bottom of the hull of this first wreck, which is situated at a depth of 40 m at the foot of the reef. Partly crushed on the north­ east under huge stone blocks, the wreck seems to extend toward the southwest of the site for more than a dozen meters. The presence of fragments of Greco-Italic amphoras, stuck to the wood by the pitch that coated them, confirms the identification of these remains. 8 Over the whole site, the material of the second wreck was, even according to the excavation jour­ nal, very clearly separated from the lower level, which contained the Greco-Italic amphoras, by a Gard) d'apres les fouilles du depotoir J 1," Revue de la Federation Archiologique de I'Herault 1 (1978) pp. 43-70. 6 V. Grace, "Timbres amphoriques trouves a Delos," BCH 76 (1952) p. 525; V. Grace, "The Eponyms Named on Rhodian Am­ phora Stamps," Hesperia 22 (1953) p. 122; V. Grace, "Revisions in Early Hellenistic Chronology," Mitteilungen des deutschen Archdologischen Institute, Athemsche Abteilung 89 (1974) pp. 193-200. 7 This carbon-14 analysis gave a terminus ante quern of 228 B.C. (Benoit 1961, p. 197). 8 The ribs of this hull bottom did not measure less than 10 cm in thickness. The thickness of the planking was 5.5 cm.

VIII, APPENDIX 1: GRAND CONGLOUE

flooring of wood. These elements of naval architecture, of which we possess sketch drawings, were misinterpreted by the divers as the deck of one single ship. In reality, they found the bottom of the hull of a second ship. Our present knowledge of ship construction removes all doubt on that point. The cargo of the second wreck, which was spread in tiers down the slope at a depth of between 32 m and 40 m, consisted chiefly of wine amphoras of Dressel Form IA (Will Type 4a). This type of jar seemingly did not appear in Gaul before the end of the third quarter of the second century B.C.9 These containers, the total number of which is estimated at about 1,200, mostly bore on the rim the stamp of Sestius, accompanied sometimes by an anchor and sometimes by a trident.10 Stoppers that were found in place also bore the stamp of a trader, perhaps the shipowner, L.TITI.C.F. 11 Fewer than twenty examples were stamped at the base of the neck by another potter: D. AV. ATEC. About twenty additional jars bore no stamp at all. Various objects belonging to the ship's crockery and provisions were also discovered in the level of the amphoras of Dressel Form IA. There were two so-called Punic amphoras, which, by comparison with finds on the Cavaliere12 and Dramont A13 wrecks, and from the Athenian Agora, 14 were datable to the turn of the second and first centuries B.C. Also found in close association with the amphoras of Dressel Form IA (Will Type 4a) were several Campanian vases of Type C, and in particular a small sampling of Campanian B ware. The latter clearly evoke Cosa Type II in clay and in glaze.15 Their appearance is thus characteristic of the last quarter of the second century. In this regard, it 9 That is, in fact, the time of the Riou 3 wreck, the amphoras of which are similar to those found on the oppidum of Entremont, destroyed about 123. These amphoras, moreover, do not occur at Carthage, which was destroyed in 146 B.C. 10 According to a count of amphoras and stamped necks in storage at DRASM, a slight predominance of anchor over trident can be observed. It should be noted, moreover, that a neck with outflaring rim bears a Sestius stamp with a new symbol. It appears to be a plant device (oak leaf with only two lobes). 11 The discovery, on the Dramont A wreck (Saint Raphael), of an anchor stock stamped with the same name (SEX-ARRI) as that found also on amphora lids is enlightening in this matter. Our thanks to Claude Santamaria, director of the excavation, for having been kind enough to furnish us this information. 12 G. Charlin, J. M. Gassend, and R. Lequiment, "L'epave antique de la baie de Cavaliere (Le Lavandou, Var)," Archaeonautka 2 (1978) pp. 9-93. As on the Sant Jordi A wreck, the finds were incomplete. 13 F. Benoit, "L'epave du Dramont," Gallia 16, 1 (1958). 14 V. Grace, "The Canaanite Jar," in The Aegean and the Near East, Studies Presented to Hetty Goldman, S. S. Weinberg, ed. (New York, 1956) cf. p. 96. ,5 D. M. Taylor, p. 71. These forms, which Professor F. E. Brown has been kind enough to permit us to study, all come

165

should be noted that certain shapes of this pottery group are very rare in Gaul from the first century B.C. on. 16 Other factors that permit us to isolate the approximate date of the wreck are, in brief: Several fragments of three vases of thin walled pottery with dotted barbotine decoration. The production of such ware has now been localized in the region of Cosa in central Italy, in the second half of the second century B.C.17 A group of commonware vases of which parallels have been found on wreck A of Colonia Sant Jordi 18 and on the Cavaliere wreck, both datable about 100 B.C.

Several fragments of Hellenistic relief bowls from the Ionian workshop of Menemakhos, which A. Laumonier dates in the second half of the second century B.C.19 Adding to the evidence in favor of the dates proposed for the diverse finds just described, the census of wrecks found off the Mediterranean coast of France20 seems to demonstrate that the amphoras of Dressel Form IB (Will Type 4b), heavily standardized jars apparently produced in the same workshops as those of Dressel Form IA (Will Type 4a),21 rather suddenly superseded the latter in the first quarter of the first century B.C. Among the various wrecks studied, no single ship carried the two types of amphoras simultaneously.22 The change in shape took place between 97 B.C., the latest consular date attested on Dressel Form IA (Will Type 4a),23 and 80 B.C., the approximate date of the Albenga wreck. 24 The change was perhaps connected, directly or indirectly, with the results of the Social and Civil from Deposit D. 16 That is particularly the case with Lamboglia Forms 1-8 and 10. As far as Form L 10 is concerned, cf. J. P. Morel, Ceramique campanienne, les formes, Ecole frangaise de Rome (1981) p. 262, Type 3450. 17 Moevs 1973, p. 49, pis. 1 and 2; Mayet, pp. 24, 126 and 127. 18 Cerda 1980. 19 A. Laumonier, La ceramique hellenistique a reliefs. 1. Ateliers "ioniens," Ecole frangaise d'Athenes. Exploration archeologique de Delos, 31 (Paris, 1977) pp. 11, 12. Although the products of this workshop are datable between 166 and 69 B.C., the author thinks that Menemakhos would represent quite well the oldest trends. 20 This census, conducted by the Department of Underwater Archaeological Research (DRASM), includes at present twentyeight cargoes of Dressel IA and fifteen of Dressel IB. 21 Peacock 1977, pp. 262-269. 22 Exclusive of older containers belonging to the ship's stores and certain hybrid shapes, variants of the "Dressel IA style." [Now, however, see recent restudy of the Spargi wreck where both types have been identified, Will 1983. (ED.).] 23 Zevi 1966, pp. 212, 214. 24 Lamboglia 1952, pp. 166 ff; Morel (cited in n. 16) p. 64.

166

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

Wars. This mass of chronological evidence permits us to date the wreck of our second ship with probability between 110 and 80 B.C. There no longer seems reason to doubt the Tuscan origin of the second ship's cargo and more particularly of the Sestius amphoras, the problems surrounding which have long been debated.25 The discovery in recent years in the town and especially in the port of Cosa of a massive quantity of Sestius stamps, with many different devices, seems to indicate it was a production center for the amphoras, or at least their port of storage and shipment. 26 Nothing, moreover, prevents our associating such an amphora factory with an ancestral branch of the family of P. Sestius, who owed a villa at Cosa.27 The remnants of wood from the second ship, partly detached at the time of the excavation in order to permit the recovery of material lying underneath, probably later suffered irreparable damage at the hands of the illegal divers, who are very numerous in this famous diving mecca. The sounding of 1980 unfortunately yielded only some sheets of the ship's lead plating. These were discovered to the northwest of the site, in a stratum where sherds of Greco-Italic amphoras were mixed with those of Dressel Form IA (Will Type 4a). The various illustrations in the excavation journal and especially the drawings of Y. Girault and J. Y. Cousteau nevertheless give us some details of the ship's construction. In addition,

wood fragments of identical size, removed as samples from above Greco-Italic amphoras in October 1954 and July 1957, confirm their observations. Covered with lead sheathing, the hull was formed of a double layer of planks, the strakes of which, 4 cm thick, measured between 18 cm and 22 cm wide. The height of the ribs varied between 8 cm and 9 cm, while that of the floorboards, at their highest point, attained 18 cm to 24 cm. This combination allowed a flooring of fixed planks, as well as stanchions 16 cm in diameter. The dimensions of the keel, finally, were between 18 cm and 20 cm in height and 12 cm to 14 cm in thickness. The general proportions of these various parts of the bottom of the hull, slightly larger than those of the La Chretienne C 28 or Cavaliere ships, signify a craft of average tonnage, the gross capacity of which must have been between 40 tons and 50 tons. These estimates, moreover, correspond quite well to the size of the cargo of Sestius amphoras. Those jars when full have a uniform weight of between 37 kg and 40 kg. To judge from its general proportions, this ship in all probability hardly exceeded 20 m in length. It could therefore easily have anchored inside the harbor of Cosa, the deep channel of which points to its having been the scene of important commercial activity.29 Translated by Elizabeth Lyding Will

Appendix 2. The Ship Remains in the Portus Cosanus JOHN DAVID LEWIS During the underwater excavations conducted in 1969, the outline of the lower portion of a ship's hull was discovered in the northwestern section of the harbor basin. Barely visible under several centimeters of sand and small rock, it was accidentally spotted 23 m off the beach area opposite villa Walls C and D in about 1 m of water (Figs. VIII-14 to 17). When the protective covering of rock and sand was hand-removed, approximately three-quarters of the

hull's lower portion was visible, including a long, pointed wooden prow. The after-section of the hull and stern had been destroyed by the action of the waves and marine borers. Although the appearance of the long, low protruding prow suggested an underwater ram and thus an identification as a warship, no metal covering for such a ram was found sheathing the wooden prow. The visible dimensions of the flat and heavy hull

25 Will 1956, pp. 224-244; Mamcorda 1978, pp. 122-131; Will 1979, pp. 339, 350. 26 The reader's attention is called to the resemblance between certain amphoras published as belonging to the Cap Roux wreck, where a Sestius stamp was found (J. P. Joncheray, "Etude de l'epave Dramont D, dite 'des pelvis,' " Cahiers d'archeologie subaquatique 3 [1974] pp. 21-48) and necks found at the site of Albinia (Peacock 1977, fig. 3, nos. 1-2).

27 Will 1956. Will first advanced the hypothesis in 1956 that that amphora factory might have belonged to P. Sestius himself. In the same article, she also suggested that there must be two or more wrecks off the Grand Congloue. 28 J. P. Joncheray, "L'epave de la Chretienne C," Cahiers d'archeologie subaquatique, supp. 1 (1975). 29 Lewis 1973, pp. 233-259.

VIII, APPENDIX 2: SHIP REMAINS

are 13.50 m long and 3.40 m wide with an estimated reconstructed length of about 15 m and a beam of 3.90 m. Twenty-nine well-shaped frames are preserved, hewn from heavy timbers and spaced on 0.33-m centers (Fig. VIII-17). The floor timbers are joined to the futtocks at what appears to be the turn of the bilge by iron fastenings securing the two; they were roughly scarfed and occasionally notched. These joints are often reinforced by the addition of a short piece of wood bolted to each of the joined members. Only short pieces of the futtocks remain above the turn of the bilge. The average frame measures 0.12 m in depth and 0.12 m in width with the longest measuring 3.50 m. The strakes appear to be fastened to the frames by means of iron spikes, the ends of which are often visible on the frames. The strakes likewise are cut from large, heavy timbers; the butts that are visible measure 0.035 m thick and between 0.195 m and 0.230 m wide. The strakes were laid flush as in carvel-built hulls. There were no signs of any planking laid as flooring, as one would expect in a cargo carrier. On the other hand, the preserved strakes are very heavy, presumably heavy enough to withstand the loading and unloading of cargo. The exact purpose of the ship's original builder may have to await a fuller examination of the entire hull. The bow is the most unusual feature of the ship (Figs. VIII-15, 16). It is almost 3 m long and consists of three short, fat frames boxed in by heavy 0.045m- to 0.050-m-thick planking—even thicker than that of the hull. Both round and square iron pins, now heavily encrusted, were used to secure the blocks of wood to the planking. The blocks are spaced between the bow planking and come to a sharp, ramlike point. Because the bow's wooden blocks were so fragile from exposure, in contrast to the well-preserved hull planking and frames, which had been covered, rocks were placed around them while the bow was measured and photographed (Fig. VIII-16). This strong, reinforced prow may well have originally been equipped with a sheath of bronze or iron, which has disappeared.1 No stempost was found. Possibly there was neither a keel nor keelson, although this was not verified by excavation under the hull. 1

To form the ram, some ancient warships used removable bronze casings over a wooden core. This allowed the bronze covering, which was an expensive item, to be removed and salvaged. It is possible that such a situation occurred here if the Cosa ship originally was built as an oared patrol craft of some kind, later to be converted into a cargo barge. See Casson, SSAW, p. 85. I am also grateful to Casson for editing this manuscript. Cf. E. Linder and Y. Ramon, "A Bronze Ram from the Sea of Athlit, Israel," Archaeology 34 (1981) pp. 62-64.

167

Because of the lack of adequate conservation means at the time, the wreck unfortunately could not be recovered. After measurements were taken and the entire hull photographed over a 3-m grid, it was again covered with rock and sand to protect the remaining timbers for future recovery and preservation. A sample of wood from one of the frames was taken for carbon-14 analysis made by the University of Pennsylvania. The results indicate that the wood is less than 500 years B.p. with a MASCA corrected date of A. D. 1588.2 The very flat and broad nature of the hull, the absence of a mast step and the possible absence of a keel and keelson indicate that the craft was probably driven by oars alone. Its heavy construction and wide beam indicate a cargo carrier. The long, sharp bow may be compared to that characteristic of Venetian lake or river galleys of the fifteenth and sixteenth centuries.3 The estimated length of 15 m would be appropriate for such a lagoon or river craft. Another possible identification to consider, however, is with the type of galley used for short, fast coastal passages or pirate raiding. 4 Could our ship have been originally constructed as such a galley and later converted to a small cargo vessel? In any case, from the amount of iron ore found throughout the hull it is quite certain that it ended its days as an ore barge grounded in the ancient Portus Cosanus. The iron ore indeed has so impregnated the timbers with iron oxide that the cargo has actually helped to preserve the wood. The carbon-14 dating of the wood in the late sixteenth century and the evidence of its cargo of iron ore suggest that the wreck may be associated with the last documented use of the Portus Cosanus in the later Renaissance during the time of Cosimo I, the Grand Duke of Tuscany (1519-1574). Preserved in the archives of Florence is a letter from the second half of the sixteenth century to the duke from an unknown local promoter, who both urges the building of a defense tower at the Tagliata and outlines a large-scale project for reclaiming the lagoon, including the construction of ironworks to smelt ore coming from the mines of Elba.5 The Torre della Tagliata was built, but of the rest of the project we For ancient types of oared harbor craft equipped with rams that allowed them to double as a combat unit, see Casson, SSAW, p. 162. 2 B. Lawn, "University of Pennsylvania Radiocarbon Dates, XV," Radiocarbon 15, 2 (1973) p. 368, no. P-1594 (COH-69-W1). 3 See Bass, History, pp. 207-210, illustration on p. 208, fig. 2; p. 215, fig. 10; p. 218, fig. 2, lower right. "Ibid., pp. 210-211. 5 Venerosi-Pesciolini, pp. 15-28; Archivio Stato Firenze, Carte

168

HISTORY, GEOGRAPHY, ARCHITECTURAL REMAINS

know nothing. Perhaps our hull loaded with ore is evidence that other aspects of the ambitious reclamation project were actually carried out, even if no evidence of smelting furnaces for an ironworks has as yet been discovered in the immediate area. A thick layer of carbon was found, however, just below the surface of the modern parking lot in the excavations of 1972, Beach Probes I-VII (Fig. III-l),

indicating that some industry was active here of post-Roman date. 6 The Cosa ship warrants excavation and preservation, both as a document of an otherwise unknown period of the history of the Portus Cosanus as well as for its archaeological value for the history of ship construction, since it is one of the few wrecks preserved from the Renaissance.

Strozziane, series I, vol. CX, c. 309-329. I am grateful to McCann for this reference and suggestion. 6 See McCann, chapter three. Another large mass of iron ore was discovered underwater some few hundred meters east of the

harbor. Perhaps another iron-ore barge from this period in the port's history met a similar fate. Investigation of this area is planned for the future.

Part II. The Evidence for Trade

'"•"Λ

1 PM 12c

SELECTION OF WILL AMPHORA TYPES REPRESENTED IN THE PORTUS COSANUS

Types are based on the typology of Roman amphoras in E. L. Will, Stamped Roman Amphoras in the Eastern Meditenanean, forthcoming in the Athenian Agora series. The authors are grateful to H. A. Thompson, Field Director Emeritus of the Agora Excavations, for permission to reproduce part of the Will typology. For illustrations of shapes not included (Types la, Ic, l i e , 12b, 16, 18b, 21c) see figures referred to in the catalogue. Scale 1:15. Id: 4a: 4b: 5: 10: 12a: 12c: 13: 14: 18a: 20: 21a: 21b: 24a:

Ansedonia, private collection, P.H., 0.698 m Agora P 8106. H., 1.05 m Delos T D 6029. H., 1.04 m Delos T D 6030. H. (restored), 1.18 m Agora SS 7319. H., 0.865 m Agora P 3464. H., 0.805 m Agora SS 10072. P.H., 0.28 m Agora SS 9468. P. H., 0.87 m Agora P 25734. H., 1.12 m Agora P 18000. H., 0.64 m American Academy in Rome. H., 0.79 m Cosa, hill site, C 70.468. P.H., 0.94 m Agora P 14618. H., 1.145 m Albenga shipwreck (Lamboglia 1952, p. 166, fig. 24)

rtffi)

N

24a

Chapter IX. The Roman Amphoras ELIZABETH LYDING WILL

INTRODUCTORY TEXT

Shipping amphoras from the Roman period have been found in large numbers in the excavations of the port of Cosa. A total of 789 jars or fragments, representing at least thirteen different types of amphoras, has been identified.1 The Roman amphoras comprise, in fact, the largest single group of discoveries in the port. More importantly, they constitute an accurate index to the chronology of the port and to the economic activity which accounted for its existence. That circumstance is the happy result of the care with which the excavation was conducted. AU the amphora finds were kept for study, and since most of them can be assigned to datable types their chronological value for the port is self-evident. But the significance of the port amphoras goes beyond the conclusions they enable us to draw about the chronology and the commercial life of a specific site. Because that site was a harbor, and because few Roman harbors have been excavated, the finds take on added interest, especially in view of the connection of the Portus Cosanus with the early Roman colony of Cosa. 2 Study of Cosa's port adds a valuable new dimension to our knowledge of Cosa itself, throwing into relief its economic importance and permitting it to be seen as part of a major export center during two centuries of its history, a matter about which our literary sources are silent. The amphoras from the port of Cosa thus add to the growing body of evidence indicating that, in establishing colonies in Italy, Rome heeded her economic interests as well as political and military considerations. The general outlines of what the shipping amphoras reveal about the port's chronology are clear 1 The thirteen types of Roman amphoras discussed in the catalogue that accompanies this chapter are numbered according to the system used in my forthcoming volume on Roman amphoras in the Athenian Agora series. I there describe twenty-three types of Roman amphoras from the eastern Mediterranean area. Of those twenty-three types, most of them from closely dated contexts at the Agora excavations, the following are represented at

and easily summarized. Broadly speaking, they suggest that the history of the port parallels that of the town of Cosa, but there are also significant differences between the two sites. As in Cosa itself, economic activity began in the port in the years immediately after the planting of the colony, if not before, and lasted through the third century A.D. The second and first centuries B.C. were the time of greatest activity in the port. The same activity is also reflected to a much smaller degree in the amphora finds from the town. During that period, the amphoras show that the port served as a major export center. An industry owned by the Sestius family shipped wine and eventually garum and perhaps other fish products to the western Mediterranean, particularly to Gaul. But a decline in activity began in the last half of the first century B.C. Exportation had apparently ceased by the end of the century, and in the port, at any rate, imports did not take the place of exports. The Portus Cosanus lay virtually dormant during most of the first century A. D. In the town of Cosa, on the other hand, the same decline was followed by intense importation, apparently via some other harbor, of the same kinds of commodities that had formerly been exported. In the second and third centuries A.D., the port revived to some extent. Amphoras imported from abroad are found, though the same types occur in far greater numbers in the town of Cosa. The port never returned to the remarkable level of commercial activity it had seen in the second and first centuries B.C., when it must have been one of the chief exporting centers of Italy and of the western Mediterranean. the Portus Cosanus: Types 1, 4, 5, 10, 11, 12, 13, 14, 16, 18, 20, 21, 23. Type 24, which is also represented in the Portus Cosanus catalogue, is not known to occur in the eastern Mediterranean. 2 The large group of Roman amphoras and fragments, numbering at present close to 1,500 pieces, from the colony of Cosa will be presented in a separate publication.

172

THE ROMAN AMPHORAS

In addition to the chronological structure that they permit us to delineate, the Roman amphoras also testify to the ways in which the port of Cosa operated. Although the port at its height was primarily a facility for the processing, bottling, and shipping of wine and fish products, there is no evidence that any commercial activity other than shipping took place there before the end of the Second Punic War. The third century B.C. is marked in our finds by only imported amphoras. Those are the early GrecoItalic wine jars discussed below as Type la, a widely distributed type of amphora, manufactured in various places throughout the Mediterranean but perhaps to the greatest degree in Sicily. Type la was most prevalent in the first quarter of the third century B.C. It occurs frequently at the major Etruscan sites. The presence in the port and in Cosa itself of fragments of these fragile jars and in Orbetello of similar amphoras (Figs. IX-I, 2) suggests that the port was being used for the receipt of goods in the earliest years of the colony, if not before, especially during the preparation of the site for the colonists, and that a certain amount of wine was being imported. While not landlocked, the unimproved port afforded anchorage, and logic suggests that some of the original colonists may have been practical enough to send belongings and supplies, as well as to make the trip themselves, by ship. The spot would, in any case, have been familiar to seafarers for centuries, long before the formal planting of the Latin colony in 273 B.C. The Type la amphoras from the Portus Cosanus thus bear witness to the early years of the colony, but about the remainder of the third century we have from the port no further amphora evidence, though Livy's statements (cited in chapter one) indicate that the port was functioning during the Second Punic War', perhaps both militarily and commercially. The presence, also, of Type l b amphora fragments in the town of Cosa may, as is discussed below, point to the existence of an early local amphora factory, one controlled by the Sestius family and dating from the last half of the third century B.C. Since such a factory was apparently located in the port during the first half of the second century (see Fig. VII-9), it seems possible, given the other evidence, that a small export operation, probably involving wine and perhaps fish products as well, had commenced in the Cosa area in the preceding century. If it existed, it was perhaps located, for safety's sake, on the hill of Cosa, above the port. The ]ES stamp (part of Sestius trademark?) on a Type lb amphora from Pech-Maho near Narbonne, a site de-

stroyed at the end of the third century B.C., hints that it was the Sestii who owned such a facility, whatever its location. So do scattered finds of Type lb amphora fragments at sites in the western Mediterranean. Although tests are not complete, those fragments appear to be made of the same distinctive fabric that characterizes the later products of the Sestius factory, in the second and first centuries B.C. The ]ES stamp is also the earliest known amphora stamp in Latin letters. If it is a broken Sestius stamp, we have evidence that the first Sestius amphoras were also the earliest truly Roman amphoras and that the Cosa area contributed to the development of Roman trade after the First Punic War. The preeminence of the Sestius amphoras of Types Id and 4a in Western commerce, beginning in the first half of the second century B.C., would thus be more easily explained. But at the moment, information on these matters is lacking from the Portus Cosanus. Our data are at present incomplete, but the dearth of amphoras in the port in the last part of the third century B.C. may not be so surprising, after all. If by that time the colony was in a position to export surplus products, the fact that Carthage was still a threat, both on land and on sea, would have made it more sensible to place any permanent processing and manufacturing structures, such as an export operation would have required, behind the town of Cosa's heavy walls rather than in the more exposed port. Loading areas for merchantmen as well as military facilities could still have existed in the port, however. Type lb was in any case a short-lived, hurriedly designed, experimental effort to respond to military demands and expanding commerce at the same time. If it was, in fact, a product of the first years of the Sestius firm, its absence so far from the port might be as much the result of "excavational chance" as of any other circumstance. Type Id, the "standard" Greco-Italic, represents the climax of the evolution of the Romanized shape, of which Type lb was the beginning (Fig. IX-3). The abundant finds of Type Id in the Portus Cosanus indicate that once the threat from Carthage had subsided, at the end of the Second Punic War, an export operation began in earnest close to the harbor wharves, where it could most efficiently be carried out (Fig. VII-9). The scale of the enterprise can be judged from the number of finds. The need to staff such an export facility could explain, at least in part, Cosa's petitioning the Senate at Rome in 199 B.C. and 197 B.C. for more colonists (Livy 32.2.7; 33.24.8-9). Of the amphoras discovered so far in the port, 9 percent belong to Type Id, and the same

IX: THE ROMAN AMPHORAS

high percentage of the type occurs in the town of Cosa. It is second only to Types 4a and 4b in frequency at both sites. Even today, in the shore area on the site of the ancient port and in the fields around Cosa, broken pieces of Type Id lie scattered in profusion, only slightly less evident than the omnipresent fragments of Types 4a and 4b. More finds of Type Id have been made in the Cosa area than anywhere else in the Mediterranean; it was a popular shape of amphora, doubtless a shipping container for wine, and among the first international commercial amphoras produced by the Romans. Like Type la, its Greek (perhaps Sicilian Greek) equivalent of a century earlier, which it superficially resembles, Type Id is found throughout the Mediterranean area. The largest concentrations are at Cosa, but examples made of different clay and thought to have been manufactured near Pompeii are probably earlier. After having been taken to Cosa, they could have influenced the evolution, if it occurred there, from Type lb to Type Id. Most finds of Type Id in the western Mediterranean, however, have "Sestius" clay and must have come from the Sestius factory in the port. Further evidence pointing to the Sestii as owners of the Type Id factory is contained in graffiti found on some of the jars and fragments of the type. Particularly helpful is a graffito, M. SE, on the shoulder of a broken jar from the town of Cosa (no. C 65.397). Stamps are rare on these early amphoras, and we are justified in interpreting these letters, scratched before firing, as naming M. Sestius, the manufacturer of the jar. The name serves as another link in the chain connecting Type Id to the Sestii, but even if the graffito did not exist, the clear relationship between Types Id and 4a in shape and in clay would point to the former as ancestral to the latter, which were once thought to be the only Sestius jars. We can no longer refer to Type 4a, then, as "Sestius-type." I suggested as early as 19563 that the Sestii owned at Cosa a factory which manufactured Type 4a, and it seemed natural thereafter to refer to the type as "Sestius-type." But by 1979 it was clear that Types 4b and 5 also occasionally bore Sestius trademarks. Our new knowledge about the earlier history of the factory now necessitates our expanding the term to include Type Id and perhaps also Type lb. The Sestius family of amphoras, in fact, affords us our first opportunity to study the development of a Republican Roman amphora factory's products over a long period. The implications 3

Will 1956, pp. 242-243.

173

are of great interest, not just for the future study of Roman amphoras but for research into Roman economic history and into the history of ports, particularly the port of Cosa. It may never be possible to identify the M. Sestius named in the Cosa graffito. Type Id seems to date from the first half of the second century B.C., and there is a known M. Sestius of Fregellae who, according to an inscription from Delos (IG XI 4. 757), was granted citizenship on the island in about 190 B.C. Miinzer in RE, s.v. "Sestius" (4), suggested he might be identical with the M. Sextilius Fregellanus who was spokesman for eighteen loyal Italian colonies in 209 B.C. (Livy 27.10.3). Fernand Benoit, the archaeologist in charge of the excavation of the underwater site off the Grand Congloue near Marseilles, even suggested that M. Sestius of Fregellae or a relative had manufactured the hundreds of Sestius Type 4a amphoras discovered in what we now know was a great double wreck. 4 The dates of the jars contradict such a theory. We can assume, however, that whether or not M. Sestius (or M. Sextilius) of Fregellae and M. Sestius of Cosa were identical, the latter was an early member of the Sestius firm, an ancestor or relative of the politically powerful Sestii of the first century B.C. It was those later Sestii who, building on the family company's long history, brought the Portus Cosanus to the height of its commercial importance. M. Sestius could have been the founder of the Sestius enterprise, however. Our assessment of his role must await further research on Type lb and its possible connection with Cosa in the third century B.C. Type Id does not seem to have been in use for very long after the middle of the second century B.C. Its evolution into Type 4a is discussed in the introduction to the catalogue of that type. By the late second century, the Sestius firm's new shape, Type 4a, an elongation and enlargement of the Type Id form, had been fully developed, arguably in response to market demands (Fig. IX-4; Color Fig. 3). The type is found with great frequency in the West (Text Fig. IX-I). Its popularity is reflected in successive enlargements of the port facilities, modifications that seem to have begun in the latter second century and to have continued, in stages, well into the first century B.C. Wine was not the only commodity processed in the port at this time, for extensive arrangements for the preparation of garum and probably of other fish products are seen not just in the combined fishery and Spring House uncovered by 4

Benoit 1961, pp. 68-70.

174

THE ROMAN AMPHORAS

the excavations, but also in the amphoras of Type 5, imitations of a Spanish form used for garum, as we argue below, that were produced by the Sestius factory from the late second century B.C. onward (Color Fig. 4). The diversification into fish was a success, to judge from the use of advanced technology in its operation, evidenced by the water-lifting mechanism discovered in the Spring House, as well as the large size of the preserved concrete fish enclosures. Pointing in the same direction is the evidence that the Sestius firm was developing additional, in some cases quite experimental, amphora forms, like Type 24a and the other shapes, discussed below, in which fish products were probably shipped. We are in a position to argue, then, that during its heyday in the latter second to mid-first centuries B.C., the Sestius enterprise in the port was at the zenith of its wine and fish businesses. The garum, an expensive commodity, would have been distributed less widely than the wine. What we know about the Sestius wine indicates that it was directed toward a mass market rather than a more discerning clientele. The size of the shipment of amphoras in the upper Grand Congloue wreck implies that the wine carried in the jars was "table grade," and our literary sources suggest that Tuscan wine in general could perhaps best be described as vin ordinaire. In A.D. 1954, Cousteau used harsher words to reflect his reaction after uncorking a still hermetically sealed amphora from the Grand Congloue and tasting its contents: "A poor vintage century, that wine." 5 The Sestii, then, used mass-production techniques in order to capture a mass market. Such techniques are economic reali-

ties in our own era, but they were new in the Roman world. They prove the rapidity and efficiency with which Roman commercial capabilities had developed during the second century B. c., and they highlight the pioneering role played by the Portus Cosanus factory in that development. Some statistics will illustrate the unique position occupied by Type 4 at Cosa from the latter second century B.C. onward. As we note elsewhere in this chapter, Type 4a is the most frequently occurring amphora in the Portus Cosanus, accounting for over 38% (300) of the finds. Type 4b, a first century B.C. descendant of Type 4a, represents some 26% of the total (208 jars or fragments). If we add indeterminate pieces, mostly belly fragments that could belong to either type, we discover that 70% of the port amphora material belongs to Type 4. The figures are convincing. The manufacture of Type 4 occurred in the port at the height of its operations. For the town of Cosa, the percentages are much lower (4a: 13%; 4b: 10%; total of both including indeterminate pieces: 24%), but are still far higher than one would expect to find at a single site, not to mention a small colonial town. Most of the pieces of Type 4 from the Cosa area are of Sestius clay. If we examine the trademarks on the stamped pieces, further, we find that 86% of the stamps found in the port (99 of 115) are Sestius trademarks, all but one of them on Type 4 and most of them on Type 4a. Twenty percent of the Latin amphora stamps from the town (41 of 201) come from the Sestius factory. Both figures represent very high, if not unique, concentrations for a single series of stamps at a single site.6

5 Cousteau 1954, p. 13. On Tuscan wine, see, for example, Horace Satires 2.3.143; Martial 1.103.9, 2.53.4, 3.49.1; Persius 5.147. 6 Totals given here for amphora stamps from the port of Cosa comprise not only the 89 stamps described in the accompanying catalogue (a number that includes 7 stamps occurring with one or more other stamps on the same fragment; cf. Cats. A64, A68, A79, A95, A199, A204), but also the 26 stamps published in Manacorda 1978, in the aftermath of an apparent excavation near the seaside parking lot of Ansedonia, within the area of the ancient port of Cosa. I had announced plans in 1974, at the annual meeting of the Archaeological Institute of America, to excavate at the same site with Frank Brown in 1975, under the permit granted to the American Academy in Rome, but, in the event, funds were not available (Will 1979, notes 7 and 11). It has seemed appropriate and useful, thus, to include Manacorda's published stamps in the totals of stamped pieces cited here, although his finds have naturally not been included, except for references to them, in the catalogue of Roman amphoras in this volume, in the overall totals given in Table IX-I or in the alphabetical list of stamps found in the official American Academy excavation (see the end of this chapter). It should be added that, in computing the totals of stamps, all devices have been considered Sestius stamps. Further, in the interests of accuracy, and because so much is now being written about the Sestii of Cosa, it should be pointed out that the

totals cited by J. H. D'Arms in a keynote address delivered at an international congress on Roman trade at the American Academy in Rome in 1978-1979, and in two recent publications, are out of date and misleading. (See D'Arms 1980, review by McCann and Will 1984, and D'Arms 1981. The texts of those two discussions of the commercial activities of some of the Sestii are almost identical, with a few changes in the later piece.) It is true that about 10 percent of the Cosa amphoras and fragments from both sites (the town of Cosa and the Portus Cosanus) bear Sestius stamps, but two factors call into question the usefulness of that assertion, for which D'Arms is in error to cite me as a source. One factor is this: not all (and perhaps rather few) Roman amphoras were stamped. Secondly, unstamped fragments, which D'Arms uses in arriving at his percentage, might or might not once have been part of a stamped jar. D'Arms thus uses incorrect figures m the wrong way. Sestius trademarks account, in fact, for 86 percent of the amphora stamps from the port of Cosa and for 20 percent of the stamps from Cosa itself. Those striking percentages are given in my 1979 article. In any case, the new information, referred to herewith in the text, permitting us to push back the beginning of the Sestius factory to the first half of the second century B.C., and possibly to the last half of the third century, makes inapplicable D'Arms's discussion of whether or not L. Sestius, the father of P. Sestius, was the "instigator of the family business." In my 1956 article, and in the unpublished texts of talks before the Archaeo-

IX: THE R O M A N A M P H O R A S

175

200 Miles 200 Km

Text Fig. I X - I . Distribution of verified Sestius stamps as o f 1981. (Revision o f map in Will 1956, p. 226, which contained only sites 1-13.) 1. Grand C o n gloue. 2. He du Levant. 3. Ruscino. 4. Ampurias. 5. Cosa. 6. Mont Beuvray. 7. Besanqon. 8. Basel. 9. Lezoux. 10. Alise-Sainte-Reine. 11. Ventimiglia. 12. Vada Sabatia. 13. Vieille-Toulouse. 14. Toulouse. 15. Pamiers. 16. La Lagaste. 17. Peyrepertuse. 18. Le Carla a Bouriege. 19. Mayne a Belesta. 20. TorroeIIa. 21. N i m e s . 22. N y o n . 23. Vienne. 24. Roanne. 25. Tournus. 26. Poitiers. 27. Cap Roux. 28. Luni. 29. Volterra. 30. Saturnia (?) 31. Narbonne. Boston University Cartographic Services Lab 1979

The number of Sestius stamps in the port is so dramatic as to make the existence of a Sestius factory in the port a matter of certainty. The Cosa figures command additional respect in view of the fact that, as of February 1982, only 61 Sestius stamps were known to me to have been found at other sites on land, especially in Gaul. The Cosa area thus accounts for 70% of all Sestius stamps found so far on land, and Cosa is the only place where almost all of the known varieties of Sestius trademarks occur. If we add to the Portus Cosanus totals the reportedly hundreds of Sestius stamps from the upper of the two Grand Congloue wrecks, the figures for the port are vastly multiplied, of course. Benoit estimated the total number of Sestius cargo jars at the Grand Congloue site as 1,500-1,700 and Long as about 1,200

(chapter eight, Appendix 1, above), but we do not yet know how many of that number bear stamps. The Sestius ship that foundered off the Grand Congloue should probably be thought of as a "floating Portus Cosanus." Nor, apparently, was it the only ship in the possession of the Sestii. Cicero mentions the navigia luculenta Sesti, and however we interpret that phrase, whether they were warships or not, it would seem to indicate, as I have suggested elsewhere, that the Sestius family possessed or had the use of ships that were distinguished or splendid or special in some way, perhaps in size as well as in appointments. 7 Cousteau felt that the many sections of three-inch lead pipe from the Grand Congloue might have been part of a plumbing or a pumping system, and it may be relevant, as I pointed out

logical Institute of America in 1974 and 1975, to which D'Arms seems to be replying, I had suggested P. Sestius as the manufacturer of the Sestius amphoras; but by the time of my 1979 article, it was clear, as I pointed out, that the extensive processing, manufacturing, and shipping industry which McCann's excavations show was owned by the Sestn in the port of Cosa was a family enterprise, one that must have developed slowly and over a number of generations. The question of which Sestius began the family business is, in any case, minor compared to the larger questions raised by the Sestius amphoras. The size and longevity of the Sestius organization force us to rethink our preconceptions not just about whether or not nobles engaged in trade, but, more importantly, about the nature of the Roman economy. The early use of mass-production, diversification, and merging techniques,

as employed by the Sestius factory, is very suggestive in the same regard (see Will 1979, pp. 348-349 and n. 37; Will 1982-3; Will 1984). And the previously unrealized commercial importance of the Portus Cosanus, and, by extension, of the colony of Cosa also points up the rudimentary state of our knowledge about Roman colonies and the role they played in trade. Many other family amphora firms similar to that of the Sestii will be discussed by me in future publications. The existence of such firms calls into question recent efforts like that of Tchernia (1983, especially pp. 99-104) to see the Republican wine trade in the west as based on exchanges rather than on the supplying of market demands. Also see above McCann, chapter one, n. 135. 7 Will 1979, pp. 349-350.

176

THE ROMAN AMPHORAS

many years ago, that the wealthy and powerful P. Sestius, praetor in 54 B.C., made at least two trips to Marseille, very probably in family ships. 8 The weight of the evidence should incline us to suppose that the upper Grand Congloue wreck is what is left of one of the navigia luculenta of the Sestius firm. The masses of Sestius amphoras from the wreck are a logical extension of the dramatic evidence from the Portus Cosanus. Only a factory the size of the one in the port could have produced such a shipment. The dates of the port's zenith and of the upper Grand Congloue wreck correspond. The wreck is now being dated between 110 and 80 B.C., according to information kindly provided me by Luc Long, who has undertaken the task of reediting Benoit's notes and republishing the two wrecks. 9 His date for the upper wreck accords with my own rather wide dates for Type 4a, stated below under the introduction to Type 4 (late second to mid-first B.C.). That date is necessitated, in my view, by the fact that amphoras of Type 4a occur in a closely dated deposit (C 9:7) at the Athenian Agora, one belonging to the last years of the second century B.C.; yet the frequency of Sestius amphoras at sites in France datable as late as the middle of the first century B.C. has long been recognized (Text Fig. IX-I). 10 Until the Sestius amphoras have been studied sufficiently to establish a chronology for them, a closer date for Type 4a and for the Sestius wreck off Marseilles cannot be reached. The late Republic must have been a time of great prosperity, a kind of Golden Age, for the Portus Cosanus, as for the town of Cosa on the hill above the port. With the wharves, the winery, the saltery and fishery, and the amphora factory, the port must have been the scene of tremendous activity (Text Fig. 1-1; Figs. VII-IO, 11, and 12). The amphoras suggest that imports were minimal and were confined to luxury grades of olive oil; thus oil must also have been produced in the area for most domestic needs. It is not surprising that our literary sources 8 Will 1956, p. 243. On the possible plumbing system, see Cousteau 1954, p. 20. 9 Long's study of the two wrecks is nearing completion. See his remarks above in chapter eight, Appendix 1. I thank L. Long for his patience, interest, and help during the preparation of this chapter. 10 As this chapter goes to press, I am informed by Luc Long that the estimated dates of certain French terrestrial sites, in particular the oppida of the region of Toulouse, are being restudied and will likely be put backward by about fifty years. If such a change in dates takes place, it will be possible to fix the date of Type 4a firmly in the late second and early first centuries B.C., the date suggested for the type by the Athenian Agora and Grand Congloue contexts. " On this topic, see also Will 1982-2.

picture the Sestii, the developers and undoubtedly the owners of the enterprise in the port, as wealthy and powerful, so powerful that Augustus forgave L. Sestius for being a chief assistant to Brutus and a lifelong guardian of Brutus' memory; so powerful that Horace gave L. Sestius a place of very high honor in his Odes.11 If the terminus ante quern for Type 4a proves correct, it will be necessary to recognize that during the last years of Type 4a another type of wine jar, descended from it, began to be manufactured in the Portus Cosanus. 12 Type 4b is second only to Type 4a in frequency in the port. A date as early as the second quarter of the first century B.C. is indicated for Type 4b, as I point out below. Manufacture of the type flourished, as can be judged from the number of jars and fragments found, and since most of the pieces from the port show early stylistic characteristics, it seems likely that the earliest known examples of Type 4b come from the port factory. But by the last half of the first century B.C., the chief sites producing Type 4b are clearly in southern Latium and Campania. The port is much less active than those areas, though the Sestius factory still existed. Why would such a shift have occurred? My own feeling is that the upheaval caused by Caesar's assassination was felt in the Portus Cosanus, and there were other reasons. We know that L. Sestius, the son of P. Sestius, and consul suffectus in 23 B.C., was one of Brutus' most militant adherents, accompanying him to Macedonia as proquaestor in 43-42 B.C. He was proscribed, but after the amnesty he returned to Italy. I have pointed out elsewhere that two stamps on rims of Type 4b from the town of Cosa seem to name L. Sestius as manufacturer,13 but in fact we have from the Cosa area only those two Sestius stamps, as well as a few devices or logograms, with which to link the Sestius family epigraphically to Type 4b. It seems arguable, then, that those stamps, which occur on rims of very wide diameters and could thus be from jars later than the 12 But see n. 10. No overlap need have occurred if the terminus ante quern of Type 4a is raised to the early first century B.C. 13 Will 1979, p. 348 and fig. 7, a and b. With regard to fig. 7a, Luc Long calls my attention to the stamp S E VE (or LE; the last two letters form a retrograde ligature), twelve examples of which were found on amphoras of Type 4b (Dressel IB) on the Plane 1 wreck off Marseille. That wreck is dated about 50 B.C. Cf. R. Lequement and B. Liou, "Ceramique etrusco-campanienne et ceramique aretine, a propos d'une nouvelle epave de Marseille," L'ltalie pri-nmaine et la zone republicaine. Melanges offerts a Jacques Heurgon (Rome: Ecole frangaise de Rome, 1976), p. 588. Those stamps, which the authors interpret as reading SELEV, seem, on the contrary, to be variations of the Cosa trademark of L. Sestius, referred to above (fig. 7a in Will 1979).

IX: THE ROMAN AMPHORAS

177

material described above, date from a time when L. Sestius had regained his property. That a factory of some sort still existed in the port in 23 B.C. is suggested by Horace's intentionally ambiguous language in his poem dedicated to Sestius in that year, Odes 1.4. Horace seems to picture pottery manufacturing and shipping activity as still going on in the port. But some of the port operations, and perhaps the chief operations, could by this time have moved to Rome. Tegulae naming L. Sestius are found only in Rome, though tiles or bricks stamped with the initials L. S occur in the Ager Cosanus. 14 At the present time, the evidence points to a dual pottery operation during the last half of the first century B.C., with the port firm reducing production and declining in importance as the factory in Rome grew. Other branches of the Sestius enterprise could also have been established elsewhere.15 It seems likely that the Sestius firm as a whole diversified during the Augustan Age, and perhaps immediately after the amnesty, into building materials and conceivably into Arretine ware. 16 As I indicate below under Type 24a, the Sestius factory in the port could even have been taken over by the Domitii, the other great landowning family at Cosa, at some time in the latter first century B.C. What is clear is the decline of activity in the port, beginning perhaps in the 40s B.C. But the decline was slow, and although there is evidence that the factory at Rome may have gone on into the first century A.D., we do not know details of the last years of the Portus Cosanus factory. As I point out under Type 16 below, a small garum operation, either a revival or a continuation of the garum industry of the port's chief period, could have existed near Cosa in the late first century B.C. That operation apparently was not located in the port, where, from the testimony of the amphoras, exportation had ceased by the end of the century. The geological evidence suggests that exportation ended at least partly because of natural causes. With the end of exportation from the port, one might expect the area to have continued as a center of importation, but such was not the case. And the same geological causes that helped to end exportation were apparently at work. Evidence shows few finds of imported amphoras until the last quarter of

the first century A.D., shortly after the building of the maritime villa (see Fig. VII-13). At the same time, however, the town of Cosa was importing wine, garum, and olive oil in quantity. We must thus assume that another port, presumably Port'Ercole across the bay, was receiving the food shipped to the Cosa area. With the destruction of Pompeii in A.D. 79, however, the manufacture of wine jars of Type 12 and a renewed exportation of wine apparently began near Cosa, but not in the port. It is not until the second and third centuries A.D. that a mild resurgence of activity (commercial or residential) is reflected in the port by the occurrence there of Spanish and North African shipping jars, for olive oil in particular. Types 16, 18, 20, and 21 testify to such a small revival of trade. But the Portus Cosanus never again achieved the prominence it had attained in the second and first centuries B . C , when it played an international role as an export center of first importance.

14 On the tegulae of L. Sestius, see Will 1956, p. 242; Will 1979, pp. 347-349. The stamped tiles or bricks (in the late Republic, the distinction between the two is difficult to draw, an observation for which I am grateful to Frank Brown) from the Ager Cosanus are noted by me on p. 348 of the 1979 article. See now Manacorda 1980, p. 174 and n. 7, in the published version of the paper to which I there refer. And cf. Manacorda 1981, pp. 33-34; and Carandini and Settis, section 35. 15 In Will 1979, p. 342, I note that 12 percent of the known

Sestius amphora stamps have been found in the Gallic isthmus, north of the Pyrenees, and I suggest the possibility that that area was a secondry manufacturing center. In a recent letter, Luc Long informs me that the amphora finds in the area of Toulouse are increasingly impressive. In addition to serving as a depot for imported Italian wine, Toulouse could also have been a production center. Future finds should clarify this matter. 16 On this matter, see Will 1979, p. 349, n. 37, and Will 1984. Cf. under Type 24a below.

CATALOGUE, CHAPTER NINE

G.D. G. W. H. T. H. W. M.D. P. H. R.D. R.H.

Greatest Diameter Greatest Width Handle Thickness Handle Width Mouth Diameter Preserved Height Rim Diameter Rim Height

A combination of numbers and letters, as 5YR 6/6, refers to the Munsell system for identification of soil colors, as described in Munsell Soil Color Charts (Baltimore, Md.: Munsell Color, 1975 edition).

CATALOGUE OF AMPHORAS

Type 1 Amphoras of Type 1 (Greco-Italics) are, as has been explained above and in Will 1982-1 passim, second only to those of Type 4 in frequency of occur-

178

THE ROMAN AMPHORAS

rence both in the Portus Cosanus and in the town of Cosa; indeed, the concentrations of Types 1 and 4 in and around Cosa are among the largest of single amphora types at any Roman sites. The evidence that the area of Cosa, and the port in particular, was a major manufacturing center for both types of containers is persuasive, and it continues to accumulate. The long history of Type 1 covers about two centuries, from the latter fourth to the last half of the second century B.C. During that expanse of time, five categories of Greco-Italics evolved: Form a, in the latter fourth and first quarter of the third century B.C.; Form b, in the last half of the third century; Form c, datable to about 200 B.C.; Form d, which belongs to the first half of the second century; and Form e (Will Type 2), datable from the first to the third quarters of the second century. Of these five categories, the first four are found at Cosa, all but Form b in the Portus Cosanus; and though Form e itself is not known to have appeared at Cosa, its descendant, Type 5 (to be described below), is found frequently in the port and in the town. Form d occurs in the port more often than any of the other categories of Type 1. Seventy-one of the seventy-six Greco-Italic fragments found in the port belong to Form d. The greatest known concentrations anywhere of Form d, in fact, are to be found at Cosa, as is the case with the Sestius amphoras of Type 4a, which seemingly evolved out of Form d in the last half of the second century B.C. The manufacture of both types of jars almost certainly took place in the port area. (On the clay of these pieces, see also de Boer, chapter sixteen, and Will 1982-1, nn. 11, 15, and 27.) Form a, the earliest variety of Type 1, reached its height of popularity in the first quarter of the third century B.C. Very few examples are found at Cosa, only two occurring at the port. The founding of the colony of Cosa in 273 B.C. would suggest that the pieces of Form a discovered there are among the latest known examples of the category, which at its height was heavily produced throughout the Greekspeaking regions of the Mediterranean, but perhaps especially in Sicily. There is no evidence to suggest manufacture of Form a at Cosa. Form a can be further subdivided into two subcategories, of which only one, Form a1, the more important of the two, is represented at the Portus Cosanus. The smallsized a1 jars, which have disproportionately wide bellies, broad shoulders, short necks, short handles oval in section and S-shaped in profile, narrow mouth-openings, and low, very outflaring rims, are distinguished also by their short, hollow, cylindrical

toes, fragile walls, and fine-textured clay (Figs. IX1, 2). The two pieces found in the Portus Cosanus are both fine-clayed hollow toes coming from amphoras of Type la 1 . Such jars must have brought wine to Cosa in the earliest years of the colony. Their presence in the port indicates that produce was arriving by ship, and that the port was functioning, from the beginning of the colony's history. Form b has not yet been found in the port area, though it occurs in the town of Cosa in a few examples. Evidence that the Sestius factory may have existed at Cosa as early as the last half of the third century B.C., and that Form b may have been a product of that factory, is discussed more fully in the introduction to this chapter and below, in the introduction to the catalogue of Type 4. It was apparently a transitional, experimental, short-lived variant, like Form c, for the manufacture of which at Cosa there is also tentative evidence (Will 1982-1, n. 15). Form c is also sparsely represented at Cosa. In the port, three toes of Form c have been discovered. The toes are quite unlike those of Form b, of which the toes are undefined, hardly more than bulges, and are also unlike the small, hollow toes of Form a. Toes of Form c are solid, not hollow, and the examples from the port show the "capped" bottom that frequently occurs with this variety of Greco-Italic amphora, hundreds of examples of which are known from the lower Grand Congloue wreck off Marseille and from the Lazareto (Puerto de Mahon) wreck off Minorca. Form c also has, like Form b, greater height than Form a, a longer neck, and longer handles, which regularly touch, and almost form a continuation of, the rim, which is higher than that of Form a and outflaring. The thicker-walled, more resistant fabric is of coarse clay that contains many inclusions. There is not yet any firm evidence to suggest where Form c was manufactured, beyond the preliminary evidence so far from Cosa. We do know, however, that in spite of the chance preservation of hundreds of amphoras of Form c on the two wrecks, the type occurs infrequently on land and was apparently much less widespread than Forms a and d, the two chief Greco-Italic categories. Form d is probably a little later than Form c, and it flourished a century or more after Form a. It can be dated with some confidence from at least the 180s to the 150s B.C. The form occurs very widely throughout the Mediterranean. Cosa has produced the largest number of finds, but strong evidence that Form d was also manufactured in Campania, in the area of Pompeii, and in the eastern Mediterranean

IX: THE ROMAN AMPHORAS

complicates the picture. At Cosa, it is the chief type of Greco-Italic, and in the Portus Cosanus, as was noted above, seventy-one of the seventy-six examples of Type 1 can be assigned to Form d. A factory for Form d was, as we have mentioned, almost certainly located in the port and almost certainly owned by the Sestii. Form d superficially resembles Form a but is larger, with longer neck and handles (Fig. IX3). Unlike Form a but like Form c, the toe is solid, the fabric is thick-walled, and the clay is coarse, though the eastern examples have somewhat finer clay than the Cosan and Campanian varieties, the clays of which are distinctively different from each other. Whether and how the eastern varieties of Form d evolved is still a matter for conjecture; but at Cosa and at a much later date in Campania, Form d developed into the larger and more commercially practicable amphoras of Type 4 (Fig. IX-4). That developmental process took place in the two areas along independent lines during the latter second and the first centuries B.C. The 49 pieces catalogued here represent something more than half of the examples of Type 1 found in the Portus Cosanus. The group includes fragments of 25 toes, 18 rims, 4 handles, and 2 necks. In addition to the catalogued objects, 27 uncatalogued finds, not described individually here, have been made in the port. They include fragments of 10 handles, 9 toes, 7 necks, and a single rim, all of them belonging to Form d. The 76 examples of Type 1 from the port comprise almost 10% of the 789 Roman amphora finds. In the town of Cosa, similarly, almost 10% of the amphoras found belong to Type Id. FORM a

Al. (PC 69-188). (Fig. IX-5). Unstamped. Harbor, underwater, Trench C l . 18.VII.69. Toe and small piece of lower belly. P. H., 0.09 m; diam. at base, 0.031 m. Buff clay (5YR 7/3); black, white, and bright orange bits. A hollow toe. A2. (PC 72-813). (Fig. IX-6). Unstamped. Lagoon, Spring House, rubble level below top of Wall a. 7.VII.72. Toe. R H . , 0.075 m; diam. at base, 0.042 m. Pinkish buff clay (5YR 6/6) with scattered small bits. A hollow toe, rounded at bottom. FORM c

A3. (PC 69-12). (Fig. IX-7). Unstamped. Harbor, underwater, Trench D l , Level III. 7. VII.69. Toe and fragment of lower belly. P.H., 0.24 m; diam. at base, 0.051 m. Pinkish buff clay (5YR 6/6); black, white, and red bits.

179

A4. (PC 69-186). (Fig. IX-8). Unstamped. Harbor, underwater, Trench C l . 23-24.VII.69. Toe fragment. P. H., 0.082 m; diam. at base, 0.045 m. Pinkish buff clay (5YR 6/6), more orange toward surface; black and white bits; beige surface. A5. (PC 69-187). (Fig. IX-9). Unstamped. Harbor, underwater, Trench C l . 23-24.VII.69. Toe fragment. P. H., 0.07 m; diam. at base, 0.038 m. Orange-buff clay (2.5 YR 6/8); black, white, and red bits; beige surface. FORM d

A6. (PC 68-60). (Fig. IX-10). Unstamped. Lagoon, Trench IH, east of Wall P, under cobblestone road. 5.VII.68. Handle fragment. H.W., 0.042 m; H. T., 0.024 m. Sandy pinkish yellow-buff clay (5YR 6/8); small black and white, and larger red, bits; grayish deposit on surface. A7. (PC 68-61). (Fig. IX-Il). Unstamped. Lagoon, Spring House, surface. 5.VII.68. Rim fragment and tiny piece of neck. R H . , 0.04 m; R.H., 0.027 m. Mouth and rim diameters not determinable. Sandy pinkish yellow-buff clay (5YR 6/8); black, white, and some large red bits; beige surface. Deep line incised around rim. A8. (PC 68-63). (Fig. IX-12). Unstamped. Harbor, below preserved surface of Wall M. 24.VI.68. Neck fragment and piece of shoulder. P. H., ca. 0.09 m; est. neck diam., ca. 0.09 m. Deep pinkish buff clay (2.5 YR 6/6); black, white, and red bits; whitish deposit on surface. A9. (PC 68-64). (Fig. IX-13). Unstamped. Harbor, Trench N, surface. 27.VI.68. Toe. R H . , 0.108 m; diam. at base, 0.05 m. Very sandy pinkish yellowbuff clay (5YR 6/8); tan surface. Gray and greenish spots on surface. AlO. (PC 68-65). (Fig. IX-14). Unstamped. Harbor, below preserved surface of Wall A. 24. VI.68. Rim fragment and small piece of neck. P. H., 0.07 m; R.H., 0.028 m. Mouth and rim diameters not determinable. Pale pinkish yellow-buff clay (5YR 7/6); large black and red, and tiny white, bits; beige surface. Raised line around top of neck under rim. A l l . (PC 68-66). (Fig. IX-15). Unstamped. Harbor, packed fill above level of Wall M and west of Wall M. 26. VI.68. Rim fragment and small piece of neck. R H . , 0.049 m; R.H., 0.024 m. Mouth and rim diameters not determinable. Very micaceous, sandy mauve clay (10R 6/3), pinkish yellow-buff (5YR 6/8) toward surface; beige surface. A12. (PC 69-74). (Fig. IX-16). Unstamped. Lagoon, Spring House, earth fill above basin plat-

180

THE ROMAN AMPHORAS

form. 22. VII.69. Toe, entire except for chips, and small piece of lower belly. P.H., 0.145 m; diam. at base, 0.047 m. Bright pinkish rose clay (10R 6/ 6); black, white, and many red bits; buff surface. Graffito on one side of toe. /*\S\J~V A13. (PC 69-156). (Fig. IX-17). Unstamped. Harbor, underwater, Trench D l , Level III. 3.VII.69. Neck fragment, preserving about half of neck and piece of shoulder. P.H., ca. 0.125 m. Yellowish buff clay (7.5YR 7/6); black and red bits; grayish buff surface. A14. (PC 69-157). (Fig. IX-18). Unstamped. Lagoon, Spring House, rubble fill above basin platform. 4-5, 7, 8.VII.69. Toe fragment. P.H., 0.061 m; diam. at base, 0.05 m. Pale pinkish yellow-buff clay (5YR 7/6); black and red bits. A15. (PC 69-158). (Fig. IX-19). Unstamped. Lagoon, Spring House, rubble fill above basin platform. 4"-5, 7, 8.VII.69. Rim fragment and tiny piece of neck. P. H., 0.038 m; R. H., 0.028 m; est. M.D., 0.12 m; est. R.D., ca. 0.14 m. Fine beige clay (7.5YR 8/4); scattered black and red bits; cream surface. Slight depression around top of rim. A16. (PC 69-159). (Fig. IX-20). Unstamped. Lagoon, Spring House, rubble fill above basin platform. 4-5, 7, 8.VII.69. Rim fragment and tiny piece of neck. P.H., 0.04 m; R.H., 0.028 m; est. M.D., 0.12 m; est. R.D., ca. 0.14 m. Coarse pinkish yellow-buff clay (5YR 6/8); black, white, and red bits. A17. (PC 69-160). (Fig. IX-21). Unstamped. Lagoon, Trench ID, surface, 11.VII.69. Toe fragment. P. H., 0.085 m; diam. at base, 0.038 m. Deep pinkish buff clay (2.5YR 6/6); black and red bits; beige surface. A18. (PC 69-161). (Fig. IX-22). Unstamped. Lagoon, Spring House, surface. 14-16.VII.69. Toe and bit of belly. P. H., 0.12 m; diam. at base, 0.05 m. Pale pinkish buff clay (5YR 7/4); black and red bits; beige surface. A19. (PC 69-162). (Fig. IX-23). Unstamped. Lagoon, Spring House, surface. 15-16, 19.VII.69. Toe fragment. R H . , 0.082 m; diam. at base, 0.04 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits; beige surface. A20. (PC 69-163). (Fig. IX-24). Unstamped. Lagoon, Trench ID, ca. 1.80 m below top of Wall P. 12.VII.69. Toe fragment. P.H., 0.073 m; diam. at base, 0.047 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits. A21. (PC 69-164). (Fig. IX-25). Unstamped. Lagoon, Spring House, earth fill above basin plat-

form. 21-26. VII.69. Rim fragment and small piece of neck. Mended. R H . , 0.046 m; R.H., 0.023 m; est. M.D., 0.13 m; est. R.D., 0.17 m. Rust clay (5YR 5/6); white and black bits; thick yellowish buff surface. Slight depression around top of rim. A22. (PC 69-165). (Fig. IX-26). Unstamped. Lagoon, Spring House, earth fill above basin platform. 22, 24-25.VII.69. Toe. P.H., 0.122 m; diam. at base, 0.05 m. Pale pinkish buff clay (5YR 7/4); black and red bits; beige surface. A23. (PC 69-166). (Fig. IX-27). Unstamped. Lagoon, Spring House, earth fill above basin platform. 30.VII; 1, 5.VIII.69. Rim fragment and piece of neck. P. H., 0.051 m; R. H., 0.033 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Pinkish tannish buff clay (7.5YR 6/6); black, white, and noticeable red bits; beige surface. A24. (PC 69-167). (Fig. IX-28). Unstamped. Lagoon, Spring House, earth fill above basin platform. 6. VIII.69. Rim fragment and piece of neck. R H . , 0.059 m; R.H., 0.036 m; est. M.D., 0.13 m; est. R.D., 0.17 m. Pinkish buff clay (5YR 6/6), darker toward core; black, red, and white bits; beige surface. A25. (PC 69-168). (Fig. IX-29). Unstamped. Lagoon, Spring House, earth fill above basin platform. 6. VIII. 69. Rim fragment, piece of neck, and start of one handle. R H . , 0.091 m; R.H., 0.036 m; est. M.D., 0.13 m; est. R.D., 0.15 m. Pinkish buff clay (5YR 6/6); black, red, and white bits; beige surface. Concave band near top of rim. A26. (PC 69-189). Unstamped. Lagoon, Spring House, surface. 14-16.VII.69. Handle fragment. H. W., 0.047 m; H. T., 0.028 m. Bright pinkish yellow-buff clay (5YR 6/8); black and red bits; beige surface. A27. (PC 69-190). Unstamped. Lagoon, Trench IF, Level IV. 18.VII.69. Handle fragment. H.W., 0.047 m; H.T., 0.027 m. Tannish buff clay (7.5YR 7/4); black and red bits; burned surface. A28. (PC 72-19). (Fig. IX-30). Unstamped. Lagoon, Spring House, surface. 6. VII.72. Rim fragment and small piece of neck. R H . , 0.039 m; R.H., 0.031 m; est. M.D., 0.16 m; est. R.D., 0.17 m. Coarse pinkish buff clay (5YR 6/6); yellowcream surface. A29. (PC 72-61). (Fig. IX-31). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 12. VII.72. Rim fragment and small piece of neck. R H . , 0.079 m; R.H., 0.037 m; est. M.D., 0.13 m; est. R.D., 0.15 m. Coarse pinkish buff clay (5YR 6/6); yellow-cream surface. A30. (PC 72-118). (Fig. IX-32). Unstamped. La-

IX: THE ROMAN AMPHORAS

goon, Spring House, rubble level north of Wall b. 20. VII.72. Small rim fragment and bit of neck. P.H., 0.056 m; R.H., 0.036 m; est. M.D., 0.13 m; est. R.D. 0.15 m. Tannish buff clay (7.5YR 7/4); tiny black, white, and red bits; cream surface. A31. (PC 72-119). (Fig. IX-33). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 20. VII.72. Small rim fragment and bit of neck. Part of one handle attachment preserved. P. H., 0.067 m; R.H., 0.031 m; est. M.D., 0.12 m; est. R.D., 0.16 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; buff surface. Slight depression around rim near top. A32. (PC 72-121). (Fig. IX-34). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 20. VII.72. Rim fragment and small piece of neck. RH., 0.073 m; R.H., 0.04 m; est. M.D., 0.13 m; est. R.D., 0.16 m. Bright peach-buff clay (5YR 7/ 8); black, white, and red bits; yellowish buff surface. Wide depression around rim. A33. (PC 72-126). (Fig. IX-35). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 20. VII.72. Small rim fragment and piece of neck. R H . , 0.07 m; R.H., 0.032 m; est. M.D., 0.14 m; est. R.D., 0.16 m. Peach-buff clay (5YR 7/8); black, white, and especially red bits; beige surface. Line incised around rim. Raised line around top of neck under rim. A34. (PC 72-259). (Figs. IX-36, 37). Unstamped. Lagoon, Spring House, surface. 28.VII.72. Rim fragment and small piece of neck. P. H., 0.064 m; R.H., 0.039 m; est. M.D., 0.12 m; est. R.D., 0.15 m. Pale pinkish buff clay (5YR 7/4); tiny black, white, and red bits. Deep line incised around rim. A35. (PC 72-263). (Fig. IX-38). Unstamped. Lagoon, Spring House, surface. 26.VII.72. Rim fragment and small piece of neck. P. H., 0.076 m; R.H., 0.035 m; est. M.D., 0.13 m; est. R.D., 0.15 m. Deep pinkish buff clay (2.5YR 6/6); black, white, and red bits; beige surface. A36. (PC 72-788 bis). (Fig. IX-39). Unstamped. Lagoon, Spring House, fill in Room 1. 10. VII. 72. Rim fragment and bit of neck. P. H., 0.043 m; R.H., 0.027 m; est. M.D., 0.12 m; est. R.D., 0.17 m. Deep pinkish tannish buff clay (7.5YR 5/6); black, white, and red bits. Deep line incised around rim. A37. (PC 72-789). (Fig. IX-40). Unstamped. Lagoon, Spring House, surface. 1. VIII. 72. Toe, chipped, and piece of belly. P. H., 0.135 m; diam. at base, 0.044 m. Coarse mauve clay (10R 6/3); black and red bits; dirty whitish surface. White deposit ton surface.

181

A38. (PC 72-790). (Fig. IX-41). Unstamped. Lagoon, Spring House, surface. 2.VIII.72. Toe. P.H., 0.097 m; diam. at base, 0.051 m. Clay hidden by coat of mortar. A39. (PC 72-791). (Fig. IX-42). Unstamped. Lagoon, Spring House, fill below rubble level north of Wall b. 21. VII.72. Toe and pieces of lower belly. P.H., 0.125 m; diam. at base, 0.055 m. Sandy pinkish buff clay (5YR 6/6), tannish buff (7.5YR 7/4) toward surface; cream surface. Mortar adheres to bottom of toe. A40. (PC 72-792). (Fig. IX-43). Unstamped. Lagoon, Spring House, rubble level south of Wall a. 3. VIII. 72. Toe and piece of lower belly. P. H., 0.11 m; diam. at base, 0.04 m. Coarse tan clay (7.5YR 7/4). Surface obscured by whitish deposit. Three concentric lines incised around lower belly. A41. (PC 72-793). (Fig. IX-44). Unstamped. Lagoon, Spring House, fill in Room 2. 4.VIII.72. Toe and pieces of lower belly. P. H., 0.125 m; diam. at base, 0.05 m. Sandy, bright pinkish buff clay (5YR 6/6); large red bits and tiny black and white sand-grains; pinkish yellow-buff surface. A42. (PC 72-794). (Fig. IX-45). Unstamped. Lagoon, Trench 20, silt level, about 0.80 m to 1.40 m below surface. 13.VII.72. Toe and piece of lower belly. R H . , 0.117 m; diam. at base, 0.046 m. Coarse pinkish buff clay (5YR 6/6). Surface obscured by blackish deposit and rust. A43. (PC 72-795). (Fig. IX-46). Unstamped. Lagoon, Spring House, sporadic find. 13.VII.72. Toe and pieces of lower belly. P.H., 0.155 m; diam. at base, ca. 0.045 m. Bright, hard pinkish yellow-buff clay (5YR 6/8); black and white bits; buff surface. A44. (PC 72-796). (Fig. IX-47). Unstamped. Lagoon, Spring House, rubble level south of Wall a. 7.VII.72. Toe fragment. P.H., 0.11 m; diam. at base, 0.058 m. Sandy peach-buff clay (5YR 7/8), pinkish yellow-buff (5YR 6/8) toward surface. Surface obscured by mortarlike deposit. A45. (PC 72-797). (Fig. IX-48). Unstamped. Lagoon, Trench 23, gray clay 1.00 m below surface. 13.VII.72. Small toe fragment. P.H., 0.062 m; diam. at base, ca. 0.046 m. Rose clay (10R 6/4); black and red bits. A46. (PC 72-798). (Fig. IX-49). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., east of Wall U. 2. VIII. 72. Toe and pieces of lower belly. P. H., 0.155 m; diam, at base, 0.048 m. Pinkish buff clay (5YR 6/6); black and red bits; beige surface. A47. (PC 72-799). (Fig. IX-50). Unstamped. Lagoon, Trench E l , surface and below. 7.VIII.72.

182

THE ROMAN AMPHORAS

Fragment of lower belly and upper toe. P. H., 0.083 m. Bright peach (5YR 7/8) to pale pinkish buff clay (5YR 7/4); black and red bits; yellowcream surface. A48. (PC 72-800). (Fig. IX-51). Unstamped. Lagoon, Trench E l , surface and below. 7.VIII.72. Toe and small piece of belly. P.H., 0.148 m; diam. at base, 0.043 m. Bright pinkish buff clay (5YR 6/ 6), pinkish at core; red and black bits; beige surface. Deep scratch on toe. Mortar adheres to fragment. A49. (PC 72-814). Unstamped. Lagoon, Spring House, fill in Room 1. 10.VII.72. Handle fragment. H. W., 0.049 m; H. T., 0.023 m. Pale pinkish buff clay (5YR 7/4); black, white, and red bits; yellow-cream surface. Type 4 In the Portus Cosanus, as in Cosa itself, Type 4 is found more frequently than any other kind of Roman amphora. In the port, finds of Type 4 number 559, or 70 percent of the entire group of amphoras and amphora fragments. Discoveries of Type 4 in the town of Cosa have also been very frequent, comprising about 25 percent of the total, an unusually high percentage, but that figure is dwarfed by the preponderance of Type 4 in the port. It is very likely that a factory for Type 4 existed in the Cosa area, and probably in the port, as seems to have been the case also with Type Id, the ancestor of Type 4 (see Text Fig. 1-1; Figs. VII-10, 11, 12) Amphoras of Type 4 belong to the general category of jars described by Dressel under his Form 1. In addition to Form 1, Dressel referred to, but did not illustrate, jars "similar to Form 1." We now know, on the basis of the painted inscriptions they bore, that the "similar" jars were earlier as a group than the ones Dressel assigned to Form 1 (Loeschcke 1942, p. 86; Zevi 1966, p. 214). I have described the earlier form in various publications since the 1950s as Type 4a (referring to the typology for Roman amphoras which is to appear in the Athenian Agora series). The later category, Dressel's Form 1, I refer to as Type 4b. Others, following Nino Lamboglia (1955, pp. 246-261), have also distinguished between the two subdivisions of the type, referring to them as Dressel IA and IB, respectively. Lamboglia also added a third subdivision, which he called Dressel IC. That category I describe below as Type 5. Contextual information at the Athenian Agora permits us to date Type 4a as early as the late second century B.C., and the tituli picti reported by Dressel

go back at least to 102 B.C. (CIL XV.4554-4555). Type 4a was in use down to about the middle of the first century B.C. (Will 1979, nn. 6, 7, and 25; but see n.10 above). It may also be somewhat earlier than the late second century. The Romans began to mark the dates of wines on amphoras in 121 B.C. (Pliny NH 14.94), the year of the consulship of L. Opimius and Q. Fabius Maximus Allobrogicus. It happens that a rather archaic-looking neck fragment of Type 4a. found near the theater of Fiesole and now in the museum there (no. 1163), bears a painted inscription naming the consuls of 121 B . C (CIL I2.659b; cf. XI.6697.1). With tituli picti, there is always the possibility of forgery, but the Fiesole neck also bears a stamp on the stump of one handle. It reads LICINV (retrograde; a misreading [CINA] of the same stamp is given in CIL I2.659a and XI.6695.25; cf. Callender 341b, but I have personally examined the stamp on two different occasions, and there can be no doubt about the correct reading). The same stamp also occurs, in different dies, on two pieces belonging to my Type 8, a hitherto unpublished amphora type datable with great probability to the last half of the second century B . C Type 8 is frequent at Delos. A stamp reading LICINVS occurs on a handle of Type 8 at Delos (TD 4264) and on another handle of Type 8 in the Benaki Collection, Alexandria (no. 586; NB 79). If the same firm manufactured all the Licinus amphoras at about the same time, the validity of the painted inscription at Fiesole is strengthened, and the likelihood that Type 4a, or an early form of it intermediate between Types Id and 4a, was in use during the last quarter of the second century B . C is increased. Hatzfeld 1912, p. 144, linked the Licinus stamp on Delos with the Orbii, many of whom, including an L. Orbius L. 1. Licinus, were on Delos, probably for business reasons, about 100 B.C. If that date is correct, and if the Orbii were in fact involved with the manufacture of Type 8, their firm apparently had a long history going back into the second century. Type 4a, in any case, is as early as the late second century B.C. The Fiesole neck may be a precious missing link between Types Id and 4a, and it may correspond in type to a whole jar of Type 4a, also in the Fiesole Museum, a piece that has the S-curve handles of Type 1 d even though its height and diameter place it with Type 4a. The damaged handles of the stamped neck fragment at Fiesole may originally have had a similar Greco-Italic curve, which I assume to be an archaic feature in Type 4a. The clay of the pieces at Fiesole (coarse, dark pinkish buff with grayish buff surface) resembles

IX: THE ROMAN AMPHORAS

that of most amphoras of Type 4a, for which the Portus Cosanus seems to have been the chief point of origin. (Peacock 1977, pp. 266-268, reports a pos­ sible kiln for amphoras similar to Type 4 at Albinia, to the north of Cosa. Other fragments from the same site were very much later in date, however, and possibly this was not a kiln for amphoras, if it was a kiln at all. See my further comments under Types 12, 16, and 21 below, and in Will 1979, n. 35.) As has already been noted in the discussion of Type Id, it, too, was probably also manufactured in the Cosa area, and there are strong circumstantial indications that it was produced both in the eastern Mediterranean and in the region of Pompeii as well. Exploration of Sabellian levels at Pompeii will, I be­ lieve, one day produce evidence to support my con­ tention that Type Id was manufactured in or near Pompeii. With the full development of Type 4a, however, the Portus Cosanus factory that produced it so dominated the market that factories elsewhere were overshadowed. At Pompeii they may have been put temporarily out of business. It is now gen­ erally accepted that the Cosa factory and probably the accompanying vineyards were owned by the Sestii, the family of which Cicero's friend, P. Sestius, and his son, L. Sestius, the consul suffectus of 23 B.C., were the most prominent members (Will 1956 and Will 1979, passim). Sestius trademarks (stamps with letters or devices or both) account for the great majority of manufacturing marks on Type 4a. Of the approximately 200 Sestius stamps reported by 1981 at sites other than the upper Grand Congloue wreck (where over a thousand amphoras with Ses­ tius stamps are said to have been found), 140, or 70 percent have been found at Cosa. Furthermore, 99, or 86 percent, of the total number of amphora stamps found as of 1981 in the Portus Cosanus are from the Sestius factory (the figure is 44 percent at present for Cosa as a whole, town and port). Evi­ dence suggests that the Sestii, earlier generations of them, also owned the Type Id factory in the port, and the reported ]ES stamp on one handle of an am­ phora of Type lb from Pech-Maho (Sigean) near Narbonne may indicate that a Sestius factory existed at Cosa as early as the last half of the third century B.C. (Will 1982-1, pp. 345-346 and Plate 85:c. For the jar and stamp in question, see also Lamboglia 1955, p. 265, and Benoit 1961, p. 41. As was pointed out in the introduction of the catalogue to Type 1, however, no examples of Type lb have yet been found in the port). Manufacture of Type Id appar­ ently ceased toward the middle of the second cen­ tury B.C. The market presumably demanded a more

183

capacious container after Delos became a free port in 166 B.C. At Pompeii, perhaps partly as a result of the competition from Cosa, the wine-amphora industry did not fully revive until the middle of the first cen­ tury B.C. (see the introduction to Type 12). In the Portus Cosanus, however, there had been a revival by the late second century B. c., and perhaps earlier. To judge from the wide distribution of Sestius stamps at sites in the western Mediterranean area, especially in France, the Sestius factory dominated the market until sometime toward the middle of the first century B.C. (Will 1979, nn. 6 and 7 and Text Fig. IX-I), when factories in Campania and south­ ern Latium gained the ascendancy with the manufac­ ture of Types 4b and 12a. Amphoras of Type 4a reflect mass-production techniques in that they are remarkably uniform in dimensions. They are very tall, generally over a me­ ter in height (Fig. IX-4). They are slender (average greatest diameter of belly, 0.28 m-0.30 m) with very long narrow necks, which seem to balance the longer bellies and also may have solved mechanical problems of pouring or shipping about which we can only guess. Both the cork and the pozzolana stopper in Type 4a were regularly set well down into the neck, the length of which therefore added little to the capacity of the jar. The necks narrow toward the bottom. The long neck is flanked by equally long, close-set handles in some cases set so close that a finger can hardly be inserted into the opening between neck and handle. The handles are wide and are thin in section (0.06 m χ 0.03 m on the average). Rims of Type 4a slope or flare out, consti­ tuting one of the type's most characteristic features. Mouth diameters average 0.14 m-0.16 m. When stamps occur (and it is difficult to estimate what per­ centage of Roman amphoras were stamped; perhaps about 10 percent of most types), they are usually found on the rim. Sestius stamps are the stamps most often found on Type 4a, all impressed on the rim. The slender bellies of the type descend from sloping shoulders, which are usually not sharply de­ fined by the right-angled joint at the outer edge that characterizes the shoulder-belly joint of Type 4b. The toes of Type 4a are rather squat and short, but solid, with an average diameter of 0.05 m-0.06 m. The clay is chiefly the "Sestius" clay (Will 1979, p. 345), coarse pinkish buff (Munsell 5YR 6/6, the ex­ act color a matter of firing time and other factors), with many black, white, and red bits, and a lighter surface wash. (On this clay, see also de Boer, chapter sixteen, and Will 1982-1, nn. 11, 15, and 27.) Type 4b was also manufactured at Cosa, along

184

THE ROMAN AMPHORAS

with Type 4a, as early as the second quarter of the first century B.C. It is second only to Type 4a in frequency of occurrence in the port, as in the town, of Cosa. But the Sestius factory seems to have de­ clined in importance by the middle of the century. Sestius trademarks are still found on Type 4b (Will 1979, p. 348 and n. 29; and cf. the devices listed in the catalogue herewith), but infrequently. The ex­ amples of Type 4b at Cosa have, also, stylistic fea­ tures (such as slightly outflaring rims) that point to their being early. The center for production of Type 4 had very probably shifted from Etruria to southern Latium and Campania by the time Type 4b reached its floruit in the last half of the first century B.C., and the Sestius factory of Cosa at about the same time may well have diversified into bricks, tiles, and per­ haps Arretine and moved its headquarters to Rome. Amphora-manufacture in the Portus Cosanus may, however, have continued at a reduced rate through the third quarter of the century, at least until L. Ses­ tius' consulship in 23 B.C. (see introductory text above and the introduction to Type 24a below, and cf. Will 1979, nn. 35 and 37). Extensive evidence of kilns for Type 4b near Tarracina was reported in the nineteenth century (CIL X. 8050; cf. Hesnard 1977, passim; Peacock 1977, pp. 262-265; Panella 1980, pp. 254-255). Type 4b may also have been manufac­ tured near Brindisi (Will 1979, n. 41). But the earli­ est examples of the type were apparently produced in the port of Cosa. They correspond in date to the jars of Type 4b that were found by the hundreds in the Albenga wreck, now generally dated in the sec­ ond quarter of the first century B.C., and in the Spargi wreck, datable about 75 B.C., a cargo (possi­ bly from the port of Cosa ?) composed of amphoras of Type 4a as well as Type 4b (Will 1983-2). A sim­ ilar date can be given the 4b amphoras from Athens, Delos, and Alexandria, which will be discussed in my volume in the Athenian Agora series. Delos, it should be noted, had all but ceased to function as a trading center by the middle of the first century B.C. But how long did Type 4b last? It does not occur at Corinth, where various Augustan amphoras, such as Type 14 (Dressel 6), mark the beginning of the life of the Roman colony. Dressel, however, reports painted inscriptions of as late as 13 B.C. on his Form 1 (CIL XV. 4539, 4575). Finds of Type 4b also occur in Augustan contexts in France, Germany, and Eng­ land (cf. the summary in Callender 1965, pp. 8-9, and the detailed bibliography in Panella 1970, pp. 116-117, and Panella 1973, pp. 492-494). Further­ more, the discovery of forty-six pieces of Type 4b in Camulodunum (founded A.D. 10; on the finds, see

C.F.C. Hawkes and M. R. Hull, Camulodunum [Ox­ ford, 1947] p. 251 and pi. 69: Form 181) indicates that the type lasted in use into the late Augustan pe­ riod. On these fragments, Professor Hawkes has been kind enough to send me the following com­ ment: "The use may by then have been only second­ ary use, and the start of the occupation of Camulo­ dunum, formerly put c. A.D. 10, may anyhow be seen now as earlier by 10 years, or even 15. For the fort in Lower Germany at Haltern, sharing types of many kinds with the British site, and of leading value to the authors for dating its start, has now a date of evacuation not at A. D. 16 (Germanicus), but firmly fixed at 9 (disaster of Varus)." On that basis, I hesitate to advance the terminus ante quern for Type 4b beyond the last quarter of the first century B.C. We are, in any case, left to conclude that export of Type 4 to the east ceased during the last half of the first century B.C., while at the same time western demand for the Falernian, Caecuban, and other wines that Dressel's painted inscriptions prove were shipped in Type 4b remained strong throughout the century. Amphoras of Type 4b have heights and greatest diameters that are similar to those of Type 4a, and the clay of 4b amphoras at Cosa is Sestius clay, but 4b differs from 4a in important respects. The necks of 4b are wider, and the handles are not set close to the necks nor are they so wide and flat as those of 4a. Their average width and thickness is 0.05 m χ 0.04 m. They are sturdier, more rounded in section, thicker, and the "curve" of the handle, where it drops vertically to the shoulder, is sometimes almost right-angled. Such handles are not difficult to distin­ guish from those of Type 4a, but it is in the rim that the clearest distinction between the two varieties lies. The rims of 4b tend to be vertical in profile, or slightly concave, though at the port of Cosa there is a slight flare in many examples, which we take to be an early characteristic. Rim and mouth diameters, however, are similar to those of 4a. The shoulders of 4b are flatter than those of 4a, with a sharp angle, as was mentioned above, separating shoulder from belly. Bellies have straighter, less sloping sides. Toes are longer, sometimes very long, thick, and massive, the diameters averaging 0.07 m-0.08 m. Very few stamps occur on the pieces of 4b in the port, and they are mostly on the rim; but from occurrences elsewhere we can state that 4b stamps are found in a variety of locations on the jar, on rim, neck, handle, belly, and toe, but shoulder stamps, especially short, businesslike ones consisting of two letters, are fa­ vored. With 4b, the muted, rather graceful lines of

IX: THE ROMAN AMPHORAS

4a, and its evocative Sestius stamps, have been exchanged for crisper, sharper, thicker lines and for stamps that as a rule seem more like anonymous factory codes than marks made by actual potters. The finds of Types 4a and 4b from the port number 300 and 208, respectively, and there is also a group of 51 indeterminate body fragments that could belong to either group. The 138 pieces of 4a and 59 pieces of 4b described herewith represent about a third (35 percent) of the total number of examples of Type 4 found in the port. AU but one of these catalogued 4a pieces are fragments of rim or upper neck. Uncatalogued pieces of 4a number 162 and include 58 rims, 53 handles, 43 toes, 6 necks, and 2 body fragments. The 59 catalogued pieces of 4b include larger pieces: 3 jars that are almost complete, 12 bellies, 2 lower bellies, 36 rims, 3 toes, and 3 shoulders. Uncatalogued examples of 4b number 149 and include 19 rims, 85 handles, 40 toes, 4 necks, and 1 jar fragment. The 51 indeterminate body fragments are all uncatalogued. FORM a

A50. (PC 69-27). (Figs. IX-52, 53, 54). Stamped: APOL. Lagoon, Spring House, fill around basin platform. 8. VII.69. Rim fragment, stamped, and piece of neck. Mended. P.H., ca. 0.103 m (excluding attached piece of mortar); R.H., 0.045 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; beige surface. Traces of possible graffiti on neck and rim. Rim profile rather rounded. Three examples of the same stamp (impressed with the same matrix) have been found at Cosa (CB 1451, CC 935, C 67.218, the last with a possible second stamp resembling PC 68-59, Cat. A201 below), all of them on fragments belonging to Type 4a. The same abbreviation also occurs on other types of amphoras and on Arretine, but the examples at Cosa and the Portus Cosanus are the only known occurrences on Type 4. A51. (PC 69-20). (Figs. IX-55, 56, 57). Stamped: HERMO[. Lagoon, Spring House, surface. 8. VII.69. Rim fragment, stamped, and small piece of neck. P. H., 0.053 m; R. H., 0.036 m; est. M.D., 0.14 m; est. R.D., 0.16 m. Buff clay (5YR 7/3); small black bits and at least one large red bit; cream surface. Piece of mortar attached. For the stamp, cf. CIL X.8050.11,12 (five examples, three of them with the ligature, from near Terracina. The type of amphora in question is not clear, but it was perhaps Type 4a or Type 4b). A52. (PC 68-7). (Figs. IX-58, 59, 60). Stamped

185

M N I MN (N's retr.). Lagoon, Trench IH, east of Wall P, surface fill. 1.VII.68. Rim fragment, stamped (the stamp probably struck twice), and piece of neck. P.H., 0.084 m; R.H., 0.05 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Pinkish yellowbuff clay (5YR 6/8); black and red bits; yellowbuff surface. Possible graffito on rim. A53. (PC 68-40). (Figs. IX-61, 62, 63). Stamped: SETS device (anchor). Harbor, area of Villa, surface. 18.VH.68. Rim fragment, stamped, and small piece of neck. P. H., 0.07 m; R. H., 0.043 m; est. M.D., 0.14 m; est. R.D., 0.19 m. Tannish buff clay (7.5YR 7/4); large black, and some red, bits; yellow-cream surface. For a general discussion of the Sestius stamps from the Portus Cosanus and elsewhere, see the introduction to this chapter. Five Sestius anchor stamps from the port are published herewith (Cats. A53-A57), and five others are reported by Manacorda (nos. 1-5). None is known from the town of Cosa, though the anchor device alone, without letters, occurs there (cf. under Cat. A97 below). A Sestius anchor stamp is also reported from Saturnia by Uenze (1958, pp. 8, 27, and pi. 4:8; said to be in the Florence Archaeological Museum, but not found there by me in 1975. For unspecified reasons, Benoit 1961, p. 64, n. 1, followed by Roman and Rancoule 1977, p. 259, and Manacorda, pp. 127-128, questions Saturnia as the source of this stamp and of a Sestius trident stamp also reported by Uenze). Far the largest known concentration of Sestius anchor stamps has been found in the upper Grand Congloue wreck, where a large number of examples, amounting apparently to several hundreds, was found (Benoit 1961, fig. 59:2, and passim in his several other published accounts of the wrecks). The stamp has been found also at the He du Levant (Will 1956, fig. 78:3; Benoit 1961, fig. 59:15) and at Besanqon (CIL XIII. 10002.470c; Thevenot 1954, fig. 105:3; Will 1956, figs. 78:1, 79:1; Benoit 1961, fig. 59:3). The related stamp, SEST anchor, in which the T is attached to the second S rather than to the E, is not found at the port but occurs at Besangon (CIL XIII. 10002.470b; Thevenot 1954, fig. 105:4; Benoit 1961, fig. 59:10), Mont Beuvray (Thevenot 1954, fig. 105:5; Benoit 1961, fig. 59:9), and La Lagaste (Roman and Rancoule 1977, p. 258, no. 5; p. 260; and fig. 1:5, a device which may, however, be a palm branch). A badly worn stamp at the Athenian Agora (SS 7173) may also belong to the group of Sestius anchor stamps. In addition,

186

THE ROMAN AMPHORAS

the anchor device alone, without letters, when it is stamped on amphoras of Type 4a, probably comes from the Sestius factory. See under Cats. A95-A99 below. A54. (PC 68-42). (Figs. IX-64, 65, 66). Stamped: SETS device (anchor). Harbor, area of Villa, surface. 18.VII.68. Rim fragment, stamped, and piece of neck. P.H., 0.075 m; R.H., 0.051 m; est. M.D., 0.15 m; est. R.D., ca. 0.21 m. Pinkish tannish buff clay (7.5YR 6/6); black and red bits; whitish surface. See under Cat. A53 above. A55. (PC 69-178). Stamped: SETS device (anchor). Harbor area, 1969. Private collection. Studied from drawing. See under Cat. A53 above. A56. (PC 72-809). (Figs. IX-67, 68, 69). Stamped: SETS device (anchor). Lagoon, "Sestius triangle"; found by John Oleson, VII. 72. Rim fragment, stamped, and piece of neck. P. H., 0.09 m; R. H., 0.053 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pinkish yellow-buff clay (5YR 6/8); many tiny black, and scattered larger red, bits; buff surface. See under Cat. A53 above. I am very grateful to Oleson for informing me of his Sestius surface finds in the port, for sending me descriptions and drawings of them, and for later lending me the pieces for study. On Cat. A56, see also Will 1979, fig. 3:a. A57. (PC 76-2). (Figs IX-70, 71, 72). Stamped: SETS device (anchor). Lagoon, "Sestius triangle"; found by John Oleson, VI. 76. Rim fragment, stamped, and very small piece of neck. P. H., 0.055 m; R.H., 0.05 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pale pinkish buff clay (5YR 7/4); many small black and red bits; buff surface. See remarks under Cats. A53 and A56 above. A58. (PC 68-2). (Figs. IX-73, 74, 75). Stamped: SETS device (trident). Harbor, beach east of Pier 1, surface. 24. VI. 68. Small rim fragment, stamped, and bit of upper neck. P. H., 0.054 m; R.H., 0.041 m; est. M.D., 0.14 m; est. R.D., 0.19 m. Rose-buff clay (10R 6/4); black and red bits; beige surface. Nine Sestius trident stamps are published herewith (Cats. A58-A66), and two others are reported by Manacorda 1978 (nos. 16-17). One example has been found in the town of Cosa (C 67.287). Another stamp was found in the Ager Cosanus in 1975 by the Wesleyan University Archaeological Survey (site 82). An example is also reported from Saturnia by Uenze (1958, pp. 8, 27; pi. 4:9; cf. Benoit 1961, fig. 56 bis, where the number of the stamp in the Florence Archaeological Museum is given as 91941. See my remarks under Cat. A53 above). As is true of the Sestius

anchor stamps, the largest number of examples of Sestius trident stamps comes from the upper Grand Congloue wreck, where many examples were found (Benoit 1961, fig. 59:1-3), though apparently the trident was found less frequently than the anchor; in fact, the Sestius trident stamps are found only on the upper Grand Congloue and in the region that includes Cosa and Saturnia. Trident stamps alone, without letters, are also probably from the Sestius factory. They occur in the town of Cosa (C 65.94, C 65.192, C 65.390) and at Mont Beuvray (CIL XIII. 10002.539), but no examples are known as yet from the Portus Cosanus. A59. (PC 68-3). (Figs. IX-76, 77, 78). Stamped: SETS device (trident). Harbor, beach east of Pier 1, surface. 24. VI.68. Rim fragment, stamped, and bit of upper neck. R H . , 0.051 m; R.H., 0.041 m; est. M.D., 0.13 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black and red bits; beige surface. See under Cat. A58 above. A60. (PC 68-34). (Figs. IX-79, 80, 81; Color Fig. 3). Stamped: SETS device (trident). Harbor, area of Villa, surface. 18. VII. 68. Rim fragment, stamped, and bit of upper neck. P. H., 0.057 m; R.H., 0.055 m; est. M.D., 0.15 m; est. R.D., ca. 0.21 m. Pinkish buff clay (5YR 6/6); black, red, and white bits; beige surface. See under Cat. A58 above. A61. (PC 69-171). Stamped: SEJS device (trident). Harbor area, 1969. Private collection. Studied from drawing. See under Cat. A58 above. A62. (PC 69-181). Stamped: SETS device (trident). Harbor area, 1969. Private collection. Studied from drawing. See under Cat. A58 above. A63. (PC 69-184). (Figs. IX-82, 83, 84). Stamped: SET]S device (trident). Harbor area, 1969. Private collection. Rim fragment, stamped, and bit of neck. R H . , 0.059 m; R.H., 0.045 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Pinkish yellow-buff clay (5YR 6/8); many small black, and a few larger red, bits; tannish surface. Marine deposit on surface. See under Cat. A58 above. A64. (PC 72-808). (Figs. IX-85, 86, 87). Stamped: SETS device (trident). Lagoon, "Sestius triangle"; found by John Oleson, VII. 72. Rim fragment, stamped (faint traces of a second impression at a slant over the first stamp), and small piece of neck. R H . , 0.083 m; R.H., 0.041 m; est. M.D., 0.13 m; est. R.D., 0.18 m. Rose-buff clay (10R 6/ 4); frequent small black, and larger red, bits; tan surface. See under Cats. A56 and A58 above. A65. (PC 76-3). (Figs. IX-88, 89, 90). Stamped:

IX: THE ROMAN AMPHORAS

SETS device (trident). Lagoon, "Sestius triangle"; found by J. D. Lewis, 18.V.76. Rim fragment, stamped, and small piece of neck. P.H., 0.081 m; R.H., 0.048 m; est. M.D., 0.15 m; est. R.D., 0.19 m. Deep tannish buff clay (7.5YR 6/4); many very tiny black bits, and scattered large and small red inclusions; beige surface. See under Cat. A58 above. A66. (PC 76-6). (Figs. IX-91, 92, 93). Stamped: SETS device (trident). Lagoon, "Sestius triangle"; found by J. D. Lewis, 18.V.76. Rim fragment, stamped, and small piece of neck. P.H., 0.071 m; R.H., 0.043 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Deep tannish buff clay (7.5YR 6/4); frequent small black, and some larger red, bits; cream surface. On Cat. A66, see also Will 1979, fig. 3:b. And see under Cat. A58 above. A67. (PC 68-37). (Figs. IX-94, 95, 96). Stamped: SETS device (caduceus). Harbor, area of Villa, surface. 18.VII.68. Rim fragment, stamped, and piece of neck. R H . , 0.09 m; R.H., 0.049 m; est. M.D., 0.14 m; est. R.D., 0.19 m. Pinkish buff clay (5YR 6/6); black, red, and white bits; yellowcream surface. Seven Sestius caduceus stamps from the port are published herewith (Cats. A67A73), and four others are reported by Manacorda (nos. 6-9). Three examples have been found in the town of Cosa (CB 1582, C 65.368, C 67.49; CB 1582 was originally published in Will 1956, fig. 80 bottom). Though the stamp does not occur on the upper Grand Congloue wreck, four examples have been found in France, at Nimes (Benoit 1961, fig. 59:4), Roanne, Le Carla a Bouriege, and La Lagaste (the last three discussed recently by Roman and Rancoule 1977, pp. 257, 259). The caduceus device alone, without letters, seems to occur only at the Portus Cosanus. See below, Cat. A102. A68. (PC 69-179). (Figs. IX-97, 98, 99). Stamped: SETS device (caduceus). Harbor area, 1969. Private collection. Rim fragment, stamped (the stamp inverted and struck twice), and piece of neck. R H . , 0.097 m; R.H., 0.048 m; est. M.D., 0.14 m; est. R.D., 0.17 m. Deep tannish buff clay (7.5YR 6/4); frequent tiny dark bits; beige surface. See under Cat. A67 above; and on Cat. A68, cf. also Will 1979, fig. 3:f. A69. (PC 69-180). Stamped: SEJ]S device (caduceus). Harbor area, 1969. Private collection. Studied from drawing. See under Cat. A67 above. A70. (PC 69-185). (Figs IX-100, 101, 102). Stamped: SETS device (caduceus). Harbor area, 1969. Private collection. Rim fragment, stamped, and

187

very small piece of neck. P. H., 0.06 m; R. H., 0.051 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Pinkish buff clay (5YR 6/6); many tiny black, and scattered larger red, bits; cream surface. See under Cat. A67 above. Deep line incised around rim, which is "flounced" below the line. A71. (PC 72-807). (Figs. IX-103, 104, 105). Stamped: SETS device (caduceus). Lagoon, "Sestius triangle"; found by John Oleson, VII. 72. Rim fragment, stamped, and small piece of neck. P. H., 0.07 m; R.H. as preserved, 0.048 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Pinkish rose clay (10R 6/6), lighter toward surface; many tiny black bits and conspicuous larger red ones; beige surface. See under Cats. A56 and A67 above. A72. (PC 76-4). (Figs. IX-106, 107, 108). Stamped: SF/TS device (caduceus). Lagoon, "Sestius triangle"; found by J. D. Lewis, 18. V.76. Rim fragment, stamped, and very tiny piece of neck. P. H., 0.051 m; R.H., 0.049 m; est. M.D., 0.18 m; est. R. D., 0.21 m. Deep tannish buff clay (7.5YR 6/ 4); frequent tiny black, and scattered larger red, bits; cream surface. See under Cat. A67 above. A73. (PC 76-5). (Figs. IX-109, 110, 111). Stamped: SETS device (caduceus). Lagoon, "Sestius triangle"; found by J. D. Lewis, 18. V.76. Rim fragment, stamped, and very tiny piece of neck. P. H., 0.059 m; R.H., 0.053 m; est. M.D., 0.15 m; est. R.D., 0.19 m. Pinkish buff clay (5YR 6/6), a tannish layer visible toward the surface; many tiny black, and scattered larger red bits; beige surface. See under Cat. A67 above. A74. (PC 68-39). (Figs. IX-112, 113, 114). Stamped: SETS device (tree or vine-prop). Harbor, area of Villa, surface. 18. VII.68. Rim fragment, stamped, and piece of neck. P. H., 0.14 m; R.H., 0.046 m; est. M.D., 0.13 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black, red, and white bits; yellow-cream surface. Rim profile rather rounded. Two Sestius tree or vine-prop stamps from the port are published herewith (Cats. A74 and A75). They are the only known examples of the stamp, the device on which does not seem to occur alone, without letters. (But see Luc Long, in chapter eight, Appendix 1 above, who describes the finding of a similar stamp on the upper Grand Congloue wreck.) A75. (PC 69-173). Stamped: SEJS device (tree or vine-prop). Harbor area, 1969. Private collection. Studied from drawing. See under Cat. A74 above. A76. (PC 69-182). (Figs. IX-115, 116, 117). Stamped: S]ETS device (small eight-pointed star). Harbor area, 1969. Private collection. Rim fragment,

188

THE ROMAN AMPHORAS

stamped, and small piece of neck. P. H., 0.089 m; R.H., 0.048 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Deep tannish buff clay (7.5YR 6/4); frequent small black, and some red, bits; tannish surface. Three Sestius stamps showing the small eightpointed star are published herewith (Cats. A76A78), and three others are reported by Manacorda (nos. 10-12). All are from the port. The stamp is also known from Alise-Sainte-Reine (Thevenot 1954, fig. 105:6, where, however, the profile given is unlike that in Uenze 1958, pi. 7:7; Benoit 1961, fig. 59:13) and Vieille Toulouse (Roman and Rancoule 1977, p. 261, who note that the AliseSainte-Reine and Vieille Toulouse stamps have different dimensions). It is important to observe that in the Sestius stamps showing the small eightpointed star device, the " T " extends to the left the top cross-bar of the " E " in the ligature and also seems to extend vertically the downward stroke of the " E , " whereas in all the other SETS stamps (the anchor, trident, caduceus, and tree or vineprop examples described above) the " T " extends to the left the bottom cross-bar of the "E." No examples of the small eight-pointed star device alone, without the letters, are known, but a large eight-pointed star is reported from the port by Manacorda (no. 25) and is known from the town of Cosa (C 65.280, C 65.381, C 65.474, C 68.310). The letters SES with five-pointed star occur on three stamps from the town of Cosa (Will 1956, fig. 80; Will 1979, fig. 3:c) and are frequent at other sites but do not occur at the port. See, however, under Cat. A89 below. A77. (PC 75-1). (Figs. IX-118, 119, 120). Stamped: S]ETS device (small eight-pointed star). Lagoon, "Sestius triangle"; found by Lawrence Davis, 22. VI.75. Rim fragment, stamped, and small piece of neck. P.H., 0.06 m; R.H., 0.054 m; est. M.D., 0.17 m; est. R. D., 0.20 m. Pinkish buff clay (5YR 6/6), lighter toward surface; many small black and white, and larger red, bits; cream surface. See under Cat. A76 above. On Cat. A77, see also Will 1979, fig. 3:d. A78. (PC 78-1). (Fig. IX-121). Stamped: SEJS[ device (small eight-pointed star). Sporadic find, 1978. Rim fragment, stamped, and piece of neck. P.H., 0.105 m; R.H., 0.044 m; est. M.D., 0.14 m; est. R.D., 0.19 m. Pinkish yellow-buff clay (5YR 6/8); many dark bits and scattered red and white bits; beige surface. See under Cat. A76 above. A79. (PC 69-6). (Figs. IX-122, 123, 124). Stamped: SES and device (lighthouse or lighted altar [?]). Harbor, underwater, near Wall D, surface. 3. VII. 69. Rim fragment, stamped twice, and piece

of neck. The stamps are contiguous, one above the other, the device being the lower stamp. P.H., 0.118 m; R.H., 0.041 m; est. M.D., 0.16 m; est. R.D., 0.20 m. Deep pinkish buff clay (2.5YR 6/ 6); black and red bits; grayish buff surface. Trace of possible graffito on rim. The lighthouse (?) occurs only here with letters but is stamped twice, without the letters, on a fragment from the town of Cosa (C 67.236). On Cat. A79, see also Will 1979, pp. 343-344, fig. 4 and n. 14 and McCann, chapter seven above, where relationship of this stamp to Cosa lighthouse is suggested. A80. (PC 68-35). (Figs. IX-125, 126, 127). Stamped: SEST device (palm branch). Harbor, area of Villa, surface. 18. VII. 68. Rim fragment, stamped, and small piece of neck. P. H., 0.058 m; R. H., ca. 0.049 m; est. M.D., ca. 0.15 m; est. R.D., ca. 0.20 m. Tannish buff clay (7.5YR 7/4); black and red bits; yellow-buff surface. Three Sestius stamps with palm-branch device are published herewith (Cats. A80-A82), and a fourth example from the port is listed below under Type 5 (Cat. A248. It is the only Sestius stamp from the Cosa area that does not occur on a jar of Type 4a or 4b, though a parallel example on Type 5 has been found at Vada Sabatia in Liguria). Four other examples from the port on Type 4a are reported by Manacorda (nos. 13-15). Four Sestius stamps with palm-branch device have also been discovered in the town of Cosa (CB 779, CB 1374, C 65.316, C 65.347; CB 1374 was originally published in Will 1956, fig. 80). The stamp is also known in many examples elsewhere: at Ventimiglia (Lamboglia 1955, fig. 14; Benoit 1961, fig. 59:11), and in France at Vienne, Vieille Toulouse, Toulouse, La Lagaste, Peyrepertuse, Pamiers, Ruscino, and Narbonne (a recent discussion of many of those occurrences is in Roman and Rancoule 1977, pp. 258, 260-261; however, the exact number of finds at La Lagaste is unclear, since the illustration in fig. 1 suggests that three Sestius palm-branch stamps were found there. On the other hand, the catalogue lists only one [no. 4] as a palm branch. The Toulouse stamp is illustrated in M. Vidal, "Necropole Toulousaine de Saint-Roch: Ie puits funeraire no. 27," Revue archeologique de Narbonnaise 6 [1973] fig. 4:24. The Narbonne stamp is listed in C. Lamour and F. Mayet, "Glanes amphoriques: I. Region de Beziers et Narbonne," Etudes sur Pezenas et I'Herault 11 [1980] no. 10). What is perhaps a large palm-branch device, without letters, has also been found in the Portus Cosanus (see below, Cat. A 103). A81. (PC 69-176). Stamped: SESJ device (palm

IX: THE ROMAN AMPHORAS

branch). Harbor area, 1969. Private collection. Studied from drawing. See under Cat. A80 above. A82. (PC 76-1). (Figs. IX-128, 129, 130). Stamped: SEST device (palm branch). Lagoon, "Sestius triangle"; found by John Oleson, VI.76. Rim fragment, stamped, and very small piece of neck. P.H., 0.055 m; R.H., 0.048 m; est. M.D., 0.17 m; est. R.D., ca. 0.20 m. Pinkish buff clay (5YR 6/ 6), tannish toward core; many small black bits, and some small white and large red inclusions; cream surface. See above, Cat. A80. For Cat. A82, cf. Will 1979, fig. 3:g. A83. (PC 68-33). (Figs. IX-131, 132, 133). Stamped: SEST device (double axe). Harbor, area of Villa, surface. 18.VII.68. Rim fragment, stamped, and piece of upper neck. P. H., 0.08 m; R. H., 0.05 m; est. M.D., 0.14 m; est. R.D., 0.19 m. Brownish clay (7.5YR 5/4); black, red, and a few white bits; tannish buff surface. Six Sestius double-axe stamps from the port are published herewith (Cats. A83-A88), and two others are reported by Manacorda (nos. 18-19). Five additional examples have been found in the town of Cosa (CF 1988, C 67.78, C 67.104, C 67.189, C 69.214; the double axes on two of those stamps, C 67.104 and C 67.189, appear larger than the devices on the other examples, perhaps as a result of smearing). The stamp has been found also at Volterra (Manacorda reports an unpublished example there, p. 127, n. 18), Lezoux (CIL XIII. 10002. 470a; Manacorda questions the existence of this stamp but does not specify his grounds for doing so), Vieille Toulouse (four examples, a recent summary of which is given by Roman and Rancoule 1977, p. 261), La Lagaste (three occurrences, described in Roman and Rancoule 1977, fig. 1 and pp. 257, 260), Narbonne (Lamour and Mayet, "Glanes amphoriques" [cited in Cat. A80], no. 9 and pi. 1:9), and Basel (this stamp often illustrated and commented upon. As I recall, M. Thevenot and I agreed some years ago that the "double axe" in this and in the other examples we knew at the time was a ligature of "TI" and not a device at all. Cf. Callender under no. 1599. My feeling now is that the double axe is used as a device but with the ligature in mind. The same may well be true of the anchor device, which resembles a ligature of "TIV," and of the sigma device illustrated in Will 1979, fig. 3:e). A84. (PC 68-41). (Figs. IX-134, 135, 136). Stamped: SEST device (double axe). Harbor, area of Villaj surface. 18.VII.68. Rim fragment, stamped, and small piece of neck. P. H., 0.07 m; R. H., 0.045 m; est. M.D., 0.16 m; est. R.D., 0.21 m.

189

Pinkish buff clay (5YR 6/6); black, red, and a few white bits; yellow-cream surface. "Double chin" between rim and neck. See under Cat. A83 above. A85. (PC 69-169). Stamped: SESJ device (double axe). Harbor area, 1969. Private collection. Studied from drawing. See under Cat. A83 above. A86. (PC 69-183). (Figs. IX-137, 138, 139). Stamped: SEST device (double axe). Harbor area, 1969. Private collection. Rim fragment, stamped, and small piece of neck. P. H., 0.09 m; R. H., ca. 0.05 m; est. M.D., ca. 0.16 m; est. R.D., ca. 0.20 m. Pinkish buff clay (5YR 6/6); many tiny dark bits and frequent red inclusions; cream surface. See under Cat. A83 above. A87. (PC 72-806). (Figs. IX-HO, 141,142). Stamped: SEST device (double axe). Lagoon, "Sestius triangle"; found by John Oleson, VII.72. Rim fragment, stamped, piece of neck, and part of upper attachment of one handle. P.H., 0.108 m; R.H., 0.045 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Tannish buff clay (7.5YR 7/4), lighter toward core; many tiny black and white, and frequent red, bits; beige surface. See under Cats. A56 and A83 above. A88. (PC 73-1). (Figs. IX-143, 144, 145). Stamped: SEST device (double axe). Lagoon, "Sestius triangle"; found summer 1973. Rim fragment, stamped. R H . , 0.052 m; R.H., 0.05 m; est. M.D., 0.12 m; est. R.D., 0.16 m. Unusually sandy pinkish buff clay (5YR 6/6); a few rather large red bits visible; whitish deposit on surface. Both clay and dimensions are unusual for the Sestius series. See under Cat. A83 above. A89. (PC 68-38). (Figs. IX-146, 147, 148). Stamped: SES uncertain device. Harbor, area of Villa, surface. 18. VII. 68. Rim fragment, stamped, and piece of neck. R H . , 0.101 m; R.H., 0.048 m; est. M.D., ca. 0.15 m; est. R.D., ca. 0.20 m. Pinkish yellow-buff clay (5YR 6/8); black and red bits. Cat. A89 is one of six Port Sestius stamps (Cats. A89-94) the devices of which cannot be ascertained owing to breakage or other damage. In the case of Cat. A89, the size and shape of the letters suggest that the stamp should be restored to read "SES device (five-pointed star)." That particular stamp is not otherwise represented at the port. See under Cat. A76 above. Five additional Sestius stamps with uncertain devices are listed as from the Portus Cosanus by Manacorda (nos. 20-24). A90. (PC 68-43). (Figs. IX-149, 150, 151). Stamped: SES[ uncertain device. Harbor, area of Villa, surface. 18.VII.68. Rim fragment, stamped, and small piece of neck. R H . , 0.085 m; R.H., 0.05 m; est. M.D., 0.13 m; est. R.D., ca. 0.17 m. Pinkish

190

THE ROMAN AMPHORAS

buff clay (5YR 6/6); black and red bits; yellowcream surface. A91. (PC 69-170). Stamped: S[ES uncertain device. Harbor area, 1969. Private collection. Studied from drawing. A92. (PC 69-172). (Figs. IX-152, 153, 154). Stamped: S[ES uncertain device. Harbor area, 1969. Private collection. Rim fragment, stamped, and piece of neck. P.H., 0.116 m; R.H., 0.042 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Sandy pinkish buff clay (5YR 6/6); many tiny black and white, and larger red, bits; worn buff surface. A93. (PC 69-174). Stamped: SES uncertain device. Harbor area, 1969. Private collection. Studied from drawing. A94. (PC 69-175). Stamped: SES uncertain device. Harbor area, 1969. Private collection. Studied from drawing. A95. (PC 72-245). (Figs. IX-155, 156, 157). Stamped: device (small anchor). Lagoon, Spring House, fill in Room 2. 27. VII. 72. Rim fragment, stamped (the stamp struck three times), and piece of neck. P.H., 0.131 m; R.H., 0.039 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Deep pinkish buff clay (2.5YR 6/6); black, white, and red bits; yellow-buff surface. Deep vertical line incised on rim. For related stamps, cf. Cats. A53-A57 above and Cats. A96A99 below. The same stamp was found in the Ager Cosanus in 1975 by the Wesleyan University Archaelogical Survey (site 98). The small anchor on Type 4a also seems to be known from Ventimiglia (Lamboglia 1955, fig. 16, where the scale of the drawing indicates that the small, not the large, anchor is illustrated). A96. (PC 69-191). (Figs. IX-158, 159, 160). Stamped: device (large anchor). Harbor area, 1969. Private collection. Rim fragment, stamped, and piece of neck. P.H., 0.082 m; R.H., 0.046 m; est. M.D., 0.135 m; est. R.D., 0.165 m. Pinkish buff clay (5YR 6/6); many tiny black, red, and white bits; beige surface. Three large anchor stamps are published herewith (Cats. A96-A99), and one example has been found in the town of Cosa (C 70.420). For related stamps, cf. Cats. A53-A57 and A95 above, and perhaps Cats. Al00-A101 below. A97. (PC 72-129). (Figs. IX-161, 162, 163). Stamped: device (large anchor). Lagoon, Spring House, rubble level north of Wall b. 21. VII. 72. Rim fragment, stamped, and bit of neck. P. H., 0.065 m; R.H., 0.044 m; est. M.D., 0.17 m; est. R.D., 0.19 m. Pinkish buff clay (5YR 6/6); black and white bits; yellow-buff surface.

A98. (PC 72-810). (Figs. IX-164, 165, 166). Stamped: device (large anchor). Lagoon, "Sestius triangle"; found by John Oleson, VII.72. Rim fragment, stamped, and very small piece of neck. R H . , 0.052 m; R.H., 0.041 m; est. M.D., 0.165 m; est. R.D., 0.195 m. Deep pinkish buff clay (2.5YR 6/6); many tiny black and white, and scattered red, bits; beige surface. See under Cats. A56 and A96 above. Although the stamp is smeared, a comparison of rubbings makes identification of the device as the large anchor very likely. A99. (PC 72-uncatalogued). (Fig. IX-167). Stamped: device (large anchor). Immured in surface of Wall L and inaccessible for study. Lagoon, Trench IA. 8. VIII. 72. Studied from photographs. See under Cat. A96 above. AlOO. (PC 72-811). (Figs. IX-168, 169, 170). Stamped: device (double anchor or thunderbolt). Lagoon, "Sestius triangle"; found by John Oleson, VII.72. Rim fragment, stamped, and very small piece of neck. R H . , 0.056 m; R.H., 0.045 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Deep pinkish buff clay (2.5YR 6/6); many tiny black, and scattered red, bits; beige surface. Two doubleanchor or thunderbolt stamps are published herewith (Cats. A100-A101). They are the only known examples of the device on jars of Type 4a. See under Cat. A56 above. AlOl. (PC 72-uncatalogued) (Fig. IX-171). Stamped: device (double anchor or thunderbolt). Immured in surface of Wall L and inaccessible for study. Lagoon, Trench IA. 8.VIII.72. Studied from photographs. See under Cat. AlOO above. A102. (PC 72-812). (Figs. IX-172, 173, 174). Stamped: device (caduceus?). Lagoon, "Sestius triangle"; found by John Oleson, VII.72. Rim fragment, stamped, and piece of upper neck, including part of one handle attachment. P. H., 0.095 m; R.H., 0.046 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pale pinkish brown clay; tiny black, white, and red bits; gray-buff surface. For related stamps, cf. Cats. A67-A73 above. See also under Cat. A56 above. A103. (PC 68-52). (Figs. IX-175, 176, 177). Stamped: device (large palm branch or fish bone). Lagoon, Trench IH, east of Wall P. under cobblestone road. 3. VII. 68. Rim fragment, stamped, and piece of neck. R H . , 0.10 m; R.H., 0.49 m; est. M.D., 0.17 m; est. R.D., 0.19 m. Coarse tannish clay (7.5YR 7/4), pale rust at core; black and red bits; beige surface. For possibly related stamps, see Cats. A80-A82 above. A104. (PC 68-6). (Figs. IX-178, 179, 180). Stamped:

IX: THE ROMAN AMPHORAS

device (pine bough or fish spine). Lagoon, Spring House, earth fill above basin platform. 5.VII.68. Rim fragment, stamped, and piece of upper neck. P. H., 0.074 m; R. H., 0.039 m; est. M.D., 0.18 m; est. R.D., 0.22 m. Pinkish yellowbuff clay (5YR 6/8); black and red bits and a few white ones. Four examples of the same stamp are published herewith (Cats. Al04-Al07), and the stamp has also been found in the town of Cosa (CF 311). A105. (PC 69-95). (Figs. IX-181, 182, 183). Stamped: device (pine bough or fish spine). Lagoon, Spring House, basin fill, 1.48 m. below platform level. 1. VIII. 69. Rim fragment, stamped, and bit of neck. R H . , 0.053 m; R.H., 0.045 m; est. M.D., 0.18 m; est. R.D., 0.22 m. Pinkish buff clay (5YR 6/6); many white bits, as well as scattered black, and a few red, inclusions; beige surface. See under Cat. A104 above. A106. (PC 72-314). (Figs. IX-184, 185, 186). Stamped: device (pine bough or fish spine). Lagoon, Spring House, wood level in Room 2. 2.VIII.72. Rim fragment, stamped, and piece of neck, including part of one handle attachment. Broken and mended. R H . , 0.105 m; R.H., 0.044 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Coarse tannish buff clay (7.5YR 7/4); red, black and white bits. Pale blue deposit on one part of fragment. See under Cat. A104 above. A107. (PC 72-323). (Figs. IX-187, 188, 189). Stamped: device (pine bough or fish spine). Lagoon, Spring House, fill in Room 2. 4.VIII.72. Rim fragment, stamped, and small piece of neck. R H . , 0.064 m; R.H., 0.048 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Sandy pale rust clay (5YR 6/ 4); black, red, and white bits. See under Cat. A 104 above. A108. (PC 68-1). (Figs. IX-190, 191, 192). Stamped: device (maze [?] nautical hook [?] triangular harrow [?] or possibly a stylized version of the Q · AF stamps on Arretine (Oxe and Comfort, no. 28 [?]). Harbor, underwater, sporadic find. 18. VI. 68. Rim fragment, stamped, and piece of neck. R H . , 0.11 m; R.H., 0.052 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black, red, and white bits: beige surface. A109. (PC 69-152). (Figs. IX-193, 194). Stamped: device (compass card [?] wind rose [?] wheel [?]). Lagoon, Spring House, earth fill above basin platform. 21-26.VII.69. Neck fragment, preserving fragmentary rim, stamped above one handle, piece of upper neck, and stump of one handle. R H . , 0.132 m; R.H., 0.048 m; est. M.D., 0.16 m;

191

est. R.D., 0.19 m; H.W., 0.049 m; H.T., 0.035 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits. This may be simply a conventional, decorative wheel design, resembling one that occurs on Attic Geometric pottery and also on Arretine (Oxe and Comfort, nos. 2558-2561). On the other hand, the specificity and unambiguity of most of the Sestius devices perhaps justify an effort to interpret those few, such as this one, that seem less clear. The same device is reported by Beltran as occurring on the lower belly, near the toe, of a jar from Badalona, Spain, an amphora belonging to the class sometimes called "Layetanian" (Beltran 1970, fig. 100:3). A twoended terracotta stamping matrix from Sardis also bears on one end the reverse impression of Cat. A109, apparently in the same size (diameter of impression ca. 0.027 m). The other end of the same matrix bears a similar design, but with three dots instead of one in each quadrant. For description and photos of the Sardis matrix, which is numbered S.60.2 (2660), I thank Ruth S. Thomas, who was also kind enough to send a drawing of another, similar piece from Sardis with a design of dots in a rectilinear grid. Both pieces will be published in her forthcoming Terracottas and Small Finds from Sardis (1958-1975). It is difficult to imagine that the matrix resembling Cat. A109 belonged to the Sestii, but it is perhaps worth noting that Publius Sestius was a promagistrate in Asia Minor in 48-47 B.C. (Broughton 1951-1952, II, p. 278) and is last heard of in Aphrodisias in 35 B.C. [Keune in RE, s.v. "Sestius" (6)]. No Sestius stamps or devices, however, have as yet been reported from Asia Minor. On the basis of our present evidence, it seems likely that the mark on Cat. A109 is a Sestius design of uncertain significance and is unrelated to the Sardis matrix. Its relationship to the Badalona stamp remains unclear. For the possible connection between the Sestius amphora industry and the Arretine ware with Sestius stamps, see Will 1979, p. 349, n. 37. There is however, no reason to connect the Arretine design like Cat. A109 with the Sestius Arretine industry, the stamps of which are listed in Oxe and Comfort, nos. 1792-1819. I connect the deisgn here with the Sestius amphora industry, since most devices on Type 4a at Cosa probably have that connection. AIlO. (PC 72-215). (Figs. IX-195, 196). Stamped: illegible. Lagoon, Spring House, fill in Room 2. 26. VII. 72. Rim fragment, stamped, and tiny piece of neck. R H . , 0.04 m; R.H., 0.036 m; est. M.D.,

192

THE ROMAN AMPHORAS

0.14 m; est. R.D., 0.16 m. Tannish buff clay (7.5YR 7/4); black, white, and red bits; beige surface. A l I l . (PC 72-309). (Figs. IX-197, 198, 199). Stamped: illegible. Lagoon, Spring House, surface. 3. VlII.72. Rim fragment, stamped, and small piece of neck. P. H., 0.056 m; R. H., 0.037 m; est. M.D., 0.14 m; est. R.D., 0.17 m. Pinkish tannish buff clay (7.5YR 6/6); black, white, and red bits; beige surface. A112. (PC 68-51). (Fig. IX-200). Unstamped. Harbor, under rocks between Wall M and Pier 1. 22. VI.68. Rim fragment, piece of neck, and part of upper attachment of one handle. P.H., 0.116 m; R.H., 0.036 m; est. M.D., ca. 0.13 m; est. R.D., 0.15 m. Coarse, sandy tannish buff clay (7.5YR 7/ 4), pinkish buff (5YR 6/6) toward surface; black, white, and red bits; pinkish yellow-buff surface. Mortar adheres to inside of fragment. A113. (PC 69-32). (Fig. IX-201). Unstamped. Harbor, underwater, Trench D2, Level II. 16.VII.69. Rim fragment and piece of neck. P. H., 0.081 m; R.H., 0.035 m (as preserved); est. M.D., 0.14 m; est. R.D., 0.17 m. Grayish rose clay (10R 6/2); black, white, and red bits. Fragment badly damaged by water action. A114. (PC 69-56). (Fig. IX-202). Unstamped. Harbor, underwater, Trench C l , Level III. 18. VII. 69. Fragment of upper neck, preserving piece of rim and stump of one handle. P.H., 0.146 m; R.H., 0.042 m; est. M.D., 0.16 m; est. R.D., 0.20 m; H.W., 0.059 m; H.T., 0.031 m. Pinkish tannish buff clay (7.5YR 6/6); black and white bits. A115. (PC 69-57). (Fig. IX-203). Unstamped. Harbor, underwater, Trench C l , Level III. 23.VII.69. Rim fragment and piece of neck. P. H., 0.066 m; R.H., 0.044 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black, white, and large red bits; surface varies in color from cream to beige to grayish. Small mark on rim is probably not part of a stamp. Al 16. (PC 69-59). (Fig. IX-204). Unstamped. Harbor, underwater, Trench C l , Level III. 22.VII.69. Rim fragment and tiny piece of neck. P. H., 0.068 m; R.H., 0.048 m; est. M.D., 0.17 m; est. R.D., 0.20 m. Pale pinkish buff clay (5YR 7/4); black and especially red bits; beige surface. Al 17. (PC 69-73). (Fig. IX-205). Unstamped. Harbor, underwater, Trench C l , Level HI. 24.VII.69. Fragment of lower belly lacking most of toe. P.H., 0.41 m; G.D., 0.288 m. Coarse pinkish buff clay (5YR 6/6); yellow-buff surface. A118. (PC 69-92). (Fig. IX-206). Unstamped. La-

goon, Spring House, basin fill. 30.VII.69. Rim fragment, piece of neck, and stump of one handle. R H . , 0.102 m; R.H., 0.037 m; est. M.D., 0.14 m; est. R.D., 0.17 m; H.W., not determinable; H.T., 0.024 m. Rust clay (5YR 5/6); black and white bits. A119. (PC 69-98). (Fig. IX-207). Unstamped. Lagoon, Spring House, fill around basin platform. 1. VIII.69. Rim fragment and small piece of neck. R H . 0.077 m; R.H., 0.051 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits. A120. (PC 72-10). (Fig. IX-208). Unstamped. Lagoon, Trench IA, top of Wall P. 5. VII.72. Rim fragment and piece of neck. P.H., 0.074 m; R.H., 0.039 m; est. M.D., 0.17 m; est. R.D., 0.20 m. Coarse pale tannish buff clay (7.5YR 8/2); frequent tiny dark bits. A121. (PC 72-18). (Fig. IX-209). Unstamped. Lagoon, Spring House, surface. 5. VII.72. Rim fragment and small piece of neck. P. H., 0.067 m; R.H., 0.057 m; est. M.D., 0.17 m; est. R.D., 0.19 m. Coarse pale tannish buff clay (7.5YR 8/2); frequent tiny dark bits; cream surface. Piece of mortar attached. A122. (PC 72-26). (Fig. IX-210). Unstamped. Lagoon, Trench IA, 1.50 m below Wall L, between piers 7 and 8. 10. VII.72. Neck fragment, preserving very small piece of rim. P.H., 0.13 m; est. M.D., ca. 0.17 m; est. R.D., ca. 0.20 m. Very coarse, pale pinkish buff clay (5YR 7/4); frequent dark, light, and red bits; cream surface. A123. (PC 72-28). (Fig. IX-211). Unstamped. Lagoon, Spring House, fill around Wall e. 7. VII.72. Rim fragment and piece of neck. P.H., 0.078 m; R.H., 0.047 m; est. M.D., 0.13 m; est. R.D., 0.18 m. Coarse pale rust clay (5YR 6/4); many black and white bits. A124. (PC 72-33). (Fig. IX-212). Unstamped. Lagoon, Spring House, surface. 4. VII.72. Rim fragment and small piece of neck. P. H., 0.09 m; R. H., 0.045 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Coarse tannish buff clay (7.5YR 7/4); many tiny dark bits; yellow-cream surface. A125. (PC 72-65). (Fig. IX-213). Unstamped. Lagoon, Spring House, level south of Wall a. 14. VII.72. Rim fragment and small piece of neck. P.H., 0.086 m; R.H., 0.041 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Coarse rust clay (5YR 5/6); many tiny dark and light bits; grayish surface. A126. (PC 72-120). (Fig. IX-214). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 20. VII. 72. Rim fragment and tiny piece of neck.

IX: THE ROMAN AMPHORAS

P.H., 0.077 m; R.H., 0.044 m; est. M.D., 0.14 m; est. R.D., 0.17 m. Rust clay (5YR 5/6); black, white, and red bits; yellow-buff surface. A127. (PC 72-122). (Fig. IX-215). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 20. VII. 72. Rim fragment, piece of upper neck, and stump of one handle. P. H., 0.10 m; R. H., 0.039 m; est. M.D., 0.14 m; est. R.D., 0.16 m; H. W., 0.054 m; H. T., 0.023 m. Rust clay (5YR 5/6); black, red, and especially white bits; yellowbuff surface. Rim rather rounded in profile. A128. (PC 72-125). (Fig. IX-216). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 20. VII. 72. Rim fragment, piece of neck, and upper attachment of one handle. R H . , 0.104 m; R.H., 0.04 m; est. M.D., 0.15 m; est. R.D., 0.17 m. Deep pinkish buff clay (2.5YR 6/6); black and especially red and white bits; beige surface. Mortar adheres to inside of fragment. A129. (PC 72-140). (Fig. IX-217). Unstamped. Lagoon, Spring House, surface. 21. VII. 72. Rim fragment and tiny piece of neck. P.H., 0.061 m; R.H., 0.042 m; est. M.D., 0.13 m; est. R.D., 0.16 m. Yellow-buff clay (7.5YR 7/6); black, white, and many red bits. A130. (PC 72-141). (Fig. IX-218). Unstamped. Lagoon, Spring House, surface. 21. VII. 72. Very small rim fragment and bit of neck. P.H., 0.064 m; R.H., 0.036 m; est. M.D., 0.14 m; est. R.D., 0.16 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; tannish buff surface. A131. (PC 72-164). (Fig. IX-219). Unstamped. Lagoon, Spring House, basin at east end of south wall of cistern. 21.VII.72. Rim fragment and small piece of neck. P. H., 0.066 m; R. H., 0.036 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits. Slanting line incised on rim. A132. (PC 72-167). (Fig. IX-220). Unstamped. Lagoon, Spring House, east end of south wall of cistern. 21. VII. 72. Rim fragment, piece of upper neck, and stump of one handle. P.H., 0.093 m; R.H., 0.037 m; est. M.D., ca. 0.14 m; est. R.D., ca. 0.17 m; H. W., 0.066 m; H. T., 0.023 m. Coarse pale pinkish buff clay (5YR 7/6); black, white, and large reddish bits; yellow-buff surface. Line incised around rim near top. A133. (PC 72-174). (Figs. IX-221, 222). Unstamped. Lagoon, Spring House, fill against south wall of cistern. 20. VII. 72. Rim fragment and piece of neck. R H . , 0.089 m; R.H., 0.038 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; beige

193

surface. Raised line around top of neck under rim. A134. (PC 72-183). (Fig. IX-223). Unstamped. Lagoon, Trench YZ, south extension below preserved surface of Wall W in layer of dark mud, 0.55 m b.s.l. and below. 21. VII. 72. Rim fragment and small piece of neck. P. H., 0.074 m; R. H., 0.049 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Tannish buff clay (7.5YR 7/4); black and white bits. A135. (PC 72-184). (Fig. IX-224). Unstamped. Lagoon, Trench YZ, south extension below preserved surface of Wall W in layer of dark mud, 0.55 m b.s.l. and below. 21. VII.72. Rim fragment and small piece of neck. P. H., 0.074 m; R. H., 0.044 m; est. M.D., ca. 0.14 m; est. R.D., ca. 0.18 m. Pinkish yellow-buff clay (5YR 6/8); black and white bits. Wide depression around rim near top. A136. (PC 72-186). (Fig. IX-225). Unstamped. Lagoon, Spring House, surface. 18.VII.72. Rim fragment and very small piece of neck. P. H., 0.057 m; R.H., 0.049 m; est. M.D., 0.13 m; est. R.D., 0.16 m. Pale rust clay (5YR 6/4), tannish buff (7.5YR 7/4) near surface; black, white, and red bits; beige surface. A137. (PC 72-187). (Fig. IX-226). Unstamped. Lagoon, Spring House, surface. 18.VII.72. Rim fragment and bit of neck. P.H., 0.064 m; R.H., 0.04 m; est. M.D., 0.14 m; est. R.D., 0.16 m. Deep pinkish buff clay (2.5YR 6/6); black, white, and red bits; beige surface. Faint depression around rim near top. A138. (PC 72-188). (Fig. IX-227). Unstamped. Lagoon, Spring House, surface. 18.VII.72. Rim fragment and small piece of neck. P.H., 0.06 m; R.H., not determinable; est. M.D., 0.14 m; est. R.D., 0.18 m. Deep pinkish buff clay (2.5YR 6/ 6); black, white, and red bits; beige surface. A139. (PC 72-189). (Fig. IX-228). Unstamped. Lagoon, Spring House, surface. 18.VII.72. Rim fragment and small piece of neck. P. H., 0.063 m; R.H., 0.039 m; est. M.D., 0.14 m; est. R.D., 0.16 m. Tannish buff clay (7.5YR 7/4); black, white, and red bits; beige surface. Mark on rim might be obliterated stamp. A140. (PC 72-194). (Fig. IX-229). Unstamped. Lagoon, Spring House, surface. 17.VII.72. Rim fragment and small piece of neck. P. H., 0.072 m; R.H., 0.048 m; est. M.D., ca. 0.12 m; est. R.D., ca. 0.18 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; beige surface. Mortar adheres to part of rim. Raised line around top of neck under rim. A H l . (PC 72-195). (Fig. IX-230). Unstamped. La-

194

THE ROMAN AMPHORAS

goon, Spring House, surface. 17. VII. 72. Rim fragment and tiny piece of neck. P. H., 0.042 m; R.H., 0.033 m; est. M.D., 0.14 m; est. R.D., 0.16 m. Pinkish buff clay (5YR 6/6); black, white, and large red bits; tan surface. A142. (PC 72-196). (Fig. IX-231). Unstamped. Lagoon, Spring House, fill below rubble level north of Wall b. 14.VII.72. Rim fragment and small bit of neck. P. H., 0.066 m; R. H., 0.048 m; est. M.D., 0.18 m; est. R.D., 0.20 m. Sandy buff clay (5YR 7/3); tiny bits, mostly black. A143. (PC 72-211). (Fig. IX-232). Unstamped. Lagoon, Spring House, fill in Room 2. 25.VII.72. Small rim fragment and bit of neck. P.H., 0.079 m; R.H., 0.035 m; est. M.D., 0.12 m; est. R.D., 0.16 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; buff surface. A144. (PC 72-212). (Fig. IX-233). Unstamped. Lagoon, Spring House, fill in Room 2. 25.VII.72. Rim fragment, piece of neck, and stump of one handle. R H . , 0.135 m; R.H., 0.05 m; est. M.D., 0.15 m; est. R.D., 0.17 m; H.W., 0.05 m; H.T., 0.047 m. Coarse deep pinkish buff clay (2.5YR 6/ 8); black, white, and red bits; yellow-buff surface. Large piece of mortar adheres to rim and handle. A145. (PC 72-214). (Fig. IX-234). Unstamped. Lagoon, Spring House, fill in Room 2. 26.VII.72. Rim fragment and tiny piece of neck. P. H., 0.052 m; R.H., 0.048 m; est. M.D., 0.16 m; est. R.D., 0.21 m. Pale pinkish buff clay (5YR 7/4); small black, white, and red bits; buff surface. A146. (PC 72-217). (Fig. IX-235). Unstamped. Lagoon, Trench YZ, south extension, about 1.00 m below preserved surface of Wall W. 21. VII.72. Rim fragment, piece of upper neck, and attachment of one handle. P. H., 0.134 m; R. H., 0.044 m; est. M.D., 0.15 m; est. R.D., 0.17 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits. Mortar adheres to surface. This fragment and Cat. A147 are probably from the same amphora. A147. (PC 72-218). (Fig. IX-236). Unstamped. Lagoon, Trench YZ, south extension, about 1.00 m below preserved surface of Wall W. 21. VII.72. Rim fragment and piece of neck. P. H., 0.14 m; R.H., 0.043 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Pinkish yellow-buff clay (5YR 6/8); black bits. A148. (PC 72-242). (Fig. IX-237). Unstamped. Lagoon, Spring House, fill in Room 1. 24.VII.72. Rim fragment and bit of neck. P. H., 0.055 m; R.H., 0.042 m; est. M.D., 0.14 m; est. R.D., 0.17 m. Pale pinkish buff clay (5YR 7/4); tiny black, white, and red bits; yellow buff surface. Faint depression around top of rim. A149. (PC 72-244). (Fig. IX-238). Unstamped. La-

goon, Spring House, fill in Room 2. 27.VII.72. Rim fragment and small piece of neck. P. H., 0.052 m; R. H., not determinable (rim damaged at top); est. M.D., 0.15 m; est. R.D., 0.20 m. Pale pinkish buff clay (5YR 7/4), lighter toward surface; black, white, and red bits. A150. (PC 72-246). (Fig. IX-239). Unstamped. Lagoon, Spring House, fill in Room 2. 27.VII.72. Rim fragment and small piece of neck. P. H., 0.075 m; R.H., 0.039 m; est. M.D., 0.12 m; est. R.D., 0.16 m. Rust clay (5YR 5/6); black, white, and red bits; yellow-buff surface. A151. (PC 72-247). (Fig. IX-240, 241). Unstamped. Lagoon, Spring House, fill in Room 2. 27. VII.72. Rim fragment and very small piece of neck. P. H., 0.061 m; R.H., 0.036 m; est. M.D., 0.14 m; est. R.D., ca. 0.18 m. Pinkish yellow-buff clay (5YR 6/8); tiny black and larger red bits. Graffito on neck under rim: XA. A152. (PC 72-250). (Fig. IX-242). Unstamped. Lagoon, Spring House, gray layer below pozzolana south of Wall a. 27. VII. 72. Rim fragment and small piece of neck. R H . , 0.058 m; R.H., 0.039 m; est. M.D., ca. 0.19 m; est. R.D., ca. 0.22 m. Pinkish tannish buff clay (7.5YR 6/6); black and red bits; yellow-cream surface. A153. (PC 72-252). (Fig. IX-243). Unstamped. Lagoon, Trench IK-IL, level of tufo foundations of Wall P. 25. VII.72. Rim fragment and small piece of neck. P.H., 0.077 m; R.H., 0.038 m; est. M.D., ca. 0.18 m; est. R.D., ca. 0.21 m. Rather fine, sandpapery tannish buff clay (7.5YR 7/4); black, white, and red bits; lighter surface. A154. (PC 72-257). (Fig. IX-244). Unstamped. Lagoon, Spring House, fill south of Wall a, below rubble. 1. VIII. 72. Rim fragment and piece of neck. P.H., 0.103 m; R.H., 0.041 m; est. M.D., 0.14 m; est. R.D., 0.17 m. Rust clay (5YR 5/5); tiny black, white, and red bits; tannish buff surface. A155. (PC 72-258). (Fig. IX-245). Unstamped: Lagoon, Spring House, fill in Room 2. 31. VII. 72. Rim fragment and piece of neck. P.H., 0.114 m; R.H., 0.04 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; yellow-buff surface. Mark on rim is probably not stamp. A156. (PC 72-260). (Fig. IX-246). Unstamped. Lagoon, Spring House, surface. 28.VII.72. Rim fragment and small piece of neck. P. H., 0.066 m; R.H., 0.046 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; yellow-cream surface. A157. (PC 72-261). (Fig. IX-247). Unstamped. La-

IX: THE ROMAN AMPHORAS

goon, Spring House, surface. 26.VII.72. Rim fragment and small piece of neck. P.H., 0.072 m; R.H., 0.042 m; est. M.D., 0.13 m; est. R.D., 0.17 m. Coarse rust clay (5YR 5/6), pinkish yellowbuff (5YR 6/8) near surface; black, white, and probably red bits. A158. (PC 72-262). (Fig. IX-248). Unstamped. Lagoon, Spring House, surface. 26.VII.72. Rim fragment and small piece of neck. P.H., 0.068 m; R.H., 0.046 m; est. M.D., 0.14 m; est. R.D., 0.16 m. Fine, hard pinkish tannish buff clay (7.5YR 6/ 6); black and white bits. A159. (PC 72-264). (Fig. IX-249). Unstamped. Lagoon, Spring House, surface. 24.VII.72. Very small rim fragment and bit of neck. P. H., 0.045 m; R.H., 0.035 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Sandy tannish buff clay (7.5YR 7/4); black and white bits and one large red piece. A160. (PC 72-290). (Fig. IX-250). Unstamped. Lagoon, Spring House, fill below rubble south of Wall a. 2.VIII.72. Rim fragment and small piece of neck. R H . , 0.071 m; R.H., 0.038 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Tannish buff clay (7.5YR 7/4); black and white bits. A161. (PC 72-302). (Fig. IX-251). Unstamped. Lagoon, Spring House, Wall e (immured). 21. VII. 72. Rim fragment and piece of upper neck. R H . , 0.088 m; R.H., 0.049 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Deep pinkish buff clay (2.5YR 6/6); black, white, and red bits; beige surface. Piece of mortar adheres to rim. A162. (PC 72-303). (Fig. IX-252). Unstamped. Lagoon, Spring House, Wall e (immured). 31. VII.72. Rim fragment and small piece of neck. R H . , 0.097 m; R.H., 0.042 m; est. M.D., 0.14 m; est. R.D., 0.17. Deep pinkish buff clay (2.5YR 6/ 6), lighter toward core; black, white, and red bits; yellow-buff surface. Mortar adheres to inside of fragment. A163. (PC 72-304). (Fig. IX-253). Unstamped. Lagoon, Spring House, Wall e (immured). 31. VII.72. Rim fragment and very small piece of neck. P.H., 0.052 m; R.H., 0.048 m; est. M.D., 0.18 m; est. R.D., 0.20 m. Deep pinkish buff clay (2.5YR 6/6); black, white, and red bits; pinkish buff surface. A164. (PC 72-306). (Fig. IX-254). Unstamped. Lagoon, Spring House, southeast buttress of cistern (immured). 2.VIII.72. Rim fragment and piece of neck. R H . , 0.104 m; R.H., 0.046 m; est. M.D., ca. 0.14 m; est. R.D., ca. 0.16 m. Pale pinkish buff clay (5YR 7/4); black, white, and especially red bits. Large piece of mortar still adheres to entire inside of fragment.

195

A165. (PC 72-307). (Fig. IX-255). Unstamped. Lagoon, Trench IKE, east end of Wall W. 31.VII.72. Very small rim fragment and piece of neck. P. H., 0.05 m; R.H., 0.032 m; est. M.D., ca. 0.17 m; est. R.D., ca. 0.20 m. Peach buff clay (5YR 7/8); black, white, and large red bits; pale pinkish buff surface. A166. (PC 72-308). (Fig. IX-256). Unstamped. Lagoon, Spring House, surface. 3. VIII.72. Rim fragment and small piece of neck. P. H., 0.056 m; R.H., 0.037 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Deep pinkish buff clay (2.5YR 6/6), tannish buff (7.5YR 7/4) toward surface; black, white, and red bits. Al 67. (PC 72-315). (Fig. IX-257). Unstamped. Lagoon, Spring House, fill below rubble south of Wall a. 2. VIII. 72. Rim fragment and piece of neck. P.H., 0.09 m; R.H., 0.043 m; est. M.D., ca. 0.14 m; est. R.D., ca. 0.17 m. Rust clay (5YR 5/ 6); black and white bits. Depression around top of rim. A168. (PC 72-317). (Fig. IX-258). Unstamped. Lagoon, Spring House, fill below rubble south of Wall a. 2. VIII.72. Rim fragment, piece of neck, and upper attachment of one handle. P. H., 0.12 m; R.H., 0.047 m; est. M.D., 0.16 m; est. R.D., 0.21 m. Pale pinkish buff clay (5YR 7/4); black, red, and white bits. Depression 0.012 in height around center of rim. A169. (PC 72-319). (Fig. IX-259). Unstamped. Lagoon, Spring House, fill below rubble south of Wall a. 3.VIII.72. Rim fragment and piece of neck. P.H., 0.095 m; R.H., 0.05 m; est. M.D., 0.13 m; est. R.D., 0.17 m. Tannish buff clay (7.5YR 7/4); black, white, and red bits. A170. (PC 72-320). (Fig. IX-260). Unstamped. Lagoon, Spring House, fill below rubble south of Wall a. 3.VIII.72. Rim fragment and piece of neck. P.H., 0.075 m; R.H., 0.044 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Very bright rust clay (5YR 5/8); black and white bits. A171. (PC 72-324). (Fig. IX-261). Unstamped. Lagoon, Spring House, fill in Room 2. 4. VIII. 72. Rim fragment and small piece of neck. P. H., 0.071 m; R.H., 0.04 m; est. M.D., 0.14 m; est. R.D., 0.16 m. Deep pinkish buff clay (2.5YR 6/ 6); black, white, and red bits; yellow-buff surface. A172. (PC 72-325). (Fig. IX-262). Unstamped. Lagoon, Spring House, fill in Room 2. 4.VIII.72. Rim fragment and small piece of neck. P. H., 0.069 m; R.H., 0.047 m; est. M.D., 0.14 m; est. R.D., 0.19 m. Pale pinkish buff clay (5YR 7/4); black, white, and red bits; grayish buff surface. Slanting line incised on rim.

196

THE ROMAN AMPHORAS

A173. (PC 72-326). (Fig. IX-263). Unstamped. Lagoon, Spring House, fill in Room 2. 4.VIII.72. Rim fragment and very small piece of neck. P. H., 0.055 m; R.H., 0.042 m; est. M.D., 0.13 m; est. R.D., 0.16 m. Pale pinkish buff clay (5YR 7/4); black, white, and red bits; beige surface. A174. (PC 72-327). (Fig. IX-264). Unstamped. Lagoon, Spring House, fill in Room 2. 4.VIII.72. Rim fragment and piece of neck. P. H., 0.105 m; R.H., 0.036 m; est. M.D., 0.13 m; est. R.D., 0.16 m. Pale pinkish buff clay (5YR 7/4); black, white, and red bits; beige surface. Mortar on rim obscures probable concavity around rim. A175. (PC 72-332). (Fig. IX-265). Unstamped. Lagoon, Spring House, white clay above wood level in Room 2. 7.VIII.72. Rim fragment and small piece of neck. P.H., 0.066 m; R.H., 0.049 m; est. M.D., 0.13 m; est. R.D., 0.15 m. Rather fine tannish buff clay (7.5YR 7/4); black, white, and red bits; yellow-buff surface. A176. (PC 72-333). (Fig. IX-266). Unstamped. Lagoon, Spring House, surface. 8. VIII.72. Rim fragment and small piece of neck. P. H., 0.074 m; R.H., 0.037 m; est. M.D., 0.14 m; est. R.D., 0.155 m. Pinkish buff clay (5YR 6/6); black, white, and especially red bits; yellow-buff surface. A177. (PC 72-334). (Fig. IX-267). Unstamped. Lagoon, Spring House, surface. 7. VIII. 72. Rim fragment, piece of neck, and part of upper attachment of one handle. P.H., 0.10 m; R.H., 0.045 m; est. M.D., 0.17 m; est. R.D., 0.20 m. Deep pinkish buff clay (2.5YR 6/6); black, white, and red bits; yellow-buff surface. A178. (PC 72-335). (Fig. IX-268). Unstamped. Lagoon, Spring House, surface. 7. VIII.72. Rim fragment and piece of neck. P. H., 0.097 m; R. H., 0.044 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Deep tannish buff clay (7.5YR 6/4); black, white, and red bits. Rim slightly rounded in profile. A179. (PC 72-338). (Fig. IX-269). Unstamped. Lagoon, Spring House, surface. 7. VIII. 72. Rim fragment and piece of neck. P.H., 0.124 m; R.H., 0.061 m; est. M.D., 0.17 m; est. R.D., 0.20 m. Pale pinkish buff clay (5YR 7/4); black, white, and red bits; beige surface. Depression around rim at center. A180. (PC 72-340). (Fig. IX-270). Unstamped. Lagoon, Spring House, below rubble level north of Wall b. 8. VIII.72. Rim fragment and very small piece of neck. R H . , 0.048 m; R.H., 0.039 m; est. M.D., 0.13 m; est. R.D., 0.16 m. Rather hard brownish clay (7.5YR 5/4); tiny black and white bits; buff surface.

A181. (PC 72-342). (Fig. IX-271). Unstamped. Lagoon, Spring House, fill in Room 2. 5.VIII.72. Neck fragment, preserving part of rim and upper neck and stump of one handle. P. H., 0.137 m; R.H., 0.045 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Tannish buff clay (7.5YR 7/4); black, white, and red bits; buff surface. A182. (PC 72-344). (Fig. IX-272). Unstamped. Lagoon, Spring House, upper level of Wall a near intersection with Wall c (immured). 3.VIII.72. Rim fragment and small piece of neck. P. H., 0.079 m; R.H., 0.051 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Tannish buff clay (7.5YR 7/4); black, white, and red bits; beige surface. A183. (PC 72-346). (Fig. IX-273). Unstamped. Lagoon, Spring House, surface. 5. VIII.72. Rim fragment and small piece of neck. P. H., 0.057 m; R.H., 0.041 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Pale pinkish buff clay (5YR 7/4); black, white, and red bits; beige surface. Faint depression around rim near top. A184. (PC 72-347). (Fig. IX-274). Unstamped. Lagoon, Spring House, white clay above wood level in Room 2. 7.VIII.72. Rim fragment and very small piece of neck. R H . , 0.053 m; R.H., 0.039 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; grayish buff surface. A185. (PC 72-360). (Fig. IX-275). Unstamped. Lagoon, Spring House, fill in Room 2. 24. VII. 72. Rim fragment and small piece of neck. P. H., 0.072 m; R.H., 0.044 m; est. M.D., 0.14 m; est. R. D., 0.16 m. Pinkish tannish buff clay (7.5YR 6/ 6); black, white, and red bits. Slanting line incised on rim. A186. (PC 72-362). (Fig. IX-276). Unstamped. Lagoon, Trench El, surface. 7.VIII.72. Rim fragment and small piece of neck. P. H., 0.068 m; R.H., 0.046 m; est. M.D., 0.13 m; est. R.D., 0.17 m. Brownish clay (7.5YR 5/4); black and white bits; yellow-buff surface. A187. (PC 72-366). (Fig. IX-277). Unstamped. Lagoon, Trench E l , surface. 7.VIII.72. Rim fragment and piece of neck. P. H., 0.099 m; R. H., 0.051 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black, white, and red bits. FORM b

A188. (PC 72-343). (Figs. IX-278, 279, 280). Stamped: DIO[. Lagoon, Spring House, surface. 5.VIII.72. Rim (entire except for chips), stamped, and pieces of upper neck. R H . , 0.103 m; R.H.,

IX: THE ROMAN AMPHORAS

0.048 m; M.D., 0.143 m; R.D., 0.175 m. Coarse pale rust clay (5YR 6/4); red, black, and white bits; yellow-buff surface. The stamp, which is inverted on the rim, appears identical to a stamp from the town of Cosa (C 70.412), also fragmentary and also inverted on the rim, reading DIOC[. It, too, is on an amphora of Type 4b. Three other stamps at Cosa (C 65.251, C 67.89, and C 68.529), on amphoras of Type 4a, read DIOCI and are perhaps forerunners of Cat. A188 and C 70.412. There are other, similar stamps at Cosa, and it should be noted that Manacorda publishes (no. 26) a stamp that he reads DION.D[. The letters of that stamp resemble those of the stamps under discussion, and one wonders whether the final letter in Manacorda's stamp, unclear in the drawing, might be an inversion of "CI" rather than a " D . " It is apparently on a rim of Type 4a, for which no dimensions are given. A189. (PC 68-36). (Figs. IX-281, 282, 283). Stamped: H[. Harbor, area of Villa, surface. 18. VII. 68. Rim fragment, stamped, piece of neck, and start of one handle attachment. P. H., 0.095 m; R.H., 0.047 m; est. M.D., ca. 0.15 m; est. R.D., ca. 0.20 m. Pale rust clay (5YR 6/4); black, and some white, bits; beige surface. Rim flares out slightly (early 4b [?]). A190. (PC 69-153). (Figs. IX-284, 285). Stamped: ]IN[. Lagoon, Spring House, earth fill above basin platform. 21-26.VII.69. Rim fragment, stamped, and piece of neck. P.H., 0.09 m; R.H., 0.041 m; est. M.D., ca. 0.18 m; est. R.D., ca. 0.21 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits. Rim flares out slightly (early 4b [?]). A191. (PC 69-127). (Figs. IX-286, 287, 288). Stamped: ]NG. Lagoon, Spring House, fill around basin platform, below platform level. 4.VIII.69. Shoulder fragment, the stamp to one side of lower handle attachment. Piece of neck also preserved. R H . , ca. 0.10 m. G.W., ca. 0.18 m. Deep tannish buff clay (7.5YR 6/4); black and red bits; yellowbuff to grayish white surface. A192. (PC 72-150). (Figs. IX-289, 290, 291). Stamped: X or A (incuse). Lagoon, Trench YZ, south extension, below preserved surface of Wall W. 21. VII. 72. Rim fragment, a stamp (or possibly a graffito) set imperfectly at the top, and piece of upper neck. R H . , 0.113 m; R.H., 0.047 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pinkish yellowbuff clay (5YR 6/8); tiny black, white, and red bits. Impression runs around rim near top. Rim flares out slightly (early 4b [?]). A193. (PC 72-1). (Fig. IX-292 to 295). Stamped:

197

S a ) A ( S retr.). Lagoon, Trench IA, between piers 3 and 4, at depth of 1.50 m b.s.l. above sand. 3. VII. 72. Neck fragment, stamped between the handle attachments. Fragment preserves rim except for chips, stumps of both handles, and almost all of neck. R H . , 0.39 m; R.H., 0.047 m; M.D., 0.152 m; R.D., 0.185 m. Coarse pinkish buff clay (5YR 6/6); reddish and dark bits; buff surface. No parallels are known for this difficult stamp, which seems to combine Latin and Greek letters. The first letter looks like a retrograde " S " ; the second, like a small omega or a sigma on its back; the third is certainly a capital delta, a letter which also occurs on three rims of amphoras of Type 4a in the town of Cosa (CB 969, C 66.461, C 67.443). The rim of Cat. A193 bears a slight depression around the center, and there is a horizontal line 1 cm wide drawn between the lower handle attachments on the other side of the neck from the stamp. Al94. (PC 68-9). (Figs. IX-296, 297, 298). Stamped: device (medallion or rosette [?]). Harbor, underwater, between southeast end of main Breakwater and Breakwater Extension A, at depth in sand of 1.0 m. 9. VII.68. Rim fragment, stamped, and piece of neck. P. H., 0.079 m; R. H., 0.054 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pale rust clay (5YR 6/4), grayish tan (5YR 7/2) toward surface; black bits and a few red ones. Five probable examples of the same stamp are published herewith (Cats. A194-A198). Rim flares out slightly (early 4b [?]). A195. (PC68-uncatalogued). (Fig. IX-299). Stamped: device (medallion or rosette [?]). Immured and inaccessible for study. Harbor, pier 1; 1968. Studied from photographs. See under Cat. A194 above. Rim seems to flare out slightly (early 4b [?])• A196. (PC 72-364). (Figs. IX-300, 301, 302). Stamped: device (medallion or rosette [?]). Lagoon, Trench E l . 7.VIII.72. Rim fragment, stamped, piece of upper neck, and stump of one handle. P.H., 0.153 m; R.H., 0.047 m; est. M.D., 0.14 m; est. R.D., 0.17 m; H.W., 0.052 m; H.T., 0.039 m. Coarse rust clay (5YR 5/6); black, red, and white bits; tannish buff surface. See under Cat. A194 above. Rim is alternately outflaring, vertical, insloping, and convex in profile (early 4b [?D. A197. (PC 72-367). (Figs. IX-303, 304, 305). Stamped: device (medallion or rosette [?]). Lagoon, Trench E l , surface. 7.VIII.72. Rim fragment, stamped, piece of upper neck, and stump of one handle. R H . , 0.227 m; R.H., 0.044 m; est.

198

THE ROMAN AMPHORAS

M.D., 0.15 m; est. R.D., 0.18 m; H.W., 0.051 m; H. T., 0.049 m. Coarse rust clay (5YR 5/6); whitish surface. See under Cat. A194 above. Rim flares out slightly (early 4b [?]). A198. (PC 72-383). (Figs. IX-306, 307, 308). Stamped: device (medallion or rosette [?]). Lagoon, Trench El, surface. 7.VIII.72. Rim fragment, stamped, piece of upper neck, and stump of one handle. R H . , 0.12 m; R.H., 0.051 m; est. M.D., 0.15 m; est. R.D., 0.18 m; H.W., 0.052 m; H. T., 0.046 m. Pale rust clay (5YR 6/4); black, white, and red bits; yellow-cream surface. See under Cat. A194 above. Rim flares out slightly (early 4b [?]). A199. (PC 72-339). (Figs. IX-309, 310, 311). Stamped: device (star [?] winch [?] ship's wheel [?]). Lagoon, Spring House, surface. 7.VIII.72. Rim fragment, stamped (the same stamp struck twice), piece of upper neck, and stump of one handle. P.H., 0.134 m; R.H., 0.044 m; M.D., 0.14 m; R.D., 0.175 m. Handle dimensions not determinable. Pinkish buff clay (5YR 6/6); tiny black and white bits; buff surface. Mended. Rim flares out slightly on one side (early 4b [?]). A200. (PC 72-345). (Figs. IX-312, 313, 314). Stamped: device (ship's wheel or star). Lagoon, Spring House, surface. 5. VIII.72. Rim fragment, stamped, and piece of neck. P. H., 0.092 m; R. H., 0.047 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Coarse pale rust clay (5YR 6/4), tannish buff (7.5YR 7/4) toward surface; black, red, and white bits; yellow-buff surface. The device resembles the "wheel-and-ray" design on Attic Geometric pottery. A201. (PC 68-59). (Figs. IX-315, 316). Stamped: device (palm branch [?]). Harbor, underwater, sporadic find. VII.68. Rim fragment, stamped, and piece of neck. P. H., 0.095 m; R. H., 0.048 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; buff surface. Stamp is unlike other known palmbranch devices. Rim flares out slightly (early 4b [?])· A202. (PC 72-361). (Figs. IX-317, 318, 319). Stamped: device (animal [?] fish [?]). Lagoon, Trench El, below offset of pier 9. 7.VIII.72. Rim fragment, stamped, and piece of neck. P. H., 0.16 m; R.H., 0.048 m; est. M.D., 0.17 m; est. R.D., 0.20 m. Coarse deep tannish buff clay (7.5YR 6/ 4); black, white, and red bits; yellow-buff surface. A203. (PC 72-7). (Figs. IX-320 to 323). Stamped: illegible. Lagoon, Spring House, fill around Wall e. 7. VII.72. Neck fragment, a two-line stamp on

the rim. Fragment preserves rim, stumps of both handles, and most of neck. P. H., 0.348 m; R. H., 0.045 m; M.D., 0.143 m; R.D., 0.177 m; H.W., 0.049 m; H. T., not determinable. Coarse pinkish buff clay (5YR 6/6); many tiny dark and light bits; buff surface. A204. (PC 72-363). (Figs. IX-324, 325, 326). Stamped: illegible. Lagoon, Trench El, surface. 7. VIII. 72. Rim, the same stamp struck twice, piece of neck, part of one handle, and stump of other handle. R H . , 0.208 m; R.H., 0.042 m; M.D., 0.15 m; R.D., 0.185 m; H.W., 0.054 m; H. T., 0.05 m. Coarse deep pinkish buff clay (2.5YR 6/6); reddish, dark, and white bits; beige surface. A205. (PC 68-16). (Fig. IX-327). Unstamped. Lagoon, Trench IH, west of Wall P. 10.VII.68. Neck fragment, preserving part of rim and upper neck and upper attachment of one handle. P. H., 0.143 m; R.H., 0.045 m; est. M.D., 0.16 m; est. R.D., 0.18 m. Mended. Pale pinkish buff clay (5YR 11 4); black and white bits; yellow-cream surface. Rim flares out slightly (early 4b [?]). A206. (PC 68-26). (Fig. IX-328). Unstamped. Lagoon, Spring House, fill around basin platform. 12. VII.68. Neck fragment, preserving one handle entire and piece of rim. P. H., 0.385 m; R. H., 0.043 m; est. M.D., 0.13 m; est. R.D., 0.16 m; H. W., 0.049 m; H. T., 0.043 m. Mended. Sandy pinkish buff clay (5YR 6/6); black, white, and red bits; yellow-buff surface. A207. (PC 69-13). (Fig. IX-329). Unstamped. Harbor, underwater, Trench D l , Level III. 7. VII.69. Rim fragment and piece of neck. P.H., 0.087 m; R.H., 0.049 m; est. M.D., 0.17 m; est. R.D., 0.20 m. Powdery tannish buff clay (7.5YR 7/4); black and white bits. Rim flares out slightly (early 4b [?])• A208. (PC 69-34). (Fig. IX-330). Unstamped. Harbor, underwater, Trench D2, Level H. 16. VII.69. Rim fragment and piece of neck. P.H., 0.083 m; R.H., 0.041 m; est. M.D., 0.15 m; est. R.D., 0.17 m. Pinkish buff clay (5YR 6/6); black and white bits; yellowish tan surface. A209. (PC 69-43). (Fig. IX-331). Unstamped. Lagoon, Spring House, earth fill above surface of basin platform. 11.VII.69. Toe and fragment of lower belly. P. H., 0.34 m; diam. at base, 0.078 m. Coarse pinkish buff clay (5YR 6/6); gray deposit on surface. A210. (PC 69-44). (Fig. IX-332). Unstamped. Harbor, underwater, Trench D2, Level II. 16. VII. 69. Neck fragment, preserving rim (chipped), most of

IX: THE ROMAN AMPHORAS

neck, stumps of both handles, and bit of shoulder. P.H., 0.35 m; R.H., 0.04 m (as preserved); M.D., 0.142 m; R.D., 0.175 m. Coarse rust clay (5YR 5/ 6); marine deposit obscures surface. Pitch inside neck. A211. (PC 69-71). (Fig. IX-333). Unstamped. Harbor, underwater, Trench C 1, Level III. 25. VII.69. Belly, preserving toe, bit of lower neck, and lower attachments of both handles. P.H., 0.84 m; G.D., 0.32 m; toe diam. at base, 0.065 m. Coarse rust clay (5YR 5/6); cream surface. Inside lined with pitch. A212. (PC 72-9). (Figs. IX-334, 335). Unstamped. Lagoon, Trench IA, just above setback of pier 7, at about 0.30 m a.s.l. 10. VH.72. Neck fragment, preserving rim, most of neck, and stumps of both handles. R H . , 0.383 m; R.H., 0.045 m; M.D., 0.153 m; R.D., 0.18 m; H.W., 0.046 m; H.T., 0.043 m. Coarse tannish buff clay (7.5YR 7/4); grayish deposit obscures surface. A213. (PC 72-12). (Fig. IX-336). Unstamped. Lagoon, Trench IA, between piers 7 and 8, above sand levels. 3. VII.72. Toe and fragment of lower belly. P.H., 0.35 m; G.D., 0.215 m; toe diam. at base, not determinable. Coarse pinkish buff clay (5YR 6/6), lighter toward core; many red bits; cream surface. A very unused-looking piece. A214. (PC 72-95). (Fig. IX-337). Unstamped. Lagoon, Trench YZ, north, level of gray clay 1.50 m below surface and above pavement of small stones. 14. VII.72. Shoulder fragment, preserving pieces of neck and belly, and lower attachment of one handle. P. H., 0.16 m. Coarse pinkish buff clay (5YR 6/6); black, white, and red bits; dirty white surface. A215. (PC 72-117). (Fig. IX-338). Unstamped. Lagoon, Spring House, surface. 20. VII. 72. Rim fragment and piece of neck. P.H., 0.11 m; R.H., 0.051 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits; beige surface. Shallow depression around upper part of rim. Rim flares out slightly (early 4b [?]). A216. (PC 72-136). (Fig. IX-339). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 21. VII. 72. Rim fragment and small piece of neck. P.H., 0.08 m; R.H., 0.043 m; est. M.D., 0.17 m; est. R.D., 0.20 m. Pinkish yellow-buff clay (5YR 6/8); black, white, and red bits; yellow-buff surface. Rim flares out slightly (early 4b [?]). A217. (PC 72-165). (Fig. IX-340). Unstamped. Lagoon, Spring House, basin at east end of south wall of cistern. 21. VII.72. Rim fragment and piece

199

of upper neck. R H . , 0.12 m; R.H., 0.043 m; est. M.D., 0.14 m; est. R.D., 0.17 m. Pale rust clay (5YR 6/4); black, white, and red bits; yellow-buff surface. Rim flares out slightly (early 4b [?]). A218. (PC 72-168). (Fig. IX-341). Unstamped. Lagoon, Spring House. Parts of jar found in three areas: (1) fill below rubble level north of Wall b; (2) fill in Room 1; (3) clearing basin at east end of south wall of cistern. 21, 24. VII. 72. Jar lacking half of rim, both handles (except for one upper attachment), part of shoulder, pieces of belly, and toe. R H . , 1.00 m; G.D., 0.303 m; R.H., 0.041 m; M.D., 0.14 m; R.D., 0.175 m. Mended from twenty-two fragments. Coarse rust clay (5YR 5/ 6); yellow-buff surface. A219. (PC 72-175). (Fig. IX-342). Unstamped. Lagoon, Spring House, fill against south wall of cistern. 20. VII. 72. Rim fragment, piece of neck, and upper attachment of one handle. P. H., 0.118 m; R.H., 0.045 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Very coarse rust clay (5YR 5/6), yellow-buff (7.5YR 7/6) toward core; black, white, and red bits. Slanting line incised on rim. A220. (PC 72-197). (Fig. IX-343). Unstamped. Lagoon, Spring House, fill below rubble level north of Wall b. 21.VII.72. Rim fragment and small piece of neck. P.H., 0.07 m; R.H., 0.047 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Pinkish buff clay (5YR 6/6); many black, white, and red bits. Rim flares out slightly (early 4b [?]). A221. (PC 72-227). (Figs. IX-344, 345). Unstamped. Lagoon, between piers on Walls Y and Z, on bedrock. 17.VII.72. Neck fragment, preserving most of rim, upper neck, and stumps of both handles. P.H., 0.145 m; R.H., 0.042 m; M.D., 0.16 m; R.D., 0.195 m; H.W., 0.061 m; H.T., 0.031 m. Sandy pale pinkish buff clay (5YR 7/4); red, dark, and white bits. Mended. Possible punch-mark on rim, and two holes. Rim flares out slightly (early 4b [?]). A222. (PC 72-240). (Figs. IX-346, 347). Unstamped. Lagoon, Spring House, fill below rubble level north of Wall b. 24.VII.72. Neck fragment, preserving pieces of rim and upper neck, and stump of one handle. R H . , 0.16 m; R.H., 0.044 m; M.D., 0.145 m; R.D., 0.18 m; H.W., 0.059 m; H. T., 0.044 m. Coarse rust clay (5YR 5/6); black and white bits; yellow-buff surface. Mended. Rim flares out slightly (early 4b [?]). A223. (PC 72-241). (Fig. IX-348). Unstamped. Lagoon, Spring House, fill below rubble level north of Wall b. 24.VII.72. Rim fragment and small piece of neck. P.H., 0.066 m; R.H., 0.046 m;

200

THE ROMAN AMPHORAS

M.D., 0.15 m; R.D., 0.18 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; beige surface. Mended. Wide depression around top of rim. Mark on rim is probably not remnant of stamp. Rim flares out slightly (early 4b [?]). A224. (PC 72-255). (Fig. IX-349). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2. VIII. 72. Jar fragment, preserving belly, toe, shoulder, piece of neck, lower stump of one handle and lower attachment of the other. P.H., 0.77 m; G.D., 0.29 m. Coarse pinkish buff clay (5YR 6/6); buff surface. Mended. A225. (PC 72-272). (Fig. IX-350). Unstamped. Lagoon, Trench U2, 1.53 m b.s.l., in fill behind and east of Wall U. 2. VIII. 72. Jar fragment, preserving belly, upper toe, lower neck, shoulder, and lower stumps of both handles. R H . , 0.85 m; G.D., 0.32 m. Coarse rust clay (5YR 5/6); cream surface. A226. (PC 72-273). (Fig. IX-351). Unstamped. Lagoon, Trench U2, 1.53 m b.s.l., in fill behind and east of Wall U. 2. VIII. 72. Jar fragment, preserving belly, toe, shoulder, lower half of neck, and lower stumps of both handles. R H . , 0.90 m; G.D., 0.29 m; toe diam. at base, 0.075 m. Coarse rust clay (5YR 5/6); cream surface. Possible trace of stamp at base of one handle. Neck appears to have been sawed off in antiquity (cf. Cat. A242 below). A227. (PC 72-274). (Fig. IX-352). Unstamped. Lagoon, Trench U2, 1.53 m b.s.l., in fill behind and east of Wall U. 2. VIII.72. Jar fragment, preserving belly, toe, shoulder, piece of lower neck, one lower handle attachment, and traces of other attachment. R H . , 0.85 m; G.D., 0.305 m; toe diam. at base, 0.07 m. Coarse rust clay (5YR 5/6); buff surface. A228. (PC 72-275). (Fig. IX-353). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2. VIII.72. Jar fragment, preserving belly, toe (chipped), shoulder, lowest part of neck, and traces of lower handle attachments. P.H., 0.80 m; G.D., 0.29 m; toe diam. at base, not determinable. Coarse rust clay (5YR 5/6); cream surface. A229. (PC 72-276). (Fig. IX-354). Unstamped. Lagoon, Trench U2, 1.53 m b.s.l., in fill behind and east of Wall U. 2. VIII.72. Jar lacking rim, toe, one handle (except for lower attachment), and bit of upper neck. R H . , 1.02 m; G.D., 0.303 m; H.W., 0.054 m; H. T., 0.041 m. Coarse pinkish buff clay (5YR 6/6); reddish, dark, and light bits; buff surface. Two holes in lower belly (0.025 m and 0.027 m in diam.). A230. (PC 72-277). (Fig. IX-355). Unstamped. Lagoon, Trench U2, 1.53 m b.s.l., in fill behind and

east of Wall U. 2. VIII. 72. Belly, preserving upper toe, shoulder, bit of neck, stump of one handle, and lower attachment of other handle. P. H., 0.68 m; G.D., 0.295 m. Worn, coarse rust clay (5YR 5/6). A231. (PC 72-278). (Fig. IX-356). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2. VIII.72. Belly fragment, lacking much of upper belly, but preserving toe (chipped) and small piece of shoulder with lower attachment of one handle. P.H., 0.81 m; est. G.D., 0.28 m. Coarse rust clay (5YR 5/6), grayish at core. A232. (PC 72-279). (Fig. IX-357). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2. VIII.72. Jar fragment, preserving belly, toe, shoulder, and lower stumps of both handles. R H . , 0.78 m; G.D., 0.29 m; toe diam. at base, 0.07 m. Rust clay (5YR 5/6); red bits; cream surface. Mended. A233. (PC 72-280). (Fig. IX-358). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2. VIII. 72. Jar fragment, preserving belly, much of neck, stump of one handle, attachment of other handle, and upper part of toe. P. H., 0.92 m; G.D., 0.295 m. Coarse pinkish buff clay (5YR 6/6); black, white, and red bits; buff surface. A234. (PC 72-281). (Fig. IX-359). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2. VIII.72. Belly, preserving piece of lower neck, stumps of both handles, and upper part of toe. R H . , 0.70 m; G.D., 0.29 m. Coarse rust clay (5YR 5/6); details of fabric obscured by surface deposit. A235. (PC 72-282). (Fig. IX-360). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2. VIII.72. Belly, preserving toe, shoulder, and stumps of both handles. P.H., 0.82 m; G.D., 0.30 m; H.W. near lower attachment, 0.048 m; H. T. near lower attachment, 0.041 m. Coarse rust clay (5YR 5/6), grayish toward core (5YR 6/1) and lighter toward surface (four distinct layers); scattered reddish bits; surface obscured by deposit. Mended. Several holes in belly. A236. (PC 72-283). (Fig. IX-361). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U., 2.VIII.72. Belly fragment, preserving upper part of belly, shoulder, lower handle attachments, and piece of lower neck. P. H., 0.47 m; G.D., 0.325 m. Coarse rust clay (5YR 5/ 6); yellow-buff surface. Two holes bored in belly (0.05 m and 0.27 m in width). A237. (PC 72-311). (Fig. IX-362). Unstamped. Lagoon, Spring House, gray level below pozzolana,

IX: THE ROMAN AMPHORAS

south of Wall a. 3. VIII. 72. Rim fragment and small piece of neck. P. H., 0.076 m; R. H., 0.045 m; est. M.D., 0.16 m; est. R.D., 0.19 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; yellow-buff surface. Rim flares out slightly (early 4b [?]). A238. (PC 72-316). (Fig. IX-363). Unstamped. Lagoon, Spring House, fill below rubble south of Wall a. 2.VIII.72. Rim fragment and small piece of neck. R H . , 0.081 m; R.H., 0.046 m; est. M.D., 0.15 m; est. R.D., 0.18 m. Rust clay (5YR 5/6); black and white bits. Rim flares out slightly (early 4b [?]). A239. (PC 72-321). (Fig. IX-364). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 1.VIII.72. Fragment of lower belly, preserving toe except for chips. P.H., 0.515 m; G.D., 0.305 m. Coarse rust clay (5YR 5/6); yellow-cream surface. A240. (PC 72-328). (Fig. IX-365). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2.VIII.72. Fragment of lower belly, preserving most of toe. P. H., 0.62 m; est. G.D., 0.28 m; toe diam. at base, 0.072 m. Coarse tannish buff clay (7.5YR 7/4), gray at core (5YR 6/1); cream surface. Mended. A241. (PC 72-329). (Fig. IX-366). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2.VIII.72. Fragment of lower belly, lacking much of toe. P.H., 0.43 m; G.D., 0.277 m. Coarse rust clay (5YR 5/6) with lighter surface. Mended. A242. (PC 72-330). (Fig. IX-367). Unstamped. Lagoon, Trench U l , 1.67 m b.s.l., in fill behind and east of Wall U. 2. VIII. 72. Fragment of lower neck and upper belly, preserving shoulder and both lower handle attachments. P.H., 0.49 m; G.D., 0.30 m. Coarse orange-buff clay (2.5YR 6/8); deposit obscures surface details. Mended. Possible trace of stamp-outline on shoulder. Neck sawed off (cf. Cat. A226 above). A243. (PC 72-336). (Fig. IX-368). Unstamped. Lagoon, Spring House, white clay above wood level in Room 2. 8. VIII. 72. Rim fragment and very small piece of neck. R H . , 0.049 m; R.H., 0.045 m; est. M.D., 0.16 m; est. R.D., 0.20 m. Deep pinkish buff clay (2.5YR 6/6); black, white, and red bits; yellow-buff surface. Rim flares out slightly (early 4b [?]). A244. (PC 72-337). (Fig. IX-369). Unstamped. Lagoon, Spring House, surface. 7. VIII. 72. Rim fragment and piece of neck. R H . , 0.083 m; R.H., 0.05 m; est. M.D., 0.17 m; est. R.D., 0.20 m. Deep

201

pinkish buff clay (2.5YR 6/6); black, white, and red bits; beige surface. Faint depression around rim below center. Rim flares out slightly (early 4b [?])· A245. (PC 72-365). (Figs. IX-370, 371). Unstamped. Lagoon, Trench E 1, surface. 7. VIII.72. Rim fragment and piece of neck. P. H., 0.082 m; R.H., 0.044 m; est. M.D., ca. 0.19 m; est. R.D., ca. 0.22 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; yellow-buff surface. Slanting line incised on rim. Rim flares out slightly (early 4b [?]). A246. (PC 72-427). Unstamped. Sporadic; 1972. Neck fragment, preserving rim, part of neck, and attachments of both handles. P.H., 0.253 m; R.H., 0.063 m; M.D., 0.147 m; R.D., 0.178 m. Coarse rust clay (5YR 5/6); many black and white bits; buff surface. Rim flares out slightly (early 4b [?])· Type 5 Lamboglia decided to identify Type 5 as "Dressel I C " (Lamboglia 1955, pp. 248-250), though the type has little but height and long neck and handles to connect it with the jars of Dressel's Form 1, which we refer to here as Type 4. Several distinctive differences can be pointed out between the two types. Type 5 has a very high rim that is often over 0.06 m, and sometimes even over 0.07 m, in height. The rim is vertical or concave in profile but flares out slightly in earlier versions. The mouth is very narrow in diameter (0.10 m-0.12 m on the average). Long, ribbed handles, narrow in section and Sshaped in profile, flank a long neck that is regularly somewhat wider at bottom than at top. The shoulder is unusually narrow, the neck joining the belly nearer the middle than the edge of the shoulder. The belly is correspondingly narrow (0.26 m-0.28 m on the average) and carrot-shaped, the toe continuing the line of the belly and often not given a separate exterior shape, though it is in fact solid, not hollow, on the interior. Stamps are infrequent but may occur on the rim, on the handle near the curve, or on the shoulder at the base of the handle. The stylistic features of Type 5 show it to be closely related to, and probably a descendant of Form e Greco-Italic amphoras (Will Type 2). That category of Greco-Italic amphoras is not found in the port or in the town of Cosa, in contrast to Type 5, which is well represented at both sites (sixty examples found so far in the town and twenty-two in the port). In a recent study of Greco-Italic amphoras

202

THE ROMAN AMPHORAS

(Will 1982-1, pp. 353-355) I argue that Type 2 originated in Spain. Contexts at the Athenian Agora permit us to date the type in the first half of the second century B.C. and as late as the third quarter of the century. It was a widespread shape, occurring as far east as Athens, Delos, and Mykonos, but it is found with particular frequency in Spain, and the clay seems close to that of the amphoras called "Tarraconese" (Tchernia and Zevi 1972). Type 5 was apparently developing from Type 2 in Spain at the same time that Type 4a was developing from Type Id in Italy. Contextual information at the Athenian Agora allows us to date both Type 4a and Type 5 as early as the late second century B.C., and, like Type 4a, Type 5 seems to have been in use into the second quarter of the first century B.C. (Will 1979, nn. 6 and 25, and see here Color Fig. 4). Not only do the two types occur together in the same closely dated context at the Agora, but a belly of Type 5 was found along with the hundreds of Sestius jars on the upper Grand Congloue wreck (Benoit 1961, pp. 45-46). Type 5 also occurs at Vada Sabatia in a context of about 100 B.C. (Lamboglia 1955, p. 249), and the frequency of the type at Delos supports a floruit for it in the first quarter of the first century B.C. Two Sestius stamps, in addition, are found on fragments of Type 5, one at Vada Sabatia and the other at the Portus Cosanus (below, Cat. A248), and a painted inscription possibly naming both Sestius and Cosa has been found on the dated piece, referred to above, from the Agora (P 6867; cf. Will 1979, pp. 346-347 and fig. 5, and here, Color Fig. 4). A possible connection with Cosa and the port is also suggested by the fact that the total number of Type 5 pieces found there, while small in comparison with the totals for Types 1 and 4, is probably greater than the total for any other known site where Type 5 occurs. But in spite of its Cosa connections, Type 5 probably originated in Spain. In addition to the Sestius fabric, two other types of clay were used for Type 5: a fragile, powdery pale greenish buff clay (Munsell 2.5Y 8/2) that also characterizes the Spanish garum jars of Dressel's Forms 7-12 (my Type 16, the Spanish connections of which are discussed below); and an unusually coarse rust-colored clay (Munsell 5YR 5/6), full of white and black bits, and covered with a peeling, dirty grayish white surface, a fabric identical to that of Type 2. It is logical to suppose that these two fabrics belong to examples of Type 5 made in two different areas of Spain, the darker-clayed jars being probably Tarraconese and the lighter-clayed ones coming from the kilns that have been discovered

near Cadiz and Algeciras in southern Spain, an area known to have been the center of one of the major fish industries of antiquity (see Peacock 1974 for an excellent discussion of these kilns). A neck fragment of Type 5 was, in fact, discovered at the kiln of Algeciras (Tchernia 1971, fig. 13), though most of the abundant pieces found in connection with that pottery establishment belonged to Dressel's Forms 7-12. Since Dressel 12 could easily qualify on typological grounds as a descendant of Type 5, the evidence of its manufacture at Algeciras adds to the argument in favor of that area as one of the sources of Type 5. But the Sestius stamps on Type 5 at the Portus Cosanus and at Vada Sabatia, the painted inscription at the Agora, and the Sestius clay both of that fragment and of many of the large group of pieces at Cosa argue for the Portus Cosanus as another area in which Type 5 was manufactured. It seems likely, also, that these amphoras were meant to serve, both in Spain and at Cosa, as containers for garum (Will 1979, n. 26), as was unquestionably true of Dressel's Forms 7-12. An industry for the production of garum was apparently growing up in the Portus Cosanus by the end of the second century B.C. (McCann 1979), and the presence of the Type 5 Sestius jars in the Cosa area and elsewhere suggests that, at the same time that it was producing wine jars of Type 4a, the Sestius pottery was also producing garum containers of Type 5. Faced with the need for a new kind of container in which to ship the port's new commodity, the Sestius organization (which may also have owned the fishery in the port, not to mention the vineyards in the Ager Cosanus) turned to a contemporary Spanish model and chose to adapt it to its own needs. That form was in turn apparently based on the secondcentury-B.c. amphora, Type 2, the Form e GrecoItalic variety. Whether jars of Type 2, the interiors of which are thickly pitched, were also used as containers for garum is not yet known. It is traditional to associate such pitched amphoras with wine (Will 1982-1, p. 354), but there is increasing evidence to show that garum amphoras were also pitched (Purpura 1975, pp. 61-63, note 15). Type 2, in any case, as we noted, is not represented at Cosa, perhaps because the Cosa fisheries at that time produced sufficient garum for local needs. But when exporting began, Spanish containers were the logical ones to copy, for Spain had been a major producer from early Greek times. Some Type 5 pieces at Cosa have, indeed, Spanish, or at least non-Sestius, clay. One such piece found in the town of Cosa (CB 1664) bears the internationally distributed trademark

IX: THE ROMAN AMPHORAS

S.C.G, which has been found at Rome (CIL XV.3417) and at Delos (two unpublished examples, TD 1051 and TD 4884), and in Spain at Tarraco (CIL 11.4967.42), at Algeciras (Beltran 1970, no. 435), and at Bolonia near Cadiz (Beltran 1970, no. 436). Perhaps the amphora that came to Cosa bearing that stamp was one of the patterns used by the Sestius factory. The eleven pieces described in the section that follows represent half of the total number of fragments of Type 5 found in the port. The group includes three necks, five rims, one handle, one shoulder, and one belly. Eleven other finds in the port, not described individually here, include eight handles, two rims, and one toe.

A247. (PC 68-8). (Figs. IX-372, 373, 374). Stamped: ALEXANDRI. Lagoon, Trench IH, west of Wall P, surface fill. 6.VII.68. Handle fragment, stamped lengthwise near curve. H. W., 0.06 m; H. T., 0.022 m. Rust clay (5YR 5/6); black and red bits; beige surface. Slightly ridged handle. The same name occurs, as might be expected, on amphoras of various types and dates, though not on Type 5. Several spellings are found in stamps on Type 4, the connections of which with Type 5 are mentioned in the introductions to Types 4 and 5, above, and under Cat. A248 below. The stamps on Type 4 include "ALEX" on a handle of Type 4a at the National Museum, Athens (EM-L 81); "]ALEXAND[" on both handles of two (?) anaphoras of "Dressel Type 1" (CIL XV.3398; one of those jars was seen by me in storage in the Market of Trajan, Rome, in 1974); "ALEXA" on the shoulder of a jar probably belonging to Type 4b at Cayla de Mailhac, Aude (on this piece, cf. Odette and Jean Taffanel, Gallia 5 [1947] fig. 4:2); and ALEXSANDI on the lower belly of one (?) amphora from the wreck of La Madrague de Giens (Hesnard 1978, p. 37, and cf. pi. 15:7 and table between pp. 41 and 42. The inventory number of the amphora in question is given there as 799; however, another and better preserved stamp with the same letters was apparently found on the same wreck and was given the number 6 954, according to Hesnard 1977, p. 160 and fig. 5. That stamp is not referred to in the later publication. Hesnard also reports two examples of the same stamp on amphoras in storage in the Terme Museum, Rome. I did not, perhaps by error, see any stamps on the jars of Dressel's Type 1 that I studied in the Terme Museum some years ago, but in

203

any case these are probably not the jars, referred to above, studied by Dressel and published by him in CIL XV.3398, since the amphoras used by Dressel are all thought to be in storage in the Market of Trajan. The examples from the Terme Museum and La Madrague de Giens are all said to add an " S " to the letters of the stamp, though that letter is hardly visible in the 1978 publication of the wreck. One hopes these uncertainties will be dispelled in a future publication. Hesnard associates the stamp with potteries found in the nineteenth century near Terracina [cf. CIL X. 8050]). A248. (PC 69-177). (Figs. IX-375 to 378). Stamped: SEST device (palm branch). Harbor area, 1969. Private collection. Neck fragment, preserving piece of rim, stamped, one handle, and part of upper neck. R H . , 0.32 m; R.H., 0.053 m; est. M.D., 0.115 m; est. R.D., 0.14 m; H.W., 0.067 m; H.T., 0.038 m. Coarse pale pinkish buff clay (5YR 7/4); frequent black and reddish bits; yellow-buff surface. Marine deposits adhere to surface. On this piece, see Will 1979, p. 346 and fig. 6. The same stamp is frequent on Type 4a. See the discussion under Cat. A80 above. As noted there, another example on Type 5 occurs at Vada Sabatia, Liguria (Lamboglia 1955, fig. 6: lower left). To date, these are the only Sestius stamps known to occur on a type other than 4a or 4b. A249. (PC 68-27). (Fig. IX-379). Unstamped. Lagoon, Trench IH, west of Wall P. 9. VII. 68. Shoulder fragment and piece of neck. P. H., ca. 0.073 m; G.W., 0.12 m. Beige clay (7.5YR 8/4); black and red bits. A250. (PC 69-97). (Figs. IX-380, 381). Unstamped. Lagoon, Spring House, fill around basin platform. 1. VIII.69. Rim fragment and very small piece of neck. R H . , 0.089 m; R.H., 0.079 m; est. M.D., 0.14 m; est. R.D., 0.18 m. Pale rust clay (5YR 6/ 4); black and white bits; beige surface. Depression around rim below center. A251. (PC 72-2). (Figs. IX-382, 383). Unstamped. Lagoon Trench IE, east of Wall N, 1.50 m below surface. 4. VII. 72. Neck fragment, preserving much of rim, piece of upper neck, and stump of one handle. P.H., 0.159 m; R.H., 0.062 m; M.D., 0.11 m; R.D., 0.145 m. Coarse tan clay (7.5YR 11 4); many black and white bits; whitish surface. Two parallel scratches on rim. A252. (PC 72-13). (Figs. IX-384to387). Unstamped: Lagoon, Trench IK, below first setback of tuff foundations of Wall P. 6. VII.72. Belly fragments and upper toe, nonjoining. Est. G.D., 0.30 m. Rust clay (5YR 5/6); many black and white bits.

204

THE ROMAN AMPHORAS

A253. (PC 72-34). (Fig. IX-388). Unstamped. Lagoon, Spring House, surface. 10. VII. 72. Rim fragment and small piece of neck. P. H., 0.098 m; R.H., 0.054 m; est. M.D., 0.12 m; est. R.D., 0.14 m. Coarse tannish buff clay (7.5YR 7/4); black and white bits; cream surface. A254 (PC 72-66). (Fig. IX-389). Unstamped. Lagoon, Spring House, rubble level south of Wall a. 14. VII. 72. Rim fragment and tiny piece of neck. P.H., 0.054 m; R.H., 0.051 m; est. M.D., 0.12 m; est. R.D., 0.14 m. Pinkish tannish buff clay (7.5YR 6/6); many red, and some dark and white, bits; yellow-cream surface. A255. (PC 72-137). (Fig. IX-390). Unstamped. Lagoon, Spring House, rubble level north of Wall b. 21. VII. 72. Neck fragment, preserving pieces of rim and upper neck and stump of one handle. P. H., ca. 0.215 m; R. H., not determinable; est. M.D., ca. 0.15 m; est. R.D., ca. 0.18 m; H.W., 0.071 m; H.T., 0.029 m. Rather sandy pale pinkish buff clay (5YR 7/4). A256. (PC 72-166). (Figs. IX-391, 392). Unstamped. Lagoon, Spring House, basin at east end of south wall of cistern. 21. VII.72. Rim fragment and bit of neck. R H . , 0.081 m; R.H., 0.067 m; est. M.D., 0.12 m; est. R.D., 0.14 m. Rust clay (5YR 5/6); black, white, and especially red bits; yellow-buff surface. Slanting line incised on rim. Depression around rim at center. A257. (PC 72-318). (Fig. IX-393). Unstamped. Lagoon, Trench IKE, surface of Wall X. 2.VIII.72. Rim fragment, piece of upper neck, and stump of one handle. P. H., 0.136 m; R. H., 0.056 m; est. M.D., 0.14 m; est. R.D., 0.17 m; H.W. as preserved, 0.063 m; H.T. as preserved, 0.049 m. Very sandy pale pinkish buff clay (5YR 7/4). Type 10 Amphoras of this type belong to the general category described by Lamboglia under Form 2 (1955, p. 262 and fig. 17) and by Baldacci under Form II b (1967-1968 [1969] pp. 18-23; 1969 [1972] pp. 128129; 1972, p. 27), though the comments of those scholars and of Beltran (1970, pp. 349-358) have raised more questions than they have answered about this very complicated type and its relationship to similar shapes (Types 3, 6, 7, 8, 9, 13, and 14 in my forthcoming Athenian Agora publication, where these matters are addressed in detail). Type 10 occurs with particular frequency in the eastern Mediterranean and Adriatic areas. Finds in the west have been sparse and have been largely underwater, at such

sites as Albenga (see the reference above to Lamboglia's discussion), La Madrague de Giens (Hesnard 1978, p. 46 and pi. 17: 1, 2, 5), and Sant Jordi, Majorca (Cerda 1980, pp. 73-77, 79, where an interesting summary of some of the western finds of Type 10 and related types is given). Type 10, which was almost certainly a container for olive oil, can be securely dated in the first quarter of the first century B. c., on the basis of its occurrence in dated deposits at the Athenian Agora, especially Deposit M 18:1, which dates from somewhat before the destruction of the Agora by Sulla in 86 BC. (cf. Will 1970, pp. 383-386, on that context and on certain amphoras of Type 10 from Delos. Research subsequent to 1970 leads me to believe that Type 10 was also in use in the late second century B.C. as well). The single find of Type 10 in the port of Cosa, a toe fragment of the fine clay that characterizes the type, is described herewith. Several other pieces, mostly stamped, from the town of Cosa will be considered in my discussion of the amphoras from that site. The paucity of finds in the Cosa area suggests that enough olive oil for domestic needs was being produced locally in the late second and early first centuries B.C. As to the origin of Type 10, Zevi 1967 and Baldacci suggest Apulia, but Istria and the northern Adriatic region are, I believe, stronger possibilities both for Type 10 and for the related shapes listed above. I shall discuss elsewhere the epigraphical evidence that bears on this matter. It is certain, at any rate, that of the related shapes referred to, Type 14 was manufactured in Istria and perhaps in Cisalpine Gaul as well (see below under Types 13 and 14). It is the possible connection of Type 10 with the famous olive oil of Istria, seen in the light of other evidence, such as the presence of the graffito OLEA on a neck of Type 10 at Delos (no. TD 1047; on this piece, cf. Hatzfeld 1912, p. 143), which leads me to suggest that Type 10 was a shipping container for olive oil. But Tchernia 1980, p. 305, argues that amphoras of "Lamboglia 2" were containers for wine. I believe part of the difficulty may lie in the fact that amphoras not belonging to Type 10 are sometimes assigned to the type (cf. McCann and Will 1984, p. 94). Perhaps the utilization of a technique such as gas chromatography could throw light on this matter (cf. J. Condamin et al., "The Application of Gas Chromatography to the Tracing of Oil in Ancient Amphorae," Archaeometry 18 [1976] pp. 195-201).

A258. (PC 72-425). (Fig. IX-394). Unstamped. Lagoon, Spring House, white clay above wood level

IX: THE ROMAN AMPHORAS

in Room 2. 2. VIII. 72. Fragment of upper part of toe, preserving piece of lower belly. P. H., ca. 0.18 m. Fine, hard tannish buff clay (7.5YR 7/4); many red bits; buff surface. Type lie This late form of Type 11a ("Brindisi-type") can be dated with confidence in the first half of the first century A.D., on the basis of contexts at the Athenian Agora (especially Deposit D 11:1, which dates to the middle of the first century A. D.) and on the basis of its frequent occurrence in the large group of amphoras found in the 1870s in the Castro Pretorio in Rome. That cache of jars was apparently datable to about the middle of the first century (Dressel 1879, pp. 194-195; Zevi 1966, p. 211). Many of those same jars are now in storage in the Market of Trajan in Rome, and one of them is doubtless the jar used by Dressel to make his drawing of Form 25, to which Type l i e corresponds. On grounds of shape, dimensions, and clay, it is very probable that Type l i e and Dressel 25 are a late equivalent of Type 11a, which I believe to have been a container for the famous olive oil of Venafrum. By extension, Type l i e would have served the same purpose in another era. Type 11a dates from the first quarter of the first century B.C. (McCann, Bourgeois, and Will, p. 295), and it belongs to the category of amphoras called "Amphorae Calabrae" by Mommsen in CIL IX. 6079, where he refers to the discovery of kilns for the type near Brindisi. Those kilns, or the remnants of them, were rediscovered by me in 1961 (Will 1962, pp. 649-650). By a regrettable confusion, Zevi (1967, n. 17), followed by Baldacci (1967-1968 [1969] p. 13, and also in later publications) and others writing in the 1970s, incorrectly identified the "Amphorae Calabrae" with the unrelated jars of my Type 10, which are not known to have had any connection with Calabria, though they may be Apulian. They could equally well, however, be Istrian or North Italian, as I point out above in my discussion of Type 10. The true Calabrian amphoras, the Brindisi-type jars of Type 11a, may have been succeeded by a shape which I call Type l i b (see again McCann, Bourgeois, and Will, pp. 294-296) and then ultimately, after the Augustan Age, by Type l i e . On the other hand, there may have been no direct connection between Types 11a and l i b , which differ from each other in fabric and in certain important stylistic details. Here, again, as I point out in the 1977 discussion, there has been confusion on the part of Tchernia (1968-1970) and other scholars who, not

205

aware of the distinctions between Types 11a and l i b , call l i b "Brindisi-type." These matters, including the question of the relationship between l i b and Type 20 (my number for Dressel's Form 20), will be examined in my Athenian Agora publication. Whatever their relationship to Type l i b , however, there seem to be close connections in style and fabric between Types 11a and lie. The jar fragment of Type l i e from the Portus Cosanus, the example described herewith, was initially published by McCann and Lewis (1970, p. 210). It is the only example of Type l i e from the Cosa area, though three pieces of Type 11a have been found in the town of Cosa. A259. (PC 68-4). (Figs. IX-395, 396). Unstamped. Lagoon, Spring House, mud fill above basin platform. 12.VII.68. Jar fragment, preserving rim, neck, handles, much of shoulder, and part of belly. R H . , 0.76 m; G.D., 0.43 m; R.H., 0.076 m; M.D., 0.122 m; R.D., 0.140 m; H.W., 0.036 m; H. T., 0.036 m. Pale pinkish buff clay (5YR 11 4); scattered reddish and dark bits; surface obscured by grayish deposit. Mended. On this piece, see McCann and Lewis 1970, p. 210. Type 12 Beginning in the second quarter of the first century B. c., the Pompeian wine production, of which there is no evidence after the demise of Type Id in the middle of the preceding century, revives to become a dominating force in world trade. There is circumstantial evidence to show that shock waves attendant on such a revival were felt even in Roman political circles. It is in any case clear that the Sestius family's wine and garum industries at Cosa declined in about the middle of the first century B.C., at the same time that, at Pompeii, an equivalent industry was being resurrected, probably in connection with the development of the newly founded Roman colony of Pompeii. Not only did the production center of Type 4b move, as we have seen, from the Cosa area to Campania/Latium, but also another type of amphora, Type 12a (Dressel's Form 3; Panella and Fano 1977, Groups 1-3), based on the shape of the double-handled wine jars of the island of Cos (Grace 1961, fig. 56), made its appearance at Pompeii. We are sure it was manufactured at Pompeii because of the many trademarks naming known citizens of Pompeii that are found on amphoras of the type all over the Mediterranean (these stamps will be published in my Athenian Agora volume). The form of Type 12a is also reminiscent not only of the Coan

206

THE ROMAN AMPHORAS

shape but also of the Pompeian jars of Type Id, and the coarse clay (Panella and Fano 1977, p. 146) is very similar to that of the Type Id amphoras from Pompeii. Dated contexts at the Athenian Agora give us supporting chronological evidence that the floruit of the earliest examples of Type 12a was the last half of the first century B.C., but examples of earlier date are known. I will argue elsewhere that the jar pattern represented by Type 12a reached Pompeii via an earlier manufacturing center in the area of Brindisi; in any case, Coan-shaped amphoras of fine, "Brindisi" clay are known to have borne stamps identical to those on Type 11a, which dates from the first quarter of the first century B.C. It seems logical to assume that these jars were the first to imitate the Coan shape in Italy and then became the immediate model for the Pompeian manufacturers. Doubtless both at Brindisi and at Pompeii the container was intended for the factitious Coan wine, that seems to have been popular among the Romans (Cato the Elder gives the recipe, de agri cultura 112). The Pompeian wine industry lasted until the destruction of the city in A.D. 79. Panella and Fano (1977, passim) present a most useful review of the evolution of Type 12 at Pompeii, while making it clear that not all the varieties were manufactured there. It may well be that it was the pseudo-Coan jars of Italy that stimulated still another industry for Coan-like jars, in Spain, where amphoras like Dressel's Form 2 (copies of Panella's "Italian" Group 8[?]) were manufactured, doubtless as shipping containers for the cheap Spanish wine deplored by Roman writers. Tchernia and Zevi (1972, passim) have argued convincingly that such Spanish jars have a characteristic and easily recognizable clay, which they describe as Tarraconese; and it is well known that kilns for these same jars have been found in the area between Tarragona and Ampurias in northeastern Spain (Pascual Guasch 1962; Tchernia 1971). These jars of Dressel 2, both the "Italian" and the Spanish varieties, are referred to here as Type 12b, which can be dated because it was in current use at the time of the destruction of Pompeii, and it occurs in contemporary contexts at the Athenian Agora (cf. Robinson 1959, p. 85 and pi. 19: [M 13]). But also included under Type 12b are amphoras, to be described below, that may have been manufactured near Cosa, as may certain examples of Type 12c, still another variety of Type 12. Type 12c jars were fractional, flat-bottomed containers resembling amphoras that occur very frequently at Pompeii and were contemporary with Type 12b. In the port of Cosa, Type 12a is represented by

only a single piece, a jar fragment of typically "Pompeian" clay. About 30 pieces of similar shape but showing widely varying fabrics have been found in the town of Cosa. With Type 12b, similarly, various fabrics can be distinguished in the fragments both from Cosa and from the port. The port pieces (2 necks, 1 handle, and 3 body fragments, catalogued herewith, to which should be added 4 uncatalogued handles) are all apparently Spanish imports, as their fabric shows the distinctive white bits of Tarraconese clay. But the much larger group of fragments (117 as of 1979) from the town of Cosa are of clay that looks typically Sestius. The same is true of the 12c pieces from the port (2 neck fragments described below). Have we in these "Cosan" pieces of Types 12b and 12c evidence that the manufacture not only of garum jars of Type 16 (see below) but also of wine jars (perhaps after A.D. 79?) occurred near Cosa? If so, these would be efforts to restore the wine and garum industries of a century earlier. Clearly such efforts, if they occurred, did not take place in the Portus Cosanus. Most of the 12b finds in the port were made underwater in the harbor. In this connection, the discovery reported by Peacock (1977, pp. 266-268) of a possible kiln site at Albinia to the north of Cosa and of pieces belonging to Type 12 and other types in the area in question (see above under Type 4) perhaps points to local manufacture of Type 12. Whether, as I point out above, the pieces found were actually made at the site or were just amphoras being used at a brick kiln is a question requiring further study. But this could have been one of the sites at which the "Cosan" pieces of Type 12 were made. At any rate, the finds of Spanish varieties of Types 12b and 16 and of pieces of the imported jars of Types 18, 20, and 21 in the port (about Types 16, 18, 20, and 21, see below) seem to point to a small-scale revival of the area as early as the late first century A.D., and certainly during the second and third centuries, a revival that brought at least a measure of activity back to a commercial center that had been largely dormant since the demise of the Sestius factory at some time after the middle of the first century B.C. FORM a

A260. (PC 69-118). (Figs. IX-397, 398). Unstamped. Lagoon, Spring House, mud fill around basin platform. 4. VIII. 69. Jar fragment, preserving much of neck and upper belly, shoulder, one handle, and lower attachment of other handle. RH., 0.47 m; G.D., 0.325 m; H.W., 0.044 m; H.T., 0.025 m. Peach-buff clay (5YR 7/8); many

IX: THE ROMAN AMPHORAS

small black bits; buff surface. Superficial " X " scratched on upper neck. Line separates upper neck from lower. FORM b

A261. (PC 69-9). (Figs. IX-399, 400). Unstamped. Harbor, underwater, Trench D l , Level III. 4.VII.69. Neck, preserving rim (chipped), both handles, and much of shoulder. P. H., 0.22 m; R.H., 0.013 m; M.D., 0.087 m; R.D., 0.114 m; H. W., 0.048 m; H. T., 0.027 m. Pale, sandy rust clay (5YR 5/6); black and white bits; beige surface. Line separates neck from shoulder, which is slightly convex in profile. A262. (PC 69-15). (Fig. IX-401). Unstamped. Harbor, underwater, Trench D l , Level III. 7.VII.69. Fragment of rim and upper neck and stump of one handle. P.H., 0.08 m; R.H., 0.015 m; est. M.D., 0.10 m; est. R.D., 0.13 m; H.W., 0.051 m; H.T., 0.024 m. Rust clay (5YR 5/6), pinkish buff (5YR 6/6) toward surface; black and white bits. Marine deposits on surface. A263. (PC 69-107). (Fig. IX-402). Unstamped. Lagoon, Spring House, mud fill around basin platform. 30. VIII.69. Fragment of shoulder and upper belly, preserving lower attachment of one handle. P.H., ca. 0.13 m. Deep pinkish buff clay (2.5YR 6/6); black and white bits; beige surface. Mended. A264. (PC 69-150). (Fig. IX-403). Unstamped. Harbor, underwater, Trench D l , Level HI. 3.VII.69. Belly fragment. G.W., 0.175 m. Finegrained deep pinkish buff clay (2.5YR 6/6), lighter toward surface; black and white bits; buff surface. Very heavily pitched on inside of fragment. Yellowish deposit over pitch. A265. (PC 69-151). (Fig. IX-404). Unstamped. Harbor, underwater, Trench D l , Level III. 16. VII. 69. Belly fragment. G. W., 0.205 m. Pale rust clay (5YR 6/4); many white bits and some black and red inclusions. Very heavy reddish to black shiny pitch adheres to inside of fragment. Yellowish deposit over pitch. A266. (PC 69-192). Unstamped. Harbor, underwater, Trench D l , Level III. 16.VII.69. Worn handle fragment. H. W., 0.04 m as preserved; H.T., 0.027 m as preserved. Very coarse rust clay (5YR 5/6); black and white bits. FORM c

A267. (PC 72-54). (Figs. IX-405, 406). Unstamped. Lagoon, Trench 23, level of gray clay, 1.10 m below surface. 14.VII.72. Neck fragment, preserving rim, part of neck, stump of one handle, and

207

attachment of other handle. R H . , 0.184 m; R.H., 0.027 m; M.D., 0.075 m; R.D., 0.117 m; H.W., 0.05 m; H. T., 0.025 m. Coarse, sandy pinkish buff clay (5YR 6/6); black, white, and red bits. Mended. A268. (PC 72-132). (Fig. IX-407). Unstamped. Lagoon, Trench YZ, south extension, below preserved surface of Wall W. 21.VII.72. Preserved fragments include most of rim and neck, stumps of both handles, and tiny piece of base. P. H. of neck fragment, 0.155 m; R.H., 0.018 m; M.D., 0.067 m; R.D., 0.083 m; H.W., 0.043 m; H.T., 0.026 m; diam. at base, not determinable. Coarse pale beige clay (10YR 8/3); sandy gray surface. Mended. Flat base was apparently hollowed out on the bottom. Type 13 Only one piece of Type 13 has been discovered in the Portus Cosanus, a handle fragment described herewith, and a single rim fragment also occurs in the town of Cosa. The rarity of Type 13 in the Cosa area accords with the type's general infrequency in the western Mediterranean. Among very few citations might be mentioned some fragments, one of them stamped, at Ostia (Zevi 1967, figs. 2, 3; cf. Panella 1970, p. 142:11), a stamped rim from Ventimiglia (Lamboglia 1955, fig. 11: bottom row), and a stamped piece in the wall of amphoras at Carthage that is datable between 43 and 15 B.C. (CIL VIII.22637.19). Type 13 is, however, very frequent in the Adriatic and eastern Mediterranean areas, particularly at Athens, Delos, and Alexandria. The finds at Delos, in fact, appear to provide a terminus post quern for the type, and the half-dozen stamped examples that occur there must be among the last Roman imports to that island, which had all but ceased to function as a trading center by the middle of the first century B.C. (Will 1970, pp. 384, 386). We must date Type 13, then, in the last half of the first century B.C., a date confirmed by contexts at the Athenian Agora. As to its place of manufacture, Zevi (1967) suggested that the examples at Ostia might be Istrian, since they resemble to some extent the amphoras of Dressel's Form 6 (Type 14 below), which are thought to have been made in Istria and perhaps elsewhere. The clay of Type 13, however, a distinctively hard fabric, usually bright peach-buff in color, with a shiny cream-colored surface, is quite unlike the typically powdery, pale buff clay of Type 14. And there are other significant differences between the two types. At present we lack information

208

THE ROMAN AMPHORAS

about the place or places of origin of Type 13, though, like Type 10, it may have been an Apulian shipping jar for olive oil, or an Istrian or northern Adriatic product from an area different from the center or centers that manufactured Type 14. A269. (PC 68-67). (Fig. IX-408). Unstamped. Harbor, Wall B, west face. 28. VI. 68. Handle fragment. L., 0.065 m. Other dimensions are not determinable owing to damage. Peach-buff clay (5YR 7/8), buff (5YR 7/3) toward surface; scattered black, white, and reddish bits. Type 14 While the shortage of examples of Type 13 in the Cosa area may parallel the general infrequency of that type in the west, the almost total absence from Cosa of Type 14 does not reflect a similar situation. The single piece of Type 14 from the Portus Cosanus, a neck fragment described herewith, is the only one so far discovered in the entire vicinity, a reflection perhaps of sufficient local production of olive oil, which was almost certainly the chief product shipped in Type 14, or perhaps of some kind of economic malaise in the area in the latter part of the first century B.C. and the first half of the first century A.D. Type 14 is, in any case, well represented in Italy, especially in Rome and in Cisalpine Gaul, as it is in the east. There can be no doubt about the identification of the neck fragment from the port. The thickness of the fabric of the rim, the narrow diameter of the mouth, the sloping attachment of rim to neck, the fine clay: all are characteristic features of the tall, long-necked, baggy-bellied, long-toed amphoras that Dressel made his Form 6. These jars, which had a long period of use, can be dated as early as the late first century B.C. through most of the first century A. D. At the Athenian Agora, they occur in several deposits datable in the late first century B.C. or early first century A. D. (cf, for example, Robinson 1959, p. 86 and pi. 19: [M 14]), and they are very frequent at Corinth and at Carthage. The type continued to be manufactured at least as late as the third quarter of the first century A. D. The stamps frequently name prominent individuals of known date, even members of the imperial family. These matters will be fully discussed in my forthcoming Athenian Agora publication. Not only have we information about who manufactured Type 14, but we know also where the jars, or many of them, were made. In 1910, Anton Gnirs published a large deposit of amphoras of Type 14 that were found in

association with a kiln at Fasana near PoIa, Yugoslavia (Istrian peninsula; see especially Gnirs's first reports in Jahrbuch for Altertumskunde 4 [1910] pp. 7988 and 184-186). Other similar finds were made in the same area. It is logical to assume that these were jars made as shipping containers for the Istrian olive oil that Pliny tells us was the chief brand on the world market in his day, although Dressel cites some painted inscriptions on these jars that may refer to wine or mulsum. On Istrian jars, see also Zevi 1967, passim; and on the possibility that amphoras of Type 14 were manufactured elsewhere in Italy, notably in Cisalpine Gaul, see the arguments of Baldacci 1967-1968 [1969], especially the addenda on p. 49. A270. (PC 72-29). (Fig. IX-409). Unstamped. Lagoon, Spring House, fill around Wall e. 7. VII. 72. Neck fragment, preserving rim (chipped) and upper part of neck. P.H., 0.145 m; R.H., 0.041 m; M.D., 0.12 m; R.D., 0.17 m. Very fine pinkish buff clay (5YR 6/6); few, scattered red, dark, and white bits; lighter surface. Type 16 This type comprises several of the most important varieties of export amphoras used for shipping garum and other fish products from the Iberian peninsula, especially from Spain, to other parts of the Roman world. Included under Type 16 are several varieties of wide-mouthed, hollow-toed amphoras: Dressel's Forms 7-12, 14, 38, and 39 (the latter two first discussed as whole jars by Pelichet 1946, Form 46), and Beltran's Form II B. Only eight small and rather nondescript fragments assignable with any probability to Type 16 have so far been discovered in the Portus Cosanus, but the finds of several varieties of Type 16 from the town of Cosa are much more frequent, accounting for 14 percent of the amphora total there. It is likely that the discrepancy between the finds in the two adjacent areas reflects the port's decline in the last half of the first century B.C. The several known subforms of Type 16 occur abundantly at Roman sites and can be dated variously from the late first century B.C. into the last half of the second century A.D. (On the date, see Panella 1973, pp. 506-521, and Manacorda 1977 [Pompeii] pp. 122-129; the early date is confirmed by dated contexts at the Athenian Agora, G 8:1 and Q 13:1, of Augustan and early first century A.D. dates, respectively, which contained amphoras of Dressel's Forms 7 and 8, shapes that are among the

IX: THE ROMAN AMPHORAS

earliest of the subdivisions of Type 16.) The Portus Cosanus was apparently largely dormant during the period covered by Type 16, except during its earliest and latest phases, and most of the Spanish garum and other fish products that reached the town of Cosa must, therefore, have entered via some other harbor, perhaps Port'Ercole (Portus Herculis) on the Argentario peninsula opposite Cosa. (See McCann, chapter three, for discussion of date of use of Portus Herculis.) Type 16 and its subcategories will be addressed in more detail in my forthcoming Athenian Agora and Cosa publications. To comment here on the fragments from the Portus Cosanus, three uncatalogued pieces (two handles with depressions down the center, and the bottom of a hollow toe) seem to belong to Beltran's Form II B, a variety of Type 16 characterized, like Dressel's Forms 7-12, by a distinctively fragile, powdery to plastery clay that varies from a pale greenish buff (Munsell 2.5Y 8/2) on the exterior surface to a peach-buff core (5YR 7/8) in some examples. This same clay is a feature also of the South Spanish jars, described above, of Type 5, though not of the Tarraconese and Cosan examples of that type; indeed, as I have explained, the South Spanish examples of Type 5 apparently came from the same kilns as many of the pieces of Type 16. The three Spanish fragments of Type 16 from the port, if they are Beltran II B, could date from the latter first or second centuries A. D., and perhaps buttress other evidence of a small-scale revival of activity in the port in the second and third centuries A. D. (see under Type 12 above and Types 18, 20, and 21 below). But in addition to the Spanish fragments of Type 16, five other Type 16 pieces of typically "Sestius" clay have been found in the port, the fragment described herewith (Cat. A271, a piece of a small jar with short neck, rim vertical in profile, and small, ridged [?] handles), as well as four other uncatalogued fragments (a neck, a shoulder, and two ridged handles). A few Type 16 pieces from the town of Cosa show the same clay, and have Augustan (?), vertical rims similar to that of Cat. A271, though most of the Type 16 examples from the town clearly come from Spanish jars. It is arguable that the "Sestius" finds of Type 16 from the port and from the town are local imitations of Spanish garum jars, as we have suggested was the case with the earliest local amphoras of Type 5. Reference was made above under Types 4 and 12 to Peacock's 1977 article (pp. 266-268), in which he describes the possible finding of a kiln site at Albinia, a few miles north of Cosa. Among his finds was

209

a ridged and grooved handle (fig. 3:14) resembling the handles of the local (?) pieces of Type 16 here described. If those pieces are in fact local, and if Peacock's find is a kiln, we may have evidence of a small-scale pottery operation, a revival of the Republican factory at a date as early as the Augustan Age. The Type 16 finds of Sestius clay might then represent an effort to restore the Type 5 and Type 24a garum industries that seem to have failed by about the middle of the first century B.C., just as we saw may have occurred, but on a larger scale, with the manufacture of Type 12b and the revival, probably in the latter first century A.D., of the manufacture of wine jars. The Type 16 pieces antedate those of Type 12b, but the two groups together, while there is no evidence of manufacture in the port, seem to point to a revival in the Cosa area of the export industries that existed in the port in the later Republic. But neither revival was to be of long duration or on anything like the scale of the Sestius firm, which had probably moved to Rome in the last half of the first century B.C. A271. (PC 72-266). (Fig. IX-410). Unstamped. Lagoon, Spring House, Room 2, white clay fill above wood level. 1. VIII. 72. Rim fragment, piece of upper neck, and stump of one handle. P. H., 0.111 m; R.H., 0.036 m; est. M.D., 0.14 m; est. R.D., 0.16 m; H.W., 0.025 m; H.T., 0.038 m (handle measurements taken near upper attachment). Coarse pale pinkish buff clay (5YR 7/4); black, white, and red bits; grayish buff surface. Type 18 This shape of amphora, which is similar to Dressel's Form 30, comprises a very widespread group of small, flat-bottomed jars with broad, short, channeled handles that are thin in section and semicircular in profile, short neck, and narrow rim. Frequent stamps on the type identify at least one place of manufacture as Tubusuctu in the province of Mauretania Caesariensis (Algeria; see most recently the detailed discussion in Laporte 1980, passim). But another area that produced Type 18 is southern France, where, as of 1978, about ten factories had been identified (Tchernia and Villa 1977; Laubenheimer 1977; Widemann et al. 1978, where the type is referred to as "Gauloise 4"). The two varieties of Type 18 show different fabrics, as might be expected, that of the French jars being generally fine-textured and beige, and that of the African jars being generally coarser and darker. Before the publication of the French

210

THE ROMAN AMPHORAS

kilns, Panella (1973, pp. 538-551, 600-605, a very thorough and useful study) had called attention to the two fabrics and to certain typological variations. In actual practice, seeing the difference between the two fabrics, and even details of shape, is difficult (cf. Laporte 1980, p. 134), as there is such a wide range of possible variations; however, scientific tests will doubtless refine the distinctions and permit exact identification of pieces found at sites other than production centers. Whether the African amphoras of Type 18, referred to here as Type 18b, are copies of the French (Type 18a) or vice versa is not yet clear, since the chronologies of the two varieties are still unsettled. Mauretania Caesariensis became a province in A.D. 40, a fact that provides a terminus post quern for Type 18b. Dated deposits at the Athenian Agora point to a lifespan extending from the latter first to the latter second centuries A.D. for Type 18a and to a date in the second century A. D. for Type 18b (as well as to a date in the third century for jars that seem to be developments of Type 18b. Those later amphoras, Type 18c, will be described in my Athenian Agora publication). In France, Type 18a is as early as the Flavian period (Widemann et al. 1978, p. 329), as it is closely associated in one kiln at Salleles with South Gaulish terra sigillata of Flavian date. Panella 1973, pp. 541-542, notes dates at Ostia for Type 18a ranging from the Flavian period to the fourth century A.D. and summarizes finds at other sites that show a similarly wide spectrum of dates; for Type 18b, the dates at Ostia seem to be largely in the third century A.D. (Panella 1973, p. 603). Finds of Type 18 (a and b[?]) in or near tombs at Meroe in the Sudan have been given dates that are perhaps too diverse to be helpful (see Laporte 1980, pp. 153-155), but we know at least that Tubusuctu became part of Mauretania Sitifensis in connection with Diocletian's reforms at the end of the third century. The late third century thus becomes a terminus ante quern for Type 18b. If the third century occurrences turn out to be later developments of Type 18b, as seems to me at present likely, the floruit for the type as a whole may well be the second century, with early examples dating from the latter first century. In that case, Type 18a would antedate Type 18b, and the jar type would have originated in France, not in Africa. As to the contents of Type 18, the evidence from France, as discussed in Widemann et al. 1978, p. 329, points to wine and garum as possible products shipped in jars originating there. Tchernia 1980, p. 306, discusses new evidence in favor of their being containers for wine. For the Algerian jars, however, olive oil might seem a likely product, owing to its

extensive production in the area around Tubusuctu in ancient and in modern times and in view of the inland position of the town, which would seem to argue against garum (Laporte 1980, pp. 140-143). Lequement 1980, p. 190, however, presents strong arguments in favor of wine as the product shipped in Type 18b. It may have been the imitation Falernian described by Galen (on which, see Tchernia 1980, pp. 308-310). Perhaps the question of the contents of Type 18 can be settled by the use of gas chromatography (see discussion above under Type 10). In the Portus Cosanus, I have identified seventeen examples of Type 18a, and they are discussed below as Form a. The catalogued group includes eight bases and two rims. In addition, four bases, two handles, and a rim are uncatalogued. Seventeen examples of Type 18b include the four pieces described below (a neck fragment and three bases), as well as thirteen uncatalogued pieces (eight handles, four bases, and a rim). As I have suggested above, the imports of Type 18, like those of Types 20 and 21, and the Spanish pieces of Types 12 and 16, imply that the port saw at least a mild resurgence of activity in the second and third centuries A.D. FORM a

A272. (PC 72-52). (Figs. IX-411, 412, 413). Stamped: M,N{ (incuse). Lagoon, Trench 23, level of gray clay, ca. 1.00 m below surface. 13. VII. 72. Base and small piece of lower belly, the stamp broken and incuse and set vertically on belly near base. P. H., 0.07 m; diam. at base, 0.08 m. Fine tannish buff clay (7.5YR 7/4); gray deposit on surface. Scratches on bottom of fragment. A273. (PC 68-58). (Fig. IX-414). Unstamped. Harbor, underwater, under rocks in Breakwater Extension D. 21. VII.68. Fragment of base, preserving piece of lower belly. P.H., 0.17 m; est. diam. at base, 0.12 m. Mauve (10R 6/3) to pinkish buff clay (5YR 6/6), lighter toward surface; few dark and reddish bits. Mended. Heavy, dark incrustation on inside of fragment. Marine deposit on exterior. An unusually wide base for Type 18a. A274. (PC 72-64). (Fig. IX-415). Unstamped. Lagoon, Spring House, rubble level south of Wall a. 14.VII.72. Nonjoining rim fragments and small piece of neck. P. H., 0.064 m; R. H., 0.023 m; est. M.D., 0.095 m; est. R.D., 0.117 m. Rather fine, pale rust clay (5YR 6/4); scattered black, white, and red bits; yellow-cream surface. Mended. A275. (PC 72-93). (Figs. IX-416, 417). Unstamped. Lagoon, Trench 23, level of gray clay 0.75 m below surface. 13. VII.72. Base, preserving pieces of

IX: THE ROMAN AMPHORAS

lower belly. RH., 0.152 m; diam. at base, 0.075 m. Fine peach-buff clay (5YR 7/8); few bits. Three holes punched through the wall of the fragment near the bottom (for reuse ofjar as fountain, probably). A276. (PC 72-133). (Fig. IX-418). Unstamped. Lagoon, Trench YZ, south extension, below preserved surface of Wall W. 21. VII. 72. Base, preserving piece of lower belly. P. H., 0.128 m; diam. at base, 0.055 m. Fine, powdery pale beige clay (10YR 8/3); few tiny dark bits; dark gray deposit on surface and breaks, but not on inside of fragment. Depression in base on bottom, outside, becomes convexity inside. Traces of pitch inside fragment. An unusually narrow base for Type 18a. A277. (PC 72-151). (Fig. IX-419). Unstamped. Lagoon, Trench 23, level of gray clay, 1.00 m below surface. 14. VII. 72. Base, preserving piece of lower belly. P.H., 0.19 m; diam. at base, 0.092 m. Fine beige clay (7.5YR 8/4); few bits; grayish black deposit on surface. A278. (PC 72-177). (Fig. IX-420). Unstamped. Lagoon, Trench YZ, south extension below preserved surface of Wall W in layer of dark mud, 0.55 m b.s.l. and below. 21. VII.72. Base, preserving pieces of lower belly. P. H., 0.142 m; diam. at base, 0.087 m. Fine pinkish buff clay (5YR 6/6); few bits. Blackish deposit, apparently pitch, in bottom. Mended. A279. (PC 72-232). (Fig. IX-421). Unstamped. Lagoon, Trench YZ, south extension, level of rubble fill from break in Wall W. 23. VII. 72. Rim fragment and small piece of neck. P. H., 0.056 m; R.H., 0.022 m; est. M.D., 0.10 m; est. R.D., 0.13 m. Fine pale tannish buff clay (7.5YR 7/4); heavy grayish deposit on surface, including breaks. A280. (PC 72-265). (Fig. IX-422). Unstamped. Lagoon, Trench YZ3, Structure YZ, silt level. 27. VII.72. Base, preserving piece of lower belly. R H . , 0.118 m; diam. at base, 0.08 m. Powdery peach-buff clay (5YR 7/8), lighter toward surface; beige surface. Base hollowed out to form shallow "false bottom." A281. (PC 72-349). (Fig. IX-423). Unstamped. Harbor, beach probe 5, level of Villa walls. 4. VIII.72. Base, preserving part of lower belly. P. H., 0.20 m; diam. at base, 0.10 m. Fine beige clay (7.5YR 8/4); few bits. FORM b

A282. (PC 69-1). (Fig. IX-424). Unstamped. Harbor, underwater, Breakwater Extension D, sur-

211

face. 1. VII.69. Neck fragment, preserving parts of rim, neck, and shoulder, and one handle. P. H., 0.137 m; R.H., 0.022 m; est. M.D., 0.10 m; est. R.D., 0.12 m; H.W., ca. 0.039 m; H.T., ca. 0.022 m. Fine dark gray clay (5YR 5/1), tannish buff (7.5YR 7/4) toward surface; few bits, if any, visible. Surface badly encrusted with marine deposits. A283. (PC 69-16). (Fig. IX-425). Unstamped. Harbor, underwater, Trench D 1, Level III. 7.VII.69. Base, preserving piece of lower belly. P. H., 0.14 m; diam. at base, 0.08 m. Striped grayish (5YR 6/ 1) and rust (5YR 5/6) clay; black, and some white, bits; gray surface. Shallow depression in center of base on outside. A284. (PC 69-135). (Fig. IX-426). Unstamped. Lagoon, Spring House, fill around basin platform, below platform level. 31. VII. 69. Base, preserving lower belly. R H . , 0.22 m; G.D., 0.28 m; diam. at base, 0.115 m. Sandy pinkish yellow-buff clay (5YR 6/8). Mended. A285. (PC 72-147). (Fig. IX-427). Unstamped. Lagoon, Trench YZ, south extension, below preserved surface of Wall W. 21.VII.72. Base, preserving piece of lower belly. P. H., 0.127 m; diam. at base, 0.105 m. Coarse rust clay (5YR 5/6); gray surface. Remnants of burned material inside fragment.

Type 20 This shape of amphora, the large globular jar of Dressel's Form 20, has been more thoroughly studied than any other category of shipping jar from the Roman period. It was a transport container for olive oil, and discoveries of workshops where Type 20 was manufactured along the Guadalquivir River in southern Spain were announced as early as the late nineteenth century. Those pottery centers were stimulated by the worldwide demand for Spanish olive oil, especially in the second and third centuries A.D., which was the floruit of Type 20, though finds of the type are datable as early as the second quarter of the first century A.D. and as late as the fourth century A.D. The city of Rome needed Spanish oil for the dole, as Monte Testaccio and finds of Type 20 all over the city attest. What more logical place to provide such oil than Baetica, of which both Trajan and Hadrian were natives? But demand for Spanish oil in the Empire extended even to the eastern Mediterranean, where dozens of stamped fragments of Type 20 have been found (these are discussed by me in 1983-1 and in the Athenian Agora volume).

212

THE ROMAN AMPHORAS

The Cosa area also imported Spanish oil in amphoras of Type 20. Three fragments from the portf are described herewith (two necks and a rim), and seven uncatalogued pieces include five bases (with attached body pieces), a neck, and body fragments of another piece. The type is equally well represented at Cosa itself, and its presence there and in the port should be thought of as reflecting not only a need for oil during the Empire but also, in the case of the port, the small-scale revival that occurred there after the period of suspended animation that apparently ended in the late first century A.D. The masses of information about Type 20 that are provided by Monte Testaccio and by extensive finds elsewhere are only beginning to be absorbed, in spite of the attention that this category of amphora has already received. A flowering of recent publications on Type 20 reflects that fact and also suggests the diverse areas of economic history and of archaeology on which the study of amphoras can throw light. Among useful commentaries on Type 20 and related matters during the recent past might be mentioned Beltran 1970, pp. 464-492; Panella 1973, pp. 522535, 627; Rodriguez-Almeida 1972, 1974-1975 [1977], 1978-1979 [1981], 1980-1, 1980-2, all passim; Broughton 1972, 1974, and 1980, all passim; Cambi 1976, passim; Manacorda 1977 (Ostia), especially pp. 134-137, 277, 365; Manacorda 1977 (Pompeii), p. 131; Colls et al. 1977 and Colls and Lequement 1980, passim; Ponsich 1974, 1979, passim; Remesal Rodriguez 1977-1978, 1980, passim; and Blazquez Martinez 1980, a rich compendium of recent research, concerned chiefly with Type 20, with contributions by scholars taking part in the First International Congress on Olive Oil in Antiquity (Blech, Blazquez Martinez, Ponsich, Rodriguez-Almeida, Remesal Rodriguez, Tchneria, Liou, Colls, Lequement, Beltran, Levi, Pascual Guasch, Mariner, Melena, Ramirez Sadaba, Garcia Moreno). Publication of the results of the Second International Congress, which took place in 1982, occurred in late 1983. In addition to continued study of Type 20 itself, future scholarly research will need to address the question of the ancestors and the descendants of Type 20 (on this matter, see Will 1977, pp. 295-296; Beltran 1980, pp. 191-192; Panella 1973, p. 533). The ancestry of Type 20 will, however, not be clarified until the confusion between Types 11a and l i b (see above) has been dispelled, a clarification that I will undertake in future publications. A286. (PC 69-31). (Fig. IX-428). Unstamped. Harbor, underwater, Breakwater Extension A, surface. 21.VII.69. Neck fragment, preserving over

half of rim, stump of one handle, and bit of shoulder. P.H., 0.143 m; R.H., 0.03 m; M.D., 0.097 m; R.D., 0.168 m. Firm, hard brownish clay (7.5YR 5/4), with a grayish layer toward surface; few black and white bits; shiny brown interior. Heavily barnacled inside and out. A287. (PC 72-249). (Figs. IX-429, 430). Unstamped. Lagoon, Trench YZ3, Structure YZ, silt level. 27. VII.72. Rim, entire except for chips, and small pieces of neck. P.H., 0.07 m; R.H., ca. 0.062 m; M.D., 0.095 m; R.D., 0.185 m. Sandy beige clay (7.5YR 8/4). Surface obscured by deposit of mortar. A288. (PC 72-285). (Fig. IX-431). Unstamped. Lagoon, Spring House, fill below rubble south of Wall a. 2. VIII.72. Neck, entire, preserving part of rim, much of shoulder, and both handles. P.H., 0.21 m; R.H., 0.042 m; M.D., 0.085 m; R.D., 0.15 m; H. W., 0.046 m; H. T., 0.041 m. Coarse tannish buff clay (7.5YR 7/4); yellow-buff surface. Mended. Graffito on one side of neck, slightly to left of center: ~J£ Two lines incised to the left of that graffito, near the upper attachment of one handle. Type 21 Zevi and Tchernia (1969, passim) made the first useful distinction between the two chief classes of long, thin, cylindrical jars that are here referred to as Type 21. These amphoras can be assigned a floruit of the middle of the third century A. D. on the basis of their frequent occurrence in several Herulian destruction contexts at the Athenian Agora (for one example, cf. Robinson 1959, p. 69 and pi. 36: [K 116]), though the type seemingly goes back to the second century. Examples found in Algeria were reused as baby coffins and found in association with a coin of Marcus Aurelius (see Serge Lancel, Bulletin d'archeologie algerienne 4 [1970] p. 251; cf. fig. 98). Type 21 divides into three probable categories, as well as into various subsidiary groups. All the categories are represented in the Portus Cosanus, a further sign of the small but genuine revival of that area in the second and third centuries A.D. The two chief classes of Type 21 are a small variety (frequently called "African I," and referred to here as Type 21a) and a large variety ("African II," or Type 21b). The two shapes are contemporary and seem to have been manufactured at the same centers in modern Tunisia, especially along the coast of Byzacium, though evidence of South Spanish manufacture also exists (Peman 1959). Zevi argues (Zevi and Tchernia 1969, pp. 185-187) that Type 21b was a container for olive oil, whereas Type 21a was used for garum; cf. Pa-

IX: THE ROMAN AMPHORAS

nella 1973, pp. 579, 585-586. Manacorda 1977 (Ostia), pp. 159, 170, however, seems unsure both about garum and about oil as contents of these jars, and Lequement 1980, p. 191, citing the presence of pitch on the inner walls of numerous fragments of Type 21, suggests that some examples of the type might have transported garum or wine. In the same discussion, Manacorda (p. 168) also cites evidence in support of a fourth century A.D. floruit for Type 21b. But, as I shall argue elsewhere, the bulk of the securely dated evidence points to the third century as the chief period of use of Types 21a and b, the most popular forms of the category. That these types continued in use into the fourth century seems entirely possible. In the Portus Cosanus, Type 21a is represented by ten fragments, of which two (a rim and a toe) are described herewith. In addition, six handles, a toe, and a belly fragment are uncatalogued. Three pieces belonging to Type 21b are also described below, a neck fragment and two toes. Three toes and a belly fragment are uncatalogued. The spatheion, the sword-shaped amphora which occurs so often in contexts of the fifth century A.D., is represented in the port by a single underwater find, from the Harbor. A few examples have also been found in the town of Cosa. On this shape, see Grace 1961, figs. 67-68; Panella 1972, pp. 105-106; Manacorda 1977 (Ostia), pp. 211-225, 281-282. It does not seem possible to deduce any chronological conclusions about activity in the port in the late Empire from this single piece. Type 21c, to which I assign the spatheion category, is considered North African by Beltran 1970, n. 1374 on p. 571 (his Form 65, B), and the clay does resemble that of Type 21 a and b, of which Type 21c may perhaps be a descendant, but the single piece here described does not permit us to conclude that the port was receiving goods from Africa at this late date. The other occurrences of Type 21, however, buttress the evidence of some activity in the Portus Cosanus as late as the third century A. D.

213

A290. (PC 72-243). (Figs. IX-434, 435). Unstamped. Lagoon, Spring House, fill in Room 1. 24. VII. 72. Rim fragment and small piece of neck. P.H., 0.075 m; R.H., 0.039 m; est. M.D., 0.10 m; est. R.D., 0.13 m. Rust clay (5YR 5/6); many white bits; buff surface. FORM b

A291. (PC 68-57). (Figs. IX-436, 437). Unstamped. Harbor, underwater, Breakwater Extension C. 4. VII. 68. Neck fragment, preserving rim, part of neck, and one handle. P. H., 0.193 m; R. H., 0.037 m; M.D., 0.08 m; R.D., 0.103 m; H.W., 0.034 m; H.T., 0.018 m. Purplish gray clay (10R 4/2), tannish (7.5YR 7/4) toward surface. Exterior covered with marine deposit. Mended. A292. (PC 72-94). (Fig. IX-438). Unstamped. Lagoon, Trench YZ, north, 1.50 m below surface, above pavement of small stones. 14. VII.72. Toe. P.H., 0.16 m; G.D., 0.061 m. Rust clay (5YR 5/ 6); black and white bits; gray surface, which becomes orange at tip of toe. A293. (PC 72-267). (Fig. IX-439). Unstamped. Lagoon, Trench YZ, south extension, level of wooden basin and rubble fill from break in Wall W. 20. VII.72. Toe and small pieces of lower belly. P.H., 0.138 m; G.D., 0.056 m. Coarse rust clay (5YR 5/6). Mortar on surface. "Combed" effect on toe. FORM c

A294. (PC 69-37). (Fig. IX-440). Unstamped. Harbor, underwater, Trench D2, Level II. 16.VII.69. Neck fragment, preserving rim (chipped), much of upper neck, and stumps of both handles. P.H., 0.145 m; R.H., 0.047 m (height of upper lip, 0.021 m); M.D., 0.10 m; R.D., 0.135 m; H.W., 0.04 m; H.T., 0.024 m. Peach-buff clay (5YR 11 8); black and white bits; yellow-buff surface. Incision marks separation between lower part of rim and neck. Marine deposit on surface.

FORM a

Type 24a

A289. (PC 72-43). (Figs. IX-432, 433). Unstamped. Lagoon, Trench YZ, north, 1.60 m below surface, above pavement of small stones. 13. VII.72. Toe and lower part of belly. P.H., 0.26 m; G.D., 0.268 m. Coarse rust clay (5YR 5/6), dark gray (5YR 5/1) near surface; many tiny black, and especially white, bits; lighter yellowish gray surface on belly, but surface of toe is pinkish buff. "Combed" effect on lower part of fragment. Mended.

This type of amphora, apparently related to Dressel's Form 28, is well represented at Cosa, especially in the town itself, where nineteen pieces had been found by 1978. In the port, there are three rim fragments, described herewith, from the Spring House, and perhaps also a base, Cat. A301 under Miscellaneous and Unclassified Fragments. Amphoras of Type 24 are small (about 60 cm in height). The examples from the Cosa area, all fragmentary, are probably earlier than Dressel 28 proper and other

214

THE ROMAN AMPHORAS

similar pieces. The Cosa fragments have high, molded rims, flat on top, that are concave in profile and thickened at top and bottom. The handles, semicircular in profile, deeply channeled, and thin in section, are attached to the neck immediately under, and touching, the rim. The neck is short and rather narrow. An incised line marks the separation between neck and shoulder. So much the Cosa pieces tell us. It seems likely that the jars from which they came resembled Dressel 28 in having a wide shoulder and ovoid belly. Unlike Dressel's shape, which is flat-bottomed, they may have rested on a flaring ring foot. The walls of Type 24a are thin, and the coarse clay is closely similar to that of the Sestius jars. Type 24 occurs at a variety of sites throughout the western sphere of Roman influence. It is frequent in early Augustan sites, such as Oberaden, Haltern, and the wall of amphoras at Carthage, datable between 43 and 15 B.C. (Loeschcke 1942, pp. 76-80; CIL VIII.22637), but the life span of the type was not limited to the last half of the first century B.C. The oldest examples of the type, which we here designate Type 24a, were found on the Albenga wreck (Lamboglia 1952, p. 166 [a neck fragment and three fragments of bases]; 1955, pp. 265-266), now generally dated in the second quarter of the first century B.C. Later examples are in contexts of the last half of the first century A.D. at Ostia and Pompeii (Panella 1970, pp. 118-119; 1973, pp. 535-537), at Herculaneum (on display in 1978 in the House of the Bicentenary), and seemingly in the Colosseum (Incitti 1977, third page of chapter on amphoras), though the Colosseum material may, like certain pieces reported from Ostia, date as late as the second century A.D. A type that extends in date from the second quarter of the first century B.C. into the second century A.D. clearly underwent major changes in shape, and these are evident in the pieces preserved to us. The Portus Cosanus and Cosa fragments correspond to the earliest known examples, those from the Albenga wreck. Their delicate lines, high, molded rims, and thin walls indicate that they should be dated, like the pieces on the wreck, in the second quarter of the first century B.C. Zevi (1966, pp. 225-226) suggested a Spanish origin for this type of amphora on the basis of painted inscriptions reported by Dressel on his Form 28. Most subsequent commentaries on Dressel 28 follow Zevi (Baldacci 1969 [1972] pp. 43-44; Beltran 1970, pp. 497-502; Tchernia 1971, pp. 64-65; Panella 1970, pp. 118-119, and 1973, pp. 535-537, to name representative ones). A Spanish origin is also suggested

by the occurrence of this type of amphora on the Port-Vendres II wreck [Colls et al. 1977]), though the finding of similar jars in kilns in southern France leads one to envision the possibility of a wider area of production (on the French jars, Laubenheimer 1977, passim, and Widemann et al. 1978, pp. 318325). The western end of the Mediterranean was very likely a major source for Type 24; but the frequency of Type 24a in the Cosa area, the similarity of the clay to that of the Sestius jars, and other factors such as the finding in the town of Cosa of two misshapen, misfired rims of Type 24a (nos. U.XXII.3 and U.XXII.4) suggest that Cosa might have been another, and perhaps the first, place in which this type was manufactured. If a garum industry developed in the Portus Cosanus during the late Republic (McCann 1979, especially pp. 40IfF.; Will 1979, n. 26; and cf. under Type 5 above), Type 24a may have been one of the varieties of containers manufactured for its export, or for the export of some other fish product. (For other possibly experimental amphora shapes, see under Cats. A302-A304 below.) The evidence in favor of Spanish or French manufacture of later examples of the type does not conflict with that suggestion, as the testimony presented above under Type 12 demonstrates. As in the case of Type 12b, provincial manufacturers may simply have copied an earlier Italian form. A further indication that Type 24a may have originated in Cosa is contained in an amphora stamp that occurs repeatedly on Augustan examples of the type. The stamp is SEX DOMITI (to which are doubtless related such variants as SEX.DOM., SEX.D, SD, and others; the chief occurrences are summarized by Callender [no. 1602, where, however, the stamp at Oberaden is incorrectly stated to be on Type 4b, "Dressel Form 1"] and by Beltran 1970 and Tchernia 1971 on the pages cited above). Such stamps are generally interpreted as referring to a manufacturer or potter, Sextus Domitius, but one wonders whether they might not also refer to the Sestius factory, and perhaps to a late period in the history of the factory, when it might have been either controlled or jointly controlled by the Domitii Ahenobarbi, who also, like the Sestii, owned estates at Cosa and whose political star was in the ascendant as that of the Sestii declined (on the Domitii Ahenobarbi at Cosa, see, most recently, Brown 1980, p. 73 and n. 11 to chapter five, and Manacorda 1980, especially pp. 178ff. On the shipping interests of the Domitii Ahenobarbi, see P. A. Gianfrotta, "Ancore 'romane.' Nuovi materiali per Io studio dei traffici marittimi," MAAR 36 [1980] p. 111). Some of the tegulae man-

IX: THE ROMAN AMPHORAS

ufactured by L. Sestius Quirinus (Quirinalis), the son of P. Sestius, spelled his name "Sextius" (CIL XV. 1445c, 1446a,b; cf. Will 1979, p. 348, n. 29 on the "L. SEX" amphora stamps. Those are also discussed by Manacorda 1980, p. 175, who notes too on p. 177 a new brick stamp from Luni naming a son (?) of L. Sestius as P. Sext[ius] Quiri[nus] [or Quirinalis]). An interesting additional observation concerns certain Arretine pieces, where the stamps of a Sestius and of P. Domitius of Arezzo occur in lozenges of similar shape (Oxe and Comfort, 607e; 1793b; 1796i.m.p), though the similarity might be just coincidence. The possibility that the Sestius amphora-holdings diversified into bricks, tiles, and Arretine has already been suggested above, in my introductory text. That the Sestius holdings at Cosa and elsewhere might have been gradually taken over by the Domitii and perhaps formed the germ of the mammoth brick empire developed by their poor relation (?) Cn. Domitius Afer (Tac. Ann. 4.66: "diu egens") and his descendants, especially his granddaughter Domitia Lucilla the Younger, the mother of Marcus Aurelius, is a fascinating possibility suggested by this stamp. The SEX DOMITI series of amphora stamps does not, however, occur at Cosa; in fact, no stamps of any kind have so far been found on Type 24a. But at some time in the last half of the first century B.C. part of the Sestius (or SestiusDomitius) pottery corporation would have moved from the Portus Cosanus. The tegula factory of L. Sestius was almost certainly in Rome (CIL XV.539; cf. Margareta Steinby, "I bolli laterizi degli Antiquari del Foro e del Palatino," Atti della Accademia Nazionale dei Lincei. Classe di Scienze morali. Memorie 7, ser. 8, fasc. 3 [1974] p. 107; Will 1979, p. 349 and n. 35). But in 23 B.C., Horace (Odes 1.4) seems to imply that pottery production is still being carried on at Cosa (Will 1982-2). As was suggested above in the introduction to the catalogue and in the discussion of amphoras of Type 4b, a reduced pottery operation may have been carried on in the Portus Cosanus through the third quarter of the first century B.C. After the end of the term of office of L. Sestius as consul suffectus in 23 B.C., his increased prestige and power may have kept him in Rome and may have contributed, along with geological causes, to the final demise of the Sestius (or Sestius-Domitius) factory in the Portus Cosanus. For further discussion of this decline, see above, Bourgeois, chapter two, and McCann, chapter eighteen.

A295. (PC 68-45). (Figs. IX-441, 442). Unstamped. Lagoon, Spring House, fill around basin platform.

215

11.VII.68. Neck fragment, preserving pieces of rim, neck, and one handle. P.H., 0.087 m; R.H., 0.036 m; est. M.D., 0.14 m; est. R.D., 0.16 m; H.W., 0.051 m; H.T., 0.017 m. Grayish rose clay (10R 6/2); black and white bits; gray surface. Mended. Incised line at base of neck. A296. (PC 69-154). Unstamped. Lagoon, Spring House, earth fill above basin platform. 30. VIL5. VIII.69. Rim fragment and small piece of neck. R H . , 0.058 m; R.H., 0.038 m; est. M.D., 0.12 m; est. R.D., 0.14 m. Pale beige clay (10YR 8/3); black and red bits; gray deposit on surface. A297. (PC 69-155). Unstamped. Lagoon, Spring House, earth fill above basin platform. 910.VII.69. Rim fragment and piece of neck. RH., 0.08 m; R.H., 0.045 m; est. M.D., ca. 0.14 m; est. R.D., ca. 0.17 m. Deep pinkish yellow-buff clay (5YR 5/8); black, red, and white bits; beige surface. Miscellaneous and Unclassified Fragments A298. (PC 68-55). (Fig. IX-443). Unstamped. Lagoon, Spring House, fill just above basin platform. 10. VII. 68. Toe fragment. P. H., 0.055 m; G.D. of upper part, 0.066 m; G.D. of lower part, 0.042 m. Hard peach-buff clay (5YR 7/8), mauve (10R 6/3) at core; conspicuous white and also dark and red bits; worn yellow-cream surface. Some "combing" visible on upper part of toe. Perhaps from an amphora similar to Beltran 1970, Forms 59-60. A299. (PC 68-5). (Fig. IX-444). Unstamped. Lagoon, Spring House, earth fill above basin platform. 15. VII.68. Belly fragment, preserving much of shoulder and part of belly. P. H., ca. 0.40 m; G.D., 0.34 m; est. diam. at base, 0.20 m. Sandy pale pinkish buff (5YR 7/4) to beige (7.5YR 8/4) clay; some large red bits; grayish deposit on surface. Mended. Interior pitched. Concentric lines incised around shoulder and base of neck. The ovoid, dropped-shoulder shape of this belly, the wide estimated diameter of the base, and the incision around the base of the neck suggest a series of containers that is perhaps to be associated with Dressel's Form 29. See discussion under Cat. A 300 below. A300. (PC 69-10). (Figs. IX-445, 446). Unstamped. Harbor, underwater, Trench D l , Level IH. 4. VII.69. Neck, preserving rim, both handles, and piece of shoulder. P.H., 0.217 m; R.H., 0.018 m; M.D., 0.068 m; R.D., 0.083 m; H.W., 0.029 m; H. T., 0.017 m. Sandpapery yellow-buff clay

216

THE ROMAN AMPHORAS

(7.5YR 7/6); scattered tiny dark bits. This piece, which is perhaps, like Cat. A299 above and Cat. A301 below, from a jar related to Dressel's Form 29, belongs in that case to a long-lived series of small, flat-bottomed shipping containers, which are perhaps Greek in origin. Early examples of the general type go back to the first century B.C. (Robinson 1959, F 72) and other pieces date as late as the middle of the third century A. D. (Robinson 1959, K 114). The type is well represented at Ostia. Cf. Palma and Panella 1968, pp. 99-100 and figs. 451-452; 1970, p. 105 and fig. 521; 1973, pp. 472-474 and figs. 369-370. Manacorda 1977 (Ostia) pp. 370-372 and figs. 440, 442, 632-634, suggests a date in the second half of the second century A.D. for the Ostia examples, and possible Adriatic origin for the type, on the basis of evidence that includes a wreck found off the coast of Yugoslavia. Panella proposes a date in the first decades of the second century (1973, p. 474). BaIdacci 1972, p. 28 and fig. 9, dates the type in the early Empire and implies Istrian origin. On the basis of the evidence at Ostia and at the Athenian Agora, the neck from the Portus Cosanus should be dated in the second century or the first half of the third century A.D. A301. (PC 69-14). (Fig. IX-447). Unstamped. Harbor, underwater, Trench D l , Level III. 7.VII.69. Fragment of base and lower belly. P. H., 0.096 m; est. diam. at base, 0.15 m. Fine tannish buff clay (7.5YR 7/4); scattered small dark and light bits; grayish deposit on surface. Mended. Concentric striations on lower belly. Heavy deposit, apparently of pitch, on interior of fragment. The wide estimated diameter of the base may suggest a connection with Dressel's Form 29 (see above under Cat. A300), but the thin walls of Cat. A301 may point instead to Dressel's Form 28 (see under Type 24a above). A302. (PC 69-110). (Fig. IX-448). Unstamped. Lagoon, Spring House, fill around basin platform. 1. VIII. 69. Rim fragment and piece of neck. P. H., 0.115 m; R.H., 0.014 m; est. M.D., 0.09 m; est. R.D., 0.11 m. Pinkish buff clay (5YR 6/6); white, black, and red bits; beige surface. Pitch-lined interior. The very low, narrow rim, molded into three layers, rests on a neck that apparently narrows toward the bottom. The dimensions of the fragment suggest that the amphora from which it came was small, and that it was also rather wide-

bellied and flat-bottomed, like later forms of Type 24a. Like Type 24a, this piece could have been a container for garum or some other fish product. Cat. A302 has, in fact, the typically "Sestius" clay of Type 24a. Only one other piece of the same type is known, a rim fragment (no. U.XIV.l) found in the town of Cosa. The two fragments may represent an experimental shape that was not ultimately produced in quantity. A303. (PC 72-284). (Figs. IX-449, 450, 451). Unstamped. Lagoon, Spring House, fill below rubble south of Wall a. 2.VIII.72. Neck, preserving most of rim, one handle, upper attachment of other handle, part of shoulder, and pieces of belly. P. H., 0.205 m; R.H., 0.016 m; M.D., 0.067 m; R.D., 0.098 m; H. W., 0.051 m; H. T., 0.032 m. Pinkish buff clay (5YR 6/6); black, white, and red bits; yellow-cream surface. Mended. Graffito at a slant on shoulder: JxL-t. This unique piece may, like Cat. A302 above, represent another "trial balloon" that was produced in the Portus Cosanus pottery only experimentally. Though the clay is typically "Sestius," the unusually narrow, low rim, bulging, "Chian" neck, and deeply channeled handles suggest a conscious effort to produce a wholly new shape, an effort perhaps undertaken contemporaneously with the development of Type 24a. The mystery of this piece is compounded by the Etruscan (?) " M " or " N " scratched on the shoulder. A304. (PC 72-305). (Figs. IX-452, 453). Unstamped. Lagoon, Spring House, immured in east face of north half of Wall c. 2. VIII. 72. Potterywaster, consisting of bases, stuck together, of three amphoras. H., 0.079 m; G.W., 0.115. Purplish brown clay (10R 3/3), black toward surface; many white bits; greenish surface. Although this set of fused bases cannot at present be associated with any known amphora shape, it may have a connection with one of the types like Cats. A302 or A303 above, the shapes of whose bases are not known. Only one other base comparable to Cat. A304 has been identified, an uncatalogued piece preserving sections of the lower belly and found in Trench IA of the lagoon. The fabric of that piece (coarse brownish clay with black and white bits and a yellow-cream surface) is also unlike that of any other known find from the port site or from Cosa itself.

IX: THE ROMAN AMPHORAS TABLE IX-I.

PORTUS COSANUS ROMAN AMPHORA TOTALS

Date

Type la Ic Id 4a 4b 4a/b 5 10 lie 12a 12b 12c 13 14 16 18a 18b 20 21a 21b 21c 24a Misc.

Catalogued

Uncatalogued

Totals

2 3 44 138 59 0 11 1 1 1 6 2 1 1 1 10 4 3 2 3 1 3 7

0 0 27 162 149 51 11 0 0 0 4 0 0 0 7 7 13 7 8 4 0 0 35

2 3 71 300 208 51 22 1 1 1 10 2 1 1 8 17 17 10 10 7 1 3 42

304

485

789

Late 4th-lst qu. 3rd B.C. ca. 200 B.C. 180s-150s B.C. Late 2nd-early or mid 1st B.C. 2nd-4th qu's. 1st B.C. — Late 2nd-mid 1st B.C. Late 2nd-mid 1st B.C. 1st half 1st A.D. 1st B.C.

1st, 2nd A.D. 1st, 2nd A.D. Last half 1st B.C. Late 1st B.c.-3rd qu. 1st A.D. Late 1st B.c.-2nd A.D. Late lst-2nd A.D. 2nd-3rd A.D. 2nd qu. lst-3rd A.D. 3rd A.D. 3rd A.D. 5 t h A.D.

2nd qu. 1st B.C. —

Total

NOTE: figures do not include the twenty-six pieces of Type 4a published in Manacorda 1978.

CONCORDANCE OF INVENTORY AND ROMAN AMPHORAS: CATALOGUE NUMBERS INV. NO. (PC)

68-1 68-2 68-3 68-4 68-5 68-6 68-7 68-8 68-9 68-16 68-26 68-27 68-33 68-34 68-35 68-36 68-37 68-38 68-39

CAT. NO. (A)

108 58 59 259 299 104 52 247 194 205 206 249 83 60 80 189 67 89 74

NV. NO. (PC)

68-40 68-41 68-42 68-43 68-45 68-51 68-52 68-55 68-57 68-58 68-59 68-60 68-61 68-63 68-64 68-65 68-66 68-67 68-Uncat.

CAT.

NO.

53 84 54 90 295 112 103 298 291 273 201 6 7 8 9 10 11 269 195

(A)

INV. NO. (PC)

69-1 69-6 69-9 69-10 69-12 69-13 69-14 69-15 69-16 69-20 69-27 69-31 69-32 69-34 69-37 69-43 69-44 69-56 69-57

CAT. NO

282 79 261 300 3 207 301 262 283 51 50 286 113 208 294 209 210 114 115

THE ROMAN AMPHORAS

218

CONCORDANCE OF INVENTORY AND ROMAN AMPHORAS:

INV. NO. (PC)

CAT. NO. (A)

69-59 69-71 69-73 69-74 69-92 69-95 69-97 69-98 69-107 69-110 69-118 69-127 69-135 69-150 69-151 69-152 69-153 69-154 69-155 69-156 69-157 69-158 69-159 69-160 69-161 69-162 69-163 69-164 69-165 69-166 69-167 69-168 69-169 69-170 69-171 69-172 69-173 69-174 69-175 69-176 69-177 69-178 69-179 69-180 69-181 69-182 69-183 69-184 69-185 69-186 69-187 69-188 69-189 69-190 69-191 69-192

116 211 117 12 118 105 250 119 263 302 260 191 284 264 265 109 190 296 297 13 14 15 16 17 18 19 20 21 22 23 24 25 85 91 61 92 75 93 94 81 248 55 68 69 62 76 86 63 70 4 5 1 26 27 96 266

INV. NO. (PC)

72-1 72-2 72-7 72-9 72-10 72-12 72-13 72-18 72-19 72-26 72-28 72-29 72-33 72-34 72-43 72-52 72-54 72-61 72-64 72-65 72-66 72-93 72-94 72-95 72-117 72-118 72-119 72-120 72-121 72-122 72-125 72-126 72-129 72-132 72-133 72-136 72-137 72-140 72-141 72-147 72-150 72-151 72-164 72-165 72-166 72-167 72-168 72-174 72-175 72-177 72-183 72-184 72-186 72-187 72-188 72-189

CAT. NO. (A)

193 251 203 212 120 213 252 121 28 122 123 270 124 253 289 272 267 29 274 125 254 275 292 214 215 30 31 126 32 127 128 33 97 268 276 216 255 129 130 285 192 277 131 217 256 132 218 133 219 278 134 135 136 137 138 139

INV. NO. (PC)

72-194 72-195 72-196 72-197 72-211 72-212 72-214 72-215 72-217 72-218 72-227 72-232 72-240 72-241 72-242 72-243 72-244 72-245 72-246 72-247 72-249 72-250 72-252 72-255 72-257 72-258 72-259 72-260 72-261 72-262 72-263 72-264 72-265 72-266 72-267 72-272 72-273 72-274 72-275 72-276 72-277 72-278 72-279 72-280 72-281 72-282 72-283 72-284 72-285 72-290 72-302 72-303 72-304 72-305 72-306 72-307

CAT. NO. (A)

140 141 142 220 143 144 145 110 146 147 221 279 222 223 148 290 149 95 150 151 287 152 153 224 154 155 34 156 157 158 35 159 280 271 293 225 226 227 228 229 230 231 232 233 234 235 236 303 288 160 161 162 163 304 164 165

IX: THE ROMAN AMPHORAS INV. NO. (PC)

72-308 72-309 72-311 72-314 72-315 72-316 72-317 72-318 72-319 72-320 72-321 72-323 72-324 72-325 72-326 72-327 72-328 72-329 72-330 72-332 72-333 72-334 72-335 72-336 72-337 72-338 72-339 72-340

CAT. NO. (A)

166 111 237 106 167 238 168 257 169 170 239 107 171 172 173 174 240 241 242 175 176 177 178 243 244 179 199 180

INV. NO. (PC)

72-342 72-343 72-344 72-345 72-346 72-347 72-349 72-360 72-361 72-362 72-363 72-364 72-365 72-366 72-367 72-383 72-425 72-427 72-788 bis 72-789 72-790 72-791 72-792 72-793 72-794 72-795 72-796 72-797

CAT. NO.

(A)

181 188 182 200 183 184 281 185 202 186 204 196 245 187 197 198 258 246 36 37 38 39 40 41 42 43 44 45

219 INV. NO. (PC)

72-798 72-799 72-800 72-806 72-807 72-808 72-809 72-810 72-811 72-812 72-813 72-814 72-Uncat. 72-Uncat.

CAT. NO.

46 47 48 87 71 64 56 98 100 102 2 49 99 101

73-1

88

75-1

77

76-1 76-2 76-3 76-4 76-5 76-6

82 57 65 72 73 66

78-1

78

(A)

THE ROMAN AMPHORAS ALPHABETICAL LIST OF LATIN AMPHORA STAMPS CATALOGUE STAMP

NUMBER

ALEXANDRI APOL HERMO[ Μ· χ (incuse) MN I M N (N's retr.) SETS device (anchor) SETS device (trident) SETS device (caduceus) SETS device (tree or vine­ -prop) SETS device (small 8pointed star) SES and device (lighthouse or lighted altar [?]) SEST device (palm branch) SEST device (double axe) SES uncertain device Device (small anchor) Device (large anchor) Device (double anchor or thunderbolt) Device (caduceus ?) Device (large palm branch or fish bone) Device (pine bough or fish spine) Device (maze [?] nautical hook [?] triangular harrow [?]) Device (compass card [?] wind rose [?] wheel [?]) Device (medallion or rosette ?) Device (star [?] winch [?] ship's wheel [?]) Device (ship's wheel or star) Device (palm branch ?) Device (animal [?] fish [?])

A247 A50 A51 A272 A52 A53-A57 A58-A66 A67-A73 A74-A75

DIO [ H[ JIN [ ]NG X or A (incuse) ScpA (S retr.) Illegible

A188 A189 A190 A191 A192 A193 Al 10-Al 11, A203-A204

A76-A78 A79 A80-A82, A248 A83-A88 A89-A94 A95 A96-A99 AlOO-AlOl A102 A103 A104-A107 A108 A109 A194-A198 A199 A200 A201 A202

NOTE: List does not include 26 port of Cosa stamps published by Manacorda 1978. See footnote 6 to Introduc­ tory Text, above. Note also that 7 of the 82 stamps listed above occur with one or more other stamps on the same fragment.

Part III. The Catalogues

Chapter X. Fine Wares

Roman Thin Walled Pottery ANNA MARGUERITE M C C A N N This class of Roman pottery 1 has received its name from its "great fragility of texture and no, or only partial conformity with the technical and stylistic aspects exhibited by contemporary ceramics of Hellenistic tradition." 2 What differentiates these small clay vessels "is the effort made to translate into clay also the thin, hard consistency of a metal sheet." 3 Indeed, the imitation of both bronze and silver prototypes constitutes one of the key sources for many of its forms and decorative techniques, which were developed during the long course of its life from the second quarter of the second century B.C. through the Claudian-Neronian period. The height of its production at Cosa, however, occurs during the third quarter of the first century B.C., when the greatest variety of forms appears. The cultural inspiration for thin walled ware comes from three major streams: the Iron Age Italian, represented in the Este or Golasecca civilizations; the La Tene culture from the Celts of central Europe; and the Hellenistic tradition. Turning away from the painted surfaces of vases from the classical period, the potters of thin walled ware prefer plain walls, largely uncoated. Beginning in the later Republic, barbotine decoration is popular and incision, rouletting, sand, coral designs and thumb impressions also form part of its decorative, unclassical vocabulary. The use of the human figure is rare. Coatings are sometimes applied to produce the black lacquered ware of the Republic, the metallic surfaces of the early Empire as well as the orange colors especially developed during the Tiberian period and continuing into the Claudian and Neronian. The term "coating" is employed here to distin-

guish this thin technique from that of the heavier black gloss of so-called black glaze wares or the redorange gloss of Arretine or later sigillata. Thin walled pottery is thus predominantly nonclassical and preclassical in appearance, a factor that argues for its original development in northern Italy. But it is found everywhere in Italy and must have been produced in many Italian centers.4 The wealth of thin walled ware at Cosa, both from the hill site and the port, strongly suggests a Tuscan center for its particular manufacture. The finds in the Iberian Peninsula and the Balearic Islands also show that it was widely exported. 5 The gray, thin walled ware produced in the sub-Alpine and Alpine regions and imported into Italy in early Imperial times further documents an active exchange between these areas and the Italian plains already before their conquest by Augustus in A.D. 14. In summary, the archaeological significance of this important class of pottery lies above all in its "documentation of Roman expansion northwards and the subjugation of the Celtic and Celticized populations in possession of the late vestiges of the Iron civilization."6

1 I would like particularly to thank Professor Maria Teresa Marabini Moevs for her expert and kind help in the study of this material. She first looked at the thin walled fragments from the port site with me at Cosa and has since then offered many helpful suggestions. While I accept full responsibility for the catalogue presented here, it could not have been completed without her book and her generously sharing with me on numerous occasions her vast knowledge about Roman pottery. In my catalogue, Cosa

numbers refer to Moevs's catalogue of the hill site material. 2 Moevs 1973, p. 35. 3 Ibid. 4 For example, see Lambogha, Albintimilium; Luni II; Ostia III; Carettoni. 5 Mayet; Almagro I. 6 Moevs 1973, p. 39.

The excavations of the hill town of Cosa established for the first time a chronology for thin walled pottery—previously recognized but never fully classified and dated. Thus, Maria Teresa Marabini Moevs's book, The Roman Thin Walled Pottery from Cosa (1948-1954), published in 1973, is both a pioneering effort and a basic contribution to the history of Roman pottery. I have adopted here her classifications and much of her descriptive vocabulary. She identified seventy-five different forms among

224

THE CATALOGUES

the material from the hill site, sixteen of which are represented at the port. In the several cases where our finds did not closely correspond to Moevs's forms, I have drawn upon those of Fra^oise Mayet, presented in her book, Les ceramiques a parois fines dans la peninsule iberique, a basic and thorough study which followed Moevs's book, in 1975, from which Mayet largely draws her chronology. From the Portus Cosanus and its fishery thirty-six pieces of thin walled ware were catalogued and many other fragments found. Besides the amphoras, the thin ware provides our largest body of ceramic material and is thus of special importance in establishing the chronology of the site. Significant is the fact that all the catalogued material, with the exception of two pieces, was found in the excavations of the Spring House, which housed a source of water for the fishery and its attached commercial operations as well as for a public or private fountain. This fact clearly establishes that many of these thin walled shapes, especially Forms 7 and 8, were indeed used for drinking vessels by the poorer people before glass came into common use. Among our finds, several new types for Cosa have been identified. Cat. TWlO, a complete goblet decorated with thumb impressions, is associated with Mayet's Form 6 (Figs. X-12, 13; Color Fig. 5). This decorative thumb technique, while found on the hill of Cosa and in other Italian sites, is not popular in Italy but apparently appealed to the Spanish export trade. Also new for Cosa is Cat. TW19, a beaker of Mayet's Form 8 C (Fig. X-28). Cat. TW23 is a fine fragment of particular interest as an illustration of the combined Hellenizing and indigenous Italic traditions, which contribute to the style of thin walled ware, a small ribbon handle of a Sovana class cup with a Hellenizing satyr's head combined with an older Italic leaf motif (Figs. X-35, 36). For establishing the chronology of the Portus Cosanus and its fishery the evidence provided by the thin walled pottery found in the Spring House is critical. As our only architectural structure— attached to both the fishery and to the harbor—it provides the key for the chronology of the entire fishery complex. Both the presence of thin ware material from certain periods as well as its absence from others when it is well represented in the hill town are significant factors. A number of conclusions may be drawn from the evidence of the catalogued material presented below, which will be discussed more fully in the chapter on chronology. First, thin walled ware does not appear in quantity in the Spring House much before 75 B.C. Only two examples of Form 1 decorated in the dotted barbo-

tine technique were found—the situlae so popular on the hill of Cosa as well as other sites beginning as early as the first half of the second century B.C. and lasting until the third quarter of the first century B. c. The main period of its popularity, however, is during the second half of the second and first quarter of the first centuries B.C. Secondly, the most popular forms of drinking vessel found are Forms 7 and 8, ellipsoidal beakers with either high swelling lips or small everted ones. These forms are precisely dated from the end of the first quarter of the first century B.C. until the early Augustan period, but their time of greatest frequency at Cosa is during the third quarter of the first century B.C. The break in our thin walled material found in the Spring House occurs at the end of the early Augustan Age, before the end of the first century B.C. This hiatus lasts until Claudian-Neronian times. Completely lacking among our finds are any examples of the metallic wares of the mature Augustan period or the orange coated vessels so popular in the Tiberian and frequently found in the town site above. Not until sometime during the Claudian-Neronian period does life seem to begin again in the lagoon fishery, at least on a small scale, judging from the three documented pieces of orange coated thin ware from this time. Only one of these was actually found in the lower levels of the Spring House excavation; the other two come from the area YZ, where a later fountain was apparently also in use. To summarize the evidence of the thin walled finds: while the spring was in use as early as the second century B.C., the Spring House structure itself appears to have been built sometime around 75 B.C. with its first phase of use continuing until about 10 B.C. The majority of the thin walled sherds fall into the third quarter of the first century B.C. From the end of the early Augustan period until about A.D. 40 or 50 the Spring House seems to have been abandoned. A second phase of use may begin as early as the Claudian-Neronian period between A.D. 41 and 68. The end of this second phase of use must be dated by other ceramic finds, for by this time the later sigillata wares and glass beakers had replaced thin walled pottery for tableware.

REPUBLICAN PERIOD: UNCOATED WARE

T W l . (PC 69-53).. (Fig. X-I). SH: rubble fill above basin platform. Fragment of base and wall of situla with dotted decoration in the barbotine technique. Orange-brown clay, fired gray on exterior.

X: F I N E W A R E S — R O M A N T H I N W A L L E D P O T T E R Y

Uncoated. MPH 0.03 m; MPW 0.02 m; D of base 0.04 m; Th 0.004 m. Small situla, Moevs Form 1, with flat base.7 Two opposing rows of superimposed dots in the barbotine technique are arranged in festoons around the base. Form 1 appears at the beginning of thin walled pottery and its shape and decoration reveal its derivation from metal prototypes in the art of the early Iron Age—particularly that of the Este civilization in the Veneto. 8 The barbotine dot decoration, which is chiefly associated with situlae of Form 1, is the most popular decorative technique used on thin walled ware during the Republican period. The finds from Cosa as well as other sites in the Mediterranean indicate that this technique may occur in thin walled ware as early as the first half of the second century B.C. and continue until the end of the third quarter of the first century B.C.9 In the recent publications of the thin walled pottery from Ostia, its use is limited to the second half of the second century10 through the first half of the first century B.C. At Luni, it does not apparently appear before the first half of the first century and disappears after 50 B.C.11 Mayet also more closely limits the appearance of Form 1 and the barbotine technique in Spain to the last third of the second century and the first third of the first century B.C.12 For comparisons with similar situlae with flat bases employing dotted barbotine decorations arranged in free curving festoons see examples from Cosa no. 5, dated in the second half of the second century B.C; 13 from Ibiza, Mayet no. 2;14 and from Athens, dated during the last three quarters of the first century B.C.15 Also see the examples from Ampurias dating between 200 B.C. and 50 B.C.16 and from Ventimiglia in a context dated between 90 B.C. and 20 B.C.17 Date: 150-50 B.C. TW2. (PC 68-47). (Figs. X-2, 3). SH: rubble fill above basin platform. Fragment of wall of situla 7

Ibid., pp. 49-58. Ibid., pp. 35-45 and pis. 92-94. 9 Ibid., p. 39. These dates for Form 1 with dotted barbotine decoration are reaffirmed in her recent restudy of the Cosa hill chronology, Moevs, "Italo-Megarian Ware," pp. 171-172, 217. A lip of Form 1 was found in the Gray Deposit dated before 150 B.C. in which two Italo-Megarian bowls were found, nos. 4 and 6, dated between 175 B.C. and 150 B.C. The main period of popularity for this form and technique, however, is also dated by stratified finds to the second half of the second century B.C. and the first quarter of the first century B.C. See Moevs 1973, p. 50, the thin walled pottery found in Level II of the Capitoleum South trench and Levels IV and III of the Dump. i» Ostia III, p. 347, p. 74, fig. 712. 11 Luni II, p. 149, pis. 90, 6-8; 91, 4-6, 9. 12 Mayet, pp. 24-25. 8

225

(?) with dotted decoration in the barbotine technique. Light orange clay, finely levigated. Uncoated, fired gray on exterior. MPL 0.06 m; MPW 0.03 m; Th 0.004 m. The flat fragment probably belongs to the vertical lower wall of a small situla, Moevs Form 1. Compare situlae from Cosa, nos. 6, 9, 10, dated in the first half of the first century B.C.18 None of these, however, shows the dotted barbotine decoration arranged in a cross pattern as in the Portus Cosanus fragment, nor have I found it on other situlae. An incised circle also appears between the dots. Whether it was an intended part of the design is uncertain. For another example of a situla with dotted barbotine decoration from the Spring House excavations compare Cat. TWl. Date: First half of first century B. c. TW3. (PC 72-757). (Fig. X-4). SH: gray layer below pozzolana south of Wall a. Fragment or rim of jar. Orange clay fired gray on exterior. Uncoated. D of rim 0.06 m; Th 0.002 m. Probably Moevs Form 3, small jar with elongated body and everted lip, dated from as early as the late third century B.C. to the middle of the first century B.C.19 For the shape of the everted lip compare especially Cosa nos. 23-25, 32, all dated in the first century B.C. before the Augustan Age. 20 Also see Mayet Form 2, no. 19,21 a shape which she dates largely on the basis of the Cosa material. Other sites as well, however, support the Cosa chronology for the form. Vases of this type are found in Las Corts necropolis of Ampurias, dated from 200 B.C. to 50 B.C.22 A terminus ante quern date is established for the form by its presence among the finds of the Carettoni excavations in the House of Livia on the Palatine in contexts dated to the end of the Republican or beginning of the Imperial period. 23 Date: First century B.C., pre-Augustan. TW4. (PC 72-683). (Fig. X-5). SH: surface, sandy 13

Moevs 1973, no. 5, pp. 49, 261, pi. 1. Mayet, no. 2, p. 26, pi. 1. 15 Robinson 1959, pi. 1, F 24. 16 Almagro I, p. 278, fig. 225, 1; p. 299, fig. 251, 2; p. 306, fig. 262, 2; p. 374, fig. 372; p. 386, fig. 390, 5. 17 Lamboglia, Albintimilium, no. 28, p. 100, fig. 48. 18 Moevs 1973, pp. 49-58, 261-262, pis. 1, 55. 19 Ibid., pp. 58-59; cf. Moevs, "Italo-Megarian Ware," p. 172, n. 7, for suggested redating to this form. 20 Moevs 1973, pp. 263-264, pis. 3, 57. 21 Mayet, pp. 26-27, pi. 3, from La Alcudia de Elche (Alicante). 22 Almagro I, no. 1 from incineration no. 81, p. 334, fig. 305, 1; no. 1 from incineration 95, p. 340, fig. 319; no. 2 from incineration no. 99, p. 342, fig. 323, 2. 23 Carettoni, p. 99, no. 89, fig. 24 c; p. 108, no. 162 h; pp. 118119. 14

226

THE CATALOGUES

fill between Walls a and b. Fragment of base of jar. Brownish, gritty clay. Uncoated. D of base 0.03 m; Th 0.002 m. Moevs Form 4, jars with ovoid body and vertical swelling lips. This shape, one of the most popular in thin walled ware of the late Republican period, may be dated from the late third century B.C. to the early Augustan period.24 Our flat base may be compared with Cosa no. 45, dated in the third quarter of the first century B.C.25 or Cosa no. 39, dated between 150 B.C. and 75 B.C.26 Inside the base impressed into the clay is a spiral design. Date: 150-25 B.C. TW5. (PC 69-58). (Fig. X-6). SH: rubble fill above basin platform. Complete base and section of wall of globular jar. Brownish clay, fired gray on exterior. Uncoated. D of base 0.04 m. Globular jar, Moevs Form 5 or Form 6, dating from the end of the third century B.C. 2 7 This basic shape, which has its origins in Iron Age forms, is widely known in sites over the Mediterraean from Spain to Greece. For example, Form 5, distinguished by its everted lip, is represented in the excavations at Vetera, Oberaden, Haltern, Populonia, Numantia, Rome, and Bologna. Form 6, distinguished by its vertical swelling lip, is also found at Ibiza, Athens, and Kourion. The form survives at Cosa until the Claudian-Neronian period. The fairly widely flaring body of our example terminates in a flat, low ring foot. Compare the shallow base of Cosa no. 64, Moevs Form 5 or 6, dated between 150 B.C. and 75 B.C, 2 8 or the base • of a jar from Ibiza, Mayet Form 3 A, no. 59, a shape especially characteristic of the second half of the first century B.C.29 For the shape of the body compare Cosa nos. 62 and 63, Moevs Form 6, dated again between 150 B.C. and 75 B.C.30 Date: Pre-Augustan. TW6. (PC 72-669). (Fig. X-7). SH: surface. Fragment of base of jar with beginning of lower wall 24 Moevs 1973, pp. 59-62; cf. Moevs, "Italo-Megarian Ware," p. 172, n. 7, for suggested redating of type into the late third century B.C. 25 Moevs 1973, pp. 59, 265, pis. 5, 58. 26 Ibid., pp. 59, 265, pis. 4, 57. 27 Ibid., pp. 62-64. For a suggested revision of the dating of Forms 5 and 6 on the basis of new evidence acquired at Cosa after her earlier publication, see Moevs, "Italo-Megarian Ware," pp. 171-172. These forms (as well as examples of Forms 1, 2, 3, 4, 13, and 35) were found in the Atrium Publicum excavations, Room 13, cesspool, Level II, known as the Gray Deposit, dated from as early as the late third century B.C. and ending before the middle of the second century B.C. 28 Moevs 1973, pp. 64, 267, pis. 6, 59. 29 Mayet, p. 33, pi. 8.

of body. Brick clay, gray on exterior, encrusted. Uncoated. D of base 0.06 m; Th 0.003 m. Moevs Form 5 or 6, globular jars with everted or vertical swelling lip, which may be dated as early as the late third century B.C. For a similar flat base with shallow, slightly beveled edge, compare Cosa no. 64, with dotted barbotine decoration, dated between 150 and 75 B.C.31 Also compare Mayet Form 3 A, no. 49. 32 Date: Pre-Augustan. TW7. (PC 72-666). (Fig. X-8). SH: surface. Fragment of base and wall ofjar. Brick clay with traces of a white thin wash on exterior. Gray on exterior and encrusted. D of base 0.05 m; Th 0.002 m. Moevs Form 5 or 6, globular jars with everted or vertical swelling lips, dating from the end of the third century B.C.33 For a shallow ring base of this shape compare Cosa no. 64, dated between 150 and 75 B.C.34 For the type of clay with a thin, white wash see Cosa no. 60, Form 6.35 Compare Mayet, Form 3 B, nos. 59-63,36 placed in the first century B.C. Date: Pre-Augustan. TW8. (PC 72-208). (Figs. X-9, 10). SH: fill below rubble level north of Wall b. Complete base of jar. Pink-beige clay. Uncoated, with some encrustation. MPH 0.01 m; D of base 0.05 m; H of base 0.005 m; Th 0.002 m. Moevs Form 5 or Form 6, globular jars with everted or vertical swelling lips, dating from the late third century B.C. and continuing at Cosa until the Claudian-Neronian period.37 The shallow concave base has a ring foot. For the globular shape of the body see no. 64 from Cosa hill excavations Form 5 or 6 dated 150-75 B.c, which has, however, a flat base, an earlier feature of the form.38 Also see Mayet, Form 3 B, no. 31, placed in the second half of the first century B.c.39 Date: Pre-Augustan. TW9. (PC 72-780). (Fig. X - I l ) . SH: gray level below pozzolana south of Wall a. Fragment of base 30

Moevs 1973, pp. 63, 267, pi. 6. Ibid., pp. 64, 267, pis. 6, 59. Mayet, pp. 29-30, 32, pi. 6, from Tarragone. Mayet takes the dating for her form largely from the Cosa material, placing the floruit of the shape in the second half of the first century B.C. before the Augustan period. 33 Moevs 1973, pp. 62-63; cf. Moevs, "Italo-Megarian Ware," p. 172, note 7. 34 Ibid., pp. 64, 267, pis. 6, 59. 35 Ibid., pp. 63, 266, pis. 6, 59. 36 Mayet, p. 30, pi. 8. 37 Moevs 1973, pp. 62-66; cf. Moevs, "Italo-Megarian Ware," p. 172, n. 7 for revised dating of Forms 5 and 6. 38 Moevs 1973, pp. 63, 267, pi. 6. 39 Mayet, pp. 29-30, pi. 4. 31

32

X: FINE WARES—ROMAN THIN WALLED POTTERY

and lower wall of jar. Brick clay, gray on exterior, some encrustation. Uncoated. D of base 0.06 m; Th of bottom of wall 0.004 m. Moevs Form 5 or 6, globular jars with everted or vertical swelling lips, dated from the late third century B.C.40 Compare also Mayet, Form 3 B, nos. 31-44.41 The base is flat, an early feature generally for this form, with a slightly turned-out edge. The thickness of the wall indicates that the jar falls into the borderline realm between thin walled ware and the thicker, utilitarian ware. Date: Pre-Augustan. TWlO. (PC 72-27). (Figs. X-12, 13; Color Fig. 5). SH: fill around Wall e. Complete profile of goblet, restored from thirty-seven fragments. Reddish orange, finely levigated clay. Uncoated; surface discolored in a few areas to gray with some encrustation. Several body fragments missing. H 0.12 m; W 0.12 m; D of rim 0.09 m; Th 0.002 m. Mayet Form 6. This squat, low carinated goblet decorated with thumb impressions combined with shallow, horizontal grooves encircling the entire wall of the body is not represented in the thin walled ware from Cosa and is not a form frequently found in Italy. It is, however, well represented in finds from the Iberian Peninsula and the Balearic Islands. Compare the goblets with thumb impressions and horizontal grooving from Palma de Majorca, Mayet Form 6, nos. 85-87.42 The use of thumb decoration, a motif deriving from metal prototypes, does appear in other forms of thin walled ware from Cosa. See the carinated cups, Moevs Form 27, nos. 132 and 133, dated in the second and third quarters of the first century B.C.43 Thumb impressions continue to be employed in later forms as well. See Cosa no. 217, Form 27, 44 a carinated cup dated in the Augustan period as well as the beaker no. 434, Form 69,45 dated in the middle of the first century A.D. For the use of combed, parallel, horizontal grooving in the thin walled ware of Cosa, see beaker no. 86, Form 8, precisely dated into the third quarter of the first century B.C.46 This evidence suggests that the combination of grooving with thumb impressions used here as well as in 40 Moevs 1973, pp. 62-64, nos. 61-65, pi. 6; cf. Moevs, "ItaloMegarian Ware," pp. 171-172, and the Portus Cosanus fragment, Cat. TW9. 41 Mayet, pp. 29-30, pis. 4, 5. 42 Ibid., pp. 37-38, pi. 11. This form is dated by Mayet into the second and third quarters of the first century B.C. on the basis of the Cosa material cited here. 43 Moevs 1973, pp. 86, 87, 273, pis. 12, 62. 44 Ibid., pp. 86, 282, pis. 22, 68.

227

the examples from Majorca probably dates from this time. This dating is supported by the evidence from the drain in the Regia in the Roman Forum, dated in the last three decades of the first century B.C. in which a fragment of a carinated cup with thumb impressions combined with horizontal grooving was found.47 The plain, everted lip is 0.02 m high and is defined by double, incised grooves around the rim. The shallow ring foot has a diameter of 0.06 m. Date: Third quarter of the first century B.C. T W I l . (PC 72-759). (Fig. X-14). SH: gray layer below pozzolana south of Wall a. Two joining fragments of base of beaker. Orange clay fired gray on exterior. Uncoated. D of base 0.03 m; Th 0.002 m. Probably Moevs Form 7, ellipsoidal beaker with high swelling lip, dated from the end of the first quarter of the first century B.C. to the early Augustan period. 48 The majority of examples from Cosa, however, may be dated to the third quarter of the third century B.C. This date for the floruit of the type is supported by material from other sites, particularly from Petrisberg near Trier, dated shortly after 45-44 B.C.49 For a similar slightly concave base with ring foot compare Cosa no. 72, Form 7.50 It is possible that this base belongs with rim Cat. TW3 found in the same context and of similar clay; but narrow, everted lips are not associated with this type of base and are rather typical of Form 3. Date: Third quarter of first century B.C. TW12. (PC 72-805). (Fig. X-15). SH: surface, below buttress of Wall a. Base with section of lower wall of beaker. Fine, deep orange clay. Uncoated. D of base 0.03 m; MPH 0.04 m; Th 0.003 m. Moevs Form 7, ellipsoidal beaker with high swelling lip, dated from the end of the first quarter of the first century B. c. to the early Augustan period.51 The shallow ring foot has a concave base with a beveled edge. An incised circle articulates the interior of the base. Compare Cosa no. 72, dated between 75 B.C. and 25 B.C.52 Date: Third quarter of first century B. c. TW13. (PC 72-60). (Figs. X-16, 17). SH: rubble 45

Ibid., pp. 238, 307, pis. 46, 86. Ibid., pp. 69-70, pis. 8, 60. Ibid., pp. 52-54, note 60 and pi. 97, no. 19. 48 Ibid., pp. 66-68. 49 Loeschcke, "Petrisberg," no. 5, pp. 96, 100, figs. 2-3, Form 4. This complete beaker establishes the full profile for Form 7. 50 Moevs 1973, pp. 66, 268, pis. 7, 59. 51 Ibid., pp. 66-68. 52 IWd., pp. 66, 268, pis. 7, 59. 46 47

228

THE CATALOGUES

level south of Wall a. Base with lower wall of beaker. Red clay fired gray on exterior; uncoated. MPH 0.01 m; D of base 0.03 m; H of base 0.003 m; Th 0.003 m. Like Moevs Form 7, ellipsoidal beaker with high swelling lip, dating from the end of the first quarter of the first century B.C. to the Augustan period. The concave base has a ring foot with a beveled edge like Cats. TWl2 and TWl5. Inside the base are three concentric, incised circles. Compare example of Form 7 from the Cosa hill excavations, no. 72, dated between 75 and 25 B.C.53 Date: Third quarter of the first century B.C. TW14. (PC 72-56). (Fig. X-18). SH: rubble level south of Wall a. Base with lower section of wall of beaker. Hard, red clay. Uncoated, encrusted and discolored. MPH 0.02 m; D of base 0.03 m; H of base 0.007 m; Th 0.002 m. Moevs Form 7 or Form 8, ellipsoidal beakers with either high swelling or small everted lips. Flat, ring base with molded edge. The closest comparison from the Cosa hill excavations is with a base of Form 7, no. 75, 54 dated in the early Augustan period, which does not, however, have a molded edge. Also compare Cat. TW25. Date: Third quarter of first century B.C. TW15. (PC 72-350). (Figs. X-19, 20, 21). SH: surface, east of Wall d. Base with section of lower wall of beaker. Fine, hard, deep orange clay. Uncoated, encrusted. MPH 0.02 m; D of base 0.03 m; Th 0.005 m. Moevs Form 7 or Form 8, ellipsoidal beakers with high swelling or small everted lips, dated from the end of the first quarter of the first century B.C. to the early Augustan period. Compare Cat. TWl 2 and the base, Form 7, from the Cosa hill excavations no. 72, dated by Moevs between 75 B.C. and 25 B.C.55 The ring foot has a concave base with a beveled edge. A shallow raised ring appears inside the base. Date: Third quarter of the first century B.C. TW16. (PC 72-356). (Figs. X-22, 23). SH: surface, east of Wall d. Fragment of base with wall of beaker. Fine, hard, orange clay. Uncoated, encrusted. MPH 0.03 m; D of base 0.03 m; H of base 0.004 m; Th 0.001 m. Moevs Form 7 or Form 8, ellipsoidal beakers 53

Ibid. Ibid. 55 Ibid. 56 Ibid. 57 Ibid., pp. 66-68. 58 IHd., pp. 66, 268, pi. 7. 54

with either high swelling or small everted lips. Compare Cosa no. 73, base of Form 7 dated in the third quarter of the first century B.C. 56 The concave base has a ring foot with a beveled edge. The interior is defined by a shallow raised ring. Compare Cat. TW15. Date: Third quarter of the first century B.C. TW17. (PC 68-54). (Figs. X-24, 25). SH: mud fill around basin platform, Level IV. Fragment of rim of beaker. Orange clay, finely levigated. Uncoated. MPH 0.03 m; MPL 0.05 m; D of rim 0.08 m; MPH of rim 0.02 m; Th 0.002 m. Probably Moevs Form 7, ellipsoidal beaker with high swelling lip, dated from the end of the first quarter of the first century B.C. to the early Augustan period.57 For example, compare Cosa no. 76 from the early Augustan period. 58 The more globular shape of the body, however, is closer to Moevs Form 659 with which it may also be associated. For similar shapes from Majorca from the second half of the first century B.C. see, for example, Mayet no. 46, Form 3 A.60 Date: Third quarter of the first century B.C. TW18. (PC 69-111). (Figs. X-26, 27). SH: mud fill around basin platform, Level IV. Seven fragments of a beaker with section of base, body and rim without the lip. Reddish beige, micaceous clay, fired gray on exterior. Uncoated. D of base 0.04 m; D of rim ca. 0.05 m; Th 0.002 m. Close to Moevs Form 7, ellipsoidal beaker with high swelling vertical lip, dating from the end of the first quarter of the first century B.C. to the early Augustan period.61 The lip is vertical and set off from the body by a grooved line in contrast to Cat. TW17, also Form 7. The rim is closest to Mayet Form 3 B, nos. 31-33, beakers from Ibiza.62 The body ends in a flat base with a slightly beveled, turned out edge. Date: Third quarter of the first century B.C. TW19. (PC 69-129). (Fig. X-28). SH: mud fill below basin platform, Level IV. Lip section of wall of body and unjoining, but complete base of beaker. Restored from five fragments with five additional unjoining pieces. Gritty, reddish clay, discolored. Uncoated. D of base 0.05 m; D of rim 0.05 m; H of rim 0.03 m; Th 0.001 m. While the swelling lip is like Moevs Form 7, this particular shape is not found on the hill site of 59

IKd., pp. 63-64. Mayet, pp. 29, 32, pi. 6. 61 Compare especially Cosa no. 78, Moevs 1973, pp. 66, 268, pis. 7, 59. 62 Mayet, p. 31, pi. 4. 60

X: FINE W A R E S — R O M A N T H I N WALLED P O T T E R Y

Cosa. The unusually high and deep lip combined with a shallow, flat base belongs with Mayet Form 8 C but lacking the decoration. 63 This form, popular in the Iberian Peninsula, is not precisely dated from archaeological evidence and is generally placed in the second half of the first century B.C. before the Augustan period. Date: Pre-Augustan. TW20. (PC 72-300). (Figs. X-29, 30). SH: mud fill above basin platform, Level IV. Fragment of rim of beaker (?). Hard reddish clay, fired gray on exterior. Uncoated. D of rim 0.09 m; Th 0.002 m. This small fragment is closest to rims of Moevs Form 13, beakers with a small, angular shoulder and a molded, slightly everted lip, primarily dated in the second half of the first century B.C., although the form can go back to the late third century B.C.64 Compare Cosa no. 97, dated in the third quarter of the first century B.C.65 Date: Second half of the first century B.C. TW21. (PC 72-81). (Figs. X-31, 32). SH: rubble level south of Wall a. Fragment of rim and wall of beaker (?). Red clay, fired gray on exterior. Uncoated. D of rim 0.11 m; Th 0.002 m. This Republican lip shape is closest to Moevs Form 13, beakers with a small, angular shoulder and a molded, slightly everted lip. 66 Compare Cosa no. 97, dated in the third quarter of the first century B.C.,67 and Cat. TW20. In contrast to the latter, this rim has a groove encircling the lip. Date: Second half of first century B.C. TW22. (PC 69-26). (Figs. X-33, 34). SH: rubble fill above basin platform. Complete base and section of body of jug, restored from three joining fragments. Gritty, light orange, micaceous clay. Uncoated. D of base 0.03 m; MPH 0.06 m; Th 0.002 m. Moevs Form 14, bag-shaped jug with everted lip and flaring ring foot, dated in the late second and the first centuries B.C., Cosa nos. 99-103.68 Our example is a variation of this form with a rounded, flat ring foot, which Moevs considers an early feature in the development of this form. 63 Mayet, p. 39. Compare especially no. 98, pi. 12, from Cadiz, and no. 104, pi. 13, from Majorca. 64 Moevs 1973, pp. 74-75; cf. Moevs, "Italo-Megarian Ware," p. 172 and n. 7. 65 Moevs, 1973, pp. 74, 270, pis. 9, 60. 66 Ibid., pp. 74-75; for evidence that this form can go back as early as the late third century B.C. see Moevs, "Italo-Megarian Ware," p. 172 and n. 7. 67 Moevs 1973, pp. 74, 270, pis. 9, 60. 68 Ibid., pp. 75-76, pis. 9, 60; also see new example of Form 14 at Cosa from SUNY House excavations, Moevs, "Italo-Megarian Ware," p. 175, dated late second or early first century B.C.

229

Compare the rounded foot of Cosa no. 103, dated between 150 B.C. and 75 B.C.69 For the round carination of the body, see the jug from the House of Livia on the Palatine, which originally had two vertical handles. 70 Moevs notes the origins of Form 14 in metal prototypes, citing especially a group of one-handled bronze Etruscan jugs beginning in the late fourth century B.C.71 Date: 150-75 B.C.

REPUBLICAN PERIOD: SOVANA CLASS

TW23. (PC 69-4). (Figs. X-35, 36). SH: backing of Wall P. Complete ring handle with portion of wall of cup with satyr's head in relief. Hard, micaceous, gritty reddish clay. Uncoated; discolored. Inner D of handle 0.02 m; Th of body 0.002 m; MPL 0.06 m. Probably a two-handled, thin walled, round carinated cup of the Sovana class. The ring handle is decorated with vertical grooves and a horizontal strip in relief at the midpoint of the loop. Below the handle is a satyr's head in relief with traces of an incised leaf design around it. For the form, see the two-handled cup found in a tomb at Sovana.72 Besides cups, this small class of pottery includes situlae and jars made of orange-red micaceous clay, so-called from the place of their provenience in Tuscany. This class of pottery has been dated from the first century B.C. until early Augustan times. 73 Similar ribbon-shaped handles are used on thin walled pottery forms from Cosa. See the two-handled cups, Forms 39 and 40, and particularly no. 201, Form 39, dated in the late Republican or Augustan period.74 This type of handle is also found in the material from the House of Livia on the Palatine also dated in the late Republican or Augustan period. 75 Furthermore, horizontal strips, an indication of a metal prototype, appear on three ribbon handles found in the drain of the Regia in the Roman Forum 69

Moevs 1973, pp. 75, 270, pi. 9. Carettoni, p. 107, no. 160 b, fig. 31 b. 71 Moevs 1973, p. 76; P. Jacobsthal, Early Celtic Art (Oxford, 1944) p. 257, from Castiglione delle Stiviere. 72 G. Pellegrini, "Tombe etrusco-romane," NSc (1903) p. 224, fig. 6, no. 3; R. Bianchi-Bandinelli, Sovana. Topografia ed arte (Florence, 1919) pp. 128-129, n. 41, pi. 37, b (c); Moevs 1973, pi. 99, fig. 2. 73 Pellegrini, NSc (1903) p. 220; Moevs 1973, p. 56. 74 Moevs 1973, no. 201, pis. 20, 67, and p. 112; also compare here Cat. RG24. 75 Carettoni; Moevs 1973, pp. 59, 112-113. 70

230

THE CATALOGUES

dated in the last three decades of the first century B.C. 7 6

For the motif of a Hellenizing satyr's head in relief, compare the outstanding example of the Sovana pottery class—a situla with a prow spout found in the tomb, Campo alia Zinza no. 8 (Fig. X-37). 77 This Sovana situla may be precisely dated from the presence in the tomb of two bronze coins, one an as of semi-uncial reduction and thus dated later than 91 B.C.,78 as well as from the black gloss pottery. These include a bowl, unguentarium, pitcher, and pyxis. The earliest piece in the group appears to be the fusiform unguentarium, Morel Form 108, which dates from the end of the third century B.C. through the first half of the second century B . C The latest piece appears to be the pyxis, which may be compared to one found at Cosa in a deposit with a terminus post quern date of 110-100 B.C. and a terminus ante quern date of about 30 B.C. A mid-second century B.C. date for the rest of the black gloss material from the tomb seems most appropriate. 79 The leaf decoration visible behind the satyr's masque and typical of Sovana class pottery rather derives from a long Italic Iron Age tradition. It is used on thin walled pottery from Cosa dated between 150 B.C. and 75 B.C.80 Relief decoration on domestic pottery from other regions in Italy also occurs at approximately this same time. For example, see the two-handled jar decorated with rosettes and corded festoons in relief from Celiniea di Pariana (La Spezia) associated with the Sovana class.81 This jar has been securely dated from the other finds in the tomb to the end of the second or the first third of the first century B.C. Our small but fine fragment of Sovana ware is thus of particular interest as it reveals the dual heritage of this special pottery class which combines both a Greek Hellenistic tradition apparent in the satyr's head with a long, indigenous one seen in the use of the leaf decoration. This particular com76 Moevs 1973, p. 113, pi. 97, nos. 30-32. For the dating of the Regia drain see ibid., pp. 52-54, note 60. 77 Bianchi-Bandinelli, Sovana (cited in n.72) pi. 40, b; G. Pellegrini, "Tombe del periode etrusco-romano," NSc (1902) pp. 508-509; Pellegrini, NSc (1903) p. 222, fig. 5. 78 M. H. Crawford, Roman Republican Coinage (Cambridge, 1974) I, pp. 77, 78, 611. 79 For the unguentarium see Morel, pp. 223-224; for the pyxis, see D. M. Taylor, pyxis E 17a, pi. 42, pp. 134-135, 160-161. I am grateful to A. R. Scott for the suggested dating of this material based on a study of the published photographs alone. 80 Moevs 1973, pp. 49, 57, 263, nos. 20 and 21, pis. 2, 56. Also see situla of Form 1 with leaf decoration found in the SUNY

bination of motifs used in this walled ware cannot be earlier than the beginning of the first century B.C. 8 2

While the Sovana pottery class continues until early Imperial times, the above evidence suggests a date for our fragment before the mid-first century B . C , most probably about 75 B.C. Date: End of the first quarter of the first century B.C.

TW24. (PC 72-385). (Figs. X-38, 39, 40). SH: surface, south of Walls d and e. Fragment of vertical, wheeled handle with section of everted rim. Fine orange clay, Uncoated, surface worn. Inner D of handle 0.01 m; Th of cup 0.001 m; MPL 0.06 m. The two-handled, thin walled cup with everted rim is like Moevs Form 39, no. 201, typical of the late Republican or early Augustan period.83 The curved handle is decorated with four vertical grooves with a thumb rest at the summit in the form of a male head in relief. The worn head of poor quality had short curls, a wide flattened nose suggesting Negroid features, and it glances to the right. This type of grooved handle employing heads in relief as a decorative element may be compared with our fragment Cat. TW23, identified with the Sovana class of thin walled ware and dated here at the end of the first quarter of the first century B.C. Also see four fragments, C 65.176179, from jars of Forms 4, 5, and 6 with human heads in relief found in the trench in the Temple of Jupiter at Cosa, Levels H-IV, dated around the end of the first quarter of the first century B.C.84 For the survival of the use of human heads for handle decorations see the two-handled, carinated, orange-coated cup in the Archaeological Museum in Barcelona, Mayet Form 38 A, no. 467, dated in the Tiberian-Claudian or Flavian periods. 85 This handle motif thus illustrates a process of assimilation not unusual in the creation of thin walled pottery forms and their decorative motifs. This handle type, which originated in the art of the House excavations at Cosa, dated shortly after the first decade of the first century B.C.: Moevs, "Italo-Meganan Ware," p. 175. Leaf patterns also appear on an unpublished beaker from Cosa, C 66.225 dating in the first century B.C. probably a variant of Moevs Form 9. I am grateful to Moevs for drawing my attention to these pieces and to R. T. Scott for the suggested dating. 81 Moevs 1973, p. 56; L. Banti, Luni (Florence, 1937) p. 165, pi. 12a. 82 Compare Mayet, pp. 16-19, for indigenous and Hellenistic influences on thin walled ware. 83 Moevs 1973, pp. 112-113, pis. 20, 67. 84 Moevs, "Italo-Megarian Ware," p. 173. 85 Mayet, pp. 73, 94, pi. 57.

X: FINE WARES—ROMAN THIN WALLED POTTERY

silversmith, served as inspiration for thin walled Republican ware, to be passed on into the decorative vocabulary of Imperial glazed ware. The rim which flares slightly outward, is decorated with a small button like attachment, 0.007 m in diameter. Date: Late Republican or early Augustan period. REPUBLICAN PERIOD: BLACK LACQUERED WARE

TW25. (PC 69-113). (Figs. X-41, 42). SH: mud fill around basin platform, Level IV. Base and beginning of wall of beaker. Hard, red micaceous clay. Black lacquer coating. D of base 0.03 m; Th 0.001 m. Probably Moevs Form 7, ellipsoidal beaker with high swelling lip.86 Compare Cosa no. 75, dated in the early Augustan period.87 For the use of this form in black lacquered ware from the hill site of Cosa see nos. 154, 155.88 This class of pottery is distinguished by its thin, resinous, black coating. It is found in the excavations in the town of Cosa in contexts ranging from the first quarter to the end of the first century B.C.89 The base is flat with a shallow, ring foot. Date: First century B.C. AUGUSTAN PERIOD: UNCOATED WARE

TW26. (PC 69-61). (Fig. X-43). SH: mud fill around basin platform, Level IV. Fragment of rim and wall of beaker. Micaceous, gritty, reddish brown clay. Uncoated. Estimated D of rim 0.04 m; Th 0.002 m. The plain rim combined with the straight wall of the body is closest to Moevs Form 33, cylindrical, low carinated beakers typical of the Augustan period. 90 Compare Cosa nos. 169-171 which are, however, larger.91 Date: Augustan. TW27. (PC 72-45). (Fig. X-44). SH: surface. Fragment of rim and wall of beaker. Fine, red micaceous clay; uncoated. D. of rim 0.08 m; MPH 0.02 m; Th 0.002 m.

The plain rim combined with a nonflaring wall of the body is like Moevs Form 33, cylindrical, low, carinated beakers typical of the Augustan period.92 Compare Cosa nos. 169-171 and Cat. TW26. 93 There is a slight indentation just below the surface of the rim. Date: Augustan. TW28. (PC 72-765). (Fig. X-45). SH: gray layer below pozzolana south of Wall a. Fragment of rim of bowl. Dark orange clay. Uncoated. D of rim 0.14 m; Th 0.002 m; MPH 0.02 m. Like Moevs Form 36, hemispherical bowls. This basic bowl shape has a long history in Roman pottery going back to the second century B.C., but its large-scale production at Cosa did not occur before the beginning of the Augustan period.94 Compare Cosa nos. 191-193, undecorated bowls of Augustan date which show grooves on the body. 95 Our fragment has a plain vertical rim with an incised line encircling the upper body below the lip. Date: Probably Augustan. TW29. (PC 72-764). (Fig. X-46). SH: gray layer below pozzolana south of Wall a. Fragment of lip and wall of bowl. Brick-colored clay, fired gray on exterior, encrusted. Uncoated. D of rim 0.12 m; Th 0.002 m. Like Moevs Form 36, a hemispherical bowl most popular in the early Empire in both uncoated and coated thin wares. 96 For shape of plain, vertical lip compare especially Cosa nos. 191-194,97 dated in the Augustan period. Just below the lip, which curves very slightly inward, two incised, encircling lines are visible. Date: Probably Augustan. TW30. (PC 72-676). (Fig. X-47). SH: surface. Fragment of vertical handle of mug (?). Reddish, fine, soft clay. Uncoated. MPL 0.06 m; W of handle 0.008 m. The long, vertical handle divided into two ridges by a central groove is like handles belonging to Moevs Form 56, mugs in the shape of a truncated cone on a carinated base found at Cosa in contexts of the Augustan and Tiberian periods. 98 Compare especially Cosa no. 294.99 Because of the type of soft clay and thin handle

86

93

87

94

Moevs 1973, pp. 66-68. IWd., pp. 66, 268, pis. 7, 59. 88 Ibid., p. 95. 89 Ibid., pp. 93-94. 90 IbId., pp. 102-103. 91 Ibid., pp. 102, 277, pis. 16, 64. 92 IWd., pp. 102-103.

231

Ibid., IWd., Ibid., 96 IWd., 97 IWd., 98 Ibid., 99 IWd., 95

pp. pp. pp. pp. pp. pp. pis.

102, 277, pis. 16, 64. 106-111. 109, 279, pis. 19, 66. 106-111. 109, 279, pis. 19, 66. 164-167. 32, 74.

232

THE CATALOGUES

rather than the harder, polished clay and thicker forms typical of Tiberian pottery, an Augustan date seems the most probable for our fragment. Date: Probably Augustan.

EARLY IMPERIAL PERIOD: GRAY WARE

TW31. (PC 69-69). (Figs. X-48, 49). SH: basin fill. Complete base and section of body of beaker. Gray clay; uncoated. D of base 0.04 m; Th 0.002 m. Ellipsoidal beaker of either Moevs Form 7 or Form 8 in gray thin walled ware of Alpine manufacture.100 This North Italian imported ware is represented in the thin walled pottery from the hill site of Cosa. For example, compare Cosa no. 414, Form 8, probably dating in the late Claudian Neronian period.101 This late Republican form was a popular one of the Alpine gray ware, which reached its peak of production about the middle of the first century A. D. Like the earlier Etruscan bucchero technique, gray ware is the result of a firing process in a reducing atmosphere. The base is flat with a molded ring foot, 0.005 m high. Date: Not earlier than the Tiberian period. TW32. (PC 72-354). (Figs. X-50, 51). SH: surface. Fragment of foot of a vase or cup. Gritty beige clay, gray on exterior with surface sheen. Wheel marks apparent. MPH 0.023 m; inner D of upper section of stem 0.02 m; inner lower D of stem 0.03 m. Probably a fragment of a flaring pedestal foot from an ovoid vase in early Imperial gray ware. Following metal prototypes, the foot is divided into three molded sections, the widest and lowest section of which is missing. For this type of pedestal foot in gray ware of La Tene tradition compare Cosa no. 400.102 Also see Moevs Form 28103 and Mayet Form 9104 which are, however, higher and more sharply molded. This form is precisely dated by the Cosa material from the first to the third quarter of the first century B. c. The pedestal foot becomes squatter by the Flavian period as evidenced by Mayet Form 46, no. 595.105 too p o r Forms 7 and 8 see ibid., pp. 66-70. For gray ware as a whole see ibid., pp. 211-227 with full bibliography. 101 Ibid., pp. 223, 305, pis. 44, 84. 102 Ibid., pp. 215, 303, pi. 83. 103 Ibid., pp. 87-88. 104 Mayet, pp. 42, nos. 109-110, pi. 14. 105 Ibid., pp. 109-110, pi. 71, from Haut Alentejo, Serrones. 2.0 mm

22.7 49.8 8.0 1.7 64.8

2.1 3.9 66.0 1.1 7.8

75.2 46.3 26.0 97.2 27.4

8.7 0.5 0.5 0.5

6.59 90.0 60.0 60.0

84.7 9.5 39.5 39.5

Free Fe2O3 %

Mineralogical Composition by Specific Gravity pH 2.8 6.3 7.7 7.9 1.9

+ +++++ + ++ + ++

2.8

XV: CORE AND SEDIMENT SAMPLES FROM LAGOON TABLE XV-3.

CHEMICAL AND MINEKALOGICAL COMPOSITION OF KEY CORES

AND TRENCH SAMPLES IN LAGOON OF COSA

Mineral Limestone Matter Gravel % >2.0 mm

Organic Matter %

CaCO 3 %

46-1 46-2 46-3 46-4 46-5

8.7 6.5 2.0 1.8 1.0

6.59 7.1 1.6 2.3 0.9

84.7 86.4 96.4 95.9 98.1

+ +

47-1 47-2 47-3 47-4

9.6 15.8 3.5 1.0

9.78 7.50 3.2 0.9

80.6 76.7 93.3 98.1

+ +

48-1 48-2 48-3 48-4 48-5

1.7 10.0 2.7 0.6 1.0

2.73 3.9 6.1 11.4 3.6

95.6 86.1 91.2 88.0 95.4

+ +

7.7

8.6

83.7

Sample

49-1 49-2 49-3 49-4

-

Free Fe2O3 %

PH

1.73 1.45 0.99 1.72 1.20

7.50 7.60 7.65 7.50 7.90

50-1 50-2 50-3 50-4 50-5

5.9 10.7 1.8 4.8 1.0

8.4 13.6 13.4 9.3 3.6

85.7 75.7 84.8 85.9 95.4

++ ++ + +

51-1 51-2 51-3 51-4

1.0 1.0 0.5 0.1

5.5 10.5 3.4 4.1

93.5 88.5 96.1 95.8

+ +

52-1 52-2 52-3 52-4 52-5

1.0 0.5 0.5 0.1 0.1

3.4

96.1

53-1 53-2 53-3 53-4

0.8

13.9

85.3

2.6

7.3

90.1

—

—

+ — -

>2.8

81.3

18.7

82.3

17.7

80.1 76.6

19.9 31.4

7.80

47.7

52.3

8.30

26.6

73.4

0.81 1.38 0.60 0.26 0.91

8.20 8.10 7.90 8.00 8.05

0.91

8.00 8.10 7.95 8.10

0.26 0.91

7.80 7.55 7.80 7.55 7.85

+ + -