Among Arabian Sands: Defining the Palaeolithic of Southern Arabia

AMONG ARABIAN SANDS: DEFINING THE PALAEOLITHIC OF SOUTHERN ARABIA Approved by: Dr. Anthony E. Marks Dr. Stanley H. Ambrose Dr. David J. Meltzer Dr. C. Garth Sampson AMONG ARABIAN SANDS: DEFINING THE PALAEOLITHIC OF SOUTHERN ARABIA A Dissertation Presented to the Graduate Faculty of Dedman College Southern Methodist University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy with a Major in Anthropology by Jeffrey I. Rose (B.A. University of Richmond) (M.A. Boston University) (M.A. Southern Methodist University) December 9, 2006 Copyright 2006 Jeffrey I. Rose All Rights Reserved ACKNOWLEDGEMENTS Of the nearly 400 pages in this dissertation, this section has been one of the most difficult to write. e COPR project was conceived on December 15th, 1995 while on vacation in the Sinai Peninsula. In the last 11 years, countless people have been instrumental in the completion of this research. Consequently, it would take a second dissertation just to thank all those that have been involved. In the interest of conserving paper, I am acknowledging only a small handful among the army of individuals who have contributed to the project. First and foremost, I thank my advisor Dr. Anthony Marks. ere is no way to acknowledge all that he has contributed, words cannot adequately express my gratitude. Tony was among the few to think this project was actually possible, and at times has had more faith in its success than me. He has been a pillar of support throughout the entire process and has been indescribably generous with his time and resources. Tony has worn many hats during the seven years that I have known him: educator, advisor, benefactor, mentor, confidante, and friend. I will spend the rest of my career striving to be a teacher such as him. I thank all the members of the 2004 COPR team, including Vitaly Usik, Daniel Richter, Yves Guichard, Teresa Medici, Ali al-Mahrooqi, and Diego Angelucci. Yves’ (Abu Matar’s) dramatic aerial kite photographs enhance this dissertation with vivid depictions of the landscape from several hundred meters in elevation. Without Vitaly’s uncanny knack for finding sites there would be very little data to present. Vitaly is also responsible for the the superb lithic illustrations that are essential to the usefulness of this work; I am truly fortunate to have him on the team. Also providing significant contributions were Adrian Parker and Mike Petraglia. Adrian’s comprehensive work reconstructing palaeoenvironmental conditions are fundamental to the Arabian Corridor Migration Model. Mike introduced me to the data from India, a part of the world in which iv I was woefully ignorant. His suggestions provided a fresh perspective and opened my tunnel-visioned eyes to the possibility of a westward expansion. I am grateful to my advising committee, Garth Sampson, David Meltzer, and Stanley Ambrose for their meticulous and thoughtful commentary on every page of this dissertation, even the most painfully boring sections, of which there are many. From the initial proposal to the final stage of formatting, I thank the legion of friends that tolerated my obsession and helped me through the darkest periods of despair: JoAnn, Chris, Gregg, Dave, Laurie, Aaron, Chrissy, Mark, John, Lia, Joe, Ron, Neil, Jimmy, Russell, Liv, Julie (the inspiration for tabula rasa), Tina, Alexia, Ellie, Steve, Bram, Joel, and Uri to name a few. I also acknowledge Katherine Monigal, who is responsible for most of the editing and formatting, as well as allowing me to use her schematic illustrations in Chapter 6. My thanks to all those in the Sultanate of Oman who have selflessly offered their time and assistance to the project, in particular Said al-Sukry, Ali al-Mahrooqi, Ahmed al-Mukhaini, and Yaqoub al-Busaidi. Petroleum Development of Oman, Historical Association of Oman, and the Ministry of Heritage and Culture provided logistical support. Finally, I thank my family: TBM, W, Stephanie, Geoff, Max, Henry, Ellie, and my cat Joseph (who taught me the importance of taking naps while writing). Of all the people mentioned on these pages, without a doubt the most vital have been my parents. Despite the fact that I opted not to become a lawyer, they have been unwavering in their support and encouragement since sending me on my very first archaeological excavation over 17 years ago. Financial support was provided by the U.S. National Science Foundation, Institute for the Study of Earth and Man at SMU, and from the Anonymous Donor’s Fund at SMU. v Jeffrey l. Rose B.A. University of Richmond, Richmond, Virginia, 1997 M.A. Boston University, Boston, Massachusetts, 2000 M.A. Southern Methodist University, Dallas, Texas, 2003 Among Arabian Sands: Defining the Palaeolithic of Southern Arabia Advisor: Professor Anthony E. Marks Doctor of Philosophy conferred December 9, 2006 Dissertation completed December 4, 2006 Until present, the Palaeolithic period in southern Arabia has been more or less terra incognita. Within the last decade, evolutionary scientists have begun to recognize the key role this region must have played in the origin of modern humans. Recent discoveries in the field of genetics underscore the significance of the Peninsula as a conduit for early human migration to and from Africa. e hypothesis proposed in this dissertation—the Arabian Corridor Migration Model— synthesizes these new genetic data with the Arabian palaeoenvironmental record. It is proposed that one or more hunter-gatherer groups expanded into southern Arabia in the late Middle and/or Upper Pleistocene (400 – 20 kya) from East Africa, during phases when favorable climatic conditions transformed the arid interior into an ameliorated savanna. ese wet periods may have caused a genetic bottleneck release that fueled demographic movements throughout the tropical belt (e.g., East Africa and the Indian subcontinent), as well as the Levant. Conversely, dry periods would have triggered large-scale desiccation throughout the Arabian Peninsula and wiped out pre-existing hominid populations—called tabula rasa events. Because of these tabula rasa events, the possibility of autochthonous development throughout the Pleistocene of Arabia can be discounted, thereby avoiding Galton’s Problem of migration versus diffusion. e Arabian Corridor Migration Model is tested using archaeological data collected during the Central Oman Pleistocene Research (COPR) 2002 and 2004 fieldwork campaigns. Based on these data recovered by COPR, three new lithic industries, sensu latu, are described: the Sibakhan, the vi Nejd Leptolithic, and the Khasfian. While the latter two are dominated by a core reduction strategy that primarily produced blades, the Khasfian is almost exclusively characterized by the façonnage manufacture of bifacial foliates. Placing these industries within a broad regional context, their technotypological elements suggest affinities with all three surrounding refugia (i.e. Levant, East Africa, and India) at different times, and support the model of modern human expansion into Arabia during the Upper Pleistocene. e directionality of this expansion remains questionable, however, and two possibilities seem equally plausible based on the archaeological evidence: an eastward movement from East Africa into Arabia, and/or a westward dispersal from the Indian subcontinent. vii TABLE OF CONTENTS LIST OF FIGURES LIST OF TABLES 1 2 INTRODUCTION OUT OF AFRICA AND INTO ARABIA Out of Africa: modern human origins, molecular anthropology, and Upper Pleistocene archaeology Theories of Modern Human Origins Genetic evidence for modern human origins East Africa during the Last Interglacial Into Arabia: geology, geography, and palaeoenvironmental conditions Geology Geography Climate Flora and Fauna The Bab al-Mandeb The Palaeoenvironment Discussion 3 THE HISTORY OF RESEARCH IN SOUTH ARABIA South Arabia in antiquity: the myth of Arabia Felix The Age of Exploration The Early 20th Century Caton-Thompson’s Hadramaut Survey Oil Discussion 4 THE PRESENT STATE OF RESEARCH IN SOUTH ARABIA Yemeni Highlands and Tihama Coast viii XIV XIX 1 10 12 12 17 21 29 29 31 36 38 41 43 50 54 55 57 60 62 66 67 69 70 Garbini and de Bayle des Hermens Italian Archaeological Mission to the Yemen Arab Republic Whalen’s Plio-Pleistocene Surveys in Southwestern Yemen Roots of Agriculture in Southern Arabia (RASA) Project Southern Saudi Arabia Eastern Province Central Province Southwestern Province Rub al’-Khali Aramco and the Desert Locust Survey Typology of the “Rub al-‘Khali Neolithic” The Faw Well Site Hadramaut, Dhofar, and the Nejd American Foundation for the Study of Man Arabian Expedition The Harvard Archaeological Survey of Oman Amirkhanov’s work in South Yemen CNRS Mission to Hadramaut Whalen’s Plio-Pleistocene Research in Dhofar The Transarabia Expedition to the Nejd Plateau Arabian Gulf, Omani Littoral, and Wahiba Sands Danish Archaeological Expedition to Qatar Royal Geographical Society’s Oman Wahiba Sands Project Saiwan Lithic Occurrences in Abu Dhabi Stone Age sites along the northern coast of Oman The Joint Hadd Project Discussion 70 72 77 79 81 81 82 89 93 93 98 104 105 105 107 110 113 115 116 117 117 121 124 126 128 131 133 ix Lower Palaeolithic/Early Stone Age Middle Palaeolithic/Middle Stone Age Upper Palaeolithic/Late Stone Age Neolithic 5 SITE LOCATIONS, GEOLOGICAL SETTINGS, AND SAMPLING STRATEGIES Ad-Dakhliyah Region Qarat al-Kibrit 1 (A5) Wadi Qilfah 1-4 (A7, A8, A9, A10) Wadi Qilfah 7 (A14) Dhofar Governorate Wadi Afur 1 (T2) Ghrain Bliss (SH1) Bir Khasfa (SH22-SH23) Dhanaqr (T21) Jibal Ardif 3 (T27) al-Hatab (T29) Discussion 6 METHODS OF ARCHAEOLOGICAL SAMPLING AND LITHIC ANALYSIS Site Sampling Strategy Lithic Analysis Debitage Tools Cores Statistical Procedures Discussion 7 ANALYSIS OF LITHIC ASSEMBLAGES Qarat al-Kibrit 1 (A5), Phase I 133 135 137 140 146 148 150 157 160 160 164 167 168 172 173 176 178 181 182 185 189 202 208 211 211 213 213 x Debitage and Cores Tools Qarat al-Kibrit (A5), Phase II Debitage and Cores Tools Qarat al-Kibrit (A5), Phase III Debitage and Cores Tools Wadi Qilfah 1 — 4 (A7, A8, A9, A10) Debitage and Cores Tools Wadi Qilfah 7 (A14) Debitage and Cores Tools Ghrain Bliss (SH1) Debitage and Cores Tools Bir Khasfa (SH22 – SH23) Debitage and Cores Tools Wadi Afur 1 (T2) Debitage and Cores Tools Dhanaqr (T21) Debitage and Cores Tools Jibal Ardif 3 (T27) 217 217 221 222 225 227 229 230 232 234 240 245 245 248 248 249 250 250 253 254 257 257 259 260 262 263 266 xi Debitage and Cores Tools Al-Hatab (T29) Debitage and Cores Tools Discussion Neolithic Khasfian Nejd Leptolithic Sibakhan T27 Summary 8 CONCLUSIONS The Arabian Corridor Migration Model Middle/Upper Pleistocene Chronology Sibakhan Sibakhan Connections with the Near East? Sibakhan Connections with the Horn of Africa? Nejd Leptolithic Nejd Leptolithic connections with the Near East? Nejd Leptolithic connections with the Horn? Khasfian Khasfian connections with the Near East? Khasfian connections with the Horn? A Passage to India? Summary 268 269 273 274 275 278 279 282 283 286 290 297 303 303 306 307 308 312 317 319 319 321 323 324 328 329 xii APPENDIX A APPENDIX B LITHIC ATTRIBUTE ANALYSIS FROM COPR 2004 ASSEMBLAGES SITES RECORDED DURING 2004 COPR CAMPAIGN 334 340 344 348 349 APPENDIX C LITHIC RAW MATERIAL FROM AREAS SURVEYED DURING 2004 COPR CAMPAIGN APPENDIX D GLOSSARY BIBLIOGRAPHY xiii LIST OF FIGURES Figure 1-1 View across the Bab al-Mandeb at sunrise, looking eastward at the Yemeni Highlands. 2-1 Physical map of Saharo-Arabian Arid Belt. 2-2 Physical map of South Arabia and major geomorphic regions discussed in text. 2-3 Aerial photograph of the Hadramaut Valley. 2-4 Photo of Wadi Andur incising the Nejd Plateau. 2-5 Photo of the Wahiba Sands longitudinal dunes and interdunal zone. 2-6 Photo of Dhofar Escarpment during monsoon season. 2-7 Photo of dense vegetation along Tihama Coast of Yemen. 2-8 Photo of terraced agricultural fields in Yemeni Highlands. 2-9 Bathymetric profile of Bab al-Mandeb at Hanish sill, with projected distance calculated at 100 meters below current sea level. 2-10 Geological profile of an exposed section in Wadi Mahwis showing coarse gravels interstratified with fine fluviatile sands. 3 31 32 33 33 34 36 37 38 41 45 71 147 148 149 150 152 152 153 156 4-1 Map of sites, towns, and drainages mentioned in text. 5-1 Provinces of Oman where survey was conducted during COPR 2004 campaign. 5-2 Aerial view of ad-Dakhliyah alluvial plain. 5-3 Major drainages on the southern ad-Dakhliyah alluvial plain and northern extent of Haushi-Huqf depression, with select archaeological sites discovered by COPR. 5-4 Qarat al-Kibrit from distance, facing south. 5-5 Aerial photograph of Qarat al-Kibrit, facing south. 5-6 Schematic profile (western) of local relief at Qarat al-Kibrit. 5-7 Northwest stratigraphic section at Qarat al-Kibrit 1. 5-8 Aerial photograph of geomorphic features around site. xiv 5-9 Rim of sabkha southeast of A7-A10 complex. 5-10 5-11 5-12 Chert and silicified limestone on surface of A7-A10 complex. Aerial photograph of A7-A10 site complex. A14 lithic surface scatter in foreground. 158 158 159 161 161 163 164 165 166 166 168 168 170 170 173 174 174 177 177 192 194 195 198 200 5-13 Chert outcropping in vicinity of A14. 5-14 Major geomorphic zones and drainage systems in the Dhofar Province, with select sites discovered during the COPR 2004 campaign. 5-15 View of Nejd Plateau, above Wadi Afur, facing west. 5-16 5-17 Rus chert exposure in the central Nejd. Mrs. Medici surveys artifacts at a high density palimpsest workshop site called Wadi Mahwis 2 (SH6). 5 m terrace adjacent to Wadi Afur at site T2, facing west. 5-18 T2 scatter in right foreground around large eboulis blocks, rockshelter at back of terrace, facing east. 5-19 5-20 5-21 Ghrain Bliss inselberg from distance. Site of Ghrain Bliss 1 (SH1) on lower terrace. SH22 Lithic scatter on surface above basin. 5-22 Tertiary crust at Bir Khasfa eroded by alluvial and lacustrine activity. 5-23 5-24 5-25 5-26 5-27 Dhanaqr (T21) lithic scatter in foreground on slope of Tertiary rock outcrop, Wadi Dhahabun in distance, view facing southeast. Dr. Usik surveys the Jibal Ardif 3 (T27) lithic scatter. Fine-grained chert plaquette outcropping in vicinity of T27. Dr. Usik examines artifacts from al-Hatab (T29) that were found eroding from an exposed section at the base of the slope. Profile of intermixed colluvium and aeolian sediments at al-Hatab (T29). 6-1 Some platform types recognized in analysis. 6-2 Midpoint cross-sections. 6-3 Longitudinal profiles. 6-4 Some dorsal scar patterns used in analysis. 6-5 Location of measurements used in analysis. xv 6-6 Orientation of core platforms. 6-7 Some core types recognized in analysis. 7-1 Horizontal distribution of artifacts at Qarat al-Kibrit 1, Level I. 7-2 Radiolarite cobbles embedded in gypsum just outside of Qarat al-Kibrit salt dome. 7-3 Blade cores from Qarat al-Kibrit 1, Level I. 7-4 Blades (a — d) and core tablet (e) from Qarat al-Kibrit 1, Level I. 7-5 Tools from Qarat al-Kibrit 1, Level I. 7-6 Perforated shell bead from Qarat al-Kibrit 1, Level I. 7-7 A. al-Mahrooqi points out evidence for halite mining at Qarat al-Kibrit cave. 7-8 Horizontal distribution of artifacts from Qarat al-Kibrit 1, Level II. 7-9 Refit multiple platform core with alternating reduction strategy from Qarat al-Kibrit 1, Level II. 7-10 7-11 7-12 7-13 7-14 7-15 7-16 7-17 7-18 7-19 7-20 7-21 7-22 7-23 7-24 Refit Kombewa core from Qarat al-Kibrit 1, Level II. Biface thinning flake from Qarat al-Kibrit 1, Level II. Unifacial tools (with refits) from Qarat al-Kibrit 1, Level II. Distal fragment of Type 4 pressure flaked lanceolate from Qarat al-Kibrit 1, Level II. Horizontal distribution of artifacts from Qarat al-Kibrit 1, Level III. Biface thinning chip (a) and flakes (b — e) from Qarat al-Kibrit 1, Level III. Bifacial preform? (with refits) from Qarat al-Kibrit 1, Level III. Early stage bifacial preforms from Qarat al-Kibrit 1, Level III. Medial fragment of pressure flaked pedunculated arrowhead (Type 1) from Qarat al-Kibrit 1, Level III. Horizontal distribution of artifacts at findspot A7. Horizontal distribution of artifacts at findspot A10. Blade refits from A10 showing removal of débordants for convexity maintenance. Unidirectional-parallel blade cores from the A7-A10 complex. Unidirectional-convergent blade cores from A7-A10 complex. Bifacial tools from the A7-A10 complex. xvi 208 209 214 215 216 218 219 219 220 222 223 224 225 226 226 227 228 229 231 231 232 233 235 236 237 239 7-25 7-26 7-27 7-28 7-29 7-30 7-31 7-32 7-33 7-34 7-35 7-36 7-37 7-38 7-39 Early stage bifacial preform with refit from the A7-A10 complex. Biface with overpassed flake refit from A7-A10 complex. Blades and points with continuous, marginal, lateral retouch from A7-A10. Blade refits from A14 showing recurrent unidirectional method. Blade core from the A14 findspot. Horizontal distribution of artifacts at SH1. Horizontal distribution of artifacts at SH22 and SH23. Neolithic Type 6 bifacial lanceolate from outside SH22-SH23 collection areas. Radial core from SH22, possibly exhausted Levallois with centripetal preparation. Sidescraper from SH22 with Quina retouch. Bifacial preform from SH22 manufactured on a blade blank. Khasfian Foliates from SH22 and SH23. Horizontal distribution of artifacts at T21. Refit Kombewa flake-core from T21. Unidirectional-parallel blade refits from T21. 241 242 244 246 247 248 251 252 254 255 256 257 261 263 264 265 266 267 269 270 271 272 275 276 277 282 284 7-40 Truncation on Kombewa flake from T21. 7-41 7-42 7-43 Miscellaneous bifacial implement from T21. Horizontal distribution of artifacts from T27. Levallois cores from T27 Phase I. 7-44 T27 éclats de taille. 7-45 7-46 7-47 7-48 7-49 7-50 7-51 Late stage bifacial preforms from T27. Early stage preform from T27. Unretouched blade (a) and unifacial tools (b — c) from T29. Unidirectional-convergent (a) and unidirectional-parallel (b) blade cores from T29. Bifacial (a) and unifacial (b) scrapers from T29. 95% confidence intervals comparing core weights among the three core industries. 95% confidence intervals comparing complete blade metrics at T21 and T29. xvii 7-51 7-52 7-52 7-53 7-53 7-54 7-54 cont. 95% confidence intervals comparing complete blade metrics in Holocene, Nejd Leptolithic, and Sibakhan Industries. cont. 95% confidence intervals comparing complete thinning flake metrics in bifacial sites. cont. 95% confidence intervals comparing complete biface metrics in bifacial industries. cont. 285 289 290 293 294 295 296 304 326 327 332 8-1 Potential routes of migration out of Africa. 8-2 High-backed centripetal cores from Station One (a,b) and Bir Khasfa (c). 8-3 Bifacial foliates (b,d,f ) from Bir Khasfa and Station One (a,c,e). 8-4 Human phylogenetic tree based on mtDNA haplogroups. xviii LIST OF TABLES Table 2-1 4-1 4-2 6-1 6-2 6-3 7-1 7-2 7-3 7-4 7-5 7-6 7-7 Synthesis of palaeoclimatic data. Radiometric dates mentioned in text. Summary of Early and Middle Holocene taxonomic schemes by region.. Provinces and wilayat of Oman where survey was conducted during COPR 2004 campaign, with corresponding abbreviation. Qualitative technological observations used in attribute analysis. Qualitative observations used in typological analysis. Artifact class by site. Cortex percentage and ratios by site. Core types by site. Blank types by site. Scar patterns exhibited on blanks and working surface(s) of cores. Frequency of tool types. Proposed lithic industries. 51 87 141 185 188 204 299 299 300 300 301 302 281 xix Chapter 1 INTRODUCTION And we’ll be saying a big hello to all intelligent life out there...and to everyone else, the secret is to bang the rocks together. —Douglas Adams, The Hitchhikers Guide to the Galaxy The exploration of South Arabia has long been my ambition. This dream was affected by two particular experiences in my life that became the catalysts for this dissertation. The first event occurred at the age of  and marks my initial encounter with Arabia. A friend recommended reading Arabian Sands, an account of Wilfred Thesiger’s journey across the Rub al-‘Khali. From the first sentence of the book, I was entranced by his vivid portrait of this harsh, desolate land: A cloud gathers, the rain falls, men live; the cloud disperses without rain, and men and animals die. In the deserts of southern Arabia there is no rhythm of the seasons, no rise and fall of sap, but empty wastes where only the changing temperature marks the passage of the years. It is a bitter, desiccated land which knows nothing of gentleness or ease. Yet men have lived there since earliest times. (Thesiger :) Thesiger describes the heat-blasted arid landscape of the South Arabian interior—sprawling dune fields, deflated gravel pavements, and, occasionally, a dried lakebed or ancient wadi that could only have formed during wetter climatic regimes. What Thesiger did not know at the time is that the climate of South Arabia is inextricably linked to the Southwest Indian Ocean Monsoon system. There have been several episodes over the past quarter of a million years that have produced a shift in the magnitude of the yearly monsoons, leading to occasional phases of pronounced amelioration throughout the interior, and, conversely, periods of hyperaridity. 1 Even today, the Yemeni Highlands that mantle the southwestern corner of the Peninsula receive over , mm of rainfall per year from the monsoon. During ancient pluvial episodes, the vast Arabian deserts became rolling grassland—an extension of the East African savanna. In this light, South Arabia is perceived as a pump connected to East Africa, drawing in myriad plant and animal species during wet cycles, and forcing them to disseminate or die out during arid periods. It is posited that hominids were among those faunal species that episodically expanded from East Africa. The question of hominid expansion touches upon my second encounter with Arabia. I was in my early ’s, studying Levantine prehistory at Hebrew University in Jerusalem. While reading an article on Pleistocene connections between Africa and Southwest Asia, there was a paragraph that immediately struck a chord: This term [Levantine Corridor] is occasionally used by palaeontologists to designate the geographic route by which Miocene fauna moved from Africa into Eurasia (e.g., Thomas ). A similar dispersal route has been noted by botanists attempting to explain the presence of tropical Sudanese or East African species in Israel (e.g., Zohary ). Both faunal and floral researches have shown that the main passage was from Ethiopia and the Afar region into the southwestern corner of the Arabian peninsula, and then into the Levant. (Bar Yosef :-) Just days after reading that quotation, with the final sentence stuck in my head, I went on vacation to the eastern coast of the Sinai Peninsula. Much of that time was spent staring across the Red Sea, longing to explore the tantalizing hills of Saudi Arabia that rise up on the other side of the water. (Figure -). The Red Sea trough that separates Africa from Arabia is a rift valley formed by the Arabian tectonic plate sliding northeastward away from the African plate. From the point at which I was gazing across the water, the distance separating the continent of Africa from the Arabia Peninsula was probably not more than ten kilometers. Just as I yearned, for no reason in particular, to swim those few kilometers eastward toward Arabia, would not our hominid ancestors have felt the same innate curiosity? Particularly when motivated by hunger or rising population densities. The optimal location to carry out such a journey is at the southern extent of the Red Sea, where it constricts at a strait called the Bab al-Mandeb, a narrow crossing between Yemen and Djibouti/Eritrea. The Bab al-Mandeb is presently about  2 Figure -. View across the Bab al-Mandeb at sunrise, looking eastward at the Yemeni Highlands (photo by Tony Farmington, http://www.antaresii.co.nz). km wide at its narrowest point, with a small island in the middle called Perim. If one factors in tectonic rifting of the Red Sea, along with lower eustatic sea levels, the crossing would not have been more than ten kilometers during certain periods of the Pleistocene. This begs the question: did our hominid ancestors stare across the Red Sea (as I found myself doing on that day) and wonder what was on the other side? It is inevitable that at least some people would have journeyed off the African continent across the Bab al-Mandeb, driven by both the curiosity to explore, as well as the availability of untapped resources. These two observations—the dramatic oscillations of rainfall in Pleistocene Arabia, as well as the proximity and accessibility of South Arabia to East Africa—form the theoretical basis of this dissertation, bolstered by several recent genetic studies. Using mitochondrial DNA, researchers isolated a specific haplogroup shared among East Africa, South Arabian, Indian, and Southeast Asian 3 populations (Quintana-Murci et al. ). The authors interpreted this haplogroup as evidence for a modern human dispersal event out of East Africa during the Upper Pleistocene that spread across the Indian Ocean rim. In the last two years, other mtDNA and y-chromosome DNA studies have corroborated and added resolution to their claim (Luis et al. ; Forster ; Macaulay et al. ; Thangaraj et al. ). The genetic and palaeoanthropological data suggest modern humans developed in subSaharan Africa roughly , years ago. They remained bottlenecked in this region (Ambrose ), cut off by the hyperarid conditions in the Sahara and Arabian Peninsula during Oxygen Isotope Stage . The sudden climatic shifts at the onset of OIS e ( kya) and OIS  ( kya) resulted in a large ameliorated zone throughout the Saharo-Arabian arid belt that housed little or no indigenous populations. With the arid cork removed from the bottleneck, modern humans could have rapidly expanded into that niche. Subsequent fluctuations throughout the Last Interglacial may have caused repeated cycles of population expansion and contraction. This dissertation proposes a model of modern human expansion in the Upper Pleistocene from East Africa into South Arabia (the southern route out of Africa will henceforth be referred to as the “Arabian Corridor”). The purported human expansion(s) probably occurred at the onset of pluvial cycles, when there was a sharp increase in the intensity of the monsoon that led to ameliorated conditions throughout the interior. The Arabian Corridor expansion model will be tested by studying the variability and development of Pleistocene lithic industries in southern Arabia, based on the assumption that the emigrating population(s) will carry with them the same or similar technologies from whence they came. The development of lithic technologies is particularly relevant, since the oscillating climate drove a continual cycle of hunter-gatherer range movement into and out of South Arabia. Therefore, Upper Pleistocene lithic technologies in South Arabia must derive from populations in one or more contiguous zones that served as refugia during arid periods. Since climatic conditions may preclude autochthonous development, four possibilities are considered: ) an expansion from East Africa, ) an 4 expansion from the Near East, ) an expansion from India, and ) an amalgamation of technological traditions from all three regions. Unfortunately, there is a paucity of data regarding Pleistocene archaeology in southern Arabia and no established chronological sequence for the Palaeolithic. In order to address the issues raised in this dissertation, it was necessary to compile a catalog of prehistoric archaeological sites and findspots in South Arabia, which, in this work, encompasses all territory south of the Rub al-‘Khali (including the Rub al-‘Khali). This region was selected because it comprises the land most directly affected during intensified monsoon cycles. In addition to cataloging the existing data, I conducted three seasons of archaeological survey and excavation. The first two projects, carried out in eastern Yemen in  and Oman in , were preliminary reconnaissance expeditions to assess archaeological potential, made possible by a grant from the American Institute for Yemeni Studies and SMU Donor’s Fund grant, respectively. After visiting the Wadi Arah drainage system in , I chose the Nejd Plateau as the most promising region to recover Pleistocene lithic assemblages on the surface and in stratified deposits. The first full season of the Central Oman Pleistocene Research (henceforth COPR) program was initiated in , with funding from a National Science Foundation Dissertation Improvement Grant, SMU Donor’s Fund grant, and a grant from the Institute for the Study of Earth and Man. The team was comprised of Mr. Yaqoub al-Busaidi (liason), Dr. Vitaly Usik (lithic illustrations, refitting, and analysis), Dr. Diego Angelucci (geoarchaeology), Dr. Daniel Richter (luminescence dating), Mrs. Teresa Medici (artifact conservation), Mr. Ali al-Mahrouqi (guide and liaison), Mr. Chameez al-Asmi (Ministry of Heritage and Culture representative), Mr. Said al-Mahri (Mahri guide in the eastern Nejd), Mr. Yves Guichard (photography), and myself. From December of  to June of , archaeological survey produced nearly  new prehistoric sites throughout Oman (Appendix B), of which approximately two dozen were collected and/or excavated to obtain a sample with which to describe the variety of lithic technologies present in southern Arabia during the Pleistocene. Evidence is presented from both the catalog of pre-existing 5 sites and results of the COPR fieldwork to begin to construct a relative sequence and identify some Pleistocene lithic industries, sensu latu. Since Palaeolithic studies in South Arabia are still in their infancy, there is not yet an established terminology with which to refer to the archaeological sequence. Therefore, it is necessary to address whether or not to employ the African progression of Early, Middle, and Late Stone Ages (henceforth ESA, MSA, and LSA), or the Eurasian sequence of Lower, Middle, and Upper Palaeolithic (henceforth LP, MP, and UP). This is not simply a matter of nomenclature; the development of lithic technologies in sub-Saharan Africa versus North Africa and the Near East follows markedly different trajectories. The question of taxonomy underscores South Arabia’s unique geographic position as a bridge between Africa and Eurasia, and touches upon the fundamental issue addressed in this dissertation: the relationship of South Arabian lithic industries to the Near East and East Africa. Because the question of nomenclature rests on the conclusions reached in this dissertation, I will refrain from using archaeological phases when addressing South Arabia (with the exception of the Neolithic), instead, using only geological phases and oxygen isotope stages as temporal markers. When discussing assemblages in Africa or the Near East, the archaeological stages will connote both chronology and geography. In other words, the ESA, MSA, and LSA will not only indicate periods of time, but will also exclusively apply to material from sub-Saharan Africa. Conversely, LP, MP, and UP will apply to North Africa and the Near East. Regarding place names and stylistic conventions, I have arbitrarily selected from a wide array of transliterations a set of standard spellings (e.g., Hadramaut, Dhofar, Bab al-Mandeb), though when quoting other scholars I have kept the original, alternate spellings. The body of water sometimes referred to as the Persian Gulf is called the Arabian Gulf on the Peninsula; therefore, I have maintained this designation. Whenever possible, I attempted to use the English equivalent for foreign terms; however, there are a few instances where there is no adequate translation and the foreign word must be used (e.g., façonnage, sabkha, débordant). A glossary of the most common terms is provided in Appendix D. Arabic plurals are used appropriately in words such as wadi/widian, sabkha/sibakh, 6 jebel/jibal. The Arabic definite article has been changed in some cases so that it is in accordance with grammatical rules, depending on the first letter of the succeeding word (e.g., ad-Dakhliyah, ashSharqiyah, al-Wusta). The dissertation is organized as follows: Chapter —“Out of Africa and into Arabia”—more fully develops the Arabian Corridor Migration model. The chapter begins with a brief review of the Out of Africa hypothesis, which represents the paradigm upon which this dissertation is based. Recent genetic evidence is presented that indicates an African origin for modern humans, and subsequent dispersal into southern Arabia during the Last Interglacial. The palaeoenvironment of Arabia is seen as one important mechanism for driving Pleistocene hominid movement out of Africa, particularly during abrupt oscillations in the Upper Pleistocene. Climatic data are presented attesting to the hyperarid conditions that set in during glaciations that probably led to the extinction and/or dispersal of local populations. There is abundant evidence suggesting there were pronounced pluvial conditions during certain phases of the Last Interglacial. Floral and faunal studies have shown that East African taxa expanded into southern Arabia during these favorable episodes; the model presented in this dissertation predicts that modern humans were among those species to move across the Arabian Corridor. Therefore, the harsh climate of South Arabia sealed human populations into sub-Saharan Africa during particularly arid phases when it was too dry for hominid occupation, and then lured populations into its interior during pluvial episodes. Unlike contiguous refugia such as East Africa or the Levant, which exhibit more or less continuous human occupation from the Lower Pleistocene onward, Arabia would have been periodically abandoned, leading to an interrupted technological lineage representing episodic expansions into South Arabia from neighboring refugia. Assuming this is the case, southern Arabia presents an excellent opportunity to approach the question of human emergence out of Africa by examining the Pleistocene archaeology of this region. Is there evidence for African and/or Levantine presence in Arabian lithic technology, and what can the Arabian sequence tell us about the timing of the modern human expansion? On a broader level, 7 the study of the peopling of Arabia provides the opportunity to examine the behavior of huntergatherers colonizing a virgin landscape. Unfortunately, the Palaeolithic archaeology of southern Arabia is more or less terra incognita. Chapters  and  compile the history and current state of archaeological research, illustrating this point. While there is a rapidly growing body of data regarding Holocene archaeology, data about the Pleistocene remain scarce. Complicating the issue, there are potentially MSA and LSA elements within the lithic collections of southern Arabia that are presently associated with Neolithic assemblages. This uncertainty within the chronological sequence is exacerbated by the fact that no stratified Pleistocene sites have yet been discovered, and there have been no detailed technological analyses of potentially early material. The geographic and geologic settings of the sites are described in Chapter . While there was only one site with radiometric dating, it is still possible to draw some conclusions regarding the age and associated palaeoenvironmental features (e.g., ancient playas or relict drainage systems) by documenting the geomorphic setting of each locality. In cases where in situ deposits were excavated, descriptions of the sediments are provided. Chapter  describes the methods used during the  COPR season for locating and sampling the artifacts. A brief background of lithic analysis is presented; how it is carried out and why it is useful for studying prehistoric groups. This is followed by a description of the lithic attribute categories used in the analysis of the COPR assemblages. Chapter  presents the lithic assemblages recovered by COPR. The reduction sequences from seven surface sites and two stratified sites are presented. A combination of qualitative and quantitative lithic observations will be used to compare technological and typological variability. Synthesizing both the results of the lithic analysis and inferences based on site setting, a tentative classification of the assemblages is proposed. This classification is then considered in the broader context of the pre-existing body of data presented in Chapters  and , culminating in the definition of some new Palaeolithic industries in South Arabia. 8 Having established a tentative chronological sequence, Chapter  will return to the question posed in the beginning of this dissertation regarding hunter-gatherer range expansion into southern Arabia. This final chapter will assess the Arabian Corridor Migration model in light of the observed Pleistocene lithic technologies. What, if anything, can be learned about human emergence from the timing and origins of these expansions into Arabia? 9 Chapter 2 OUT OF AFRICA AND INTO ARABIA Weary bands of travelers in some shady haunt Among Arabian sands —William Wordsworth, The Solitary Reaper This dissertation concerns a particular group of travelers among Arabian sands. It is proposed that one or more neighboring hunter-gatherer groups expanded into southern Arabia during the late Middle and/or Upper Pleistocene ( –  kya), when favorable climatic conditions transformed the desiccated interior from desert into savanna, thereby facilitating a genetic bottleneck release(s) in East Africa that ultimately led to the modern human colonization of the globe by the end of the Pleistocene. This proposed scenario will be referred to as the “Arabian Corridor Migration Model.” Climatic ameliorations during the Last Interglacial in South Arabia began abruptly, and, at certain points, were quite pronounced in magnitude (Higginson et al. ). The inhabitants of nearby refugia—East Africa to the west and the Levant to the north—found themselves suddenly presented with a presumably uninhabited grassland, dotted with a plethora of playa lakes (McClure ). It is expected that groups from one or both regions moved into South Arabia at this time, based on the assumption that populations spread as the habitat in which they can survive expands (Lahr and Foley ). Genetic evidence points to an East African origin for the Upper Pleistocene inhabitants of southern Arabia (Quintana-Murci et al. ; Maca-Meyer ; Underhill et al. ; Forster ; Lovell et al. ; Forster and Matsumura ; Macaulay et al. ; Thangaraj et al. ). This is not surprising, given the geographic proximity and similar environmental conditions. Coming from the Horn, the expanding population would have moved into an environmental niche to which they 10 were already adapted. Floral data indicate that much of the modern vegetation in Dhofar is derived from East African taxa (Patzelt, n.d.), which probably spread into South Arabia during episodic ameliorations. This is corroborated by the presence of several species of medium to large carnivores in southern Arabia, whose origins can be traced back to East African lineages (Harrison ). To penetrate the peninsula, the African population needed only to cross the Bab al-Mandeb Strait; the southern outlet of the Red Sea, which constricted to less than ten kilometers during periods of lower eustatic sea levels (Sirocko ; Siddall et al. ). Whether ten or twenty kilometers, the voyage from Africa to Arabia required less knowledge of seafaring than the nautical technology necessary to colonize the island of Flores during the Middle Pleistocene (Morwood et al. ). Recent mtDNA studies regarding the evolution of Papio hamadryas baboons in Yemen estimates this species branched from a lineage in East Africa sometime between  –  kya (Wildman ; Wildman et al. ; Fernandes et al. ). In other words, baboons underwent a range expansion into South Arabia during the Last Interglacial. If monkeys found a way to navigate the Bab alMandeb, presumably modern humans could too (perhaps carrying said baboons, or vice versa). This chapter is divided into two sections: ) Out of Africa, and ) Into Arabia. The first section outlines the underlying paradigm upon which the Arabian Corridor Migration model is based; that is, Homo sapiens sapiens speciation in sub-Saharan Africa, and subsequent dispersal into Australasia during the Upper Pleistocene. Genetic evidence is presented in support of this model. There is then a review of the archaeology of Upper Pleistocene groups in the Horn of Africa, because if modern humans expanded from this region, the same or similar techno-typological elements are expected in South Arabia. Section two, “Into Arabia,” focuses on the backdrop of southern Arabia during the Upper Pleistocene. Geographic, geologic, climatic, and palaeoenvironmental data are presented. These data are relevant, as modern human dispersals into South Arabia were presumably linked to the timing and magnitude of pluvial events. By synthesizing these various lines of evidence, the proposed model of hominid range expansion(s) out of Africa and into Arabia is fully articulated. 11 Out of Africa: modern human origins, molecular anthropology, and Upper Pleistocene archaeology THEORIES OF MODERN HUMAN ORIGINS The nature of modern human origins has long been debated in the contrasting theories of total replacement (e.g., Bräuer and Rimbach ; Stringer ) versus multiregional evolution (e.g., Wolpoff et al. ; Wolpoff and Caspari ). Two other models worthy of note have been suggested that fall somewhere between the two extremes: ) “hybridization and replacement,” which accepts an African origin for all moderns, but allows for some degree of interbreeding between expanding modern humans and indigenous archaic populations (Bräuer ; Trinkaus ); and ) “assimilation,” which also assumes an African origin, but, rather than solely envisioning a demographic movement of populations, this model emphasizes a greater degree of gene flow and similar selective pressures to explain the widespread distribution of Homo sapiens sapiens (Smith ). Regardless of the degree of admixture with local populations, the case for a primarily African origin is indicated by the palaeoanthropological data. The earliest reliably dated Homo sapiens sapiens fossils (albeit with some archaic features) include:  –  kya at Omo-Kibish in Ethiopia (Butzer et al. ; Day ; McDougall et al. );  –  kya for the recently discovered Herto specimen, also from Ethiopia (Clark et al );  –  kya at Singa, Sudan (McDermott et al. );  –  kya on the teeth discovered at Mumba Cave, Tanzania (Bräuer and Mehlman ); and  –  kya on the anatomically modern human fossil remains at Klasies River Mouth in South Africa (Singer and Wymer ; Lam et al. ; Vogel ). Outside of Africa, remains from Skhul and Qafzeh in the Levant indicate modern humans had spread into that nearby region around  kya, possibly as early as  kya (Schwarcz et al. ; Stringer et al. ; McDermott et al. ; Mercier et al. ). Sometime prior to  kya, modern humans arrived in Australia, indicated by the Lake Mungo  and  modern human specimens (O’Connell and Allen ). The recent Homo sapiens sapiens remains from Petera cu Oase in Romania yielded  dates of around  kya, representing the earliest evidence for modern humans in Europe (Trinkaus et al. ). 12 There are, however, some inherent problems in the use of fossil record to address human origins: …the term modern human encompasses a variety of human biological groups, which vary geographically in the recent world and have continued to evolve since their emergence in the later Pleistocene. Moreover, it may be difficult to draw clear distinctions between them and late archaic humans, depending on the criteria and approach employed to delimit “modern humans.” (Trinkaus :) What does it mean to be anatomically modern? Which metric variables are germane for discerning morphological diversity on a species level? There is not necessarily universal agreement on this point, which leads to differing taxonomic classifications and interpretations for some of the specimens (e.g., Wolpoff and Caspari ; Stringer ; Wolpoff et al. ). Since the Arabian Corridor Migration Model describes events well after the appearance of the first anatomically modern humans, the timing of their speciation is of no concern to this dissertation. It is assumed the primary locus was sub-Saharan Africa, overwhelmingly supported by the genetic evidence presented below. What is of concern are the demographic shifts that culminated in the nearly complete modern human colonization of the globe by the Terminal Pleistocene. When did moderns first leave Africa, how many waves of dispersal were there, and what route(s) did they take on their way out? It is predicted that the first pulse out of Africa—a genetic bottleneck release—occurred sometime around OIS e, at the onset of the Last Interglacial (Ambrose ). Prior to this, climatic conditions during the Penultimate Glaciation (~ –  kya) were manifest in parts of Africa by intense aridity that caused the Saharo-Arabian Arid Belt to act as a cork sealing mammalian fauna into sub-Saharan Africa (Tchernov a, b, c, ). Scholars have observed that during interglacials, when more humid conditions prevailed, the Ethiopian faunal range shifted northward to encompass the Sahara, northern Africa, and the Levant (Payne and Garrard ; Tchernov a). Modern human remains from Skhul and Qafzeh, dated between  –  kya, provide evidence for an early wave of dispersal during this bottleneck release; described as “populations tracking the fluctuations of their native habitats” (Lahr and Foley :-). 13 The success of this expansion has been the subject of debate, due to certain taxonomic ambiguities within the Levantine fossil record. Hominid remains classified as Neanderthal were uncovered at Kebara Cave that date to between  and  kya (Rak and Arensburg ; Valladas et al. ), although it is generally agreed upon that these Levantine Neanderthals possess more gracile features than their European cousins. Since they post-date the Skhul/Qafzeh modern humans, a model has been proposed in which Neanderthals were forced southward into the Levant during OIS , replacing the modern humans that once occupied this region (Bar-Yosef ). A recent statistical analysis of cranial morphology among the Skhul/Qafzeh modern humans and Levantine Neanderthals was conducted to test whether or not they truly represent different populations. The results of this analysis suggest that there was no separate Neanderthal clade; “the traditional ‘Neanderthal’ versus ‘modern human’ groupings in the Levant may not be as distinct as often thought” (Kramer et al. :). From these findings, the authors propose a model based on admixture, in which modern humans left Africa during OIS  and interbred with indigenous Levantine populations. This model is the subject of much debate, as ancient DNA studies suggest otherwise. At the heart of the issue is the fact that the Skhul/Qafzeh fossils are yet fully modern; therefore, this is probably not the population that gave rise to the haplogroup M expansion from East Africa. Following OIS , there is another bottleneck around , years ago, apparent in the genetic history of both modern humans and eastern chimpanzees (Ambrose ). Certain scholars correlate the sudden and drastic population contraction at this time with the volcanic super-eruption of Toba in northern Sumatra, which ejected over  cubic kilometers of ash and millions of tons of sulfur gas into the atmosphere, triggering a , year “volcanic winter” (Rampino and Self , ; Ambrose ; Rampino and Ambrose ). As a result, they speculate the female population in Africa would have dropped to approximately , individuals. Following this demographic catastrophe, there was a second bottleneck release associated with OIS , which eventually led to the successful colonization of Eurasia. There is a problem with this scenario, however, as Toba ash does not occur in the coldest portion of the ice cores, as the model predicts. 14 Aside from the Skhul/Qafzeh fossils, there is no direct evidence for modern human migration out of Africa. On the contrary, comparative analysis of Northeast African and Levantine MP assemblages along the Levantine Corridor suggests there were no compelling technological connections between these two regions during the Upper Pleistocene (Marks ). Examination of lithic technological trajectories in the Levant demonstrates a single blade tradition from the late Lower Palaeolithic to the Upper Palaeolithic: An intrusive influx of a small group of Aurignacians carrying this Amudian/Pre-Aurignacian is belied by the context of the Yabrudian, in which both the Amudian and Pre-Aurignacian occurs, and which, at least at Tabun, shows no demonstrable discontinuity with preceding or succeeding lithic industries...overall similarities in the blade reduction systems and their concomitant typological similarities through time in the Levant, however, indicate that there was an unmistakable Leptolithic tradition with roots in the Lower Palaeolithic. (Monigal :-) As for expansion beyond the Near East, there was no successful modern human movement into Eurasia until  –  kya. Klein () attributes this , year lag between modern human speciation and expansion to the fact that anatomically modern ancestral groups were not “behaviorally” modern until ~ kya; therefore, they were not equipped to expand into more challenging environmental niches or replace indigenous archaic groups. The aforementioned problems in the identification of “anatomically” modern humans outlined by Trinkaus (), also apply to the concept of “behavioral” modernity. Raw material procurement strategies, mobility patterns, and non-lithic remains are crucial lines of evidence to understanding the transition to behavioral modernity, unfortunately these data are not available within the archaeological record of southern Arabia. As such, the issue is given little weight in this dissertation, and the only recognized Upper Pleistocene behavioral distinction is the shift from MSA/MP to LSA/UP technologies, simply because this is the archaeologically visible material from South Arabia. While Klein’s () hypothesis may be germane to the peopling of Eurasia, the colonization of the Arabian Peninsula was not subject to the process he proposes. There was no technological advance that allowed hunter-gatherers to expand into Arabia; rather, their known habitat simply spread there. In this light, the Arabian Corridor Migration Model predicts the influx of MSA/MP 15 elements into South Arabia (although later LSA/UP technologies may certainly be present), because the expansion(s) are associated with pluvials during OIS  and OIS , which precede the LSA transition in Africa. Which route(s) did the early modern humans take during their exodus from Africa? In addition to the Levantine Corridor, there are two other potential conduits leading off the continent: the Strait of Gibraltar and the Bab al-Mandeb. Regarding Gibraltar, Straus (:) writes “peopling of Iberia from northwest Africa during the mid-Upper Pleistocene seeming (ironically) to be out of the question, as southern Spain and Portugal were one of the last refugia of Neanderthals…” So, the archaeological evidence (debatably) argues against significant movements across the Levantine Corridor or Gibraltar after OIS , leaving only the Arabian Corridor as a viable route for the mid-Upper Pleistocene expansion out of Africa. Consequently, a number of scholars have speculated about human emergence along this conduit, employing genetic, anthropological, and biological evidence, as well as computer simulations (e.g., Tchernov b; Lahr and Foley ; Cavalli-Sforza ; Rose ; Stringer ; Rose a, b; Petraglia and al-Sharekh ; Field et al. ; Mellars ). One simulation program entitled STEPPINGOUT was used to examine statistical probabilities for hominid migration routes out of Africa. The simulation found that, when the Arabian Corridor was excluded as a possible exit point, the likelihood that hominids would leave Africa was cut in half, leading the analysts to conclude “this route [the Arabian Corridor] was of greater significance than the Nile Valley for hominid dispersal” (Mithen and Reed :). A more recent GIS-based computer simulation assessed the most parsimonious expansion route into South Asia during OIS . Using what they term “friction surface” tests to model the pace and direction of hunter-gatherer movements, the analysis concluded that populations originating in Africa and the Levant “would have been more likely to enter South Asia if they took a southern route” (Field et al. ). There is not yet enough evidence to adequately evaluate the Arabian Corridor from an archaeological perspective. Clearly, neither the current body of palaeoanthropological nor 16 archaeological data are enough to explore the timing and route of the modern human exodus from Africa. With the recent application of genetic research, however, scholars have been granted a new and effective means with which to address these questions. GENETIC EVIDENCE FOR MODERN HUMAN ORIGINS Nearly  years ago, the field of palaeoanthropology was augmented by a new and vastly useful tool for examining human origins and diversity. The team of Cann, Stoneking, and Wilson applied genetic data to probe the question of modern human emergence (Cann et al. ). Theirs and subsequent researches are based on the premise that mitochondrial DNA is a potent tool for measuring the variability and antiquity of human populations because: ) it has a copy rate of . substitutions per base pair per million years, which is ten times that of nuclear DNA ; ) a high substitution rate; ) there is no recombination; and, most importantly, ) it has an exclusively matrilineal mode of inheritance. Their analysis concluded that all modern human groups can be traced back to a small ancestral population that evolved in sub-Saharan Africa some , generations ago, roughly between , and , . This pioneering work was soon followed by additional publications that corroborated and added further resolution to the timing and location of the ancestral population, by incorporating Y-chromosome markers, microsatellites, and chromosome- markers to carry out phylogenetic identification and to measure the timing of evolutionary events.(e.g., Vigilant ; Horai et al. ; Tishkoff et al. ; Hammer et al. ; Pandya et al. ; Tishkoff et al. ; Wallace et al. ; Jin et al. ; Ingman et al. ; Ke et al. ). In addition to mtDNA, these recent analyses have Collectively, they predict the ancestral population lived between , and , years ago in East Africa. This application of genetics is not without criticism. Some scholars (e.g., Templeton ; Tamura and Nei ; Templeton ; Wolpoff and Caspari ) point out that the extreme variation in substitution rates between groups, as well as parallel mutations that cause difficulties in the estimation of genetic distance, call into question the phylogenetic inferences (Tamura and Nei 17 ). Furthermore, they argue that the method once used for these analyses—restriction-fragment length polymorphism (RFLP)—is ill-suited as a tool for calculating mutation rates and, therefore, the timing of evolutionary events. This argument against RFLP’s is outdated, as more reliable methods are now used such as sequencing hypervariable segment I (HSV-I). Despite the critiques against the use of genetic data, this dissertation has been written under the assumption that molecular anthropology is a reasonable tool for examining human origins. In fact, the archaeological data presented and analyzed in this work will serve as an independent line of evidence with which to verify a genetically-predicted human dispersal event (Quintana-Murci et al. ; Maca-Meyer ; Underhill et al. ; Forster ; Lovell et al. ; Forster and Matsumura ; Macaulay et al. ; Thangaraj et al. ). The recognition and analysis of genetic haplogroups make it possible for molecular anthropologists to examine ancient demographic movements. Haplogroups are groupings of chromosome branches with the same genetic characteristic, such as restriction enzyme recognition sites or deletions at the same haplotype location on the DNA; they are defined on the basis of unique mutation events or character states. Variability within a specific haplogroup is then used to measure the age of that phylogenetic group, and, consequently, the point at which it diverged from an older population. This method of analysis has been used to assess modern human dispersals through the Levantine (Kring et al. ; Luis et al. ) and Arabian Corridors (Quintana-Murci et al. ; Maca-Meyer ; Underhill et al. ; Forster ; Lovell et al. ; Forster and Matsumura ; Macaulay et al. ; Thangaraj et al. ). All of these dispersal analyses begin with the recognition of superhaplogroup L: an mtDNA haplogroup ubiquitous among all Eurasians and the preponderant haplogroup within East African groups, used to trace the coalescence from the ancestral population (Excoffier and Langaney ; Vigilant ; Watson et al. ). Branching from superhaplogroup L are haplogroups M, N, and R: they are basal non-African clades that are found among Eurasians, but absent in Africa. Haplogroup M is an East Asian lineage (Quintana-Murci et al. ), while haplogroup N branches off into western Asia and Europe (Forster and Matsumura ). Haplogroup R is a rapid offshoot from 18 haplogroup N, and appears to be a lineage originating on the Indian subcontinent (Metspalu et al. ). There is one interesting exception to this; a recent genetic analysis of populations in Ethiopia revealed a high incidence of haplogroup M, which is, to date, the only reported evidence for this haplogroup in sub-Saharan Africa (Quintana-Murci et al. ). The authors describe three possible scenarios to explain the presence of haplogroup M in Ethiopia: ) it evolved through independent mutation; ) it was acquired through recent exchange with Asian populations; or ) it was originally present among the ancient Ethiopians and subsequently expanded into Asia. The team used both RFLP and sequencing of the mtDNA hypervariable segment I (HVSI) to analyze a sample of Ethiopian and Indian populations. These methods were used to calculate the coalescence time of haplogroup M within the two groups. They concluded that “haplogroup M already existed in eastern Africa approximately , years ago,” prior to the divergence that gave rise to the Indian population (ibid.:). Furthermore, the authors note that haplogroup M is conspicuously absent in the Levant, supporting the notion that there was minimal mixture between populations living within the Arabian and Levantine Corridors. A subsequent study examined modern Egyptian and Omani Y-chromosome DNA to evaluate the extent of human migrations across the Levantine versus the Arabian Corridors. Researchers concluded that the Levantine Corridor provided most of the genetic input into Eurasia, while “genetic flow through the Horn of Africa during these demic episodes was very limited” (Luis et al. ). I argue that any attempt to apply information obtained from modern urban South Arabian populations to Pleistocene demographic movements will be inherently flawed, as all archaeological and climatological data point to discontinuous habitation during the Terminal Pleistocene. The southern portion of the peninsula appears to have been abandoned due to extreme hyperarid conditions at the last glacial maximum, then repopulated during the early Holocene; therefore only supplying genetic evidence from the last ~, years. Furthermore, there is ambiguity from where early Holocene groups originated, although some have pointed out affinities with PPNB inhabitants 19 of the Levant (Edens ; Potts ). For this reason, the findings of Luis et al. () should be discounted in respect to Upper Pleistocene demographics. There are three related groups in Arabia that may be of greater antiquity than the Holocene: the Mahri, Shahri and Soqotri peoples who occupy pockets of Dhofar, eastern Yemen, the Nejd Plateau, and the island of Soqotra. They exhibit a different morphology from other Arabian peoples and speak an offshoot of proto-Semitic that is only distantly related to modern Arabic. If the Dhofar Mountains were a refugium during the last glacial maximum, the elusive evidence for occupation in Arabia during the last glacial maximum may be buried within the caves that riddle the escarpment. These facts beg the questions: ) when did the Mahri, Shahri, and Soqotri speaking groups enter Arabia, and ) what is their genetic relationship to modern groups? It would be useful to conduct genetic analyses of these populations to address these questions. Two recent analyses of haplogroup M were carried out on aboriginal populations in Malaysia (Macaulay et al. ) and the Andaman and Nicobar Islands (Thangaraj et al. ). Because these groups are considered remnants of the founding populations and have undergone minimal genetic exchange with more recent immigrants to the region, they provide useful data for exploring the origins of haplogroup M. Researchers report two previously unknown clades derived from haplogroup M: M and M, concluding that these clades developed independently in the Andaman Islands from other southeast Asian lineages. They calculate M and M were isolated from the original founder population between  and  kya (Thangaraj et al. , ), which supports the timing proposed by Quintana-Murci et al. (). Analysis of haplogroup M among the Orang Asli indicates these indigenous inhabitants of Malaysia diverged from the founder population some  to  kya, further bolstering the predicted age of coalescence (Macaulay et al. ). In light of these genetic data, the theoretical issue of whether or not early modern humans could cross the Bab al-Mandeb has seemingly been resolved. However, questions regarding the timing of the expansion, as well as the characteristics of the migrating population can only be addressed by incorporating archaeological data. Therefore, the following section examines MSA and early LSA sites in East Africa—the loci and timeframe of the genetically-predicted ancestral modern humans. The 20 chronological sequence and lithic techno-typologies will be reviewed, in order to construct a rough picture of the emigrating population(s). EAST AFRICA DURING THE LAST INTERGLACIAL The sub-Saharan lithic sequence follows its own trajectory of technological development, considerably different from that of North Africa, the Levant, and Eurasia. The MSA begins as early as a quarter of a million years ago (Wendorf et al. ; Deino and McBrearty ), and lasts until ~ –  kya (Ambrose ). East Africa encompasses a large territory covering several ecosystems; Upper Pleistocene lithic assemblages reflect this geographic expanse in their techno-typological variability (Clark ). There appear to be regional differences between sites stretching from the Horn of Africa (e.g., Clark ; Wendorf and Schild ; Clark et al. a) and those further to the south in Kenya and Tanzania (e.g., Merrick ; Anthony ; Ambrose ), the most prominent of which is the high frequency of unifacial/bifacial points within assemblages from the Horn, versus their rarer occurrence further to the south. As such, this review of the East African archaeological record will focus on MSA/ early LSA sites primarily from the Horn of Africa (Somalia, Eritrea, Ethiopia, and Djibouti) given their proximity to the Bab al-Mandeb. While there are a number of well-preserved, stratified Upper Pleistocene sites from this region such as Porc Epic (Teilhard de Chardin et al. ; Breuil et al. ; Perlès ; Clark and Williams ; Clark et al. a), K’One (Kurashina ); Gorgora (Moysey ; Leakey ), Midhishi  (Brandt and Brook ; Brook and Gresham ; Gresham ), Bur Eibe (Graziosi , Clark ; Graziosi ), and Bur Hakaba (Clark ), very few have produced reliable absolute dates. Instead, the chronological sequences have been based on associated geomorphic contexts and stratigraphic correlations. Another major problem that plagues African archaeology is nomenclature. A variety of industries were named and defined in the first half of the th century, but by the ’s many of these lithic entities were demonstrated to be spurious. The past few decades have been characterized 21 by mass “un-naming” of industries (e.g., the Stillbay question discussed below), although without creating alternative definitions or groupings. At present there is a taxonomic vacuum in East Africa; each assemblage is considered invidually but not placed in a broader context. Therefore, this dissertation has chosen to resurrect some of the discarded names that may still possess germane elements. That is not to say, for instance, that the Stillbay mentioned in this dissertation has any relationship to the South African Stillbay, or Magosian to the type site in northern Uganda (later shown to be a mixed assemblage). Somaliland Stillbay might be called Somaliland Joe, but rather than reinvent the wheel, it is simpler to use designations that are already familiar to most readers. Clark’s () was the first major attempt to construct an MSA/LSA chronological sequence for the Horn. Three MSA “culture complexes” were defined: Acheulo-Levalloisian, Levalloisian, and Somaliland Stillbay; and four LSA complexes: Magosian, Hargeisan, Doian, and Wilton. The latter two LSA entities date to the Terminal Pleistocene and are beyond the scope of this dissertation; therefore, they are excluded from this discussion. Since the Somaliland Magosian appears to evolve locally from the Somaliland Stillbay and exhibits different characteristics then its coeval namesake in Kenya, for the purposes of this review the Magosian of the Horn will be considered a separate and unrelated entity. The earliest phase under consideration, Acheulo-Levalloisian/Abyssinian Fauresmith, is characterized by handaxes, cleavers, and some Levallois core reduction. Dates for this ESA/MSA transitional industry are unknown in the Horn, although work at the Kapthurin Formation in Kenya suggests the Final Acheulean in East Africa occurred sometime around  kya (Deino and McBrearty ). If one accepts the earliest date for the MSA at  kya (Wendorf et al. ), then the ESA/ MSA transition would fall within the range of  –  kya. The subsequent “Levalloisian” is characterized by Levallois and disc cores, platform faceting, core-scrapers, handaxes, and a low frequency of unifacial and bifacial points. There are no dated Levalloisian sites, and the taxonomic category has since been proven invalid (Clark ), which explains why recent excavators “have failed to recognize this inadequately defined complex” (Brandt :). 22 The Somaliland Stillbay is the most prominent MSA complex, both in terms of geographic distribution throughout the Horn and relatively long life spanning the Penultimate Glaciation and Last Interglacial. Presentation of the data from this phase is difficult, however, because no commonly accepted methodological conventions have been used in lithic analysis. A germane example of this is the problem of classifying the variety of point types that occur in the late Middle/Upper Pleistocene, which has led to some typological ambiguity in the literature. Leakey () first applied the terms “proto-Stillbay points” and “Stillbay points” to refer to unifacial/partly-bifacial/bifacial implements with extensive flat, invasive retouch that are common in MSA East African assemblages. Clark () refined this tool type by adding “Lower Somaliland Stillbay” and “Upper Somaliland Stillbay,” based on relative stratigraphic correlations of his findings throughout the Horn. The types are morphologically distinguished by diminishing size and increasing quality of manufacture, although the technology itself remains the same. Wendorf and Schild () classify nearly identical implements from Gademotta/Kulkuletti as “bifacial points” and “Mousterian points.” Kurashina () calls morphologically/technologically tools similar Clark’s “proto-Stillbay/ Lower Somaliland Stillbay” tools from Gadeb “Acheulean handaxes” (albeit flat and symmetrical). He uses the terms “bifacial/parti-bifacial point” to describe the later, more diminutive MSA variety at K’One , which is mirrored by Mehlman () in his description of the Mumba Cave assemblage. On paper they are given different names, although illustrations and descriptions indicate they are all the same, distinct tool type. Conversely, the “unifacial/bifacial points” identified by Merrick () and Anthony () at Prospect Farm do not exhibit the same technology used in production; rather the retouch is more crude, semi-steep, and they are typically more triangular in shape—in this case same name, different beast. In addition to points, Somaliland Stillbay assemblages exhibit Levallois cores (both centripetal and unidirectional-convergent), flat and biconical discoids, some backed pieces, a high frequency of scrapers, and occasional burins. Clark (:) argued that “tools exhibiting undoubted evidence of pressure flaking are found,” which, if true, is among the earliest evidence for use of this knapping technique. At Porc Epic, Somaliland Stillbay points account for  of all shaped pieces, 23 the highest frequency in East Africa and presumably a testament to their importance within this MSA toolkit (Clark and Williams ; Clark et al. a). Obsidian hydration dates on Somaliland Stillbay artifacts from Porc Epic cluster between  and  kya (Michels et al. ), although the technique of obsidian hydration dating has repeatedly been shown to be ineffective (e.g., Ridings , ; Anovitz et al. ; Hull ); therefore, this dissertation considers these dates with skepticism. Potassium-Argon measurements on the sediments at Gademotta and Kulkuletti provide an age range between  –  kya (Wendorf and Schild ; Wendorf et al. ; Wendorf et al. ). Both radiocarbon and thermoluminescence dating at Midhishi  indicate an age greater than ,  (Brandt and Brook ; Gresham ; Brandt et al. ; Brandt ). Outside of South Africa, the term Stillbay is no longer used. Anthony () addressed the efficacy of the “Stillbay Culture” as a valid taxonomic unit by examining Stillbay sites from South Africa to the Horn. She concluded that: …it can be said that not only do the assemblages contained within it differ decidedly, but also the very concept underlying the term Stillbay Culture is out of date…One cannot, in any way, justify the sorting out from this general mass a feature entitled the “Stillbay Culture,” for its assemblages resemble one another no more closely than they do those that have been excluded from this pseudo-culture…the Stillbay Culture does, in fact, not exist. (Anthony :) Although Anthony dismisses the concept of an all encompassing South and East African techno-complex (rightly so), she explicitly states that her analysis “excludes the Somaliland Stillbay, so ably defined by J.D. Clark in his volume on the Horn of Africa” (ibid.:). Unfortunately, many ignore this point and have since thrown the baby out with the bathwater. For this discussion, I choose to resurrect the taxonomic category of Somaliland Stillbay as a valid, distinct MSA cultural unit, albeit completely unrelated to the South African Stillbay. It is considered an industry, sensu stricto, based on: ) the morphology of the finished tools, ) the relative frequency of finished tools, ) technological processes involved in their production, which includes all stages of the reduction sequence from raw material procurement to discard, and ) these same or similar techno-typological 24 characteristics must occur repeatedly within various assemblages that are more or less coeval and bounded in space (Marks, n.d.). The Somaliland Stillbay is geographically circumscribed: to the west is the SangoanLupemban complex, which some have associated with the climatic boundary between humid Central Africa and the drier savanna of East Africa (Isaac ; McBrearty ); to the north the Nilotic industries of Egypt and the Sudan are almost exclusively characterized by prepared core reduction (e.g., Marks a, b); as one moves southward down the Rift the frequency of unifacial/bifacial points and extent of invasive retouch used in their production decrease considerably (e.g., Merrick ; Anthony ; Ambrose , ). Upon examining the MSA material from Units V and VI at Mumba Cave in northern Tanzania, Marks observed a complete absence of bifacial tools, along with a high degree of Levallois and bipolor reduction. In contrast to core reduction strategies from the Horn, there is little to no unidirectional convergent method of convexity maintenance in any strata at Mumba (Marks, personal communication). Somaliland Stillbay appears temporally bounded as well, developing out of the “protoStillbay,” which was marked by large, flat bifaces at early MSA localities such as Melka Kunture (Bailloud ; Hours ; Chavaillon et al. ), the Kapthurin Formation (Deino and Mcbrearty ); and Gadeb (Kurashina ). Moreover, this same tradition of unifacial/bifacial point production carries through to the early LSA; morphologically similar points, although more diminutive, are found among derivative Hargeisan and Magosian industries in the Horn (Clark ). There is growing evidence that the northern boundary of the Somaliland Stillbay extended into the Sahara at certain periods of the Upper Pleistocene. Assemblages with Somaliland Stillbay techno-typological characteristics have been reported from the Eastern Desert in the Sudan (Rose b), Sai Island (Van Peer et al. ), and the Western Desert in Egypt (Wendorf et al. ), which suggests these findspots represent occasional range expansions. While exploiting savanna ungulates, hunter-gatherers in the Horn shifted northward as the  –  mm rainfall zone expanded to northern Sudan and transformed the desiccated region into grasslands. Conversely, 25 Nubian Type I cores were reported from K’One, suggesting links with Northeast Africa were bidirectional (Kurashina ). An MSA site was recently reported from the Red Sea coast of Eritrea, where lithic tools, seemingly Somaliland Stillbay in character, were found embedded in a marine terrace (Walter et al. ). Called the Abdur Reef Limestone, this relict shoreline ranges from  –  masl, and is comprised of shells and corals deposited during a phase of marine transgression. Uranium-Thorium measurements on the fossil corals indicate an age of around  kya. Flakes, blades, and handaxes were reported from a number of loci within the exposure, manufactured primarily on obsidian, with occasional chert and quartz specimens. Most noteworthy was a quartz bifacial tool found at findspot AN- (ibid.: fig. c), which appears to be a diagnostic Somaliland Stillbay point. Although the Abdur Reef Limestone is located on the coast, a marine subsistence strategy should not necessarily be assumed (contra Stringer ). There are no fish bones within the deposit, it is unclear if the shellfish found with the artifacts are naturally-deposited or associated with human occupation, and the presence of fauna such as elephants, hippopotami, rhinoceros, and various bovids within the terrace indicate the availability and probable exploitation of terrestrial game. Regardless of their subsistence strategy, this site provides evidence for human occupation within the East African littoral during the Last Interglacial. Hence, it is likely that coeval groups inhabited the nearby Djibouti coast to the south, which straddles the west side of the Bab alMandeb. Lithic data regarding the MSA/LSA transition in the Horn is scarce, and there are no absolute dates directly associated with any of the assemblages. The earliest LSA industries are Magosian and Hargeisan, distributed throughout the southern and northern halves of the Horn, respectively. These industries are considered transitional based on the presence of elements from both the MSA (e.g., unifacial/bifacial points, sidescrapers, discoids) and LSA (e.g., backed pieces, microliths, blade/lets). Their attribution as early LSA rather than terminal MSA is arbitrary; theoretically, they could belong to either phase. 26 Regarding the Magosian, Clark (:) argued it “may be considered to be directly derived from the Somaliland Stillbay, which immediately underlies it at some sites.” This industry is best represented at the stratified sites of Bur Eibe and Bur Hakaba in southern Somalia (Graziosi , ; Clark ; Brandt ). A Magosian-like industry overlying the Somaliland Stillbay was reported at Porc Epic (Breuil et al. ); however, subsequent work demonstrated this material to be the result of vertical mixing (Clark and Williamson ). The Somaliland Magosian toolkit is comprised of unifacial/bifacial implements similar to the Somaliland Stillbay point, although typically smaller and with finer-retouch. Shapes include lanceolate, foliate, limace, sub-triangular, and triangular. There are also backed blades, microliths, outils écaillés, endscrapers, sidescrapers, thumbnail scrapers, and some burins. Magosian core types include blade cores (single-platform, multiple-platform, and bidirectional), as well as discoids and partial-discoids. In addition to the lithics, ostrich eggshell beads, hematite, and utilized bone have been reported. In the absence of radiometric dates, the timing of the Magosian is unknown, although based on its relative stratigraphic position at the sites where it has been found, it would appear to date sometime between  –  kya. The buried site of Aladi Springs in Ethiopia yielded an assemblage with “MSA points,” microliths, sidescrapers, endscrapers, Levallois cores, and microlithic cores. The artifact-bearing stratum, a thick calcareous loam, is correlated with the OIS  humid phase (Williams et al. ; Clark and Williams ; Gasse et al. ), although subsequent work suggests the deposit may be mixed (J.D. Clark and K. Williamson, personal communication reported in Brandt :). One of the most interesting, anomalous elements of the terminal MSA/initial LSA is the Hargeisan Industry: A local and probably hybrid form…at first glance it would seem that these northern Somaliland industries are but a local form of the Magosian, but a detailed study shows that in the angleburins and end-scrapers (which are as good as any found with the true North African Capsian) we have forms which are entirely foreign to the Magosian, and clearly demonstrate that we are dealing with a distinct cultural complex. (Clark :-) 27 The industry was first discovered overlying Somaliland Stillbay in an exposed section of red sandy alluvium near the city of Hargeisa, for which it was named (ibid.). The artifacts were found in situ, in sediments indicating arid/semi-arid climatic conditions, correlated with the end of the Gamblian and First post-Pluvial wet-phase (OIS  and OIS ). Since its discovery at Hargeisa, the only other significant contribution to our understanding of this industry comes from Midhishi , a cave within the Golis Mountains of northern Somalia (Brandt and Brook ; Brook and Gresham ; Gresham ). Hargeisan lithic assemblages consist of prismatic blades and blade cores (single platform and bidirectional), large backed blades, bladelets, microliths, well-made burins (bec-de-flute), endscrapers on blades, and the occasional biface resembling a diminutive Somaliland Stillbay point with extensive flat, invasive, soft hammer retouch. There are flakes removed from discoidal or radially prepared cores that exhibit platform faceting (i.e. Levallois). Ostrich eggshell is often associated with these findspots; at location H.R at Hargeisa, a cache of several hundred pieces was found (Clark ). Two other noteworthy findspots were reported east of Boosaaso in northeastern Somalia (Grasiozi ). There are few published details regarding these assemblages, although Brandt () identified microliths, endscrapers, unifacial/bifacial points, and blades within the assemblages, tentatively associating it with the Hargeisan. The uppermost layer of Midhishi  is described as Hargeisan, with an associated radiocarbon date of , ±   (presumably a minimal date for the industry based on its correlation with an interglacial phase at Hargeisa). Excavators describe the assemblage as “transitional” in character, due to the presence of typical MSA tools manufactured on blanks from Levallois and discoidal cores, microliths, endscrapers, and blade cores. Comparing the Somaliland Stillbay and Hargeisan strata, they note a reduction in the frequency of “MSA [Somaliland Stillbay] points” and Levallois cores over time, while endscrapers become more frequent. In addition, the Hargeisan unifacial/bifacial points are smaller than their Somaliland Stillbay predecessors. The enigmatic nature of the Hargeisan allows for creative speculation regarding its origins. It represents the only Upper Pleistocene entity in East Africa with a “true” blade technology—the 28 hallmark of the Upper Palaeolithic outside of sub-Saharan Africa, but not a particularly common feature of the LSA, which is primarily based on the production of microliths. There is no previously established leptolithic tradition in East Africa predating the Hargeisan. Furthermore, the distribution of the Hargeisan in northern Somalia, in proximity to the Bab al-Mandeb, raises the possibility of “cultural” admixture of populations on either side of the Red Sea. Either the Hargeisan represents a local development that happens to develop concurrently with other Upper Palaeolithic trajectories elsewhere in the world, or it may be that the Hargeisan is an intrusive element, indicating movement of either people or technology into Africa. Such bidirectional movement(s) along the pathways out of Africa have previously been suggested (emphasis added): There is, however, no reason a priori to exclude the possibility that intercontinental contacts occurred on a two-way street, especially at Suez, via Sinai, or across the shallow Bab al-Mandeb, so close to that corridor to sub-Saharan Africa, the Nile. (Straus and Bar-Yosef :) Having presented the underlying Out of Africa model, genetic evidence supporting this theory, and potential archaeological characteristics of the purported ancestral population(s), this chapter will now shift across the Red Sea to examine the climate, geography, and geology of South Arabia in the Upper Pleistocene—the backdrop of the proposed human expansion. Into Arabia: geology, geography, and palaeoenvironmental conditions GEOLOGY The Arabian Peninsula is divided into two main geological provinces: the Arabian Shield and the Arabian Shelf. The Arabian Shield encompasses the western portion of the Peninsula, consisting primarily of exposed outcrops of Precambrian crystalline rock. Most of this rock is derived from diorite, quartz diorite, granodiorite, or granite. Localized regions of the shield are covered in Tertiary and Quaternary volcanic rock (Chapman ; Bayer ). The Arabian Shelf, mantling the eastern portion of the peninsula, is made up of shallowwater marine sediments overlying the basement rock, remnants of the Tethys Sea. These sequences, 29 sometimes reaching , m in thickness, are comprised of sandstone, shale, and limestone (Chapman ). Much like the early modern human population that is the subject of this dissertation, the Arabian Peninsula is itself derived from Africa. For most of geological history, both landmasses formed part of a pan-Afro-Arabian continent. Sometime around  mya, the Arabian plate broke off from the African Shield and began to slide northeastward (rotating counter-clockwise), which triggered a chain-reaction of morphological transformations throughout the landscape (Bowen and Jux ). The most significant modification was the formation of the Red Sea trough. This narrow, elongate depression is more than , km long and varies in width from  to  km along the main channel. Bayer () estimates that the rift has opened  -  km within the past  -  mya and Kellogg and Reynolds () calculate that the Red Sea has opened some ° since the initial rifting. The separation of Arabia from Africa triggered volcanic activity along the western edge of the Peninsula, primarily in the Hijaz region of southwestern Saudi Arabia and western Yemen. Volcanic rock such as basalt, rhyolite, and commendite accumulated on the underlying Palaeocene sandstones, leading to the formation of a mountain chain along the western coast of Arabia (Bohannon et al. ; McGuire and Bohannon ). The southern highlands—Hadramaut and Dhofar—are comprised of Arabian shelf sedimentary beds that began to uplift in the Middle Cretaceous. Together, Hadramaut and Dhofar form a steep escarpment that runs parallel to the coast for some , km. The escarpment rises abruptly from the coastal plain, between  and  km inland, and reaches , – , masl before leveling off onto a heavily dissected tableland. Due to the high amount of precipitation in Dhofar that causes particularly active erosion, combined with the thick, horizontal limestone beds that comprise the local geology, the Dhofar escarpment is heavily karstic, containing some of the largest caves and sinkholes in the world (Platel et al. ; Hanna and al-Belushi ; Shaw ). The genesis of the Hajar Mountains is also linked, in part, to Arabian tectonic activity. As the Arabian plate moves northeastward, the Indian plate is sliding under it, forcing uplift throughout 30 Figure -. Physical map of Saharo-Arabian Arid Belt (false color landsat image courtesy of Google Earth™). the southeastern portion of the Peninsula. Some of the uplifted strata are comprised of oceanic crust (ophiolites) from the ancient Tethys Basin, representing one of only a few instances in the world where ophiolites are exposed within continental rock (Hanna ). Compression from the Arabian plate pushing against the Eurasian plate caused it to tilt eastward, creating a depression on the eastern side of the peninsula that is now the Arabian Gulf. The pressure of Arabia pushing against Eurasia also caused the Mesozoic and Cenozoic sedimentary strata within the Arabian Shield to buckle into a series of north-south folds, which now form the world’s richest oil reservoirs (Thompson ). GEOGRAPHY The Arabian Peninsula is the nexus of Africa, Europe, and Asia. The landmass is part of the expansive Afro-Asiatic arid belt, a chain of deserts that stretches from northwest Africa to northern India (Figure -). Geologically, Arabia is a giant plateau of Precambrian crystalline rock overlain, 31 Figure -. Physical map of South Arabia and major geomorphic regions discussed in text (false color landsat image courtesy of Google Earth™). 32 Figure -. Aerial photograph of the Hadramaut Valley. in certain areas, by limestone, sandstone, and basalt. Today, much of the landscape is blanketed by massive aeolian deposition (El-Baz ). The peninsula is bounded on the west by the Gulf of Aqaba and the Red Sea, on the south by the Gulf of Aden and the Arabian Sea, and on the east by the Gulf of Oman and the Arabian Gulf (Figure -). Arabia measures , km from north to south along the Red Sea coast, and nearly , km across at its maximum width from southwestern Yemen to the easternmost point in Oman. Figure -. Photo of Wadi Andur incising the Nejd Plateau. 33 Tropical and sub-tropical littoral conditions are found along the perimeter, while the basin-shaped interior is dominated by alternating steppe and desert landscapes. Three major dune fields are found in South Arabia: the Rub al’ Khali, Wahiba Sands, and Ramlat as-Sabatayn (ibid.). Mountainous terrain runs along the western and southern edges of the peninsula. The ‘Asir Highlands stretch north-south along the western border of Saudi Arabia, renamed the Yemeni Highlands where they extend into Yemen. These mountains reach nearly , masl in the south— the highest point on the entire Peninsula. As a result of its elevation, the region receives up to , mm of rainfall per annum, the most in Arabia. As one travels eastward from the mountains, there is a gradual descent into the Ramlat as-Sabatayn desert, which is actually the southwestern extremity of the massive Rub al-Khali basin. Further east, beyond the Ramlat as-Sabatayn, is the Hadramaut Plateau, or simply jol, as it is known in Arabic. This heavily dissected limestone tableland is incised by the dramatic Wadi Hadramaut, a canyon that extends for several hundred kilometers, reaching up to several kilometers in width, and bounded by steep canyon walls that rise between  to  m above the valley floor (Figure -). To the east of Hadramaut are the Nejd Plateau and Dhofar mountain range in southwestern Oman. Here, the narrow coastal plain sharply rises to , masl at the Dhofar Escarpment, which Figure -. Photo of the Wahiba Sands longitudinal dunes and interdunal zone. 34 levels off northward onto the Nejd Plateau. The entire region is comprised of uplifted Tertiary limestone that gradually slopes into the Rub al-‘Khali basin. The top of the Dhofar escarpment marks the watershed divide; the drainages flowing south of this line are seasonally active under present conditions, incising the limestone cliffs at a steep angle and creating springs and lagoons as they pool onto the coastal plain. Presently, the northward flowing drainages receive almost no stormflow, but, during pluvial cycles, the magnitude of the monsoon was sufficient to produce relatively high-energy fluvial systems (Figure -). The Hajar Mountains—rising to , masl—are situated in northeastern Oman. This relatively long chain of mountains stretches from Ras al-Hadd in eastern Oman to the tip of the Musandam Peninsula at the Strait of Hormuz, a distance of over  km. An extensive network of widian flow inland from the Hajar Mountains across a wide bajada plain and into both the Rub al-‘Khali and Haushi-Huqf basins. Haushi-Huqf is a relatively shallow depression in eastern Oman that is carpeted by a series of sibakh. Presumably, these sibakh were freshwater lakes during humid phases—poorly drained basins fed by aquifers and stormflow runoff. The Wahiba Sands are also located in eastern Oman, northeast of the Haushi-Huqf Depression. This desert is comprised of linear dunes oriented on a north-south axis and running parallel to one another for several hundred kilometers (Figure -). The dunes reach up  m in elevation and are separated by swales one to three kilometers wide (Glennie and Singhvi ). The most recent aeolian deposits in Wahiba formed during the Last Glacial Maximum, at which time the entire bed of the Arabian Gulf basin was exposed, providing abundant unconsolidated carbonate sands available for aeolian transport. Strong prevailing winds blowing southeastward carried these grains to eastern Oman, where they formed the Wahiba dunes (Glennie ). Encompassing nearly , square kilometers, most of the interior of southern Arabia belongs to the Rub al-‘Khali. This basin slopes from an elevation of approximately , masl in the west to nearly sea level in the east. The dunes of the Rub al-‘Khali include a variety of types, resulting from the alternating wind patterns and diverse sources of sand. Dunes are largest in the southwestern portion of the basin; some specimens reach up to  m high. Similar to Wahiba, the current Rub 35 Figure -. Photo of Dhofar Escarpment during monsoon season. al-‘Khali deposits formed during the late Pleistocene, consisting of reworked Pleistocene sediments above a bed of Pliocene alluvium (McClure ). CLIMATE Roman geographer Ptolemy divided the Arabian Peninsula into three zones: Arabia Petraea, Arabia Deserta, and Arabia Felix—designations that, to this day, are still applicable. The first two zones, ‘Rocky Arabia and Desert Arabia’, refer to the northern and central reaches of the Peninsula, part of the Saharo-Arabian arid belt. The third area, Arabia Felix, translates as ‘Fortunate Arabia’. The region was termed thus due to the monsoon rains that cycle through the Indian Ocean and deposit precipitation in the highlands of southern Arabia, creating tropical and sub-tropical environments (Brunner ). The peninsula is subject to two different weather regimes: ) Atlantic late-winter storms called “Northwesterlies” that move eastward over the Mediterranean Sea, down the Arabian Gulf, and eventually dissipate over the Rub al-‘Khali, bringing cool gentle winds and light precipitation; 36 Figure -. Photo of dense vegetation along Tihama Coast of Yemen. and ) summer storms brought by the Southwest Indian Ocean Monsoon system, lasting from June to September, in which the highlands of Yemen and Oman are bombarded by relatively heavy rainfall (Hugh and Mason ; Lézine et al. ; Glennie and Singhvi ). The mountainous terrain of southern Arabia traps moisture from the monsoon (Figure -), as a result the ‘Asir and Dhofar Mountains receive anywhere between  —  mm annually; while areas closer to sea level seldom experience more than  —  mm per year (Schyfsma ). Biogeochemical, lithogenic, and biological evidence from cores taken in the Arabian Sea suggest this weather pattern was established in the Middle Miocene (Léclaire ; Kroon et al. ). Annual temperatures vary widely throughout Arabia, affected by season and elevation. Interior (lowland) weather stations have recorded temperatures in the shade as high as °C, with mean summer temperatures ranging from .° to .°C, depending upon elevation. During the winter months, temperatures often drop below freezing in central and northern Arabia. Mean winter 37 Figure -. Photo of terraced agricultural fields in Yemeni Highlands. temperatures in the interior range from ° at higher elevations to °C in the lowlands. The coastal regions are slightly warmer in the winter, averaging .°C (Schyfsma ). FLORA AND FAUNA The sparse vegetation of South Arabia is grouped within the following biotopes: coastal habitat; interior basin; ‘Asir-Yemen Highlands; Dhofar Mountains; and Hajar Mountains. Flora such as Cressa cretica (a small flowering herb), Nitraria retusa (a small thistle-like bush), and Funcus maritimus (rush) are found growing on the marine shores and salt flats. Wild rue, mangrove, indigo, date-palms, henna, tamarinds, mistletoe, and ilb prosper along coastal wadi banks, particularly along the Tihama (Red Sea) coast (Figure -). In the interior, plants such as tamarisks, poplars, acacias, and several species of reeds, grasses, and small shrubs are found scattered near depressions and seasonal drainage systems that receive a limited degree of moisture (Hugh and Mason ). Modern pollen samples collected from surface deposits on the Ramlat as-Sab’atayn steppe yielded  Cyperaceae (sedges),  Tribulus (shrub),  Dipterygium (annual flowering herb),  Aerva lanata (polpala herb), and  Gramineae (grasses) (Lézine et al. ). 38 There is a wide variety of flora in the ‘Asir-Yemen Highlands, due to the heavy precipitation during the summer monsoon. Here wild figs, leguminous trees, tamarisks, date-palms, indigo, qat, myrrh, and a variety of flowering bushes and herbs are found along the wadi banks and planted on sprawling terraced fields (Figure -). Forests of juniper cover the mountain slopes between , and , m elevations (Hugh and Mason ). Pollen samples taken from the wooded Yemeni highlands show a predominance of acacia, Zygophyllum (Syrian bean caper), and several species from the family Chenopodiaceae, or goosefoot (Lézine et al. ). A similar distribution of flora has been identified in the Dhofar Mountains, with the addition of Boswellia carteri, or frankincense bush. The Hajar Mountains in eastern Oman are home to taxa such as primula, violet, juniper, and honeysuckle in the higher elevations. In lower lying areas grows Moringa peregrina (ben tree), Pularica (fleabane), and Gallionia aucheria (bedstraw)—a distribution bearing similarities to western India (Hugh and Mason ). As for mammalian fauna, animals belonging to the family Bovidae are by far the most prominent on the Peninsula. This includes one species of oryx, three species of Gazella, two species of Capra, one belonging to genus Hemitragus (wild goat), and one of genus Ovis (wild sheep). These animals typically occupy the highlands mantling southern Arabia—Yemen, Hadramaut, Dhofar, and Hajar Mountains. Gazelle have been noted from more arid settings such as the high plateaus, while the desert-adapted oryx were once ubiquitous throughout the interior, even within the Rub al-‘Khali. There are no members of families Cervidae (deer) or Suidae (pigs) in Arabia (Harrison ). Small mammals include various species from family Soricidae (shrews), order Rodentia (rodents), and order Chiroptera (bats). There are carnivores such as mongooses, genets, dogs, wolves, and foxes (Harrison ). A variety of felines are present, including Felis silvestris (wild cat), Felis margarita (sand cat), Caracal caracal (Caracal lynx), Panthera pardus (leopard), and Acinonyx jubatus, i.e. cheetah (Harrison and Bates ). Like the Bovidae, these animals are typically found in and around the montane zones. There are some examples of Lagomorpha (rabbits, hares, and pikas) reported from the interior—on the desert plateaus and flanks of the Rub al-‘Khali (Harrison ). 39 There are clear faunal links between East Africa and southern Arabia. Fernandes et al. () report mtDNA evidence for a recent genetic divergence between African and Arabian genets. He notes several other small and medium-sized carnivores (e.g., mongooses, desert foxes, honey badger, caracal, jungle cat, golden jackal) that occur on both sides of the Red Sea and may also be genetically linked. Faunal connections among shrews from the Horn are noted: One of the six other species for which substantial range extensions are recorded here, Crocidura somalica, has definite NE African affinities, in keeping with so much of the flora and fauna of Dhofar, while the South Arabian sub-species of Genet, G. g. granti, is also more closely related to the Ethiopian races of the species… (Harrison :) Also indicating faunal connections across the Red Sea is the presence of the primate Papio hamadryas—Sacred Baboon—in Yemen. Presently, they are indigenous to the rocky hill country of Somalia, Ethiopia, and Yemen. Papio hamadryas are arid-adapted creatures that forage protein-rich insects, hares, and other small mammals; they obtain water from shallow pools, going so far as to dig wells in desert regions with a high water table (Nowak ). Analyses of Papio hamadryas mtDNA lineages on both sides of the Red Sea indicated that they originated in East Africa sometime between , and , years ago, and subsequently migrated into Arabia (Wildman ; Wildman et al. ; Fernandes ). The fact that the East African baboon lineage is older than the Arabian implies that there must have been a demographic bottleneck release sometime in the Upper Pleistocene. THE BAB AL-MANDEB So, it is imperative to ask: was the Bab al-Mandeb navigable or did it represent an insurmountable geographic boundary? The name itself gives the impression of a daunting passage: Bab al-Mandeb translates from Arabic as “the gate of tears.” Presently, the narrowest point is an  km crossing called the Perim Narrows. The shallowest point is the Hanish sill, not exceeding a depth of  m and central channel width of  km (Maillard and Soliman ; Fairbanks ; Murray and Johns ). Scholars agree that the strait was never 40 fully exposed at any point during the Last Interglacial cycle; there was never a land bridge. The bathymetric profile of the rift floor is such that it has a shallow depth of around  m for several kilometers, before plunging into a deep, narrow trough at the central channel. Therefore, based on a projected drop of  m during the Last Glacial Maximum, Africa and Arabia would have been separated by a strait  m deep and  km wide at the Hanish sill (Figure -). The morphology of the Red Sea basin—an elongated trough with a narrow gap at the southern end—is such that it is extremely sensitive to changes in global sea levels, thereby serving as an effective tool for reconstructing fluctuations throughout the Pleistocene. Due to limited exchange flow between the Red Sea and Gulf of Aden, the Red Sea is particularly saline. As sea levels decline and the Bab al-Mandeb strait constricts, Red Sea salinity increases even more. Scholars are able to measure oscillations in palaeo-salinity by analyzing the oxygen isotope record from planktonic foraminifera in Red Sea cores (e.g., Luz ; Thunell et al. ; Rohling and Zachariasse ; Siddall et al. ; Sirocko ; Siddall et al. ). There are at least three aplanktonic periods at , , and  kya, indicative of salinity levels high enough to prevent planktonic growth. To achieve this level of salinity, Siddall et al. () estimate the Hanish sill must Figure -. Bathymetric profile of Bab al-Mandeb at Hanish sill, with projected distance calculated at  meters below current sea level. 41 have been reduced to  m in depth. Conversely, particularly low periods of salinity occur at  and  kya, in which raised sea levels caused marine transgression, thereby expanding the Bab al-Mandeb. The Bab al-Mandeb was at its widest at the onset of the Upper Pleistocene: approximately  km. The gap gradually narrowed over the next  kya, constricting to less than  km by the last glacial maximum. Whether  or  kilometers apart, it is the position of this dissertation that neither distance was an obstacle for the expanding modern human population(s). At both maximum and minimum sea levels, the shores of Arabia were visible to groups living across the Red Sea. There is archaeological evidence attesting to a Middle Pleistocene hominid group colonizing the island of Flores, which lies far off the continental shelf of Indonesia, well beyond line of sight (Morwood et al. ). Since the voyage to Flores, which occurred over half a million years ago, required greater nautical technology than that required to cross the Bab al-Mandeb, it is assumed that the narrow span was not a problem for Upper Pleistocene groups. This is corroborated by a recent experiment in replicating Pleistocene seafaring vessels, in which the feasibility of creating usable rafts with only the toolkit and resources available in the Middle Pleistocene was demonstrated (Bednarik ). There is still the problem of navigating the fierce currents that flow through the gap, which give the strait its name “gate of tears.” Examination of annual Red Sea inflow and outflow at the Bab al-Mandeb indicate that during the winter months, between November and June, dangerous current flows southward into the Gulf of Aden, known as the hindi current. During the monsoon, upwelling in the Gulf of Aden forces a change in the direction, pushing water into the Red Sea between June and October, called the shami current (Hugh and Mason ; Siddall et al. ). Presumably, navigating the straits during the shami season would be much safer than crossing during the hindi season, when one runs the risk of being washed into the open sea. Shami currents were more frequent during periods of reduced sea level; Siddall et al. () demonstrate that at > m lower than present, there was a much greater volume of water entering the Red Sea from the Gulf of Aden per annum. 42 In considering these data, it is reasonable to conclude that Upper Pleistocene hominids could have navigated across the Bab al-Mandeb, as long as there was a shami current, during phases of both high and low sea levels. THE PALAEOENVIRONMENT J. D. Clark () argued that palaeoenvironmental conditions in the Sahara—oscillating between ameliorated savannas and desiccated deserts—acted as a pump in disseminating populations to and from sub-Saharan Africa throughout the Pleistocene, pulling groups in while fertile, and pushing them out when arid. The Arabian Corridor Migration model predicts similar dynamics in hunter-gatherer movements into Arabia; groups were “pulled” onto the peninsula during pluvials, as opposed to being “pushed” out of neighboring regions during arid phases. The model envisions a series of demographic/geographic bottlenecks sealing hunter-gatherers within surrounding refugia, and subsequent bottleneck releases sending groups into southern Arabia. These bottleneck releases would have occurred at the onset of pluvial conditions, as soon as population sizes surrounding Arabia began to increase and the Peninsula became habitable. Therefore, in order to predict the timing of human expansions, it is necessary to review the climatic oscillations in southern Arabia. Most of the precipitation in southern Arabia is brought by the Southwest Indian Ocean Monsoon system, considerably more so than from “Northwesterly” winter storms. Consequently, the environmental fate of the region—amelioration or desiccation—is particularly predictable because it rests upon the intensity and duration of the monsoon, which has been in flux for at least the last quarter of a million years (Clemens et al. ; Muzuka ). Palaeoclimatologists studying the history of the Southwest Indian Ocean Monsoon system link increases in magnitude to a greater degree of solar radiation in the Northern Hemisphere, in combination with a contraction of glacial conditions. Analysis of dinoflagellate cyst content from Arabian Sea deep sea cores during the last glaciation reveals an abrupt fluctuation at . kya (Zonneveld et al. ). This spike is attributed to the disappearance of snow and ice cover over central Asia, Tibet, and the Himalayas, which led the authors to conclude that one of the primary 43 mechanisms driving monsoon fluctuations are climatic conditions at glacial-interglacial boundaries. Biogeochemical and lithogenic data from Arabian Sea cores spanning the last , years also support the notion that monsoon winds were sensitive to changing glacial climates. The retreat of ice sheets, the rise in continental albedo (solar radiation reflected off the earth’s surface), and the increase of water surface temperatures in the western Indian Ocean triggered shifts to the north and spikes in magnitude of wind and rainfall (Clemens et al. ). Computer simulations have been used to estimate the average wind speed of the Southwest Monsoon during intensification phases. Speeds are currently at about  m/sec, while during certain periods the winds would have reached  m/sec. In addition, precipitation would have been  greater than its present value—from  mm/day to . mm/day. Increased intensity in airflow drove the monsoons up further into the Arabian Peninsula, with evidence as far north as Bubiyan Island in the northern Arabian Gulf (Sarnthein ; Kutzbach ). Researchers are able to predict not only the timing of oscillations, but also the pace of these shifts. Analysis of nitrogen isotope ratios was conducted on an interval of Arabian Sea core spanning the bracket of time between  and  kya—a known stadial/interstadial boundary. Significant changes in mean strength of the monsoon occurred within  years—relatively instantaneous on a Pleistocene time scale (Higginson ). While the onset of the intensified monsoon was rapid, evidence with comparable resolution suggests the shift back toward aridification was a more gradual process occurring on a millennial scale, at least during the most recent wet/dry shift (Lückge et al. ). Episodic increases in monsoon strength over the Arabian Sea were identified during OIS , , and  by analyzing C isotopes from sedimentary organic carbon obtained in deep sea cores (Muzuka ). The authors speculate that the increase in coastal upwelling during OIS  and , which are both glacial phases, may be due to an intensification of the “Northwesterly” storm regime rather than the Southwest Indian Ocean Monsoon. The stable oxygen isotope record of various planktonic foraminiferal species (i.e., Globigerinoides ruber, Globigerina bulloides, and Neogloboquadrina dutertrei) were used to examine 44 temporal variations in marine palaeoproductivity during the most recent interglacial/glacial cycle. These analyses unanimously concluded that palaeoproductivity in the Arabian Sea is directly linked to the intensity of the monsoon. Scholars note the onset of intensified monsoon episodes can lag up to , years after shifts in glacial conditions; this lag represents the threshold necessary for sufficient melting of snow and ice cover over central Asia, Tibet, and the Himalayas to affect Indian Ocean insulation patterns (Reichart et al. ; Petit et al. ; Ivanova et al. ). Figure -. Geological profile of an exposed section in Wadi Mahwis showing coarse gravels interstratified with fine fluviatile sands (photo courtesy of D. Angelucci). 45 Fluctuations in monsoon intensity have left a considerable fingerprint on the landscape of southern Arabia. Palaeoenvironmental signals indicate a number of significant wet-phases throughout the Quaternary. Immediately leading up to the Pleistocene, Arabia was hot and humid with subtropical forest cover over much of the Peninsula (Thompson ). Massive alluvial gravel deposits have been identified along three major east-west water courses: Wadi Dawasir, Wadi as-Sahba, and Wadi ar-Rima. These rivers, comparable in size and flow to the Nile or Amazon, were carved out between . and . mya (Anton ). Further attesting to this prolonged Plio-Pleistocene pluvial are a series of alluvial fans bounding the western edge of the Wahiba Sands. These unique structures are characterized by a complex network of palaeochannels that form superimposed, cemented gravel ridges. The fans have been subject to intense aeolian deflation, leading to landscape-lowering and consequent exhumation of the cemented channels; in other words, the palaeochannels are in bas relief on the landscape. Researchers posit the earliest channels were deposited during the late Pliocene/early Pleistocene (Maizels ). The uppermost sediments covering the bajada plain south of the Hajar Mountains in Oman were deposited during this pluvial as well (Rodgers and Gunatilaka ). Finally, there have been a series of red palaeosols identified in the Yemeni Highlands, which appear to stratigraphically correlate with the Plio-Pleistocene boundary (Hötzl and Zötl ). Following the late Pliocene/early Pleistocene, researchers associate the latter half of the Lower Pleistocene and Middle Pleistocene with increasing aridity. This is signaled by widespread wadi infilling; in addition, basalts deposited at this time are not dissected by any drainage systems. Middle Pleistocene sediments are rich in gypsum, suggesting limited fluvial activity. Although there are some aeolian sediments associated with the Middle Pleistocene, it does not appear that aridity had reached the threshold necessary for pervasive dune formation comparable to the present (Anton ). Evidence for occasional humid phases can be found in fluviatile sands and gravels (Figure -) interstratified in the wadi fills (Hötzl and Zötl ). The only absolute dates for Middle Pleistocene pluvials were obtained from speleothems in Hoti Cave, northern Oman. U/Th measurements 46 indicate periods of increased growth between  –  kya and  –  kya, which coincide with OIS  and OIS  (Burns et al. ). Of primary interest to this dissertation are climatic conditions during the late Middle Pleistocene and Upper Pleistocene: was the climate arid enough during the Penultimate Glaciation to preclude hominid habitation, and, conversely, were conditions humid enough to facilitate human occupation during the Last Interglacial? Unfortunately, there are no environmental signals from the late Middle Pleistocene, so very little can be said about the climate at that time. Considering that monsoon intensity tracks closely with the global oxygen isotope curve, it is assumed that the environment was predominantly hyperarid between  and  kya. During OIS  ( –  kya), there are indications of at least two significant wet-phases. Palaeosols were identified in the western Omani piedmont, which yielded U/Th dates clustering around  and  kya (Sanlaville ). Other soils were noted in the ad-Dahna Desert of northern Arabia, where late Pleistocene dunes overlie two separate pedogenic strata that could only have formed on stabilized dunes with a dense cover of vegetation (Anton ). The aforementioned Plio-Pleistocene bas relief gravel channels west of the Wahiba desert are superimposed by thinner fluviatile gravels that were tentatively associated with the OIS e and OIS a humid episodes (Maizels ). Finally, OIS  pluvials are signaled by the Hoti Cave speleothems, which produced U/Th dates indicating rapid growth at  –  kya and  –  kya. It is interesting to note that speleothem growth was particularly pronounced during OIS e, more so than all subsequent pluvials (Burns et al. , ). There are a paucity of palaeoenvironmental data from OIS  and early OIS , which, based on the genetic evidence, represents the bracket of time when haplogroup M diverged from its founding East African lineage. Analysis of summed probabilities based on monsoon activity within this timeframe suggests increasingly hyperarid conditions culminating around , , followed by an oscillation back toward wet conditions from approximately , to , years ago (Parker, personal communication). Claims for hyperarid conditions at  kya are further bolstered by evidence from the Toba super-eruption at this time, which purportedly caused massive global 47 desiccation during the , year long “volcanic winter” mentioned previously (Ambrose ; Rampino and Ambrose ). The only physical evidence for arid conditions at the onset of OIS  is inferred from stratigraphic profiles in the Rub al-‘Khali, which indicate the first stage of aeolian accumulation began sometime greater than , . This deposition, however, is relatively minor compared to the immense aeolian structures that accumulated during the terminal Pleistocene (McClure ). The first western accounts of the Rub al-’Khali from the early th century all noted a series of small buttes that stand out in stark white or gray against the seemingly endless wasteland of monotonous rust-colored sand (Clark ). During his journey across the desert in , explorer St. John Philby determined that these localities are small eroded lake basins comprised of marl terraces and hardened evaporitic crusts, noting associated freshwater shells and lithic implements (Philby ). More recently, radiocarbon dating of mollusk shells and lakebed marls determined the lakes were active as early as ,  (McClure ), providing direct evidence for the OIS  pluvial; which is not to say this was the earliest phase that the lakes existed, earlier sediments may have been erased by subsequent reworking. The palaeolakes ranged from ephemeral puddles to pools up to ten meters deep, and numbered well over a thousand. They are primarily distributed along an east-west axis across the center of the Rub al-‘Khali basin, covering a distance of some , km (McClure , ), and have also been reported from the Ramlat as-Sabatayn desert in Yemen (Lézine et al. ), as well as the an-Nefud in northern Arabia (Schulz and Whitney , ). These lakes were short-lived; the poorly drained basins would have been filled by occasional torrential stormflow runoff and disappeared within a few years if not recharged. Playas from the Mundafan Depression are an exception to this, because the region is situated on the eastern slope of the Tuwaiq Escarpment; as such, it has a wide catchment area due to drainage from the escarpment. The thickest lacustrine deposits were found here and were estimated to have an -year span for a single lake period (ibid.). Fossilized floral and faunal remains from these deposits yielded an array of large vertebrates including: Oryx, Gazella, Equus hemionus, Alcelaphus buselaphus, Capra, Hippopotamus, Camelus, Struthio (ostrich), and possibly Bos primigenius (McClure ). This variety 48 of species could only have existed if there was light to medium rainfall distributed evenly over the Rub al-‘Khali, facilitating expansive grasslands over most of the terrain (McClure , ). Ostracodes and freshwater mollusks indicative of low salinity are present, as well as species of foraminifera that attest to highly brackish conditions (McClure and Swain ). The presence of grass, shrubs, and herbs are evidenced by tubule scree—thin, tubular fragments of fossilized material scattered in the sand around the basins. They were formed when dissolved calcium carbonate in the water precipitated onto plants as the lake evaporated. In addition, there are phytoliths within the lacustrine sediments, indicating abundant grasses. Evidence of fish remains are conspicuously absent, because lakes were rarely refilled and became too alkaline too quickly to develop a population (McClure ). In addition to the palaeolakes, other signals of this mid-Upper Pleistocene wet-phase include depositional terraces in the Wadi Dhaid that produced radiocarbon dates between , – ,  (Sanlaville ), palaeosols in the ad-Dahna desert stratigraphically correlated with the later Upper Pleistocene (Anton ), and calcite deposits from ancient hyperalkaline springs in Oman, radiocarbon dated between , and ,  (Clark ). Two soil horizons were discovered around the central plateau of the Yemeni highlands, characterized as mollisols—soils that form on landscapes covered in savannah vegetation. Uncalibrated radiocarbon dates yielded a measurement of , ±  for the lower stratum, and , ±  for the upper horizon (Brinkmann and Ghaleb ). Researchers speculate the dry-phase that set in during the Terminal Pleistocene was more arid than the peninsula had ever before experienced (Anton ). Dating of the dune formations in the Rub al-‘Khali (McClure ), an-Nefud (Anton ), and the Wahiba Sands (Gardner ; Glennie and Singhvi ) indicate a major phase of aeolian deposition between approximately , and , . Calcite fractures in Oman corroborate the evidence for increasing aridity, indicating there was considerably less moisture in the environment beginning around ,  (Clark ). Following OIS , the Terminal Pleistocene hyperarid phase ends with a fluctuation back toward more humid conditions at the onset of the Holocene. This sub-pluvial period lasted until 49 ~, , at which time the present climatic regime was established (Overstreet and Grolier ; Cleuziou et al. ; Sanlaville ; Brunner ; Wilkinson ; Parker et al. ). Discussion This chapter has presented a scenario for modern human migration(s) out of Africa via the Arabian Corridor. These expansions are based on the premise that hunter-gatherers, adapted to the savanna in the Horn of Africa, periodically expanded their range eastward into the interior of Arabia as the savanna phytogeographic zone spread onto the Peninsula at the onset of Upper Pleistocene pluvial episodes. The Arabian Corridor Migration Model is strongly supported by recent genetic analyses, which indicate that there were multiple bottleneck releases from Africa during the Last Interglacial (Ambrose ). Studies of haplogroup M suggest the colonization of Australasia resulted from an expansion out of the Horn of Africa roughly between  –  kya. These genetically-predicted expansions probably occurred during pluvial episodes in southern Arabia. A number of environmental signals have been presented that attest to periodic phases of intensified monsoon activity, leading to amelioration of the interior deserts (Table -). Significant wet-phases correlate with OIS e, OIS a, and OIS , when retreating glacial conditions altered the Indian Ocean insulation patterns and dragged the monsoon northward into the Arabian interior. A product of these relatively frequent oscillations was the constant fluvial reworking of aeolian sediments during pluvials, and subsequent aeolian deposition during arid phases. Over time greater and greater amounts of fluvial sands became available for aeolian transport, and by the end of OIS , the massive dunes that blanket southern Arabia had begun to form. One ironic product of these dune formations was the creation of a vast network of playas during OIS  that formed within interdunal depressions. These freshwater lakes dotted the Rub al-‘Khali basin and supported populations of hippopotami, ostrich, and various Bovidae, among other taxa. Many of the species associated with these faunal assemblages required not only freshwater lakes to survive, but vast open grasslands, further evidence of the interior transformation to savanna. 50 Table -. Synthesis of palaeoclimatic data. 51 South Arabia served a unique role in the region due to these environmental extremes, in combination with its geographic position as the nexus of three continents. Arabia was a bridge connecting Africa with Eurasia. During arid phases the bridge was impassable: an insurmountable geographic barrier that sealed populations in East Africa and prevented any movement eastward. During pluvials, Arabia facilitated genetic bottleneck releases via hunter-gatherer range expansions onto the Peninsula. Given the archaeological record of East Africa during the MSA and early LSA, it seems likely that the expanding population would have moved into the interior of southern Arabia (contra Stringer ), or, at least, employed an adaptive strategy that incorporated terrestrial resources. East African MSA sites are much more prominent in the hinterland than the littoral, with associated faunal assemblages indicative of a population adapted to hunting ungulates rather than marine resources. This is corroborated by a lithic toolkit that appears suited for killing and skinning terrestrial game; none of the implements associated with acquiring and processing marine resources are present (e.g., scaled pieces, hooks, barbs, harpoons, net sinkers), with the exception of Katanda (Brooks et al. ). In sum, the interior of South Arabia was ideal for savanna-adapted hunter-gatherers migrating from East Africa. The same or similar fauna roamed across the vast plains, playa lakes and seasonally active streams provided enough water, and, in certain regions, there were abundant and high-quality lithic raw material sources. The Arabian Corridor Migration Model is testable through the use of archaeological data, based on the assumption that the lithic techno-typology of the migrating population(s) will resemble that of the source area. This is particularly true for the Arabian Corridor, as the early moderns were expanding into the same niche, and, hence, employing the same or similar adaptive strategies. Therefore, the model predicts MSA/early LSA elements will appear in Upper Pleistocene lithic assemblages from South Arabia. Lithic industries from the late MSA/early LSA of the Horn are easily distinguished by the presence of Somaliland Stillbay points, and derivative Magosian/Hargeisan forms. Examination of the East Africa archaeological record has shown these distinct implements to be geographically bounded, 52 and falling generally within the range of  –  kya. The only visible change within this tool type from the MSA to the LSA is their decreasing size and frequency, yet the range in shapes, as well as the characteristic flat, invasive soft hammer retouch producing a lenticular cross-section remains the same. MSA assemblages are distinguished from the early LSA based on the presence of Levallois and discoidal cores. LSA elements such as frequent backing, burins, endscrapers, and microliths are first introduced in “transitional” Hargeisan and Magosian assemblages; unfortunately these assemblages are undated and poorly understood. So, the proposed model of modern human migration(s) predicts one or more lithic technotypological complexes were present in South Arabia during the Upper Pleistocene resembling the Somaliland Stillbay, Somaliland Magosian, and/or Hargeisan. What archaeological data has been uncovered from South Arabia during this time period? Is there evidence for MSA/early LSA elements in South Arabia? The next two chapters will attempt to address these questions by reviewing the history and present state of archaeological research in southern Arabia. It is argued that Somaliland Stillbay points, indeed, are present in Arabia, though they have been lumped with Early/Middle Holocene assemblages because they are only documented within palimpsest surface collections. 53 Chapter 3 THE HISTORY OF RESEARCH IN SOUTH ARABIA Who controls the present now controls the past Who controls the past now controls the future Who controls the present now? Now testify —Rage Against the Machine, Testify Southern Arabia has been described as “archaeology’s last frontier” (de Maigret :). Explorer Bertram Thomas called it “almost the last considerable terra incognita” (Thomas :xiii). The region has historically languished in fabled anonymity due to its isolated geography, inhospitable climate, difficult terrain, and, at certain times, xenophobic policies. As a result of South Arabia’s isolation—prior to the th century their only contact with the outside world was restricted to commercial interaction within the various port cities along the coast— the outside world has often associated this region with the unknown. Southern Arabia has spawned fanciful stories of distant, exotic civilizations resplendent in legendary wealth, what de Maigret (: ) calls “the myth of Arabia Felix.” This inaccessibility has pervaded prehistoric research. Too few scholars have carried out comprehensive investigations, and the vast deflated, desiccated landscape has severely hindered the location and excavation of buried, in situ material. As a result, the current chronological sequence for prehistoric Arabia is a tangled knot of disparate threads and problematic assumptions. The following two chapters will attempt to untangle this knot. The current chapter deals with the early history of research in southern Arabia, documenting traditional perceptions of the region and the gradual infiltration by outside scholars from the rd millennium  through the midth century. During these five millennia, Arabia was an impenetrable fortress closed to the outside world, whose walls were slowly eroded by the curiosity of foreign merchants, explorers, and scientists. 54 These fortifications were ultimately toppled on December , —the day Bertram Thomas set forth from the Dhofar Escarpment up into the Rub al-‘Khali Desert. He was the first foreigner to traverse the entire expanse of the world’s largest sand sea. It was during his and subsequent voyages through this wasteland that people began to take note of ancient lakebeds, the remains of fauna such as hippopotami, and stone tools indicating prehistoric human habitation. It was also during this time that massive oil reservoirs were discovered within the tectonically folded limestone beds beneath the surface of Arabia, triggering dramatic changes and heralding the sudden growth of infrastructure throughout the Peninsula. The following chapter will continue documenting the history of research, presenting the Pleistocene and Early/Middle Holocene archaeological occurrences discovered in the petroleum era. South Arabia in antiquity: the myth of Arabia Felix The earliest written references to South Arabia come from Old Akkadian texts dated to ca. , . Cuneiform tablets speak of a distant trading partner called Magan, known as an important source of copper and diorite. Sargon the Great wrote of ships from Magan docked at the port in Agade, and lists the goods brought from this kingdom. His successors Manishtusu and Naram-Sin recorded accounts of their assault on Magan (Newton and Zarins ). Archaeological work within the past  years has produced evidence verifying that Magan refers to the southeastern corner of the Arabian Peninsula, where a number of Bronze Age copper and diorite mining settlements have been excavated throughout the Hajar Mountains, and Sumerian cylinder seals appear at several port sites (Potts ). In addition to exporting mineral and stone, aromatics were harvested from the resin of frankincense trees that flourished throughout the Dhofar Escarpment, and were subsequently transported to the coast for export (Zarins ). It is also clear that Magan interacted extensively with the Indus Valley, supported by finds of Harappan-style pottery, weights, and cylinder seals that were made locally along the coast of southeastern Arabia (Potts ). The rise of Magan is clearly linked to its role as a way station for trade between Mesopotamia and the Indus Valley. 55 These international relationships across the Arabian Gulf and Indian Ocean grew and flourished over time. The region appears in several Neo-Assyrian and Achaemenid texts. In one instance, the modern Omani city of Izki is mentioned in a letter written by Assurbanipal II, in which he refers to the settlement as “the city of the King of Qade” (Potts ). The term Qade, which is Semitic, appears as Maka in Old Persian, and eventually became the modern Persian name for Oman and the United Arab Emirates. One of the most well-known early references to South Arabia appears in the Old Testament, in the story of the Queen of Sheba’s visit to King Solomon. The tale describes the queen’s arrival in Jerusalem with “a very great retinue, with camels bearing spices, and very much gold, and precious stones” ( Kings :-). Sheba is the Biblical name for Saba, an Iron Age kingdom known to have existed in southwestern Arabia at least as early as   (de Maigret ). South Arabia is mentioned in classical Greek and Latin texts, attesting to the region’s burgeoning prosperity. The earliest reference comes from Herodotus, in which he refers to the land as Arabia Felix—Happy Arabia. This title was in reference to the vast wealth perceived to exist there. Herodotus described the famed incense groves of Dhofar, detailing the growing, harvesting, and processing of the highly-valued resins, claiming “the whole country is scented with them, and exhales an odor marvelously sweet” (Historia, Book III:-). According to Herodotus, exposure to Greek society led the inhabitants of South Arabia to adopt certain aspects of classical culture—it is ironic that particular veneration was given to Dionysus/Bacchus. The Romans were among the greatest consumers of incense; Pliny the Elder wrote that Nero burned a year’s production of Arabian incense at his wife’s funeral (Historia Naturalis, Book XII). Increasingly accurate descriptions of South Arabia became available following Alexander’s conquest of the Near East. The most detailed of these works were authored by Theophrastus, Eratosthenes, and Agatharchides, describing, with some measure of accuracy, the land and its peoples. The first mention of the region in Latin texts is found in the writings of Strabo, who based his account on the military expedition of Aelius Gallus in  ; a campaign launched to gain control of the North Arabian incense trade. Strabo’s seventeen volume work, Geographia, provided descriptions 56 of locations on the Peninsula and even chronicles a lineage of rulers in the Gulf region (de Maigret ). An overseer in the Great Library of Alexandria, Claudius Ptolemy drew a comprehensive map of the Arabian Peninsula around  . His map was compiled from the earlier works of Strabo, Eratosthenes, as well as the anonymous Peripluses, which were travel accounts provided by Greek and Roman merchants as they sailed around the Peninsula. These documents supplied Ptolemy with information about the hinterland, albeit via secondhand accounts garnered from local tribesmen. Though the placement of interior towns was way off, it was by far the most advanced rendition of Arabian geography at that time (Clapp ). Following the rise of Islam, even Muslim authors regarded South Arabia as a land existing on the periphery of their known world. The Abbasid historian Ibn Ishaq was among the first to compile a history of ancient South Arabia in the th century, followed by al-Tabari, Ibn Kalbi, and al-Hamdani who produced Jazirat al-‘Arab, one of the most accurate texts dealing with Arabian geography. These scholars attempted to document South Arabia during a period known as jahilliyah (the time of ignorance), which serves as a general reference to all pre-Islamic history (de Maigret ). The Qur’an itself weighs in on the ancient history of South Arabia, telling the story of the People of ‘Ad. A once prosperous nation, this culture is said to have occupied the interior of southern Arabia, reveling in immense prosperity from the spice trade. Because of their gluttonous transgressions, their capital city, Ubar, was destroyed by Allah and the people “removed from sight” (Surah Hud:-). The tale of Ubar spawned an archaeological expedition throughout the Nejd Plateau and southern fringe of the Rub al-‘Khali in the early ’s. The team used remote sensing, textual evidence, folklore, and archaeological survey to search for the fabled city, which led to excavations at the Iron Age fort at Shisur (Clapp ). The Age of Exploration The intellectual revival during the Renaissance rekindled Western interest in South Arabia, which had lain dormant for nearly one thousand years. In their passion for classical texts, scholars 57 rediscovered the accounts of Ptolemy, Pliny, Strabo, as well as the Islamic historians who kept these classical texts from disappearing in the intellectual void that pervaded Medieval Europe. These new Renaissance scholars sought to verify the ancient accounts, giving birth to an age of exploration that brought a number of intrepid adventurers to the Arabian Peninsula. The first European to set sail for Arabia was an Italian named Ludovico di Varthema at the end of the th century. An account of his travels was published in his work Itinerario in , and was used by mapmakers for the next  years. After visiting Mecca and Medina, di Varthema sailed south down the Red Sea, passed through the Bab al-Mandeb Strait and reached the port of Aden. Upon arrival he was arrested by the vice-sultan of Yemen, though eventually set free after befriending one of the vice-sultan’s wives. He traveled up into the Yemeni highlands to Ta’izz, Dhamar, and eventually the capital at Sana’a. De Varthema was careful and precise in the descriptions of his journey, providing a thorough account of Yemen’s geography and customs (de Maigret ). Not long after di Varthema’s journey, a Portuguese colony was established in southeastern Arabia, one of the only two instances of western colonization on the Peninsula. Portuguese traders recognized the importance of the Strait of Hormuz, which controls the flow of traffic through the Arabian Gulf. Led by Alfonso de Albuquerque, the Portuguese navy carried out a series of campaigns along the Trucial Coast, achieving victory over the city-state of Hormuz in . This episode of colonization lasted nearly  years; it came to a close in  with the Portuguese defeat at the hands of the Ottoman Empire. During the entire period of occupation, the Portuguese built an extensive series of forts along the southeastern coastline of the Peninsula, although, like so many that had come before them, they never penetrated inland (Ramerini ). One of the most important episodes in establishing long-term Western contact with southern Arabia was brought on by a terrible storm at sea. Two Jesuit missionaries on their way to Ethiopia, Montserrate and Paez, were shipwrecked off the coast in . They were taken prisoner and brought to Dhofar for trial, after which the two were transported on a long and difficult journey westward to Sana’a. On the way, the missionaries stopped at Marib, providing the first eyewitness description of the ancient Sabaean capital. They were kept prisoner in Sana’a for five years, and eventually sent 58 to the Red Sea port of al-Mokha, where they were set free. Paez mentions a Yemeni drink from an “infusion made with the skin of a fruit called bunna.” This is the first known reference to coffee, which flourishes in the high elevations of the Yemeni Mountains (Pirenne ). Paez’s account unwittingly revived trade with South Arabia, which had been dormant for a millennium. Beginning in the th century, European merchants flocked to the Tihama coast of Yemen to establish emporiums for coffee export. Like the limited Periplus accounts, however, these caffeinated entrepreneurs were restricted to the coast and information regarding the interior remained scarce. The modern era of scientific exploration in Arabia began on October th, . On that day, six members of a Danish expedition crossed the Suez for what was the start of an ill-fated investigation of the Arabian Peninsula. One year earlier, Semitic scholar J.D. Michaelis from the University of Göttingen had submitted to King Frederick V of Denmark a number of academic questions regarding South Arabia. In response, the king commissioned a team of naturalists to travel to Arabia to address these inquiries; this expedition represented the first purely scientific trip to the Peninsula. The journey lasted just eleven months, ending in the death of five of its six members. The only member to make it back was Carter Niebuhr, who, after his return, penned Travels through Arabia and other Countries in the East. His work was published in German, French, and English, and was immensely popular because it offered the West a comprehensive and objective description of little-known South Arabia and its seemingly-alien inhabitants. Unfortunately, the doomed expedition failed to answer the queries put forth by Michaelis, and offered very little new data regarding the archaeology of Arabia. Its primary contribution was that it served as a catalyst for sparking Western scholarly interest in the region (Hansen ). Niebuhr’s publication coincided with growing antiquarian excitement in the archaeology of Egypt and the Near East. East India Company merchants traveling to South Arabian ports began to take note of ruins and ancient inscriptions along the coastline. They were aware of recent breakthroughs in Mesopotamia decoding ancient cuneiform tablets, and sought to replicate these 59 successes in a related part of the world. Would-be epigraphers dutifully copied down undeciphered texts and disseminated these finds to scholars upon their return, thus launching efforts to decode Himyaritic, the proto-Semitic language of ancient South Arabia. Exploration throughout the th century was dominated by scholars probing southwestern Arabia for inscriptions, each new line of text provided clues to the ancient history of this enigmatic region. Among the most successful of these epigraphers were T. Arnaud, J. Halévy, and E. Glaser. With the blessing of the emir of Yemen, they were able to enter the hinterland and investigate the ruins at Marib. They brought back a plethora of new data from the ancient capital of the Sabaean Kingdom. This served to further pique the interest of academia, eventually leading to archeological excavations in the beginning of the th century to verify historical accounts documented in the Himyaritic texts (Doe ). The Early th Century The earliest archaeological excavation began in  by von Wissman and Rathjens, two geologists from Hamburg University. They were invited to Yemen by the Imam of Sana’a, though restricted from visiting the Jawf region, where Marib is found. Instead, von Wissman and Rathjens were permitted to excavate the site of Huqqah, not far from Sana’a. They uncovered the foundations of a large temple to the sun goddess Dhat Ba’adan, with inscriptions placing it in the st century . After five years of investigations, the two geologists published Vorislamische Altertümer in , which documented the material culture of a civilization that had previously only been known through myth and inscriptions. By the ’s the world’s explorers were rapidly running out of new territory. The recent defeat of the Ottoman Empire had opened up previously restricted areas in the Middle East, namely the harsh sand and gravel deserts of South Arabia. This initiated a great race to claim “exploration’s last great prize” (Clapp :), which involved crossing the Rub al-‘Khali Desert, the largest sand sea in the world. T.E. Lawrence had thrown down the gauntlet when he warned “nothing but an airship can do it” (Lawrence ). 60 The primary contenders were Bertram Thomas and Harold Philby. Both adventurers spent the late ’s positioning themselves, both geographically and politically, to cross the “horrid wastes of Rub al-’Khali” (Thomas :xii). Philby was in Riyadh, hoping for consent from King ‘Abdul ‘Aziz ibn Saud to conduct a southward crossing. His friend and protégé, Thomas, waited for permission from the wali of Salalah, plotting his expedition route from south to north. Thomas claimed victory, ascending the Dhofar Escarpment and points beyond on December , . Beyond simply an intrepid adventurer, Thomas diligently recorded geographic and ethnographic data throughout his journey. He observed that inhabitants of the Dhofar Mountains and Nejd Plateau spoke Shahri and Mahri, two dialects of a language distantly related to Arabic. These culturally and morphologically distinct groups considered themselves descendants of a mythical race from the distant past known as “the People of ‘Ad.” This oral tradition persists to the present day; during the COPR  survey, local Mahri tribesmen recounted this same folklore to us, contending that they belong to the oldest lineage on the Arabian Peninsula. Along his journey, Thomas charted Iron Age forts that once served as waystations along the incense route, and noted a camel track that Beduin folklore told was a route to the mythical city of Ubar (Thomas ). This last observation is what eventually led N. Clapp to organize the first archeological investigation of the Nejd Plateau in the early ’s to look for Ubar. Two years after Bertram Thomas’ accomplishment, Philby was finally successful in launching his expedition from Riyadh across the Rub al-‘Khali. During his travels, he inadvertently discovered a large geological impact crater, of which only a small handful are known throughout the world. Philby collected botanical specimens, as well as pottery and lithic tools. Discovery of these artifacts led him to conclude that the climate in the Rub al-‘Khali had been considerably more humid in antiquity (Philby ). Philby conducted a second expedition in  to explore the upper portion of the Wadi Hadramaut and to locate Shabwa, the ancient capital of Hadramaut. During the height of prosperity in the late Iron Age, the Kingdom of Hadramaut stretched from central Yemen to the Dhofar Escarpment, controlling both the production and export of frankincense. His journey successfully 61 located the ruins of Shabwa that had been described in detail nearly two millennia earlier by Pliny, though he was disappointed to find the actual extent of the city was more modest than he had been led to believe from Pliny’s account (Philby ). It is a testament to the opening gates of South Arabia that Freya Stark, a female traveler, could enter so deeply into Muslim society on her expedition through the Hadramaut Valley in Yemen. In January of , the explorer sailed to Mukalla, and then traveled up the Wadi Du’an, into the legendary and previously inaccessible Hadramaut Valley. She visited the major cities in the valley, including Shibam, Seiyun, and Tarim. Her book; The Southern Gates of Arabia: A Journey in the Hadhramaut described the architectural and potential archaeological treasures throughout Hadramaut, bringing this little-known region to the attention of scholars. Until that point, historians had focused primarily on the Yemeni Highlands, where the ancient Sabaean and Himyaritic kingdoms flourished in antiquity. CATON-THOMPSON’S HADRAMAUT SURVEY A prominent archaeologist specializing in the prehistory of northeastern Africa, G. CatonThompson’s interest was piqued by Stark’s travelogue of Hadramaut, which led her to carry out a series of investigations in Yemen during the late ’s. Her work was conducted at the same time Leakey’s fossil hominid findings were bringing world attention to East Africa. In response to these new data, Caton-Thompson traveled to Arabia in order to address the question of Pliocene and Pleistocene connections across the Red Sea. She excavated an Iron Age temple of the Hadramaut moon god, Sin, at Hureidha in the Wadi ‘Amd (Caton-Thompson ). From the perspective of prehistoric archaeology, Caton-Thompson’s most significant contribution was her survey throughout the tributaries of the upper Hadramaut drainage system (Caton-Thompson ). This project was the first consideration of Arabian prehistory, and the first to recognize the presence of Pleistocene-era stone tools. Even today, it is one of the most thorough documentations of ancient lithic technologies in South Arabia. 62 Lithic scatters were observed in several regions including: the spur of the jol overlooking Hadramaut above Tarim, in cliff screes of the Wadi Hadramaut and Wadi ‘Amd, in the ten, five, and three meter gravel terraces of Wadi ‘Amd, and in the aeolian silt deposits that fill the main channel of the Wadi ‘Amd. She describes the general nature of the collected assemblages as: a pebble-casual flake culture feeling its way towards a sort of half-completed Levalloisian technique handicapped from advance to a better level by its continued addiction to unfaceted high angling [of the platform] and poverty of retouch and [tool] types. (Caton-Thomspon : ) All of the artifacts collected were manufactured from low-quality chert or silicified limestone. The material from the silt/gravel sequence filling the main channel was comprised of flaked pebbles, which Caton-Thompson classified as chopping-tools. The flakes were all struck from straight or cortical platforms, with a simple unidirectional core reduction strategy. The material exhibited a mix of light and heavy rolling, indicating it was in secondary position. Because of the paucity of artifacts on or within the wadi sediments, Caton-Thompson posited that Palaeolithic human activity was probably restricted to a nearby lateral ravine that had a perennial stream, rather than in the main channel (ibid.:-). Along with non-descript chopping-tools, similar to those found in the Wadi ‘Amd aeolian silt, sub-discoidal Levallois tortoise cores were reported from the ten meter gravel in the same wadi. These specimens exhibited radial or lateral flaking around the core margins. This method of reduction was apparent from the debitage, as well; flakes typically showed radial or longitudinal scars. Again, the platforms are primarily straight or cortical. Very few tools were reported, consisting of chopping-tools, sidescrapers, and discoidal scrapers (ibid.:-). Only seven tools were collected from the five meter terrace, most of which were nondiagnostic flakes and cortical flakes from a unidirectional, single-platform reduction technique. The material was in slightly better condition than those specimens collected on the ten meter terrace (ibid.:). 63 Lithics on the three meter terrace were greatly rolled. Two cores are split discoidal cobbles that exhibit convergent scar patterns on flat working surfaces. Like the other terraces, the debitage have plain and cortical butts. The single tool was a steep-ended scraper (ibid.:-). Several lithics were recovered from the surface of the aeolian silt filling Wadi ‘Amd, and were therefore considered chronologically later than the artifacts occurring in the terrace gravels. These surface finds were considerably less worn than the terrace artifacts, though they showed variable degrees of surface wear. One of the most notable pieces from this locus was a flat Levallois core reduced from an ovoid chalcedony pebble. The core had centripetal preparation on both faces, and the ovoid-shaped scar of a Levallois flake removed from the working surface. The striking platform appears to have been faceted. In addition, there was a sidescraper with obverse, denticulated retouch along one lateral edge. There also was a flake that showed platform faceting, a converging scar pattern, and marginal retouch along one lateral edge (ibid.:). A relatively large assemblage of material was collected near a modern spring in the Wadi Djeda, a steep, rocky ravine within the southern cliffs of the Hadramaut, just east of Seiyun. The lithics came from the terrace on a scree slope some  m above the floor of the pool. The quality of the chert was quite bad, though arêtes are sharp and there was little evidence for transport. One piece was particularly noteworthy—a bifacially worked tool reduced from a large flake. The dorsal face exhibited semi-steep, heavy invasive retouch and the ventral face was partially reduced via flat, invasive flaking. The piece had a heavier patina and greater abrasion than other specimens from the scree. There were close to  artifacts collected from this area, though most were described as very poor quality of “misshapen, mistruck flakes.” The few tools that were reported are borers and sidescrapers (ibid.:-). A moderate density of material was recovered from around a large spur on the southern edge of the Hadramaut Valley near the town of Tarim. The material was found in the wadi gravels at the base of the spur, on the lower edge of the scree slope, and on the top of the spur. Artifacts from these loci were divided into three groups: Tarim a, b, and c respectively. 64 The Tarim Group a assemblage was comprised of chopping tools, radial cores, and two bidirectional flake-blade cores. Manufactured from poor quality chert, the debitage indicates a predominance of bidirectional cores with longitudinal preparation. One blade was recovered with a well-faceted platform, in every other case they were straight or cortical. Group b was collected on the scree about  m above the wadi gravels, and was in good condition with sharp arêtes. Of particular note was a flat, triangular Levallois core with centripetal preparation and faceted platform. In addition, there were pieces described as chopping-tools, discoids, and at least one flake with a well-faceted butt. The final assemblage from Tarim, Group c, comes from the summit of the scree. Again there was some evidence for platform faceting, discoids, Levallois cores, choppers, and a “spindle-shaped” core similar to a specimen from Group a. Caton-Thompson concluded that the material from Tarim and Wadi Djeda “are so similar that, whatever the age range, which must surely be considerable, they are certainly homogeneous aspects of a single continuous unchanging industry” (ibid.:-). The final area surveyed by Caton-Thompson around the port of al-Mukalla, approximately  km south of Hadramaut. Artifacts were noted on a gravel plain that dates to the Pleistocene and had been subsequently scoured by lateral erosion from two neighboring wadi systems. Therefore, Caton-Thompson posited that the artifacts from this assemblage were Neolithic or later in age. One artifact is a small, pointed oval core that may, in fact, have been a bifacial preform. There are diminutive sub-spherical cores with radial reduction, plano-convex naviforms, and a trihedral rod with regular, parallel pressure flaking (ibid.:-). Caton-Thompson concluded that this material was not related to Africa; however, her Hadramaut survey preceded most MSA archaeological work in East Africa. Thus, her region of comparison was restricted to Northeast Africa, which, at the time, was primarily represented by the Kharga Oasis assemblage with “true” Levallois technology (e.g. radially prepared working surfaces and well-faceted platforms). What Caton-Thompson did not know at the time, which has been subsequently demonstrated by a number of archaeological expeditions to the Horn of Africa, was that Upper Pleistocene material exhibits a mix of façonnage and radial technologies, with occasional blade 65 and point core reduction. In the regions contiguous to South Arabia, the production of bifacial points in the MP/MSA is restricted to East Africa and a few sites in the eastern Sahara, there is not one example of façonnage technology from the Near East or Zagros Mountains. Oil Throughout its history, Arabia Felix was the purveyor of highly-valued trade goods. It supplied copper and diorite to Mesopotamia and the Indus Valley, incense and spices to the Mediterranean World, and eventually coffee to Western Europe. The most recent chapter in Arabia’s commercial history began in , when the Shah of Iran temporarily revoked the Anglo-Persian Oil Companies concession to exploit oil in his country. This act set off a flurry of activity by U.S. and British companies desperately seeking alternate sources of cheap Middle Eastern petroleum. An American corporation—Standard Oil Company of California (SoCal)—discovered oil in Bahrain, and one year later SoCal obtained a concession from the Kingdom of Saudi Arabia to conduct petroleum exploration in their country. After six years, the search proved more successful than they could ever have imagined; the largest oil reserves in the world were discovered beneath the kingdom. One of the myriad consequences of abundant oil reserves was the rapid development of interior Arabia. Suddenly, there was global interest in this often ignored region, which had previously only received the attention of fringe scholars and intrepid explorers. Fueled by petroleum development and political maneuvering, South Arabia was abruptly dragged into the modern world. The pace of exploration dramatically increased. Wilfred Thesiger, a member of the British Foreign Service, was dispatched to Dhofar in  to investigate breeding sites of desert locusts that periodically swept out of Arabia and plagued crops in East Africa. Not long after his arrival, he fell under that same enchantment cast by the the Rub al-‘Khali that had previously seduced Philby and Thomas. He organized his own crossing of the Empty Quarter, and like his predecessors, documented prehistoric materials throughout the desert. In particular, he described a series of unique ancient stone structures that occurred at regular frequencies along the caravan track (Thesiger ). 66 These installations, referred to as “triliths,” are comprised of a series of three upright slabs of rock, aligned along a linear path, typically found on the ridges of prominent hills. They are common throughout southern Arabia, though their function remains a mystery. One of the most plausible explanations is that they were Iron Age waystations along the incense route (Zarins ). In an account of his travels, written some  years after his journey, Thesiger noted the fundamental changes that had befallen the region, and accurately prophesied what would soon come to pass: I went to Southern Arabia only just in time. Others will go there to study geology and archaeology, the birds and plants and animals, even to study the Arabs themselves, but they will go about in cars and will keep in touch with the outside world by wireless. They will bring back results far more interesting than mine, but they will never know the spirit of the land nor the greatness of the Arabs. (Thesiger :xiii) Discussion This chapter has described the early history of research in southern Arabia. Since the Bronze Age, South Arabia has been enshrouded by “the myth of Arabia Felix.” For the first five millennia of its history, interaction with the outside world was limited to commercial endeavors along the coast. These interactions spawned fanciful stories of the interior; winged vipers protecting hidden groves of frankincense and ancient cities buried beneath the sand. Ancient scholars perceived of this region as lying on the periphery of known civilization; the body of knowledge was, to a large degree, comprised of secondary accounts and unverified stories. By the early th century, facilitated by British colonialism, adventurers looking to explore the world’s last frontiers turned their sights on Arabia. Many sought to claim the ultimate prize, which was the first complete crossing of the Rub al-‘Khali. At the same time, the first prehistoric research campaign to South Arabia was initiated by G. Caton-Thompson in the Hadramaut Valley. The net result of these combined works was the discovery of a variety of lithic surface scatters throughout the interior of southern Arabia, and the realization that the Peninsula was occupied further back in time than any had previously considered. This also marks the beginning of the problems that 67 plague current research—evidence for the Arabian Pleistocene is only known from palimpsest surface scatters, which are often mixed with Holocene materials. The veil enshrouding South Arabia was abruptly ripped off upon the discovery of oil reservoirs beneath the surface of the Peninsula. One byproduct of this tectonic socio-political shift was the sudden infiltration by armies of outside scientists. The next chapter continues this presentation of the state of prehistoric research by reviewing the work that ensued following the discovery of oil. Archaeological projects that were conducted subsequent to World War II are presented (when oil exploitation began in earnest), and used to explain the foundations of the Arabian prehistoric chronological sequence. 68 Chapter 4 THE PRESENT STATE OF RESEARCH IN SOUTH ARABIA Palaeolithic industries, legion in number and distressing in nomenclature. —Gertrude Caton-Thompson, Climate, Irrigation, and Early Man in the Hadramaut The preceding chapter reviewed South Arabian archaeology from the earliest written history until the ’s. During this time, research on the Peninsula, for the most part, was a byproduct of political and commercial interests. Early seafarers from the Near East and the Aegean wrote travel accounts based on their trade experiences up and down the expansive coastline. A handful of intrepid explorers penetrated the hinterland, providing fleeting glimpses of the vast deserts that sprawl across much of the interior. By the middle of the th century, the dismantling of the Ottoman Empire and burgeoning oil industry fueled a flurry of amateur archaeological activity; geologists prospecting for petroleum began to note the myriad lithic sites littering the landscape. What was lacking was a research agenda directed by scholarly interest in the prehistory of Arabia. This situation began to shift in the late ’s and early ’s, in part due to changing socio-political conditions throughout the Middle East. The Iranian Revolution, Iran-Iraq war, and Soviet intervention in Afghanistan diverted a number of projects southward into the Gulf countries, who provided a solace from the turmoil that was ravaging the northern Middle Eastern countries (Tosi ). As a result, the nature of Arabian archaeology shifted from opportunistic, random data gathering to theoretically driven research. This chapter presents the state of Arabian prehistoric archaeology in the modern era. In the last few decades, several international projects have been carried out throughout southern Arabia, producing a number of new lithic assemblages. Unfortunately, the paucity of stratified sites continues to impede much of our understanding of the chronological sequence. Potentially earlier artifacts 69 have been prematurely placed in a Neolithic framework, despite the fact that none of the material in question has been found in a datable context. Therefore, it is necessary to review the entire body of published Pleistocene and Holocene lithic material, in order to catalog the variety of industrial complexes found in Arabia. The chapter attempts to answer the question: which elements can be accurately dated and which techno-typological complexes are chronologically unverifiable. Although there are a large number of Holocene lithic industries that have been identified in the literature, only those with a bifacial component will be presented, as they are the only Holocene assemblages that directly relate to this dissertation. The presented projects are organized geographically within five geomorphic zones (roughly from west to east): ) the Yemeni Highlands and Tihama Coast; ) southern Saudi Arabia; ) Rub al-‘Khali; ) Hadramaut, Dhofar, and Nejd tablelands; and ) Arabian Gulf, littoral Oman, and adjacent Wahiba Sands Desert. Figure - presents a general map of South Arabia with key archaeological sites and drainage systems discussed in the text. Yemeni Highlands and Tihama Coast GARBINI AND DE BAYLE DES HERMENS The first Palaeolithic material discovered in northern Yemen was found by Garbini during a brief expedition to the Yemeni Highlands in  (Garbini ). The artifacts were reported near Bayt Na’am, some  km west of the capital Sana’a. The collection was made on the surface, and there is little description of the material other than having a general Middle Palaeolithic character. Five years later, more Palaeolithic material was discovered by de Bayle des Hermens at occurrences just outside of Sana’a and in the region around Marib. The site near Sana’a was on a plateau above Wadi Dahr, not far from Garbini’s original findspot. Fifty-one lithic artifacts were collected including cores, flakes, sidescrapers, and denticulates, all attributed to the Middle Palaeolithic. At Jebel Milh, near Marib, only ten artifacts were collected, which are described as Lower Palaeolithic pebble tools (de Bayle des Hermens ). While not particularly useful for 70 71 Figure -. Map of sites, towns, and drainages mentioned in text. understanding the nature of these industries, the mere presence of Palaeolithic material in northern Yemen is noteworthy. ITALIAN ARCHAEOLOGICAL MISSION TO THE YEMEN ARAB REPUBLIC In the early ’s, an Italian project was initiated in the Yemen Arab Republic (North Yemen), led by Alessandro de Maigret (de Maigret , , ). The Italian Archaeological Mission included both Pleistocene and Holocene survey and excavation, as well as geoarchaeological, palaeobotanical, palaezoological, and ethnoarchaeological research. The team investigated a number of geomorphic zones, including drainage systems and upland plateaus within the jagged basalt highlands, the coastal piedmont, and the Ramlat as-Sabatayn—a massive dune field that marks the southwestern fringe of the Rub al-‘Khali. They successfully located a number of Neolithic and Palaeolithic findspots, casting light on this previously unknown region of South Arabia. Yemeni highlands and coast There was abundant evidence for Palaeolithic occupation throughout the Hawlan region of the Yemeni highlands, although no in situ sites were located. The team discovered a high concentration of Lower Palaeolithic artifacts in the foothills of Hayd Ahmad. This density of material on the surface of the gravel plain is due to a high degree of erosion, both from truncation of the Pleistocene sediments by surface runoff and anthroturbation by present day agricultural activity. Another concentration of Lower Palaeolithic artifacts was identified at the base of Jebel Gawl, in a similar geomorphic setting. Unfortunately, the description of these collections is limited to “cores, flakes and a bifacial of the Acheulean period” (de Maigret :). A second collection was made in the foothills of Hayd Ahmad during a subsequent field season, producing “flakes, side-scrapers, notched tools, cores and a few bifacials [handaxes]” (de Maigret :). Of particular interest was a site discovered on the Dhamar Plain in the Yemeni Highlands a few kilometers south of the city of Ma’bar. The region consists of Tertiary volcanic formations filled with interstratified beds of loess and alluvial gravels. More than  artifacts were collected 72 on a deflated surface in proximity to a small volcanic cone. The collection consisted of flakes, cores, choppers, chopping tools, sidescrapers, and denticulates. There were two bifacial implements manufactured from tuff and rhyolite that were flat, ovate, and exhibited regularized edges. Bulgarelli () classified these specimens as Upper Acheulean, analogous to the well-made ovate and cordiform bifaces reported by the Comprehensive Survey of Saudi Arabia, as well as to those specimens from the site of Habarut on the Yemeni/Omani border. The Italian project reported Middle Palaeolithic material, though in secondary position, and/or without stratified context on deflated surfaces. During a preliminary reconnaissance in , de Maigret’s team located the site of Humayd al-‘Ayn in the region of Hawlan. The findspot was on the surface of a Mesozoic Plateau, where flint slabs were eroding from tabular exposures. A sample from an area of  m was collected for analysis, which yielded a high percentage of primary material, some flakes and cores, and a limited number of finished tools (sidescrapers, denticulates, and notches), leading the authors to conclude that the site served as a factory for the extraction and early preparation of local flint slabs and nodules. In describing the technological and typological characteristics of the collection, the surveyors speculated “the lithic artifacts could be compared on a morphological basis with industries of the Middle Palaeolithic” (Bulgarelli :). Not far from Humayd al-‘Ayn, on a similar plateau, the site of al-Masanna was discovered; it was also posited to be a factory site for the extraction and early reduction of flint nodules, and exhibited a similar Middle Palaeolithic industry (ibid.). In , the site of Hammat Gawl an-Numayri (HGNiv) was discovered on the bank of the Wadi Hababid in the Yemeni highlands. The site was in proximity to a raw material outcrop and relatively large drainage system. These factors would have provided favorable conditions for human occupation. The collection from HGNiv was characterized by “a large number of cores, chips and flakes, and by a small amount of tools” that were reported to have Middle Palaeolithic characteristics (de Maigret :); however, it is unclear as to what those characteristics were exactly. A second site exhibiting a “Middle Palaeolithic type industry” was discovered on the northern part of the Suhman Plateau in the highlands, named Jebel al-Humaymah (HUi). The local situation is similar to HGNiv, 73 with water, raw material, and topographic relief in the vicinity. Again, no details were provided regarding the technology or typology of the HUi lithic collection (ibid.). After determining that Palaeolithic sites in the highlands were either on deflated surfaces or buried under meters of sediments and thus inaccessible, the Italian team refocused their efforts on the piedmont of the Tihamah Coast. They explored the thalweg of several widian that drain from the highlands onto the coastal plain, noting multiple phases of aggradation in the exposed profiles, which geomorphologists argued were deposited during Pleistocene wet phases. The layers are comprised of alternating beds of cobbles and consolidated gravel colluvium. Artifacts were discovered within these Quaternary deposits, most frequently in the upper portion of the conglomerate. Again, the description of the collected material is limited to “a large flake” and “a chopper,” though their position within the profile indicates that they are almost definitely Palaeolithic (de Maigret :; Bulgarelli :). The Italian Archaeological Mission also documented a wealth of Neolithic sites, including several localities with buried, in situ material. The sites were attributed to the Neolithic based on the occurrence of lithic artifacts without pottery, and their association with simple stone structures, such as terraced walls and elliptical huts. A high density of findspots was recorded along the upper courses of the Wadi Danah on the eastern plateau, at approximately , masl. Large-scale excavations were conducted at the site of Wadi at-Tayylah (WTHiii) over several seasons, providing a plethora of data. Large stone structures were the most conspicuous feature at WTHiii, described as elliptical and ovate houses. In situ material was discovered inside one of the houses, including a hearth, lithic artifacts, and faunal remains. In addition, there was a fragment of a stone bracelet, which the authors compare to late th millennium specimens in the Nile Valley and Levant. Though the bones were badly eroded or crushed, it was possible to identify most of the sample. Nearly  of the bones consisted of domesticated cow, suggesting a community of cattle pastoralists. Considering the subarid conditions that currently characterize this region, evidence for pastoral subsistence provides further corroboration for an Early/Middle Holocene wet phase. 74 The lithic assemblage was characterized by a high frequency of naturally backed cutting tools, unifacial pieces with denticulate retouch, well-made endscrapers with both rounded and straight retouched edges, “bill-like” perforators, heavy-duty rabots, trihedral drills, lageniformes (thick, elongated ovate bifacial tools), and small, asymmetrical unifacial and/or bifacial pieces that ranged in shape from ogival to ovate. These latter bifacial forms exhibited heavy scars, pronounced arêtes, and irregular edges, indicating that they were probably preforms, and therefore not analogous to Eden’s Type  forms from the Rub al-‘Khali (Edens ). Among the debitage, éclats de taille (thinning flakes) are frequent, while blades are virtually absent. Based on sedimentary and climatic findings from the region, the authors bracketed the Neolithic occupation at WTHiii between the th and th millennia  (de Maigret ; de Maigret ). A Neolithic shell midden, Jahabah, was discovered in the lower reaches of Wadi Ruman on the Tihama Coast, not far from the city of al-Hudaydah. The site was located on the eastern extremity of the coastal plain; its position far inland suggests it was contemporary with the Early Holocene marine transgression. The lithic assemblage associated with the midden exhibited a variety of local volcanic raw material, including obsidian, rhyolite, and basalt. Obsidian is by far the most prominent material at Jahabah, representing over  of recovered artifacts. The assemblage included one diagnostic form—a tanged, pressure flaked arrowhead analogous to the Neolithic projectile points found throughout the interior of South Arabia. A Terebrailia palustris shell recovered from the top layer of the midden yielded a radiocarbon date of , ±  , thereby adding further resolution to the dating of the Neolithic Arabian Bifacial Tradition (de Maigret ). Ramlat as-Sabatayn In the final year of the Italian mission to northern Yemen, the team traveled to the Ramlat as-Sabatayn dune field, at the fringe of the Rub al-‘Khali, to investigate desert Neolithic sites in this region. The expedition focused around the area of Wadi Harib, near the village of al-Haglah al-Hama. A number of surface scatters were recorded, including two particularly rich findspots named HARi and HARii (de Maigret ). 75 HARi was found on the right bank of Wadi Harib, near a present day well. The ground surface was noted for its dark-gray coloring, thought to be the remnant of a deflated palaeosol. In addition to lithic material, marsh gastropods (Melania tuberculata) and a fragment of Cypraea were recovered from the site. Lithics were manufactured on (listed in order of frequency) flint, quartz, granite, basalt, and obsidian. The flint occurs in nodular form throughout the wadi channel, and, presumably, the volcanic material was obtained from sources in the Yemeni Highlands. Debitage accounted for over  of the collection, including simple flakes reduced by hard hammer percussion, and small, amorphous cores. The toolkit was comprised of endscrapers, drills, and retouched flakes. The largest site discovered by the Italian expedition was HARii, located approximately two kilometers upstream from HARi. The scatter was situated on a slightly elevated dune above the wadi bank. There was a concentration of charcoal in the northern portion of the site; surveyors systematically collected from a six meter transect centered on this feature. A test-pit revealed that the archaeological material rested on a sandy veneer not exceeding two centimeters in thickness. Similar to HARi, the raw materials observed in the assemblage include flint, obsidian, quartz, and basalt. The majority of debitage were flakes, with a small handful of blades, though these appeared incidental rather than indicative of a true blade industry. Striking platforms were typically straight (), with some cortical (), and a small number of dihedral (.) and punctiform (.). The cores are similar to those collected at HARi, which were intensely exploited and as a result appear small and amorphous. The toolkit was comprised of projectile points, bifacial foliates, drills, and miscellaneous unifacial tools. Every one of the projectile points exhibited pressure flaking and showed the typical range of shape such as tanged, barbed, and rhomboidal forms. The bifacial foliates were quite small (not exceeding  mm in length) and most exhibited pressure flaking. There were two specimens that were categorized as broad foliates made with soft hammer percussion. These pieces are asymmetrical, have deep scarring, and irregular edges; therefore, it is likely they are, in fact, preforms [Type  of Eden’s typology]. One bifacial tool was a lanceolate ( cm long) with a pointed tip and convex base. 76 There is pressure flaking at the very end of this specimen. The unifacial tools consisted of endscrapers, sidescrapers, notches, denticulates, and retouched flakes. In addition to chipped stone artifacts, there were also grinding stones, polished axes, steatite bowl fragments, beads, and a fragment from an alabaster bowl, as well as floral and faunal remains. The identifiable bones included wild species of Equidae, Gazella, and one fragment of a wild goat (Di Mario ). WHALEN’S PLIO-PLEISTOCENE SURVEYS IN SOUTHWESTERN YEMEN By , North Yemen and South Yemen reached a peace agreement, ending their civil war that had lasted for over two decades. N. Whalen, who had served on the Comprehensive Survey of Saudi Arabia in the late ’s and early ’s, immediately took advantage of this opportunity to access archaeologically important areas of the country that had previously been restricted because of their position on the old border that divided the two Yemens. He conducted two surveys in  and  around the southwestern portion of the country, in proximity to the Bab al-Mandeb Straight (Whalen and Pease a; Whalen and Schatte ). The survey was primarily concerned with the Pleistocene occupation in southwestern Arabia, investigating the question of early hominid migration out of Africa across the Bab al-Mandeb. In , the team focused their efforts on the southern extent of the Tihama coastal plain and piedmont at the base of the Yemeni Highlands. This region receives heavy seasonal rains brought by the Southwest Indian Ocean Monsoon system. Consequently, the southern Tihama is covered by deep layers of sand, gravel, and silt deposited by coastal flowing widian that drain from the Yemeni Highlands. Descending directly from the western slopes of the mountains are large Lower Pleistocene alluvial fans, which have subsequently been dissected by erosional channels. These braided fans are mantled by a deflated surface of gravel and cobbles, comprised of pyroclastic rocks such as basalt, andesite, trachyte, and rhyolite. Whalen conducted a survey on several of these alluvial fans, reporting thirty sites with artifacts that exhibited characteristics from every phase of the Palaeolithic. All of the material 77 was collected on the surface, and was classified based on size, technology, typology, and degree of patination. Not one artifact was heavily rolled or abraded, indicating there was minimal secondary transport. Most of the sites were multicomponent, ranging in size from  m to , m. Acheulean specimens were most prominently represented among all the artifacts collected, characterized by large, hard hammer debitage, polyhedrons, discoids, picks, cleavers, and rare handaxes. The material is heavily patinated by thick black desert varnish. The authors note that this facies of Acheulean differs from that of Saudi Arabia, where handaxes are much more frequent at Lower Palaeolithic sites. Middle Palaeolithic artifacts were somewhat smaller than Lower Palaeolithic pieces, and exhibited some degree of soft hammer percussion (Whalen and Pease a). There was evidence for Levallois technology, and tools included scrapers, denticulates, and notches that had a patina ranging from pale yellow to light tan. There were Upper Palaeolithic artifacts recorded as well, which were smallest in size. These collections are characterized by the presence of blades that had been truncated or retouched into scrapers. Pressure-flaking is present, and the pieces exhibit little or no patina (ibid.). Whalen returned to Yemen in  to continue his work in the southwestern portion of the country, this time on the southern coast along the Gulf of Aden. During the survey, he focused on low foothills at the base of the mountains—flat escarpments that emanate from the highlands and run parallel to broad widian that drain southward into the ocean. The surfaces of the surveyed areas were entirely deflated, precluding the recovery of any stratified material. The collections of entirely surficial material were categorized based on the Mode  and Mode  system of classification used by some Africanists (sensu Clark ; Toth ). Mode  industries consist of large core tools such as choppers, polyhedrons, discoids, core scrapers, and some scrapers on flakes. Mode  industries encompass the same variety of tools as Mode , with the addition of handaxes, cleavers, and picks (these industries are more commonly referred to as Oldowan and Acheulean, respectively). The material from Whalen’s  survey was classified according to raw material, artifact dimensions, patination relative to other pieces from the same site, and presence/ absence of diagnostic tool types. 78 Whalen found six Mode  findspots on a single escarpment overlooking Wadi Shahar. The scatters averaged , m, with a relatively high density of artifacts per square meter. Specimens were typically made on quartzite (), with a predominance of choppers and cores. The material showed no evidence for rolling or size sorting, suggesting there was minimal post-depositional transport (Whalen and Schatte ). Mode  material was much more prevalent throughout the survey area; sites were ubiquitous on sloping fans between the base of hills and the bed of Wadi Ghadin. The surface of these fans was a hardened mantle of blackened basalt pebbles overlying a sandy clay loam. Scatters averaged , m, somewhat larger than the Mode  findspots, though were considerably less dense. Artifacts were coated with a thick desert varnish and badly weathered, though there was minimal evidence for rolling or size sorting, again indicating there was only minor post-depositional transport. The lithics were primarily made from basalt () and rhyolite (), and large scrapers were the predominant tool category, followed by a smaller percentage of cleavers, handaxes, and picks (ibid.). Whalen draws analogies between his Mode  [Oldowan] material and the material excavated from al-Guza Cave by Amirkhanov (Amirkhanov ). He posits that the presence of pre-Acheulean in southern Arabia, in conjunction with recently established early dates of . mya at Dmanisi in Georgia and Mojokerto and Sangiran in Java, suggests there was an early migration of Homo erectus out of Africa via the Bab al-Mandeb Straight sometime during the Plio-Pleistocene boundary (ibid.). ROOTS OF AGRICULTURE IN SOUTHERN ARABIA (RASA) PROJECT During the late ’s, a team led by J. McCorriston carried out systematic survey and excavation throughout the Wadi Sana, in the Yemeni Highlands (McCorriston ). The multidisciplinary project was successful in generating detailed models of Early and Middle Holocene climatic conditions, as well as adding significant resolution to the description and dating of Neolithic industries. The earliest material reported by McCorriston were Fasad points found in alluvial gravels and scree or beside Early Holocene fluvial and aeolian deposits. The characteristic implements are 79 unifacial points, often manufactured on blades, with a tang at the base achieved by obverse pressure flaking. These tools were collected in conjunction with drills made on flakes, as well as tabular scrapers. Though there were no absolute dates for these collections, McCorriston posits that they antedate the local bifacial tradition, based on the geological proveniences in which they were found and their marked contrast to dated bifacial assemblages, thus placing them sometime greater than ~, . A series of four rockshelters, named Khuzmum, were located on the eastern exposure of the Khuzma as-Shumlya bedrock outcrop, at the confluence of Wadi Shuymlya and Wadi Sana. Test pits were excavated in one of these shelters, producing over one meter of cultural materials, which were capped by a layer of eboulis postdating the occupation. A number of radiocarbon dates were obtained from ashy material within the stratified deposit, yielding dates ranging between , and ,  (calibrated). Over , lithics were recovered from the test pits, comprised exclusively of artifacts made on chert obtained locally from small wadi cobbles. Most of the assemblage was classified as debitage, of which  were éclats de taille. There were approximately  diagnostic tools, primarily bifacially worked projectile points with wings and barbs that were triangular in crosssection. The excavators also report bifacial preforms, trihedral rods and drills, and a low frequency of scrapers, borers, and retouched flakes. There were no significant changes in artifact types throughout the sequence, leading the authors to suggest that this was a homogenous industry throughout the duration of human occupation at Khuzmum. A number of open-air findspots were recorded along the middle course of the Wadi Shumlya, not far from Khuzmum. These localities are characterized by hearths with associated faunal and lithic material, situated on or in wadi silts. Several radiocarbon dates were obtained from charcoal fragments within the hearths, ranging from , to ,  (calibrated). These dates are corroborated by the presence of a large mammal bone identified as Bos sp., which could only survive in this region during the Early/Middle Holocene wet phase. A pressure flaked, tanged projectile point diagnostic of the Neolithic Arabian Bifacial Tradition was recovered from one of the hearths (McCorriston ). 80 Southern Saudi Arabia The transformation triggered by the discovery of oil caused rapid development and modernization throughout the Kingdom of Saudi Arabia. This resulted in the establishment of the Department of Antiquities and Museums in the early ’s. One of the first acts of the fledgling organization was to organize the Comprehensive Archaeological Survey of Saudi Arabia, a decadelong project aimed at documenting prehistoric and historic archaeological sites throughout Saudi Arabia. The project was massive in scale, covering a territory of over two million square kilometers that had not previously been reconnoitered. In order to handle this vast area, the country was divided into six primary zones, including: ) northwestern, ) northern, ) eastern, ) central, ) southwestern, and ) western (Masry ). For the purposes of this summary, only findspots falling within the southern half of Arabia—the southwestern, central, and eastern zones—will be discussed. EASTERN PROVINCE The first area investigated by the project was the eastern zone, led by Robert McC. Adams from the University of Chicago. The oldest evidence for human occupation in this region comes from the Yabrin Oasis on the northern periphery of the Rub al-‘Khali. Yabrin is a large depression surrounded by low, rugged escarpments. Many of the ridges and knolls overlooking this basin are capped by chert of varying qualities; scatters of lithic debris are often associated with these outcrops. At three particular occurrences, assemblages were found that were dominated by “Acheulean” implements, based on the presence of handaxes, cleavers, large scrapers, and blades, with most tools showing some degree of bifacial flaking. The presence of cleavers is noteworthy, as this tool type is typically associated with the Early Stone Age of Africa, the Near East, and India. The appearance of this tool form at Gesher Benot Ya’aqov in the Levant has been used to argue for an early hominid radiation out of Africa (e.g., Goren-Inbar et al. ). Another series of Palaeolithic findspots were located in the Wadi al-Sahba, near the town of Harad. Seven surface scatters were recorded on a Pleistocene terrace above the wadi channel. The assemblages typically occurred on low knolls and gravel ridges overlooking the wadi, and on alluvial 81 remnants protected from deflation by massive collapsed limestone blocks. They consist of heavy cleavers, retouched flakes, flake-blades, scrapers, and tortoise cores, leading the authors to classify the assemblages as Middle Palaeolithic. A number of aceramic sites were recorded near al-Hasa Oasis, some  km northeast of Yabrin Oasis. The lithic material is thought to date to the early post-Pleistocene, based on the presence of pressure flaked projectile points, blades, and scrapers. Unfortunately, no further detail is given regarding the attributes of these assemblages. The scatters were commonly found near springs or exposed blowouts within stabilized dune deposits, as well as adjoining shallow channels and interior drainage basins (Adams et al. ). CENTRAL PROVINCE Following work in the Eastern Province, the Comprehensive Survey of Saudi Arabia shifted its attention to the Central Province. The surveyors began to recognize problems inherent in Arabian prehistory: The identification of stone tool industries within Saudi Arabia remains at a very rudimentary level. The number of industrial complexes have been tentatively identified and placed in relative chronological order on the basis of some degree of similarity with the lithic sequence fairly well worked out in other parts of the Near East. (Zarins et al. :) The latter half of this statement is fundamentally problematic, as it presumes prehistoric industries on the Peninsula should resemble Near Eastern material. It is equally plausible there are affinities with East Africa, which has a markedly different techno-typological lineage than the Near East. This assumption haunts Arabian prehistoric studies to this day. In order to deal with growing uncertainty regarding surface collections, the Central Province team began by dividing the material into three groups without attributing them to any particular chronological phase: handaxes sites, flake sites, and “Neolithic” sites (this latter group is based on similarities with the abundant pressure flaked bifacial sites reported throughout the Rub al-‘Khali). 82 The team documented a number of isolated handaxes throughout the Wadi Dawasir; findspots are on a low terrace within the drainage system and on inselbergs that rise above the gravel plain outside of the wadi. Most of these implements are ovate and elongated ovate, with a small percentage of lanceolates. They are typically thick biconvex in cross-section and have sinuous edges. Often the butts are left unmodified, exhibiting rounded cortex, indicating that the handaxes were manufactured from cobbles, most often sandstone, coarse-grained quartzite, and ferruginous sandstone. Some unifacial tools are associated with the handaxes, although most are simply retouched or utilized flakes and chunks. There were a few scrapers with steep retouch, sometimes inverse. A small number of cores were found, classified as biconical and discoidal. A separate group of assemblages was collected, also from the Dawasir drainage system, distributed throughout the same geological settings as the handaxe sites. The technology is characterized by prepared flake cores; the most frequent type is flat discoidal with a single working surface and cortex on the underside. The surveyors note that the specimens are derived almost exclusively from tabular ferruginous sandstone, which is conducive to this method of core reduction. These discoids are technologically uniform throughout several assemblages. The striking platform is formed by short, broad removals around the margin of the core perpendicular to the working surface. A second type of larger discoid also occurs that exhibits radial reduction on both surfaces. The authors note that the high-backed discoids occur on a wider range raw material, and, hence, attribute morphological differences between the two types of discoids to the initial condition of the raw material—tabular versus globular. Cores classified as Levallois occur in low percentages at these sites, showing three different methods of convexity maintenance. These include radially prepared “tortoise” cores, longitudinally prepared bidirectional cores, and an elongated flake core with non-invasive radial removals. Finally, there were a large number of unprepared cores associated with these scatters. The most frequent are those with opposed platforms utilizing the same working surface, though other bidirectional types were found with adjacent working surfaces. In addition, there were a number of single platform prismatic blade cores. 83 Tools at these sites are scarce, the most frequent type being sidescrapers and endscrapers. There are low percentages of backed flakes, perforators, and small chopping tools. A small, bifacial lanceolate was collected at one site, though it was thought to be intrusive. In light of this thesis, however, the lanceolate may be coeval with the rest of that assemblage. Another category of sites was located on the surface of the Wadi Dawasir channel and on undulating sandy ridges outside the channel. These findspots are characterized by the presence of bifacial implements. The authors divide these bifacial sites into two groups based on technology. The first group is similar to the Rub al-’Khali pressure flaked bifaces, in that it exhibits small (< cm) shouldered and tanged points. Rhomboid forms, common in the Rub al-‘Khali, are absent in these assemblages. The pieces are derived from chert and fine-grained quartzite, vary in quality of workmanship, and are most often lenticular in cross-section. A series of blades were found at one particular findspot that shows various stages of bifacial reduction, indicating these implements were made on blade blanks. The second group is comprised of bifacial foliates that are larger in size and produced by percussion flaking. In addition, there are a small percentage of unifacial foliates that are of similar shape and dimension to the bifacial specimens. Most pieces are manufactured on chert, and the authors note these are probably not preforms for pressure flaked tools (Zarins et al. ). Pressureflaked and percussion bifacial implements rarely occur together; thus it is reasonable to suppose they belong to separate industries. Following their work in the Dawasir drainage system, the surveyors shifted their attention to the western portion of the Central Province, along the flanks of the ‘Asir mountain chain. The authors divided these sites into Acheulean, Mousterian, Mousterian of Acheulean Tradition, and Upper Palaeolithic. These sites are all surface scatters, so temporal attributions are based purely on techno-typological affinities with dated assemblages elsewhere in the world (Zarins et al. ). Sites classified as Acheulean were most often located on high terraces in proximity to sources of granite, rhyolite, andesite, and/or quartzite, which were the preferred raw materials amongst these assemblages. The most common tool types within this group were handaxes, cleavers, and picks. 84 Similar to the Lower Palaeolithic material from the Wadi Dawasir, the handaxes often have rounded, cortical butts. According to the surveyors, the majority of sites found at the base of the ‘Asir Mountains fall into the category of Mousterian. Similar to the distribution of Acheulean sites, these findspots most frequently occur on higher elevations overlooking widian and plains below. Quartzite was the preferred raw material, followed by andesite and rhyolite. In regions where chert was abundant, Acheulean and Mousterian sites do not occur. Tools within the Mousterian assemblages include scrapers (most often transverse), burins, borers, denticulates, notches, retouched blades, and bifacial knives. The prepared cores from these sites include both radial Levallois and biconical discoids. Sites with similar core technologies and toolkits, though with the added component of handaxes, were classified as Mousterian of Acheulean Tradition. Both Mousterian and Levallois points are conspicuously absent amongst Mousterian and Mousterian of Acheulean Tradition sites (ibid.). A small number of Upper Palaeolithic sites were tentatively reported, though in every case they occur in conjunction with earlier assemblages (and may, in fact, belong to these earlier components). They were classified as such based on the presence of burins, borers, blades, notches, and scrapers, though many classic Upper Palaeolithic features are absent, such as blade tools, points, and carinated pieces. The authors wisely caution that “Upper Palaeolithic industries in Arabia may be expected to differ from those that evolved elsewhere, and attempts to identify the one by the other may indeed be futile” (ibid.:). A number of sites extending from Riyadh to the ‘Asir Mountains were attributed to the Arabian Neolithic bifacial tradition; they are comprised of the typical array of prismatic blades, notches, scrapers, and pressure flaked, basally-tanged arrowheads. In addition, ground stone artifacts, bone fragments, ostrich egg shell, imported sea shells, beads, and steatite are occasionally found with these surface scatters. The sites occur at the top of inselbergs, headlands overlooking valleys, and on shuquq surfaces in interdunal zones. In two cases, structures were found that are possibly associated with the lithics, characterized as large stone circles. There is no report of percussion-flaked bifacial foliates in this region (Zarins et al. ). 85 Researchers returned to the Central Province in  to explore the area immediately surrounding the capital city of Riyadh (Zarins et al. ). Their system of classification followed that of previous years, dividing sites into Acheulean, Middle Palaeolithic, and Neolithic time periods based on techno-typological correlates in the Near East and Europe; there is no mention of prehistoric industries in sub-Saharan Africa. One of the most prominent series of Acheulean sites from this region was discovered in proximity to a quartzite outcrop and represents a complex of factory sites for the exploitation of this resource. The outcrop overlooks a large alluvial plain comprised of Mio-Pliocene undifferentiated alluvial sediments. Overlying this mantle of alluvium are exposed terraces which belong to a later period of fluvial activity, tentatively dated to the Middle Pleistocene based on posited phases of morphogenesis. The lithic scatters were ascribed to the mid to late Acheulean based on correlates in the Azraq Basin in Jordan, as well as other collections recovered during previous phases of the Comprehensive Survey. The toolkits consisted of handaxes produced by both soft and hard hammer percussion, cores, flakes, trihedral picks, backed knives, and burins. Surface collections of similar makeup were recovered on the Aruma Plateau at the base of a sandstone inselberg, as well as below a small quartzite outcrop within a tributary feeding the Wadi ‘Atk (ibid.). Some  km due west of Riyadh, a particularly rich findspot was located in the Saffaqah area. At the time of its initial discovery, an area of  m was collected in  x  m units, yielding some three thousand “Middle Acheulean” artifacts. The Saffaqah area is just southeast of the village of Dawadmi, in an east-west oriented valley that extends for approximately . km. The site, -, occurs on a small terrace about  m above the current valley floor; occupying the northern slope and descending fan of an andesite dike. Adjacent to the site are the remnants of two ancient waterfalls that evidence alternating cycles of wet and dry climatic phases. The volume of water apparently discharged by these waterfalls during the Pleistocene strongly suggests the existence of an ancient lake inside the enclosed basin (Zarins et al. ; Whalen et al. ; Whalen et al. ). Archaeologists returned to - in  and  to further investigate the findspot; special attention was given to this site because of the presence of subsurface deposits, making it the first 86 Table -. Radiometric dates mentioned in text. 87 Lower Palaeolithic site to be excavated on the Arabian Peninsula. The depth of the deposit ranged between  and  cm, comprising a single unit of soil derived from chemical weathering of the underlying granite bedrock. Many of the artifacts had calcareous concretions on their underside, due to the percolation of alkalis in the soil during moist intervals throughout the Pleistocene. These concretions were subsequently used for U/Th isotope dating. Consequently, the dates, ranging between , and , ( -) should be considered minimal, as they indicate the formation of the concretions rather than the production of the artifacts. The authors note that the two different date ranges cluster around Pleistocene wet phases, and, based on techno-typological correlates, were deposited around a quarter million years ago (Whalen et al. ). The lithics recovered from - were manufactured primarily on local andesite obtained from the dike, followed by granite, quartz, and rhyolite. The toolkit consists of hard hammer bifacial forms typical of the Acheulean, such as handaxes, cleavers, picks, trihedrals, polyhedrons, and spheroids. The smaller flake tools include borers, burins, notches, chisels, and unifacial knives. Among the debitage, there was evidence for Levallois technique, though no details regarding platform faceting or convexity maintenance are provided (ibid.). As previously noted, the presence of cleavers is particularly interesting, as this is typically a type common in Africa and mostly absent elsewhere during the Middle Pleistocene. Their appearance in the Levant has been used to argue for an early hominid expansion out of Africa (Goren-Inbar et al. ). In addition, cleavers have been reported from Galeria Pesada (Marks et al. ) and Atapuerca (Carbonell ) on the Iberian Peninsula; they are ubiquitous throughout India as well (Pappu ). Surveyors noted several Middle Palaeolithic occurrences throughout the Riyadh environs, recognized by the presence of Levallois technique, “tortoise” cores, and demonstrable differences in patina from Neolithic collections. The distribution of these sites mirrors that of the preceding Acheulean findspots, typically found below quartzite outcrops within the Wadi ‘Atk drainage system and at the base of sandstone inselbergs on the Aruma Plateau, suggesting similar moist environmental conditions during phases of the Middle and Upper Pleistocene. In contrast to the Acheulean material, flint is used for the manufacture of lithic implements. Middle Palaeolithic sites from the Riyadh area 88 are often found in proximity to nodular flint outcrops that are eroding from the limestone strata that comprise the local Jurassic-Triassic beds. The authors describe one site east of Riyadh on a limestone terrace as “easily the most spectacular of the period” within the Central Province (ibid.:). The site is situated on a flint outcrop and covers an area of approximately , m. Unfortunately, the only description given is “a dense scatter of blades, flakes, cores, and debitage.” The Neolithic sites within the Riyadh region fit easily into the previously described bifacial industries that are ubiquitous throughout the interior of Arabia. The types most frequently represented include pressure flaked, tanged or barbed arrowheads. Within the area immediately surrounding Riyadh, twenty-three sites of this type were recorded from four different geomorphic settings. The first zone is on fine silt wadi terraces immediately above present-day channels. These sites are typically diffuse scatters with no finite boundaries; artifacts are often found in isolation at these findspots. In addition, fire-cracked rock appears in abundance at these occurrences. Neolithic sites were also found on the shores of small relict playas. The third ecological situation for Neolithic findspots is on top of sand dunes, particularly on the Erq al-Banban just northeast of Riyadh. Again, fire-cracked rocks appear at these sites, as well as grinding stones and ostrich egg shell. Pieces of granite were found, suggesting a trade network of at least  km. At one such dune site, bead fragments were recovered derived from both Conus sp. and Dentalium shells, indicating connections with the Red Sea and/or Arabian Gulf. Finally, a small number of Neolithic sites were found at the base of sandstone inselbergs on the Aruma Plateau, in the same setting as Middle Palaeolithic and Acheulean findspots. SOUTHWESTERN PROVINCE Investigations during the  season of the Comprehensive Survey focused on the Southwest Province, which encompasses five distinct geomorphic zones: ) the Rub al-‘Khali, ) the ‘Asir highlands, ) the sandstone hills that lie between the Rub al-‘Khali and ‘Asir highlands, at the 89 interface of the Arabian Shield and Arabian Plateau, ) the Wadi Tathlith drainage system, and ) the Red Sea coastal plain (Zarins et al. ). As in past surveys, investigators reported several Acheulean findspots. In general, these scatters are not aligned with present day watercourses (with the exception of material from the Wadi Fatimah), rather they occur outside the drainage systems, typically at the base of granite inselbergs among quartzite outcrops. Material was also found along the Red Sea coast, the first Palaeolithic material reported from this zone. These coastal sites are associated with remnant coral terraces cut by marine transgression. Assemblages are typically comprised of handaxes with rounded, cortical butts, unifacial choppers, backed knives, multi-directional cores, and large flakes with prominent bulbs indicating hard hammer percussion (ibid.). Several “Middle Acheulean” findspots were located in the Wadi Fatimah, a drainage system that originates in the mountains just north of Mecca and flows southwestward into the Red Sea. Of the  sites associated with Wadi Fatimah,  were found on the northern rim above the channel, in proximity to raw material suitable for producing stone tools such as basalt, andesite, and rhyolite (Whalen et al. ). Acheulean sites were recognized based on the presence of diagnostic forms, including handaxes, cleavers, picks, trihedrals, discoids, polyhedrons, and spheroids. Handaxes were primarily lanceolate in shape, followed by amygdaloid, ovate, and subcordiform. Based on the deep scars and sinuous edges, they all appear to have been produced by hard hammer percussion. The Acheulean collections were distinguished from later material by their large size and heavy patination. The surveyors noted a similarity between this material and the subsurface assemblage excavated at . A test pit was excavated at findspot - in Wadi Fatima; though the sediments were sterile, a calcareous nodule was sampled that yielded a U/T date of around ,  (ibid.:). Mousterian industries in the Southwest Province were identified on the basis of the Levallois technique, the production of smaller blanks than the Acheulean, and the presence of “tortoise” cores and discoids. Several sites fitting this description were collected throughout the survey, found along wadi terraces or in small embayments. These sites have a similar distribution as the Acheulean 90 findspots, indicating settlement under comparable climatic conditions, which differ from present day hydrological configurations. Raw material consists almost exclusively of rhyolite and quartzite in the ‘Asir and interior locales, and basalt along the coastline sites, which are in proximity to large lava fields. A small number of sites included handaxes among the toolkit and were therefore classified as Mousterian of Acheulean Tradition (Zarins et al. ). Four Mousterian sites were located in the Wadi Fatimah region near Mecca, yielding discoids, biconical cores, convergent scrapers, a limace, and scrapers exhibiting Quina retouch. One site, -, yielded a handaxe in association with Mousterian implements, which Zarins (ibid.) classifies as Mousterian of Acheulean Tradition. No diagnostic Upper Palaeolithic sites were discovered in this region, though Zarins (ibid.) point out their absence may be due to the fact that the assemblages do not necessarily resemble those from other parts of the world; therefore scholars are not yet equipped to recognize them. There were several findspots, from a variety of settings, similar to the other pressure flaked bifacial assemblages ubiquitous throughout South Arabia and apparently dating to the Neolithic. Sites of this type were found in small embayments surrounded by rock outcrops within the sandstone foothills east of the ‘Asir highlands. Artifacts were manufactured primarily from fine-grained chert, followed to a lesser degree by quartzite and ferruginous quartzite; they are typical of the Neolithic, including the tanged projectile points that are the hallmark of this phase. While a few specimens were shaped by percussion flaking, their diminutive size and crude shape suggests these are preforms for pressure flaked implements, rather than completed foliates analogous to the bifacial tools represented at Habarut, among other sites. In addition to chipped stone material, ostrich eggshell fragments, bone chips, and a polished awl made on an exotic white quartzite were recovered. In a few instances, features were found associated with these scatters, described as small stone circles, grouped slabs, and hearths. At one site, obsidian blades and debitage were collected, the closest source of which is found hundreds of kilometers to the south in the Yemeni highlands (ibid.). Neolithic sites were also discovered in the Wadi Tathlith drainage system, also occurring within small embayments and at the base of rock outcrops. Artifacts include tanged and rhomboid 91 projectile points, finely retouched endscrapers and sidescrapers, small conical cores, and blades. Most of the implements are derived from the same fine-grained chert as in the sandstone foothills, with a small number of obsidian pieces. Ground and pecked celts were also discovered, as well as steatite bowl fragments. The highest concentration of Neolithic sites was documented in the ‘Asir highlands, found in a similar setting as the Wadi Tathlith and sandstone foothill findspots. Artifacts are typical of the Southwest Province material, and architectural structures similar to Wadi Tathlith were found at several of the findspots. At one site in particular, over  circular structures were recorded, as well as a wall built of granite boulders. Sparse occupation was noted on the coast, suggesting the Neolithic inhabitants of this region were not adapted to marine subsistence. In general, the authors observe that the Southwest Province Neolithic projectile points differ slightly from the rest of Arabia in that they do not exhibit the same level of craftsmanship, though forms and technology are more or less comparable (ibid.). The Comprehensive Survey returned to the southern Tihama Plain in  to further investigate prehistoric coastal occupations. They discovered a handful of indeterminate aceramic sites along the coastal edge of a sabkha flat. The largest of these findspots yielded a high density of lithic material knapped from local fine-grained igneous and metamorphic basement rock, including implements on high-quality quartzite, as well as a small number of exotic pieces on flint and obsidian. The toolkit was comprised of Neolithic bifacially pressure flaked, tanged and barbed projectile points, large choppers, and drills. In addition to these chipped stone artifacts, there were fire-cracked rocks, sandstone grinding fragments, ostrich egg shell, and abundant marine shells. A second site, -, was located on an interdunal shuquq and contained a large concentration of quartzite debitage with a few pieces of obsidian and flint. There is little description of the assemblage, though the brief mention of snapped blades and backed microlithic lunates is noteworthy, as this type parallels Middle Holocene developments in the Levant and Egypt, in which these diagnostic implements appear during the th millennium , or, alternately, are typical of the Mesolithic/Epipalaeolithic. 92 The third Neolithic site discovered during the  season was situated on a small sand dune along the sabkha/dune interface. The findspot consisted of lithic debris and a series of small hearths marked by circular installations of fire-cracked rock. The chipped stone material is comprised of quartzite cores, flakes, borers, scrapers, and awls, with a few specimens made on flint and obsidian. In addition to the flaked material, there were sandstone and basalt grinding stones, and a small greenstone polished axe (Zarins and Zahrani ; Zarins and al-Badr ). The Comprehensive Survey returned to the southern Tihama Plain the following year, locating more prehistoric sites, including one that is similar to - discovered the previous year. The findspot covered approximately , m, and was marked by a series of small, discrete shell middens resting on an alluvial/aeolian aggradation overlooking the coastal sabkha. Lithic material is similar to the - assemblage, consisting of quartzite choppers, scrapers, blades, flakes, awls, “groovers,” burins, and sickles, as well as a distinct “transverse projectile point.” Implements on flint were rare and included a blade, a pressure flaked projectile point, and an exhausted core. Several lunates (without backing) and a microlithic core were manufactured from obsidian. There were polished greenstone and basalt tools, as well as a polished stone bracelet that closely resembles specimens in the Sinai and Egyptian Chalcolithic. The authors note that the position of these coastal sites would have placed them on the projected shoreline during the Flandrian transgression (Zarins and al-Badr ). Rub al’-Khali ARAMCO AND THE DESERT LOCUST SURVEY A veritable army of geologists, biologists, and petroleum engineers invaded the Rub al-‘Khali during the ’s and ’s. These scientists were primarily employed by the Arabian American Oil Company (Aramco), founded in  to exploit the plethora of oil traps beneath Saudi Arabia’s Tertiary mantle, as well as the British Desert Locust Survey, which spawned Thesiger’s trek across the Rub al-‘Khali described in Chapter . As researchers traversed the landscape, they frequently stumbled 93 upon numerous and vast lithic surface scatters. Their collections supplied prehistorians with material for analysis, and their training in geomorphology allowed for more accurate documentation of geological setting and provenance. Unfortunately, most of the material collected by these avocational archaeologists consisted of bifacially worked arrowheads, foliates, and handaxes, reflecting the bias of non-specialists in collecting only bifacial tools. This tendency toward dihedramania—the unnatural passion for bifaces—has been pervasive throughout the history of research in South Arabia, and has often led scholars to ignore core technologies associated with bifacial tools. The following is an inventory of sites collected by amateur archaeologists throughout the ’s and ’s, and subsequently studied by lithic specialists. Among those asked to examine these collections was F.E. Zeuner at the University of London (Zeuner ). Lithic material was collected from a findspot on the western fringe of the Rub al‘Khali by members of the Desert Locust Survey, at a locality named Shaqqat al-Khariyta. The material was found in two spots: ) on the surface of an interdunal zone that was once a wadi channel draining northward from the Yemeni highlands, and which was subsequently buried by migrating dunes, and ) just south of the first locus, along the margins of the ancient channel, embedded in a thin veneer of surface sediments. The lithics from both loci were manufactured from an array of raw materials, including very fine-grained chert (approaching chalcedony), obsidian, fine-grained quartzite, and, less often, quartz. Aside from the quartz, which was derived from local pebbles, the raw materials came from non-local sources. Typologically, the assemblage from the surface of the interdunal zone appears Neolithic, based on the presence of pressure flaked projectile points that range in shape from lanceolate to ovate, and with basal tangs. Unifacial tools included thumbnail scrapers, borers, and retouched spalls. The second findspot yielded both lithic material and a dense scatter of bone fragments that had been impregnated with calcium carbonate and were thereby preserved in good condition on the surface. The bones were Gazella sp., and included a horn-core that may have been used for soft hammer percussion. Tools from this assemblage differ from the first group in the absence of foliate forms, and in an abundance of tanged/barbed pieces. Most tools are bifacial, though some forms are 94 only partially flaked on the ventral surface or are entirely unifacial. Like the first group, all specimens exhibit some degree of pressure flaking. Based on typological affinities in Africa, Zeuner (:) concludes that “the South Arabian Neolithic is derived from the west and that there may be both Magosian and Aterian elements in its ancestry.” This conclusion is at odds with the currently accepted scenario that posits an expansion of Pre-Pottery Neolithic B groups southward from the Levant (e.g., Tosi ; Potts ). Similar materials were collected in  by the Desert Locust Survey; they were stored in the National Museum of Kenya and eventually examined by Richard Gramly. Unfortunately, the provenance of these artifacts is not well recorded, and it is unclear if the material is mixed. The collections come from a series of surface scatters in the western Rub al-‘Khali, in proximity to Shaqqat al-Khariyta. In general, the artifacts consist of small, pressure flaked bifacial implements that are either leaf-shaped or tanged. Endscrapers and borers are prominent, as well. Like Zeuner, Gramly, noted that the percussion-flaked foliates are typical of East Africa, drawing analogies with the “Somaliland Doian” industry (Gramly ). A Near Eastern prehistorian working at the Harvard Peabody Museum, Henry Field examined a collection of surface material from four findspots that had recently been recovered by Aramco workers: ) One of the most notable sites was G-, where  elongated points of “Solutrean” type were identified. The assemblage has typological elements similar to Zeuner’s material from the first locus at Shaqqat al-Khariyta. The tools exhibit parallel pressure flaked ribbon-shaped scars, and are made on high-quality chert. In general, Field noted a tremendous wealth of surface material extending along the fringe of the Rub al-‘Khali, from Jiladah to G-. There was no raw material in proximity to these surface scatters; thus they are posited to be hunting camps (Field ; Field ). ) An assemblage of similarly well-made “Solutrean” type bifacial tools was reported from Station , also on the northern margin of the Empty Quarter. Lithic material was collected on a broad gravel plain at the edge of an ancient lake deposit, surrounded by large moving sand dunes. The tools were manufactured from fine-grained quartzite and chert. They were made on thin blades, 95 reduced by pressure flaking, and show a range of shapes including elongated triangular, elongated cordiform, and limande naviform (sensu Bordes ) bifaces. A small percentage of specimens have basal tangs. Along with these bifacial tools, there is one unifacial discoidal scraper with continuous, semi-steep obverse retouch along the entire perimeter (Field ). ) The site of Hairat al-‘Ul yielded an assemblage that is typologically identical to the other “Neolithic” sites in the Rub al-‘Khali. The lithic material was collected from a shuquq surface between parallel longitudinal dunes, just north of the Hadramaut. There are freshwater shells and ostrich eggshell fragments associated with occasional patches of silt that dot the shuquq. Chert and finegrained quartz artifacts were scattered across the surface. Like the other sites reported in this region, the bifacial tools were made on thin blades and exhibited fine pressure flaking (Field a). ) Three localities were identified just southeast of Nuhaydayn al-Qawnasah, along a dune ridge known as ‘Erq ar-Runaylah. These assemblages are typical of the Rub al-‘Khali pressure flaked bifacial industry, with finely-made arrowheads and points on thin blades (Field ). A small unnamed assemblage was collected slightly northeast of G- by Aramco workers, in which a charcoal sample was collected from a hearth associated with this lithic scatter that yielded a radiocarbon date of , ±   (Field b). Smith and Maranjian () analyzed a collection recovered near the Aramco camp known as Sharorah, in the southern Rub al-‘Khali near the Yemeni border. The findspot is on a large sandstone outcrop that rises above the surrounding gravel plain. Nearly  artifacts were recovered, manufactured from chert and quartzite. Three quarters of the assemblage are bifacial tools, most of which are pressure flaked points and foliates. Tanged points are common as well and the tools are generally elongated. There were some bifacial foliates among the artifacts that exhibited cruder flaking from percussion rather than pressure, and were generally broader in shape. A small percentage of unifacial tools were collected, including a variety of scrapers. Unretouched blanks and burin spalls also appear in the collection, though too few pieces of debitage were recovered to discuss technology at the site (Smith and Maranjian ). 96 A second findspot was examined by Smith and Maranjian, about  km north of Shaqqat al-Khariyta. The site of Jiladah occurs on an isolated gravel patch between large aeolian ridges. Of the  pieces collected,  are projectile points. This category includes bifacial foliates shaped by percussion, and various forms of pressure flaked bifaces that are ubiquitous throughout the interior of South Arabia. The few unifacial tools collected are small fan-shaped endscrapers, retouched blades, and perforators. The authors mention one bifacial specimen with a flute originating from the tip of the point, which they attributed to an impact fracture (ibid.). This is noteworthy, as one scholar has recently identified more of these “fluted” projectile points—all originating from the tip—and argues this is an intentional technology used in the manufacture of Arabian armatures (Charpentier ). Habarut is an oasis on the Yemen-Oman border, originally discovered by a British political officer stationed in Aden. The locale is some  km north of the coast on the Nejd Plateau. The Wadi Habarut is one of a several parallel widian draining northward from the Dhofar Escarpment. At the oasis, the wadi reaches approximately  m in width, running between abrupt sandstone hills. Next to the wadi is a perennial freshwater spring that supports a large grove of date palms. Flint implements were collected from the surface of a low hill  m from the spring, and from a nearby flat hilltop overlooking the oasis (Payne and Hawkings ). The first assemblage is comprised of varying types of chert bifacial tools including knives with diagonal parallel pressure flaking, trihedral rods with extremely fine pressure flaking across all three faces, tanged arrowheads, and plano-convex naviforms (sensu Rose ). While the assemblage is roughly comparable to the Rub al-‘Khali Neolithic material, as far as the technology employed in bifacial production, the finished forms are unique in the high degree of trihedral pieces (ibid.). Only two pieces were collected from the second findspot at Habarut, both described as “small handaxes.” They are made on high-quality chert, are elongated cordiform in shape, and have flat, invasive retouch achieved via direct percussion, rather than pressure flaking. The butts are thin and rounded, and there is a small amount of cortex on both faces, indicating the tools were reduced from tabular chert or plaquettes (ibid.). Van Beek et al. () speculate this distinct bifacial assemblage dates to the Late Stone Age, drawing comparisons with the “Somaliland Doian” industry, 97 although the Habarut assemblage lacks the concave-base forms that appear in East Africa. Analogies are also made with “Upper Acheulean” industries in northern and eastern Africa. In addition to the Upper Pleistocene/Holocene material discovered by Aramco workers and the Desert Locust Survey, Field identified a potential Lower Palaeolithic assemblage, the first of its kind in Arabia. The findspot comes from a locality called Nuhaydayn al-Qawnasah in central Arabia. About  artifacts made from chert and fine-grained quartzite, classified as handaxes and cleavers, were collected. The tools have heavy desert varnish and a deep mahogany patina. They exhibit rounded arêtes, indicating considerable aeolian abrasion (Field ). The U.S. Geological Survey documented similar artifacts, while conducting an investigation of Saudi Arabia in the early ’s. They reported a series of large, heavy handaxes, as well as crude chopping tools/cores. The material, collected from wadi gravels, is typically abraded, rolled, and has heavy desert varnish. While there are extensive descriptions of raw material lithology and artifact measurements, the published specimens were not examined by a lithicist, thus, data regarding technology and typology are absent (Overstreet ; Overstreet ). TYPOLOGY OF THE “RUB AL-‘KHALI NEOLITHIC” One of the hallmark works in defining the Arabian Bifacial Tradition is that of Christopher Edens (), who synthesized all preceding research in the Rub al-‘Khali and analyzed the variability among these various bifacial industries. Edens’ work represents the most comprehensive summation of lithic industries discovered in interior Arabia prior to the ’s, and has been used as a foundation for much of the current state of prehistoric research. Furthermore, it is one of the only accounts that goes beyond typological analysis and presents technological observations regarding the bifacial reduction sequence. Edens examined collections from four findspots in the western Rub al-‘Khali including Jiladah and Sharorah (originally published in Zeuner ; Smith and Maranjian ; Zarins et al. ), al-Mutabthat, and Mundafin. With the exception of Jiladah, which is located deep in the Rub al-‘Khali, the other three sites are situated along the desert fringe. The sites are all surficial; the 98 material was collected from deflated environments on gravel ridges, interdunal flats, and/or sandy depressions. Raw material used at these sites is diverse, including a wide range of cherts, quartzite, quartzitic sandstone, quartz, obsidian, silicified limestone, pertrified wood, and volcanic rock. With the exception of obsidian and volcanic rock, the sources for these materials are fairly close to the site, occurring on gravel pavements that are common throughout the western Rub al-‘Khali. The closest source of obsidian is far to the south, in the Yemeni highlands. The material is found in both nodular and tabular packages; the al-Mutabthat collection included several unmodified plaquettes. Edens notes that the quartzite and quartzitic sandstone appear in significantly higher frequencies among the bifacial tools versus informal tools, suggesting conscious selection of raw material. Based on typological analysis of the tools from these four collections, the author constructed a type list based on common attributes. His typology is as follows: ) Stemmed bifacial points. Pressure-flaked forms that are triangular to sub-triangular in shape, with stems that are either straight-sided with rounded bases or taper to a sharp point. They usually have short barbs or are shouldered. The cross-section is lenticular, and they range from - mm long, - mm wide, and - mm thick. ) Miscellaneous small projectile points. A variety of fairly homogenous forms including points of similar shape but smaller or larger than the dimensions presented in type , points that are triangular to plano-convex in cross-section, partially-bifacial points, and unifacial points. ) Rhomboidal points. Pressure-flaked, rhomboid shaped points with a lenticular cross-section. ) Narrow foliates. Pressure-flaked, slender foliates that are longitudinally symmetrical, with rounded butts and sharp points (occasionally both ends are rounded). The lateral edges range from straight to convex, and the cross-section is lenticular. The dimensions range from - mm long,  mm wide, and - mm thick. There is often a burin spall removal originating from the tip of the piece and running down one or both margins, possibly a technology used for resharpening (cf. Tixier ; Wendorf and Schild ). The author identifies lateral snaps, as well as fluting spalls that emanate from the tip and plunge into the center of the piece. Edens is in agreement with Smith and 99 Maranjian (), asserting that these breaks result from impact fractures rather than an intentional technology (contra Charpentier ; Charpentier and Inizan ). ) Broad foliates. These forms are distinguished from type  by the use of soft hammer technology, and greater breadth relative to length. Two sub-types are included in this category: those with the greatest width near the base of the piece, as opposed to those widest near the center. Dimensions are generally - mm long, - mm wide, and - mm thick, though Edens notes that these forms have a much greater variety in size than the other types. He interprets some as roughouts for stemmed points, though adds “there is no definitive evidence for this” (Edens :). ) Lanceolates. Relatively large, soft hammer and pressure flaked bifaces with straight to gently convex sides, rounded butts, and fairly acute tips. They range in size from - mm long, - mm wide, and - mm thick. Several of the lanceolates have margins that angle sharply from the midpoint to form a sharp tip. The cross-sections range from lenticular to rhomboidal. ) Unidentified broken bifaces. This category includes tip, medial, or butt fragments, most often of Type  and Type  pieces (because of their larger size and, thus, greater tendency to break. ) Unfinished bifacial pieces. This determination is based on the regularity and straightness of the edge, relative thickness of the cross-section, and the nature of the retouch technique. ) Endscraper on flake. These forms have convex scraping edges opposite to the proximal end of the flake. Some are carinated or semi-carinated, though the majority are simple scrapers. In many examples, there is also shallow or blunt retouch along one or both lateral edges. A series of endscrapers from the site of Jiladah exhibit bulbar thinning. ) End scraper on lamellar flake. This type is distinguished from form  by the use of blade blanks in manufacture. The author notes that these blades are not true blades in the sense of recurrent flaking from a prepared prismatic blade core, they simply fall into the blade category based on their dimensions. There is often steep lateral retouch and, in some cases, bulbar thinning. ) Peripheral scraper. These scrapers are manufactured on thin flakes. The scraping edge is convex, formed by steep marginal retouch. Like form  and form , lateral retouch is often present. 100 ) Simple sidescraper. This category is defined by the presence of a scraper edge along one side of the flake, ranging from sub-parallel to scalar retouch. In some cases, there is rough backing, and one specimen is essentially a uniface, with soft hammer retouch entirely across the dorsal face of the piece. ) Miscellaneous scrapers. These miscellaneous forms include composite scrapers (with both distal and lateral retouch), convergent bilateral scrapers, circular scrapers, and unidentifiable fragments. ) Trihedral drill. These implements (mèche de forêt) are either straight-sided or have curved lateral edges, with a triangular cross-section. Sometimes the ventral face is only partly modified, whereas in other instances all three faces are shaped by finely executed pressure flaking. ) Drill on flake. A narrow bit is formed on a flake or blade. The bit itself is bifacially worked, while the rest of the implement is parti-bifacial, or simply exhibits ventral thinning. The pieces are formed by pressure flaking, and cross-section ranges from triangular to plano-convex. ) Miscellaneous small tools. This category includes notches, backed pieces, geometrics, and knives. ) Retouched flakes. Informal tools or simply utilized flakes. ) Axe/adze. Large tools with a straight, beveled working edge. Most have rounded butts and convex lateral edges, and are relatively broad. Almost all examples are bifacially or parti-bifacially formed by soft hammer percussion. ) Chopping tool. As the name implies, these are choppers typically manufactured on large, rounded cobbles. ) Disc. Relatively thin, circular pieces that average around - mm in diameter, though can reach up to  mm in maximum dimension. Typically, one face is extensively worked, while the opposite side has marginal retouch. Edens groups these  tool types into four general categories: Group I) Types -—bifacial and unifacial points Group II) Types -—scrapers 101 Group III) Types -—other light tools Group IV) Types -—heavy duty tools Comparison of the four sites on the group level yielded remarkably similar compositions, with - falling into group I, - into group II, - into group III, and - into group IV. At the level of individual tool types, there is much greater variability within the bifacial forms. Sharorah and al-Mutabthat had similar structures, characterized by high frequencies of type  pieces (), and notably low occurrences for types - (-) and - (-). At Jiladah, type  forms are virtually absent, while type  points account for  of the toolkit, and type  forms, comprising , are relatively frequent. The fourth site, Mundafin has moderate occurrences of type  points (), type  narrow foliates (). Type  broad foliates and type  retouched flakes are considerably higher at Mundafin than any other site considered in the analysis. In interpreting this patterning, Edens points out a number of limiting factors such as the lack of temporal resolution (these are all surface sites on deflated landscapes), the small sample size of the four sites, and the lack of systematic collection, thus creating a certain degree of sampling bias. Regardless, the author suggests regional styles and site function as factors that may contribute to inter-site variability. In addition to typology, there was sufficient debitage from Jiladah and al-Mutabthat to conduct intra-site technological analyses. Approximately half of the blanks are classified as éclats de taille, exhibiting evidence for soft hammer percussion. Of these éclats de taille, some  have faceted and pseudo-faceted platforms, and a slightly larger percentage show platform abrasion. The dimensions of the soft hammer flakes average approximately  mm long,  mm wide, and . mm thick at Jiladah, and  mm long,  mm wide, . mm thick at al-Mutabthat. There are few cores present, showing two general reduction strategies. The first type is simply nodules split in half, with removals from the unprepared fracture plane surface. Other examples are biconical discoids, some with a steeply convex working surface, while others are flat. Among this latter group, the striking platforms are often faceted. 102 Using these data, particularly from Jiladah, which had a high percentage of type  broken bifaces and type  unfinished pieces, the author presents a model for the bifacial reduction sequence: ) hard hammer blocking out, ) soft hammer thinning and shaping, and ) regularization of the edges and flattening of the piece by either soft hammer or pressure flaking. Stage one is considered optional, and is presumably related to the thickness and form of the initial material (thin blanks or plaquettes do not require this primary step). The specimens that show use of hard hammer technique are thick in cross-section and have sinuous edges with deep bulbar scars. The second step of soft hammer thinning is initiated by platform preparation around the margin of the piece. This is achieved by non-invasive, steep flaking, followed by edge abrasion to round and strengthen the platform. In general, soft hammer thinning begins from the tip of the piece and works down the edges, and is subsequently repeated for the opposite face of the preform. The third and final stage of reduction follows a pattern similar to the previous stage, with either an additional round of soft hammer thinning from the tip, or careful platform repreparation followed by pressure flaking (Edens :-). There are many laterally snapped, discarded preforms. In some cases, the final removal from to breakage was an overpassed flake. Charpentier () posits these overpassed flakes belong to an intentional “plunging” technology unique to South Arabia. Given the ubiquity of mistakes such as this in any foliate-producing assemblage, it is more probable that the “plunging” flakes and corresponding laterally-snapped preforms represent common errors in late stage thinning of foliates. In addition to the lithic assemblages, groundstone artifacts were present at all four sites, though detailed descriptions are omitted, as it is beyond the scope of this dissertation. There was faunal material from Jiladah and al-Mutabthat, although its condition is uniformly poor, being extremely fragmented and heavily mineralized; it was only possible to conduct identification on the level of genera. Present at the sites were Gazella sp. and Equus sp. In addition, there were a small amount of Cypraea sp. cowrie shells (ibid.:-). 103 THE FAW WELL SITE While analyzing presumably Neolithic surface collections from the Rub al-‘Khali that had been stored in the Department of Antiquities, Saudi Arabia, Edens () encountered two chipped stone collections that exhibited a technology radically different from the otherwise ubiquitous Arabian Bifacial Tradition. The collections came from an area labeled “The Faw Well Site,” although no other information regarding provenance was provided. The artifacts are made primarily on fine-grained gray-brown flint, with a small amount of yellow quartzite. There were a total of  artifacts, of which  are classified as blades,  flakes, three cores, and  retouched pieces. Of particular interest are the blade attributes, which suggest a prepared core technology. There were two crested blades and six blades struck once the crested blade had been removed, as indicated by a negative blade scar on one side of the dorsal face, and core preparation flakes on the other side of the dorsal face. Nearly a third of the unbroken late stage blades are plunging. Striking platforms are typically small and lipped; the bulbs are shallow and diffuse. Blades struck from the edge of cores are quite frequent, suggesting that the working surface of the cores were relatively narrow. The later stage blades average less than  mm wide, placing them within the category of bladelet. The three cores from the collection are consistent with the blade blanks, though all vary in their morphologies. There is a single platform unidirectional blade core, a flat opposed platform core with faceted striking platforms, and a single platform core with radial preparation around the entire margin of the working surface, from which only one blade had been struck. The toolkit is comprised of four backed/double-backed blades, one endscraper on a blade, one denticulate on a blade, one notch on a blade, one burin on a flake, six retouched blades/flakes, and one bifacial arrowhead. The author argues that the single arrowhead does not belong with the rest of the assemblage, and is intrusive. The Faw Well material provides evidence for a previously unknown bladelet industry in southern Arabia. The author draws tentative correlations with the late Upper Palaeolithic or 104 Epipalaeolithic of the Levant. The double-backed bladelet is a microgravette, which is diagnostic of the late Ahmarian; potentially placing the collection in the late Pleistocene. Hadramaut, Dhofar, and the Nejd AMERICAN FOUNDATION FOR THE STUDY OF MAN ARABIAN EXPEDITION A large, multi-disciplinary American expedition led by Wendell Phillips descended on South Arabia in the early ’s (Phillips ). The team of over  archaeologists, geologists, epigraphers, photographers, and physicians surveyed historic sites in the highlands and along the Tihama coast, and briefly excavated at Marib and Timna. Unfortunately, the expedition to Yemen was curtailed following a disagreement with government officials. To continue the research, Phillips hastily fled to Muscat and secured permission from the Sultan to transfer his project to Oman. Within days, scores of expedition vehicles and equipment were loaded onto a cargo ship and carried from Aden to Salalah. They spent a season excavating al-Bilal on the coastal plain, which served as the primary port for Dhofar in the Iron Age. Phillips’ work at al-Bilal was the first formal archaeological campaign in the Sultanate of Oman, a country which, until that point, had restricted access to Western researchers. His publication made scholars aware of the archaeological potential in this little known corner of the Arabian Peninsula. One member of Phillips’ truncated expedition to Yemen was Gus Van Beek, who returned to the country in  to pick up the thread of prehistoric research begun by Caton-Thompson. Frustrated by the lack of information regarding this time period, he sought to expand the meager body of data by further survey in the Hadramaut. He chose this region for its potential role as a refugium in antiquity, hoping to find a continuous stratified sequence (Van Beek et al. ). Van Beek and his team located a number of surface findspots, surveying the main Hadramaut channel and associated tributaries between Tarim and Henin. The sites, all lithic scatters, were grouped into seven geomorphic settings: ) upper remnants of the Jol; ) the main plateau surface; ) talus slopes; ) benches cut into the limestone along the lower canyon wall; ) flat spurs and outlying remnant terraces; ) low-lying surfaces within the wadi channel; and ) rockshelters. 105 The team observed a series of sites associated with outcrops of high quality chert at the top of buttes on the Jol. At one findspot, estimated to cover an area of roughly , m, the density of chipped stone debris was so great the artifacts formed a solid pavement. Lithic material was systematically collected within a two square meter unit, yielding  pieces. The material is relatively fresh and lightly patinated, though no details were reported regarding typology or technology of these artifacts. The authors noted there was no lithic material on the main surface of the Jol, attributing this absence to the lack of flakeable raw material. They recorded frequent scatters, however, along the edge of the plateau just above the Hadramaut. These findspots are most often located at the head of small tributaries that have slopes gradual enough to allow access into and out of the Hadramaut channel. Raw materials of varying qualities were found close by, in the underlying limestone strata exposed along the access widian. Artifacts were occasionally found on the upper talus slope of the Hadramaut canyon walls. In situations where good-quality raw material was present, there was often abundant workshop debris. A large number of scatters were found on the erosional terraces at the top of the lower limestone cliffs. In certain localities, such as wadi confluences, these terraces become quite wide and form a large, flat surface. In almost every instance where such a bench occurs, artifacts are found on the surface; the few times lithics were not observed can be attributed to the absence of immediately available raw material. Lithic material was frequently collected from the flat, breccia-capped surfaces of spurs protruding into the main wadi channel. Raw material is also abundant in these settings, occurring in nodular form within the breccias. These localities were optimal for hunting, as they commanded an excellent view of the surrounding terrain, as well as providing easy access to raw material. The assemblages collected from these five geomorphic settings above the main channel exhibited a similar reduction sequence, and are described as a single unit (Van Beek et al. : -). The technology is characterized by a Levallois technique in which convexity was achieved through unidirectional and bidirectional flaking. There were also triangular blanks produced from 106 convergent core preparation, as well as discoids. In general, striking platforms are straight, with a low percentage of crude platform faceting. The completed tools are simply described as a variety of scrapers and retouched points. Material that appears more recent was observed on the tops of mounds rising some ten meters above the active wadi channel. These surfaces presumably post-date the higher-elevation localities in the canyon; the associated lithic assemblages must be later. This is corroborated by the minimal or complete absence of patination on these specimens. In general, the material is much smaller (approaching microlithic in proportion), and diagnostic implements such as tanged points and pressure flaked bifacial foliates occur in these assemblages. A second, separate group of artifacts is reported from these low-lying localities. The material is almost completely unpatinated, and is described as a crude, non-descript flake industry. The lithics are commonly found in association with circular stone features on gravel fans. These seemingly ad hoc assemblages were probably produced by more recent Holocene inhabitants of the region. Nearly every rockshelter investigated by the survey lacked evidence for prehistoric human activity. The single exception was a large shelter beneath the lower limestone cliffs near Henin. The rockshelter yielded a small assortment of pressure flaked bifacial pieces that are foliate, pointed, and rhomboid in shape, as well as tanged arrowheads. A few of these tools were manufactured on obsidian; the nearest source of which is in the Yemeni Highlands over  km to the west. THE HARVARD ARCHAEOLOGICAL SURVEY OF OMAN While other countries throughout South Arabia were developing infrastructure and rapidly accumulating vast oil wealth in the ’s and ’s, the Sultanate of Oman lagged behind due to its isolationist tendencies under the rule of Sultan Said bin Tamur. The era of xenophobia came to an end on July  , when Sultan Tamur’s son, Qaboos bin Said, led a bloodless coup against his father, thus heralding a renaissance that catapulted the once backward Sultanate into one of the most developed, wealthy, educated, and peaceful countries in the Middle East. Oman’s immediate 107 acceptance and encouragement of outside scholars opened the floodgates of scientific research. Since , much of what is known about South Arabia comes from data collected in Oman. Among the first archaeological expeditions to Oman was the Harvard Archaeological Survey, conducted by J. Pullar in the early ’s (Pullar ; Pullar and Jäckli ; Pullar ). With logistical aid from Petroleum Development Oman (PDO), the Harvard team was able to carry out reconnaissance survey throughout much of the interior of Oman; an area that had not been traversed by Westerners since the desert explorers of the early th century. Some  aceramic sites were recorded from the hinterland of Oman; the only evidence of prehistoric coastal occupation was at the cape of Ras al-Hadd. The inland sites are all surficial lithic scatters located along major watercourses that drain from the Dhofar escarpment in the south and Jebal al-Akhdar in the north. In most cases, it is difficult to determine if the findspots are mixed assemblages, due to their position on deflated landscapes and lack of diagnostic implements. There was a small handful of sites that garnered special attention in the literature. Several collections were made at the site of Fahud, which is a small limestone syncline southwest of Jebel al-Akhdar along the fringe of the Rub al-‘Khali basin. The lithics, primarily made on flint, are spread across a gravel surface, covering some , m. Pullar (:) speculated there were two separate archaeological phases represented at the site; the first was characterized by pressure flaked arrowheads with small barbs and tangs. Based on parallels with the “Beduin Microlithic” near Kharga, she attributed this material to the late Upper Palaeolithic. The second period identified at Fahud consisted of bifacial foliate fragments and numerous sidescrapers. There was a high ratio of debitage to completed tools, leading Pullar to conclude this locale served as a workshop site. A collection of similar artifacts, also attributed to the late Upper Palaeolithic, were recovered near Natih, another syncline less than  km northeast of Fahud. Within this collection were trihedral rods, similar to those reported by Caton-Thompson at Mukalla, along the southern coast of Yemen. The Harvard Archaeological Survey was the first to explore the Nejd Plateau in southern Oman, observing an incredibly dense amount of lithic material throughout the landscape. This is probably due to its unique microclimate that facilitated its role as a refugium during arid phases, as 108 well as the high-quality chert that is ubiquitous all over the region. One of the richest areas was the well of Bir Khasfa in the Wadi Arah, where heavy brush within the wadi bed suggests a high water table and possible ancient lake (a test-pit excavated during the COPR  season confirmed the presence of ancient lacustrine sediments). Two findspots from Bir Khasfa were given special attention by the surveyors—JC- and JC-. JC- covered an area  m long and  m wide. The artifacts consisted almost entirely of elongated blanks manufactured on very fine-grained opaque brown chert. Pullar classified these blanks as “flake-blades” instead of true blades, because, though they were blade-proportioned, the dorsal scar patterns were predominantly convergent rather than parallel. Striking platforms are not faceted, and there were no tools manufactured from these types of blanks. The densest findspot at Bir Khasfa was JC-, scattered across approximately , m. Six stone circles, measuring three meters in diameter, were found in this area. The collection of artifacts appears to be homogenous, including: debitage (), blades (), bifacial foliates (), scrapers (), points (.), retouched flakes (), and knives (). The bifacial tools often have lenticular cross-sections, and are ovoid, cordiform, and lanceolate in shape. The author draws parallels with tools from the Aterian levels at Kharga Oasis, and with D-Group material from Qatar (Pullar : ). The bifaces are made on fine-grained chert, and are formed by flat, invasive soft hammer percussion. There were also retouched points manufactured on unifacial blanks that have noninvasive semi-steep convergent retouch. The pressure flaked arrowheads common throughout South Arabia were found as well, exhibiting tanged and barbed forms. The Harvard team discovered Stone Age materials at Shisur, which is one of only two permanent natural water sources in the Nejd. The artifacts differ from those collected at Bir Khasfa, consisting of flake and blade cores with a high percentage of large primary flakes and “flake-blades,” often with faceted platforms. Like site JC-, there were very few completed tools, although Pullar notes several large oval-shaped blanks with flake scars on both surfaces, presumably early stage bifacial preforms. This disparity between Shisur and Bir Khasfa may simply be a factor of raw material availability—the COPR  survey observed plaquettes and small chunks of shattered tabular chert 109 scattered across the surface of Bir Khasfa, which are not particularly conducive to core reduction because they are so thin. In considering the range of collections in Oman, Pullar recognizes a number of similarities to Northeast and East Africa. Specifically, she points out parallels with the Aterian from Kharga Oasis, although Oman lacks the microlithic material, and there is “an extraordinary dearth of cores” (Pullar :). There are superficial similarities to Levantine Neolithic sites such as Munhata and Beidha, though specific diagnostic forms are missing. Like a few of her Africanist predecessors who had examined South Arabian lithic collections, analogies are made with the “Somaliland Doian,” although hollow based arrowheads are absent (ibid.) The findings of the Harvard Archaeological Survey were corroborated by Smith () and Villiers-Petocz (), who examined various collections stored at the Department of Antiquities in Muscat. In most cases, the findspots were located and collected in the ’s by geologists and petroleum engineers working for PDO. Like the sites documented by the Harvard Archaeological Survey, the findspots are all located in or near interior-draining widian, and in proximity to finegrained raw material. Few conclusions can be reached based on these collections, as they are all surface sites and may be mixed. Because the artifacts were collected by avocational archaeologists, there is a bias toward completed bifacial tools. The variety of forms described by Pullar are all represented, including bifacial foliates, ovates, cordiforms, lanceolates, picks, as well as pressure flaked arrowheads and trihedral rods. Villiers-Petocz (:) concludes from her study of the technology and typology exhibited in these collections that “a substantial temporal span…from the Lower-Middle Palaeolithic to the post-Palaeolithic” is represented, which is contrary to the currently accepted thinking that classifies all the bifacial material as Holocene. AMIRKHANOV’S WORK IN SOUTH YEMEN Following the discovery of oil, a flood of new research was carried out in the Kingdom of Saudi Arabia and the Gulf Countries, where petroleum was most prevalent. It is ironic that Yemen, which was once the heart of Arabia Felix was left behind during this phase of expansion. The most 110 fertile corner of the Arabian Peninsula was barren of oil, and thus did not enjoy the same wealth as its neighbors. Furthermore, Yemen became a Cold War battleground beginning in the late ’s, following a socialist uprising against British colonial rule in the south. The country was split in two, North Yemen was a democracy and South Yemen became the first and only Marxist Islamic state. Yemen’s schism launched a protracted civil war that lasted from  to . Compounded by numerous border disputes with Saudi Arabia and Oman, the strife kept most scholars away from the troubled region. Yemen was not completely cut off from scientific exploration, however. There was a Soviet project during the ’s, led by Hizri Amirkhanov, that located and excavated prehistoric materials throughout South Yemen. The expedition surveyed several areas of the country, including upland localities, drainage systems, and the coast (Amirkhanov , , , ). Amirkhanov reported both deflated surface scatters, as well as stratified deposits. In situ materials were recorded from three cave sites and six preserved open-air deposits. Most of these openair sites are single occupation, though two distinct archaeological phases were recognized at Meshed / and as-Safa, both situated along the middle course of the Wadi Du’an. The three caves were near the border of Oman. A sequence of archaeological layers was excavated at al-Guza Cave. The collections from these deposits are presented as a homogenous assemblage, described as a pebble industry with choppers, well developed, large single-platform cores, and a low frequency of sidescrapers, “awl-like” implements, and discoids. Bifacial tools are notably absent. Amirkhanov classified the al-Guza Cave material as “Oldowan or at least pre-Acheulean” (Amirkhanov :). Acheulean material was reported from four buried open-air sites and eighteen surface scatters from the flanks of the Yemeni highlands, as well as on terraces in the Wadi Du’an. The technological and typological characteristics of these assemblages include an abundance of handaxes that exhibit a variety of forms, choppers, single-platform cores with sub-parallel removals, blade blanks, and the absence of cleavers (contrasting with the findings of the Comprehensive Survey, which reported Acheulean findspots with cleavers). 111 Amirkhanov recorded eight Middle Palaeolithic surface scatters, classified as such based on the presence of Levallois technology “in a broad sense” (ibid.:). All of these findspots were found in the western Hadramaut drainage system and related tributaries, primarily the Wadi Du’an. Typologically, these Middle Palaeolithic assemblages included denticulates, sidescrapers (predominantly transverse), and “the occurrence of Upper Palaeolithic forms” (ibid.), though no further details are provided. The Soviet expedition recognized Upper Palaeolithic artifacts at four stratified and nineteen surface sites, also in the western Hadramaut. The stratified sites were not pristine; they were secondary deposits of artifacts within alluvial aggradation. The predominant technology at these sites is the use of flat cores with parallel removals along the working surface. Although they are similar to specimens from earlier archaeological phases, Amirkhanov argued these cores are “very standardized metrically compared with their Middle Palaeolithic counterparts” (ibid.). These “Upper Palaeolithic” toolkits are comprised of endscrapers, points, awls, and knives. The team obtained a sample of datable material from an archaeologically sterile profile that was tentatively correlated with the basal layer of one of their Upper Palaeolithic sites. The material yielded a radiocarbon date of , ± , , thus establishing a terminus ante quem for these collections. There were  Neolithic sites discovered by Amirkhanov’s team, most found within Mahra Province. Most of these findspots are from the escarpment that rises abruptly above the coastal plain, and in valleys that drain inward toward the Rub al-‘Khali basin. This region is essentially the western extent of the Nejd Plateau in Oman, separated by an arbitrary political border. Two well-defined buried sequences were excavated at Khbek and Msabig Caves in this region, and a particularly deep open-air sequence (> m) was excavated at Habarut . At least ten geologic layers were recognized at Habarut , the earliest of which were soils that yielded radiocarbon dates of , ±  and , ±   (ibid.). The lithic collections from these presumably in situ deposits are characterized by bifacial tools manufactured on flakes. The blanks were removed from simple single-platform unidirectional cores with unfaceted platforms. The unifacial tools are comprised of endscrapers, micro-endscrapers, 112 sidescrapers, knives, and awls. The bifacial component consists of trihedral rods and plano-convex naviforms (sensu Rose ), as well as large bifacially worked cutting tools (axes). Unlike the Arabian Bifacial Tradition recognized throughout the interior of South Arabia, tanged and barbed projectile points are absent from the Mahra assemblages. The excavators did not recognize any significant change in technology or typology through this Neolithic sequence (ibid.). CNRS MISSION TO HADRAMAUT In  and , M.L. Inizan of the Centre Nationale de Recherche Scientifique, France, conducted two prehistoric surveys around the region of Shabwa in western Hadramaut (Inizan and Ortlieb ). This region was chosen due to its potentially important geographic location, at the western edge of the Jol, where it abuts the Ramlat as-Sabatayn desert. Four geomorphic zones are encompassed within this small region: ) the top of the Jol, ) the piedmont at the base of the escarpment, ) the widian that drain from the plateau into the Ramlat as-Sabatayn depression, ) and ancient lacustrine deposits within interdunal zones in the Ramlat as-Sabatayn. At least two findspots with Middle Palaeolithic technological and typological characteristics were collected from the plateau just above Wadi Muqqah. The artifacts exhibited a heavy brownishblack patina, and were exclusively manufactured from chert. Three types of cores were observed within the assemblage, including Levallois flake and point cores, flake cores, and blade cores. The Levallois cores exhibit two different techniques for convexity maintenance: centripetal and bidirectional preparation (ibid.:-,fig.-). In one case, the bidirectional point cores are quite steep at the distal end which is modified by a non-invasive supplementary platform (ibid.:,fig. :), reminiscent of Nubian Type I cores (Marks, personal communication). Among the debitage, there were a small handful of Levallois flakes. Most have straight platforms, with only one example of a dihedral platform. There were also thick blades, some with cortical backs, suggesting they were removed from the edge of the working surface. The authors report a few examples of technologically diagnostic waste—there were occasional lames à crêtes identified at the site. 113 Inizan and Ortlieb () raise two points based on this qualitative analysis of the debitage: ) there is evidence for Levallois flake and point production exhibited by the cores, but actual Levallois points are absent and flakes are rare; and ) there are some laminar elements, indicated by the presence of blade cores and blade-proportionate debitage, however, in this regional context, laminar technologies do not appear to form part of the lithic tradition. Based on the definition of lineal and recurrent Levallois methods put forth by Boëda (), the authors interpret the laminar elements observed at Wadi Muqqah to be the product of a recurrent Levallois technique, and not “true” blades, since no prismatic blade technologies have been reported in southern Arabia. Tools are low in frequency and dominated by bifacial pieces. They are all asymmetrical and reduced by hard hammer percussion. The dimensions are varied, but are typically small limaces. Scrapers and denticulates are rare; some are manufactured at the end of robust blades, while a small percentage is made on éclats de taille. Regarding chronological attribution, the authors place the Wadi Muqqah industry in the late Lower or Middle Palaeolithic, based on similarities with assemblages from Europe and the Near East. They point out, however, that this tradition of bifacial tools associated with Levallois core reduction spans hundreds of thousands of years, and, thus cannot be used to establish a more precise time frame. Another observation made by Inizan and Ortlieb is that there is an absence of a true blade technology anywhere in Arabia. According to the model of European or Middle Eastern technological development, Middle Palaeolithic industries are almost always succeeded by Upper Palaeolithic blade industries. Why, they ask, is this not the case in Arabia? Their answer—that Upper Palaeolithic occupation in Arabia may not be represented by laminar technology—addresses the heart of this dissertation. The authors remain open to the possibility that lithic development in South Arabia does not adhere to the same lineage as that seen in Europe and the Near East; rather, it may follow the developmental tradition from a different area of the world, namely East Africa (ibid.:). 114 The CNRS expedition to Yemen also collected assemblages of artifacts manufactured on quartzite at Tuheifat,  km north of Shabwa, and at the entrance to Wadi Thib. The assemblages include large flakes with some blades and coarse bifaces. Finally, there were a series of sites with pressure flaked Neolithic material reminiscent of the Arabian Bifacial Tradition from the Rub al‘Khali. The tools are manufactured on chert, and exhibit the classic pedunculate and foliate forms typical of this industry. One findspot, Site E, yielded several trihedral rods and an extremely wellmade point, possibly dating to a later phase of the Neolithic. WHALEN’S PLIO-PLEISTOCENE RESEARCH IN DHOFAR In , Whalen carried out a survey in the Dhofar Governorate of southwestern Oman, continuing to search for evidence of Plio-Pleistocene hominid occupation in southern Arabia. The research focused on terraces and escarpments at the southern portion of the Nejd Plateau, near the town of Thumrait. Palaeolithic sites were rare in the western portion of this region, probably because of the high, precipitous canyon walls that hindered access to the widian below. As one travels eastward, in drainages where the hills above the channel were lower and gently sloping, sites became much more prevalent. Sites were most common along tributaries emptying into the primary drainages. Every site encountered during the  survey was multicomponent, representing two or more phases of occupation (Whalen et al. ). The artifacts ranged from Early Acheulean to Neolithic; identification was based upon the extent of weathering, degree of patination, and presence of hard hammer percussion. The oldest artifacts had a heavy patina of black desert varnish, which became progressively lighter on more recent samples. Neolithic tools were completely lacking in patina. The artifacts at all sites were exclusively manufactured from chert derived from Rus Formation exposures that are ubiquitous throughout the Nejd Plateau. At the  sites that had Early Acheulean material, choppers were the most prevalent category, constituting  of all artifacts, with only  handaxes represented. There were also low percentages of scrapers, polyhedrons, picks, 115 spheroids, cleavers, knives, burins, notches, and awls. Cores and flakes comprised  and  respectively, though there is no data regarding the technique(s) of core reduction. Middle Acheulean sites had a somewhat different distribution of artifacts. Bifaces were much more prevalent at these findspots, accounting for nearly  of the total collection, while the frequency of choppers dropped to . Unfortunately, there is no description regarding the morphology of these bifaces or technology used to produce the implements. Among the other miscellaneous tools, there is a similar distribution of frequencies (although awls are absent), as well as for the flakes and cores. Again, there is no information given as to the specific methods of core reduction employed during this archaeological phase (ibid). THE TRANSARABIA EXPEDITION TO THE NEJD PLATEAU When Bertram Thomas crossed the Rub al-‘Khali in , his Beduin guides pointed out a camel track that supposedly led to the legendary city of Ubar. Half a century later, scholar Nicholas Clapp stumbled across this note in Thomas’ travel account and became determined to locate the mythical Ubar. Clapp organized the Transarabia Expedition, which was carried out between  (Clapp ). The project represented the first comprehensive investigation of the Nejd Plateau, during which a plethora of lithic findspots were recorded. Aceramic lithic sites recorded by the Transarabia Expedition span multiple phases, although all appear Holocene. The purportedly earliest sites are characterized by unifacial pedunculates manufactured on blades [Fasad Points]. Thirteen occurrences are classified as such; architectural remains are conspicuously absent from these findspots, though fire-cracked rock, shell, and bone fragments are present at some of the sites (Zarins ). Subsequent to the “Fasad Point” phase are assemblages correlated with the “D-group” phase of the Neolithic. These collections are defined by bifacially worked, pressure flaked tanged and barbed points. Again, there is no architecture associated with these localities. The most noteworthy site was Ibn Hamuda, found on the fringe of the Rub al-‘Khali, not far from the Yemen/Oman border. It was located on a small wadi channel and contained several large grinders, slabs, mortars, pounders, 116 and limestone vessels. The assemblage was characterized by pressure flaked lanceolates, rhomboids, trihedrals, and pedunculates. There were also typical endscrapers on blades (ibid.). A final Neolithic phase on the Nejd Plateau is defined by the presence of trihedral rods. While this point type is common throughout the Nejd, they are rare outside this zone. Like the previous two phases, these sites all lack structural remains. The trihedral rod sites are bracketed within the th and th millennia , based on correlations with artifacts from the middle and uppermost strata at Habarut  (Amirkhanov ; Zarins ). Arabian Gulf, Omani Littoral, and Wahiba Sands DANISH ARCHAEOLOGICAL EXPEDITION TO QATAR One of the most influential projects in South Arabian archaeology was conducted by a Danish expedition from  to , under the auspices of the Forhistorisk Museum in Aarhus. The team, headed by P. Glob, T. Bibby, and H. Kapel, carried out survey and excavation throughout the Sheikhdom of Qatar, recording over  prehistoric sites. They identified four Holocene industries, which were placed in a relative chronological sequence based on seriation of the lithic industries. Although many of their conclusions have “proved worthless with the end of the s” (Tosi : ), elements of their nomenclature have remained prevalent within the Holocene chronological sequence (Kapel ). There have been recent revisions to terminology and sub-groups proposed by Inizan (, ) and Potts (, , ), based on more recent data from reliable stratigraphic contexts, as well as associated chronometric dates. The Danish sequence has been both useful and problematic because it is based on typological seriation of artifacts rather than on absolute dates from the sites. Even at the time, the original authors recognized the difficulties posed by their methods: …the construction of an absolute certain sequence for the various Stone-Age cultures meets with considerable difficulties. At no site have we been able to ascertain any trace of a trustworthy stratigraphy. An estimation of the chronological sequence of the cultures based soley on typology runs, as we have said, a risk of subjectivity and misinterpretation. (Kapel :) 117 The Danish team classified their sites into four groups: A, B, C, and D. They purposely avoided using terminology with temporal connotations because, at the time, so little was known of this period in South Arabia. Nearly every artifact found in Qatar is made on flint, which is plentiful throughout the country and often of high quality, with occasional specimens made on fine-grained quartz and quartzite. In addition to the A – D Groups, markedly different material was collected from the upper limestone cliffs, described as “large, unhandy, primitive implements resembling hand-axes” (ibid.:). Gebel () conducted some brief follow-up work on this material, which was collected near Hili on an anticline of the Omani Mountains. He described the material as a heavyduty tool assemblage with sidescrapers and cleavers, and concluded it is probably Middle Palaeolithic. Almost every findspot of the A – D Groups were situated along the coastal plain, which is critical for dating, as interbedded formations of calcareous sandstone were observed at the base of the escarpment along the eastern coast of Qatar. Geologists interpreted these features as raised offshore bars deposited during the last prolonged phase of marine transgression. Radiocarbon dates indicate the formations are greater than , , consequently providing a terminus ante quem for all A – D Group sites. One radiocarbon date was obtained from a small B-Group site at the base of a limestone cliff on the coastal plain. There were small patches of charcoal within the sand, indicating the presence of ancient hearths dug along the beach. The sediments yielded a radiocarbon date of , ±   (ibid.). Based on surface scatters that are assumed to be homogenous and pristine, descriptions of the A – D industries are as follows: A-Group.  sites were placed into this category. These findspots most frequently occur on the rocky plateaus or directly at the base of the limestone cliffs that rise above the coastal plain, in proximity to flint outcrops. Most of the sites appear to be workshops, with a high percentage of cortical flakes and large blanks; completed implements are rare. The authors describe “hand-axes or blanks for hand-axes” as the dominant tool, which are often leaf-shaped and exhibit percussion flaking. Because of their location at the highest points above the coastal plain and absence of pressure 118 flaked technology, researchers attributed these findspots to the earliest industry (“primitive” upland sites notwithstanding) in the sequence. B-Group. While numbering only eight sites, the B-Group is by far the most distinct of the Qatar sequence. These sites are characterized by the production of well-made blades used to manufacture pressure flaked arrowheads and unifacial pedunculates. In terms of both technology and typology, this industry closely resembles the findspots with finely-made projectile points that are ubiquitous throughout South Arabia. The majority of B-Group sites are found where low cliffs lie close to the present day coastline. As previously mentioned, a single radiocarbon date places this industry at the end of the th millennium , corroborating Field’s (b) date from the Rub al‘Khali (both measurements are uncalibrated). C-Group. The C-Group industry is comprised of  sites. The authors speculate it is derived from the A-Group, based on the presence of bifacial implements that resemble artifacts of the earlier industry. The dominant tools are finely-shaped scrapers and awls, as well as knapped flint spheroids that were interpreted as ammunition for slings or bolas. Though less frequent, there were crudelyknapped tanged arrowheads with short barbs. This “scraper culture” is often found on rocky surfaces, crags, and hillsides approximately - masl, and generally close to the present day coastline. D-Group.  sites were found attributed to this industry, characterized by tanged and barbed arrowheads, and other pressure flaked tools of various forms. Some of these armatures are described as “tiny masterpieces of technique, which bear comparison in shape and in technical refinement with the finest flint implements known from anywhere else in the world” (Kapel :). In addition to pressure flaked arrowheads, there are larger implements described as triangular and sub-triangular percussion-flaked adzes and axes. These pieces have plano-convex cross-sections, suggesting that they were reduced from large blanks. The authors also describe scraper-like implements fashioned on flat flint “tiles” that range between four and six millimeters in thickness. These tools are shaped by semisteep unifacial and/or bifacial reduction. The larger D-Group sites occur on flat, sandy areas near the coastline, though there are often isolated finds of tanged-arrowheads from this industry throughout the country. 119 The Danish sequence was refined in the late ’s by a French expedition that returned to Qatar to investigate a particularly rich D-Group site recorded by the Danish team near the fishing village of Khor, along the coast of southwestern Qatar, which had potentially stratified material (Inizan , ). The findspot was initially recognized by Kapel due to its dense surficial bifacial assemblage, situated on a raised hill on the margin of a coastal sabkha. The first season of excavation, conducted between  and , excavated a test-pit within the hillock, revealing three distinct, stratified layers that yielded lithics, pottery, burnt stone, shell, and bone. The lithic assemblage was comprised of éclats de taille, percussion-flaked bifacial fragments, a pressure flaked arrowhead fragment, sidescrapers, notches, denticulates, and a “pic.” These artifacts were made on chert derived almost exclusively from exposures that occur in proximity to the site, often in the form of large blocks and amorphous nodules. In addition, there is a small percentage of artifacts produced from local fine-grained limestone. The shells are often from edible species (Turbo sp., Pinctada sp., Gafrarium calipygum, Asaphis deflorata, and Cerithium); none are worked. There were a few shark vertebrae found within the sequence, and the only evidence for terrestrial fauna was a single equid molar. A hearth was excavated as well, which was about  cm in diameter and consisted of a hollowed out basin filled with ashy sediment. Two samples were submitted for radiocarbon dating: ) sediments from within the hearth yielded a measurement of , ±  , and ) a marine shell measured , ±   (Inizan ). The French team returned to Khor in  and  to further investigate the stratified deposits, obtain more dates, and increase the size of the collection. They opened two wider excavation areas at the top of the hill and at the base. An assortment of lithics, pottery, shell, and fish bone similar to the previous season was unearthed, as well as a burial consisting of cremated remains in secondary position found at the top of the hill. A third absolute date, obtained from shell within the deposit at the bottom of the hill, yielded a measurement of , ±  . The authors expanded the lithic assemblage to over , pieces, analysis of which indicates the site may have served as a secondary workshop for the manufacture of bifacial implements. While there was a high frequency of éclats de taille, very few cortical flakes appear in the collection. 120 The unifacial tools are predominantly notches and denticulates. Bifacial tools are percussionflaked, triangular and sub-triangular in shape, plano-convex in cross-section, and primarily reduced from large blanks. One specimen found just outside the burial was a rhomboidal, pressure flaked arrowhead. Associated with the lithic assemblage were sherds of painted pottery, many of which are diagnostic of the Ubaid period in southern Mesopotamia (Inizan ). ROYAL GEOGRAPHICAL SOCIETY’S OMAN WAHIBA SANDS PROJECT In the late ’s, an interdisciplinary expedition was launched in the Wahiba Sands, a desert comprised of longitudinal dunes in east-central Oman (Edens a). The survey focused on three geomorphic zones: a strip of coastline on the eastern border of the dune field, a transect between two massive dunes in central Wahiba Sands, and portions of the Wadi Andam-Halfayn drainage system just west of the Sands. Because this was the first archaeological work conducted in the region, the team carried out exploratory rather than systematic survey, in order to gather basic data regarding the types of human occupation in the Sands. Along the coastal zone and within the Sands, lithic material occurred on the surfaces of inselbergs and their corresponding scree slopes, on flat interdunal exposures, atop the surface of stabilized dunes, and on sabkha surfaces. The findspots begin approximately half a kilometer from the present shoreline and continue into the desert. The author divides the findspots into low-density clusters of fewer than ten artifacts, typically comprised of a point plus debitage, and higher density clusters with several points, other tool forms, and associated debitage. The artifacts are made from fine-grained, red radiolarian chert cobbles of the Hawasina group that outcrops along the margins of the Sands. Thus, raw material from within the desert must have come from localities up to  km away. At both types of sites, points are a distinctive typological component and were divided into two general groups. The first type is a unifacial tool manufactured on a wide blade with naturally pointed distal end formed by a convergent dorsal scar pattern, which is thin and triangular in cross-section (though in some cases there is marginal retouch to shape/resharpen the tip). These 121 implements are characterized by a proximal tang formed by pressure flaking. Although pointed and having blade dimensions, debitage analysis suggests they are not the product of a true blade technology or prepared core technique, but rather were selected products of unspecialized flake production. Edens names these distinct, homogenous tools Fasad points, based on the locality in Dhofar where they were first discovered. The second group of points collected from the coastline of the Wahiba Sands is the ubiquitous pressure flaked bifacial arrowhead, so common throughout southern Arabia. Like the specimens documented in the Rub al-‘Khali, this type is fairly heterogeneous, and consists most prominently of the stemmed variety. They are usually partially-bifacial, made on flakes, the dorsal surfaces of which are extensively shaped by pressure flaking, while the ventral surfaces have minimal retouch at the proximal end to form the tang, converging at the distal end to shape the point. The completed tools are typically triangular in shape, with a wide tapering stem, and are plano-convex in cross-section. There are also examples of fully bifacial types, lanceolate-shaped points, and narrow foliates, each sub-type exhibiting the same technology in their manufacture. There were a small number of other tool types found in association with the Fasad and bifacial points, including sidescrapers and endscrapers on thin flakes, notches, drills, geometric microliths, and pièces esquillées. There was minimal evidence to reconstruct reduction sequences; cores are rare among the assemblages. In most instances, they are multiple platform informal flake cores. Flakes are frequently ad hoc pieces detached from chert chunks, and are small, ranging in length from two to four centimeters. Bulbar characteristics show evidence for either hard hammer percussion or pressure flaking; there is minimal indication of soft hammer technique. The author notes three localities with lithic assemblages exhibiting different technological and typological characteristics from the material described above. These collections were focused on the production of bifaces “in the Palaeolithic sense” (Edens a:). The specimens are flat, biconvex foliates with invasive retouch; they are all broken so it is difficult to determine shape, although they are asymmetrical and have irregular edges. The associated debitage are generally éclats de taille showing different types of flaking— were hard hammer,  soft hammer, and  with 122 “possible” pressure flaking. In addition to this different material, there were a few examples of small unifacial, partially-bifacial, and bifacial stemmed points. The artifacts from these three anomalous findspots are made on the same red radiolarian chert as all other Wahiba Sands sites. The scatters are situated on water-laid sediments, presumably coeval with the other artifact-bearing surfaces throughout the coastline and desert. These observations, in conjunction with the presence of a tanged point at one of the sites, led the author to posit these localities are not Palaeolithic, but contemporary with the later material, merely representing a functional variation within the larger industry already described. Edens concludes that the localities are intermediate in both the technological stage of reduction and the geographical distance between raw material source and the final area of use (ibid.). The lithic sites located just west of the Wahiba Sands, in the Wadi Halfayn-Andam drainage system, display typological differences from those sites on the coastline and in the Sands. The findspots in this region were primarily found on “raised palaeochannels” (see Maizels ) or on coeval gravel spreads that mantle the aeolianite terraces. Sites were also found, to a lesser degree, on or around aeolianite remnants and on landscapes covered by siliceous conglomerates. A major lithological element of the aforementioned gravels is chert, which has a relatively large average maximum dimension in the northern portion of the drainage system. There is no other locally available chert source, making these channels the only outcrop with raw material suitable for knapping. The palaeochannels form dense linear chert sources characterized by good quality raw material found in large blocks. Most of the knapped material on these surfaces is workshop debris dominated by cores, unretouched flakes, and other debitage. The cores are classified as either singleplatform unidirectional or bidirectional opposed platform. A good portion of striking platforms are cortical or plain, due to the blocky nature of the raw material. The material is almost exclusively reduced by hard hammer percussion, with minimal core maintenance and shaping. Because of the general lack of preparation, there is a high frequency of knapping errors, indicated by distal terminations that show  hinging and  overpassed. Cores are generally used to detach a small 123 number of flakes and blades, and are then discarded. The author observes that clast size diminishes significantly as one travels downstream of the raised palaeochannels. The smaller nodules seem to affect methods of core reduction; the southerly assemblages exhibit more careful control of striking platforms, preparation of working surface, and fewer knapping errors (ibid.). While low in frequency, the Wadi Halfayn-Andam assemblages have a bifacial component alongside the core technology. These pieces are thick, typically ovoid or irregular in shape, asymmetrical, with sinuous or even jagged edges. Unifacial tools consist of scrapers, flakes with marginal retouch, and burins on thick flakes, which are alternately classified as cores on flakes (ibid.). The author describes a group of collections found on the banks of Wadi Andam and on gravel spreads away from the wadi system with different techno-typological features. The raw material at these localities is smaller than the gravels from the raised palaeochannels, resulting in more diminutive core, flake, and tool sizes. Similar to the cores noted at sites upstream, there is no platform preparation and little control of the working surfaces. Tool forms at these localities exhibit a wider range of variability, including several different types of scrapers, notches, drills, truncated flakes, retouched pieces, and chopping tools. These collections were often associated with additional features, such as circular stone installations and stone piles, as well as some shell objects. Despite the additional characteristics, Edens argues that these sites are coeval with the other localities within the Wadi Halfayn-Andam drainage system based on “the same general industrial tradition” (Edens a: ). He posits that the assemblages are roughly coeval with the industry reported within the Wahiba Sands, all belonging to the Arabian Bifacial Tradition. SAIWAN During a field excursion investigating sources of bitumen, the Italian Archaeological Expedition to Oman stumbled upon a Palaeolithic findspot called Saiwan, located the junction marking the northern extent of the Haushi-Huqf Depression, southern end of the ad-Dakhliyah alluvial plain, and southwestern corner of the Wahiba Desert. The site is approximately six kilometers from the shores of a relict quaternary lake basin (Biagi ). Saiwan was initially recognized by 124 a dense concentration of artifacts lying on the surface, which stood out in sharp contrast to the surrounding landscape. Approximately  artifacts were collected, all manufactured from a fine-grained, pale yellowish-brown chert. The patina is thick on most pieces, and one or both surfaces is partially abraded by aeolian activity. Nearly two-thirds of the collection is comprised of debitage, predominantly flakes and elongated “blade-like” flakes with wide, high-angled, plain striking platforms that indicate hard hammer percussion. Half the blanks exhibit some degree of cortex, suggesting primary reduction was carried out on site. The cores consist of both discoids and single-platform, unidirectional blade cores. The discoids are thick and biconical with centripetal blows around the entire perimeter. Some specimens have marginal retouch along part of the edge, suggesting they may have served as tools as well as cores. The blade cores have striking platforms that range from plain to oblique, and primarily show parallel scar patterns on the working surface(s). Among the tools there were eleven bifaces, reduced from either chert nodules or large flakes. They are typically flat, biconvex, and sinuous in profile, with arêtes that suggest a mix of both soft and hard hammer reduction. In some cases, the bifaces exhibit minor edge-grinding for platform preparation. These tools fall within band IV of Borde’s handaxe diagram, described as ovals-discoidslimandes (Debenath and Dibble ). Two unique specimens were lagéniforme handaxes with rounded butts and pointed distal end. Scrapers are also prevalent within the toolkit. Retouch is often marginal, inverse, and located on the lateral edge. The author argues that the collection is homogenous, based on technological characteristics, patina, and degree of aeolian abrasion. Precise dating is impossible because the material was found exclusively on the surface, and there are no analogous assemblages from southern Arabia (at least none systematically collected and analyzed in this detail). Based on the bifaces that fall within Borde’s band IV, together with a predominance of sidescrapers, Saiwan was tentatively attributed to the Upper Acheulean (Biagi ). 125 LITHIC OCCURRENCES IN ABU DHABI McBrearty (, ) collected chipped stone artifacts from four localities in Abu Dhabi, along the Trucial Coast of the Arabian Peninsula. In each case, the lithic material was found covering the surface of Baynunah Formation exposures. The Baynunah Formation is a late Miocene sedimentary bed capped by a thick layer of tabular chert. The formation is horizontally oriented, often capping the jibal throughout the western region of Abu Dhabi. Isolated artifacts were found on many of these jibal, though the largest numbers were collected at Shuwaihat, Jebel Barakah, Hamra, and Ras al Aysh. Initial collections were uncontrolled, in order to obtain a sample representative of the range of material present, followed by a small systematic collection to determine artifact density and potential distribution patterns. Shuwaihat. The artifacts in proximity to Shuwaihat were found scattered over an area of approximately , m, positioned on the sea cliffs and wave-cut platforms at about  masl. The predominant raw material used in stone tool manufacture is a low-quality silicified limestone. The assemblage from Shuwaihat consisted of cores and debitage; there were no formal retouched tools. Of the  cores, seven were classified as radial, including discoidal, subradial, and high-backed types. While the raw material is not conducive to blade manufacture, two of the cores are bidirectional blade cores and one is a crescent-shaped blade core. These types are identical to the “naviform” blade cores known from the Levantine PPNB. They exhibit radial preparation around their circumference, with blades subsequently removed from an axis perpendicular to the radial striking platform. Blades detached from these cores have either a dorsal crest, which is the remnant of the radial preparation, or parallel scars from later stage blade removals. McBrearty (:) argues that the primitive appearance of the Shuwaihat assemblage is a result of the poor quality of the raw material. Hamra. Chipped stone material from Hamra was found strewn about a large jebel over an area of approximately , m. A thick tabular chert outcrop occurs at the summit, where heatshattered, high-quality yellowish flint with a fairly heavy patina is abundant. 126 The author observes that the Hamra assemblage is similar to that reported from Shuwaihat, though the artifacts are manufactured on a much higher quality raw material. Again, there were no formal tools within the collection. Cores include unidirectional and bidirectional prismatic blade cores, as well single and multiple-platform flake cores. Of the  cores from the collection,  show a radial flaking pattern. Over half of these radially prepared cores exhibit blade removals from a platform perpendicular to the radial flaking axis, mirroring some of the naviform specimens from Shuwaihat. The debitage consists predominantly of flakes with simple or radial dorsal scar patterns. Ras al Aysh. Lithic artifacts were collected over an area of , m at Ras al Aysh, both on the surface of the jebel slopes and within a thin subsurface veneer of outwash sediments. The artifacts are manufactured on a fine-grained chert that outcrops at the summit of the jebel. Like the other two collections, formal tools are absent at Ras al Aysh. The cores include unidirectional, bidirectional, and radial types. Although none of the cores exhibit blade removals perpendicular to the radial working surface (like those recorded at Hamra and Shuwaihat), there was a ridge blade among the debitage. The blanks show simple, radial, parallel, and bidirectional dorsal scar patterns. There are both cortical flakes and exhausted multidirectional cores, indicating that early and late stage core reduction was carried out at Ras al Aysh. Jebel Barakah. Chipped stone material was collected on the flat bluffs on the side of Jebel Barakah, covering an area of , m. The artifacts were scattered on the lower talus of the hill, extending from the sea cliffs to just below the summit. The raw material is a fair-quality flint derived from the Baynunah Formation bed that caps the summit of the jebel. The collection exhibits a consistent homogenous technology, characterized by radial cores () and flakes with radial scar patterns (.). There was no evidence among the cores or debitage for the blade component present at the other findspots. Two formal tools were collected: a biface tip that was formed by soft hammer percussion, and a broken flake with marginal unifacial retouch. While only a fragment, the dimensions of the bifacial specimen suggests that the tool was fairly large; its maximum width is  mm, and thickness is  mm. Based on these sizes, McBrearty (:) speculates that the maximum length must have been greater than  mm. 127 McBrearty divides the collections into two groups, the first is comprised of Shuwaihat, Hamra, and Ras al Aysh, and the second includes Jebel Barakah. Group One is characterized primarily by the presence of a blade technology, which the author compares the material to sites collected near Khor, on the northeastern coast of Qatar. Blade production at these occurrences consists of core preparation followed by the detachment of blades perpendicular to the working surface. Unlike the material from Abu Dhabi, the Khor cores are not radially prepared, though it is noteworthy that the raw material there comes in the form of tabular flint plaquettes that probably did not require the same degree of preparation. The Khor assemblages include pressure-flakes points reduced from blade blanks, which are notably absent from Abu Dhabi sites. Because of the lack of formal tools and different techniques of blade core reduction, it is impossible to draw any definitive connections between these two complexes. According to McBrearty, the Group One assemblage is linked more directly with PPNB material from the Levant. The crescent-shaped blade cores “show remarkable similarity” to naviform cores observed at Jericho and Abu Hureyra; therefore, may date between , and ,  (ibid.: ). Group Two is characterized exclusively by radial core reduction with a façonnage element. Unfortunately, neither of these technologies are diagnostic of any particular period in time. Without the characteristic pressure flaked points, Group Two cannot be confidently classified as belonging to the Neolithic Arabian Bifacial Tradition. Thus, it is not possible to provide a temporal classification of the Jebal Barakah artifacts, though McBrearty notes “nothing in their technical execution or state of patination would exclude them from the Acheulian or Middle Stone Age” (ibid.:). This perspective is particularly germane given that McBrearty is one of the few scholars with experience in East Africa Pleistocene archaeology that have worked in Arabia. STONE AGE SITES ALONG THE NORTHERN COAST OF OMAN One byproduct of the renaissance that so drastically transformed the Sultanate of Oman in the early ’s was the movement of the nation’s capital from Nizwa, in the interior, to Muscat, on the northern coast. This resulted in the rapid development of infrastructure throughout this region, 128 which, until that time, had been comprised of small fishing villages strung along the coastline. By the late ’s, archaeological investigations commenced; the work stretched from the capital Muscat eastward to the cape of Ras al-Hadd. These expeditions, many of which continue to the present day, were spearheaded primarily by Italian, French, and German researchers. Investigations at Wadi Wutayya  in Muscat produced the oldest Holocene radiocarbon date from southern Arabia. The date of , ±   comes from ash from a fireplace, which rests on Pleistocene gravels and is overlain by over two meters of Holocene sediments. The associated lithic material was comprised of just three non-diagnostic specimens (Uerpmann and Uerpmann ). Most of the lithic collections around the capital region represent specific coastal adaptations and have little to no bifacial component; therefore, they are beyond the scope of this research. Levels  and  at Wadi Wutayya  produced elements of bifacial reduction, albeit in low frequencies. There were informal objects with bifacial retouch in level , and two trihedral rods from level , leading the excavators to classify these strata as a facies of the Arabian Bifacial Tradition. Ash from a fireplace in level  was  dated to approximately , , and a mollusk shell yielded a measurement of , . After calibration, these two dates cluster around ,  (ibid.). Among those coastal sites with significant façonnage technology is the assemblage from Saruq. This occurrence was located on the summit of a limestone hill near the village of Saruq, on the outskirts of Muscat. The site is a surface scatter with a dense covering of both lithic artifacts and mollusk shells. Three radiocarbon dates on mangrove snails from different areas of the site yielded measurements between approximately , and , . Because of the narrow clustering of these dates, as well as the uniformity of the lithic industry, Uerpmann and Uerpmann () consider the site of Saruq a homogenous unit. The most frequent raw material at the site was a good-quality, fine-grained flint of unknown origin, followed closely by red and green radiolarian chert. About one quarter of the artifacts are manufactured on a clear crystalline quartz, and a very low percentage are made on opaque white quartz and various chalcedonies. While the sources of these raw materials are unknown, they are presumably derived from local outcrops. 129 There were a number of cores in the assemblage, typically quite small and most are amorphous or globular. The poorest samples were made on coarse, brittle raw material, often exhibit a single flake removal, and can alternately be classified as debris from shatter. There were a small number of fairly regular specimens that were either discoids, or have one, two, or multiple striking platforms (ibid.). Most debitage are flake blanks; there is an insignificant number of blades and bladelets. The core reduction technique is described as “opportunistic”—there was poor control over the striking platforms and working surfaces, and, based on the prominent bulbs of percussion, knapping was carried out with a hard hammer (Uerpmann :). The toolkit from Saruq exhibited a wide variety of types, which Uerpmann and Uerpmann () placed into six general categories including: attrition tools, cutting tools, scraping tools, piercing tools, microliths, and bifacials. Attrition tools primarily include pièce esquillée, which the authors posit may have been used as punches for opening mollusks. Cutting tools are a non-standard category comprised of any blank in which one or more edges is retouched to form a cutting surface. Also in this category are tile knives made on thin tabular flint with heavy, semi-steep retouch carried out by hard hammer percussion. This type is significant, as similar tools were previously reported from Qatar (Kapel ) and represent a diagnostic form in southeastern Arabia. Scraping tools are infrequent at Saruq, and consist of sidescrapers and endscrapers, as well as denticulates. Piercing tools are the most numerous at the site. The most prominent form is a tiny borer with forming a nose at the tip, named “Saruq drills.” These types are often made on quartz, and, presumably, are associated with bead working, indicated by the perforated shells collected at Saruq. There were just two microliths, possibly used as insets in composite tools (Uerpmann and Uerpmann ). Most of the bifacials are described as simple foliates, which are subtriangular in shape. They are formed by hard or soft hammer percussion, followed by varying degrees of pressure flaking. In some cases, the retouch leaves part of the dorsal or ventral face unmodified, showing that the tools are reduced from thin blanks. One fragment was assigned to the category of foliate blades, which are narrow, elongated, and have parallel edges. The authors note that well-shaped arrowheads, ubiquitous throughout the interior of South Arabia, are absent at Saruq; however, they caution that 130 this patterning may be a result of the site’s proximity to a densely populated urban center, therefore, the tools may have been picked up by collectors. In addition to chipped stone artifacts, there were also net sinkers, pitted crushing stones, grinding stones, and perforated disc beads made on shells (Uerpmann and Uerpmann ). Of all the sites discovered in the region around Muscat, Saruq and Wadi Wutayya  (levels  and ) represent the only collections with significant bifacial components, perhaps because they predate the other marine-adapted communities that had access to more advanced technologies (e.g. metallurgy). Radiocarbon dates from both sites place the assemblages roughly within the same bracket of time, leading the authors to conclude that they represent a unique element of the Arabian Bifacial Tradition. Uerpmann () named it the Saruq-Facies, based on the site in which it is most prominently represented. This facies is viewed as an element of the ABT, given that it is coeval with Neolithic sites from the interior and exhibits similar techno-typological characteristics. The SaruqFacies differs from other ABT collections in the absence of tanged and barbed points, though the foliates fit comfortably within Type  and Type  of Eden’s proposed typology. It should be noted, however, that the trihedral rod from level  at Wadi Wuttaya  was not found in conjunction with typical ABT foliates, and may represent an earlier industry or different facies. THE JOINT HADD PROJECT One of the longest ongoing missions to the Sultanate of Oman is the Joint Hadd Project, led by S. Cleuziou from the CNRS in France and M. Tosi from the University of Bologna in Italy. The project, consisting of survey and excavation in the Ja’alan region of Oman, was initiated in  to investigate the development of Holocene communities along the coast. Ja’alan is the easternmost point on the Arabian Peninsula, stretching from the cape of Ras al-Hadd southward to as-Suwayh. In two decades, the Joint Hadd Project has produced over  archaeological sites that date from the th to th millennia . At least three different lithic complexes are represented in Ja’alan: ) a group of collections typified by unifacial and bifacial pedunculates, as well as perforators (Charpentier ); ) an 131 industry with pressure flaked points (without tangs), perforators, and diminutive multiple platform cores (Charpentier et al. ); and ) assemblages characterized by bifacial foliates formed by soft hammer retouch (Charpentier ). Unifacial pedunculates from the first industry can be characterized as Fasad points, and closely resemble the finds reported by Edens (a) in the nearby Wahiba Sands. Pressure-flaked projectile points from the second industry are typical of the coastal ABT (e.g. D-group), with well-made armatures lacking tangs or barbs. These points were found in association with worked shell beads at the site of SWY-, which provided  measurements of , ±  and , ±  . The third industry is particularly germane to this dissertation, as it appears to be a unique element of the Arabian Bifacial Tradition (Charpentier ), probably related to the Saruq-Facies defined by Uerpmann (). Seven sites belonging to this third industry have been reported from the cape of Ras alJinz (RJ-, RJ-, RJ-, RJ-, RJ-, RJ-, and RJ-) and one from the area around as-Suwayh (SWY-). These sites are typically located between two and five kilometers inland, just below the Jebel as-Saffan mountain chain. There are a number of high-quality raw materials in this region that were available to the occupants of Ja’alan. The most heavily exploited material was a fine-grained Tertiary flint that outcrops around Jebel as-Saffan, which was probably selected for the production of bifacial foliates because it occurs in large angular slabs. There is also a high-quality translucent chert locally available around RJ-///, which outcrops in small tabular plaquettes coated in black cortex. The generally diminutive nature of the bifacial foliates discovered at RJ-/// is due to the small size of this local translucent flint. Another material that is available in Ja’alan is a very fine-grained radiolarian chert that derives from the Hawasina geological group. It is found in small rounded nodules scattered throughout the wadi channels of this region. Some artifacts from Ras al-Jinz are made on a good-quality quartzite that is not found in Ja’alan; the closest source is more than fifty kilometers away (Charpentier ). All of the sites are surface scatters on deflated gravel pavements. Two sites were associated with architectural remains: RJ- and RJ-. In the case of RJ-, however, the site was reoccupied 132 during the Bronze Age and the cairns and hearths located at the site appear to be vestiges of this later occupation. RJ- contained a large structure some . meters in diameter with a hearth in the center. The bifacial foliates from the sites around Ras al-Jinz and as-Suwayh exhibit a wide variability in size, ranging from four to fourteen centimeters in length. The pieces are biconvex in cross-section, relatively thick, and appear to have been reduced from plaquettes and tabular slabs, rather than blanks. This is indicated by small patches of cortex on both faces of some of the specimens. Different forms are represented, including subtriangular, ovate, leaf-shaped, and lanceolate. Most of the artifacts are shaped by flat, invasive soft hammer percussion, confirmed by striking platforms on the debitage that are characteristic of this method of percussion. In rare cases, there is some pressure flaking along the margins of the piece. One artifact from SWY- exhibits parallel, ribbon-shaped pressure flaking scars that completely cover both faces of the piece. Although there are no radiometric dates associated with any of these sites, the material closely parallels Saruq-Facies and Khor D-group assemblages that date to the th millennium . Discussion This chapter has presented the lithic technologies, associated features, organic remains, geological context, and authors’ reason(s) for dating their sites, in order to understand why the South Arabian chronology has been organized as such. The following discussion will synthesize these data by archaeological time period and summarize scholars’ current opinions regarding the organization of lithic industries in South Arabia. LOWER PALAEOLITHIC/EARLY STONE AGE Possibly the earliest archaeological evidence of human occupation in South Arabia comes from al-Guza Cave in the Hadramaut. Amirkhanov reported a pebble industry from a stratified deposit of purportedly Plio-Pleistocene sediments. Based on the geological context and absence of handaxes, he classified the assemblage as some form of pre-Acheulean, either Oldowan or Developed Oldowan. This proposition is tentative, because the artifacts were not excavated from datable 133 sediments and, by their very nature, pre-Acheulean artifacts are ambiguous. Tool types characteristic of this industry (e.g. sidescrapers, awls, and choppers) are found in assemblages ranging from Lower Palaeolithic to Chalcolithic. Whalen’s work in southwestern Yemen has produced archaeological materials potentially corroborating Amirkhanov’s supposition of pre-Acheulean habitation in southern Arabia. He described several surface findspots characterized as a pebble industry, which he classified as “Mode ,” using a nomenclature employed by some Africanists. The evidence cited by both Whalen and Amirkhanov is consistent with Oldowan/Developed Oldowan industries of East Africa; therefore, has important implications for the initial peopling of the Arabian Peninsula. Unfortunately, the material is undated and Whalen’s methodology is problematic because he uses patination, weathering, and size as variables for organizing the lithic remains into different time periods. There are a plethora of sites with evidence for Acheulean material; however, they too, are undated. These assemblages are reported in the Yemeni Highlands, Hadramaut drainage system, Nejd Plateau, and interior drainage systems in Saudi Arabia. In most cases, Acheulean sites occur as surface scatters, though Amirkhanov excavated four buried open-air sites in Yemen. South Arabian Acheulean assemblages are typically characterized by an abundance of handaxes, choppers, single-platform cores with sub-parallel removals, and large scrapers made on flake and blade blanks. In some instances cleavers are reported, which are diagnostic of the Early Stone Age in Africa; hence, suggesting Lower/ Middle Pleistocene connections across the Bab al-Mandeb. The earliest radiometrically dated material from Arabia comes from the site of -, near the village of Dawadmi in the Central Province of Saudi Arabia. Lithic artifacts were discovered both on the surface and buried within approximately one meter of sediments. Calcareous nodules attached to the artifacts yielded U/Th dates clustering around OIS , OIS , and/or OIS . Regardless of the extreme variability, - indicates there is Pleistocene material in Arabia that occurred during an interglacial phase. Based on the “Middle Acheulean” appearance of the assemblage, and that the radiometric measurements provide only minimal dates, the occupation fits more comfortably within OIS ; by OIS  tool makers in East Africa had developed Levallois reduction. 134 MIDDLE PALAEOLITHIC/MIDDLE STONE AGE The Middle Palaeolithic/Middle Stone Age is one of the most problematic archaeological periods in the Arabian prehistoric sequence because scholars traditionally expect the material to resemble European or Near Eastern, rather than African industries. Middle Palaeolithic assemblages, characterized by prepared and pebble core technologies, sidescrapers, and denticulates, were recognized by Amirkhanov’s Soviet Expedition to South Yemen (Amirkhanov , , , ), de Maigret’s survey in the Yemeni Highlands (de Maigret , , ; Bulgarelli ), Whalen’s work in southwestern Yemen (Whalen and Pease ; Whalen and Schatte ; Whalen et al. ), and the CNRS expedition to western Hadramaut (Inizan and Ortlieb ), as well as evidence discovered by the Comprehensive Survey of Saudi Arabia and Caton-Thompson’s work in Hadramaut. In a recent synthesis of the Arabian “Mousterian,” Petraglia and Alsharekh () point out that Levallois technology, a defining characteristic of the Levant, North Africa, and Eurasia, is not well represented within Arabian assemblages. Therefore, it is misleading to apply the term “Mousterian” to these collections, which falsley implies techno-typological affinities with Middle Palaeolithic industries to the north. Petraglia and Alsharekh (ibid.) identify three Middle Palaeolithic entities in Arabia: “Mousterian of Acheulean Tradition,” “pebble Mousterian,” and “Aterian.” They argue that MAT assemblages exhibiting small bifaces in conjunction with core technologies may represent a transitional mode between the Upper Acheulean and early Middle Palaeolithic. Rather than a local technological variant as argued by Parr et al. (), the “pebble Mousterian” is probably more closely tied with clast size and form. The third variant, the “Aterian,” is based on a findspot from the southwestern edge of the Rub al-’Khali (McClure ). The site yielded a series of unifacially woked pedunculates, including tanged points, scrapers, denticulates, awls, and other miscellaneous types. While McClure () proposes a late Pleistocene date based on Aterian correlates in North Africa, others have classified similar tanged unifacial arrowheads as “Fasad Points,” associated with early Neolithic occupation in southern Arabia. 135 Inizan and Ortlieb () provided a detailed analysis of the Middle Palaeolithic core technologies discovered during their work in western Hadramaut. They identified Levallois point cores, Levallois flake cores, and blade cores at a findspot in Wadi Muqqah. These types exhibit both centripetal and bidirectional techniques for convexity maintenance. There were also thin bifacial handaxes found in association with these Levallois cores. There were some cores exhibiting a bidirectional technique similar to Nubian Type I cores, which are found in Africa as far south as K’One in Ethiopia (Kurashina ). Similar to the Wadi Muqqah collection, there are elements of façonnage reduction at K’One. Of particular interest to this dissertation are handaxe collections that have been classified as Upper Acheulean. In contrast to the preceding Lower and Middle Acheuelan handaxes, these artifacts are thinner, symmetrical, are less sinuous in cross-section, and exhibit more controlled flaking, which, in some cases, was achieved through soft hammer percussion. Biagi () reported an Upper Acheulean occurrence at Saiwan, central Oman. The bifaces are flat and biconvex in cross-section with moderately sinuous edges; they are ovoid, discoid, or limande in shape. The core technology associated with these tools consists of both discoids and singleplatform, unidirectional blade cores. Based on this description, the Saiwan collection bears similarity to early Middle Stone Age assemblages in the Horn of Africa (e.g. Wendorf and Schild ). Zarins et al. (, ) and Bulgarelli () make brief mention of Upper Acheulean bifaces from various surface sites in Saudi Arabia and Yemen, often classified as “Mousterian of Acheulean Tradition” due to their association with radial Levallois cores and discoids. Payne and Hawkins published handaxes from Habarut that, based on the illustration, could easily fit into the European Micoquian or African Stillbay Industries (Payne and Hawkins :fig. ). After examining the material from Habarut, Van Beek et al. () drew comparisons with bifacial tools from Somalia. The “Group Two” assemblage from Jebel Barakah is comprised of radial cores and the tip of a large biface with flat, invasive percussion retouch. McBrearty () points out similarities between this assemblage and material from the African Middle Stone Age. The assemblages from A-A, presented in Chapter , also fall within this category. 136 Thus, although uncertain and poorly dated, there appears to be an industry in South Arabia characterized by radial Levallois cores, discoids, and in at least some places, large, thin, well-made bifacial ovoids, discoids, and limandes. Based on radiometrically dated African correlates at Gademotta and Kulkuletti (Wendorf and Schild ), these assemblages may indicate an early Middle Stone Age occupation in southern Arabia sometime during the penultimate glaciation/last interglacial. The name Sibakhan Industry is tentatively proposed and will henceforth be used to refer to collections that fit this description. It has recently been argued that the bifaces from Saiwan are Neolithic bifacial preforms (Zarins :); this is unlikely due to their large dimensions, regularized edges, associated core types, and complete absence of anything resembling a Neolithic implement at the site. Furthermore, there have been no preforms similar to the Saiwan bifaces reported from any South Arabian Neolithic site. Neolithic core technologies are markedly different, and there are often other types of artifacts and features associated with Neolithic findspots such as hearths, structures, or ground stone artifacts, none of which was found at Saiwan. UPPER PALAEOLITHIC/LATE STONE AGE In too many discussions of lithic material dating to the Late Pleistocene, authors reference Upper Palaeolithic industries, citing the absence of a prismatic blade industry and typical Upper Palaeolithic tool forms (e.g. carinated pieces, endscrapers, Emireh points). The exclusive use of Upper Palaeolithic terminology is symptomatic of the fact that some researchers base their expectation on the industries of Europe, the Near East, and Northeast Africa rather than industries in sub-Saharan Africa. This insistence on these particular features is based on the assumption that Arabia must parallel lithic techno-typological trajectories in the Near East or Europe. Inizan and Ortlieb () first proposed the possibility that the Upper Palaeolithic of South Arabia may not resemble the European/Near Eastern sequence. To date, however, there has been no expectation that the Late Stone Age forms that typify the Horn of Africa (i.e. well-made, diminutive soft hammer foliates and backed microliths) may be found in the South Arabian UP/LSA. Many of 137 the ambiguous, undated soft hammer percussion, foliate assemblages discovered throughout Arabia could potentially fall into this category; although, at present, they may be misclassified as elements of the Rub al-‘Khali Neolithic. One such group of potential LSA bifacial foliates are those Edens refers to as Type  in his proposed typology—diminutive, thin, biconvex tools formed by soft hammer percussion, ranging from ovoid to foliate (Edens ). Specimens fitting this category have been reported at Bir Khasfa (Pullar ; Rose a), Fahud (Pullar ), and at scattered surface findspots published by Smith () and Villiers-Petocz () from museum collections stored in Muscat. In the interest of simplicity and clarity, artifacts belonging to this group will henceforth be referred to as Khasfian Foliates, based on the site of Bir Khasfa where they were first reported. Khasfian Foliates are problematic because nearly every occurrence of a biface (with the exception of large handaxes) in South Arabia has been crammed into the Arabian Bifacial Tradition, which has potentially led to oversimplification within the sequence. Edens () places Type  within the Neolithic and speculates that they may be preforms for pressure flaked artifacts. While this may be true for some specimens, it is the position of this dissertation that symmetrical artifacts with regular edges are not preforms, but completed products. Charpentier correlates the Holocene bifacial foliate industry from Ja’alan (Saruq-Facies) with Khasfian Foliates, despite the fact that he draws clear morphological differences between the two types: proche de ce que la literature préhistorique a dénommée ‘feuille’, elles n’en ont toutefois ni la finesse ni la minceur” [though similar to what the prehistoric literature calls ‘leaf-points,’ these specimens are neither as delicate nor as thin]. (Charpentier :) There is no reason to associate the Khasfian Foliates, which all occur in the interior, with Saruq-Facies artifacts that are exclusively found in coastal assemblages typified at Ja’alan, Saruq, and Khor. While they all share the same general classification of “leaf-point” or “foliate,” Charpentier notes that the tools have clear morphological differences from classic foliates. Furthermore, while Khasfian Foliates are sometimes reported in low frequencies from Rub al-‘Khali Neolithic sites (Edens 138 , b), they have never been found in a secure geological context with classic pressure flaked forms diagnostic of the Arabian Bifacial Tradition. It would not be surprising if the Rub al-‘Khali sites, which are typically found in interdunal zones associated with ancient playas, are palimpsest assemblages representing multiple phases of habitation. Radiocarbon dates and geologic analyses indicate these basins experienced two major phases of activity: approximately , – ,  and , – , (McClure ). There is a distinct possibility Upper Pleistocene elements are mixed with the “Desert Neolithic” material. Virtually every researcher with experience in East African Palaeolithic archaeology has suggested potential affinities between Khasfian Foliates and UP/LSA industries such as Magosian, Doian, and Aterian (Caton-Thompson ; Van Beek et al. ; Gramly ; Pullar ; and Villiers-Petocz ). In light of this ambiguity, it is the position of this dissertation that the Khasfian Foliates should be removed as a typological element of the Arabian Bifacial Tradition. They should be considered an undated typological group that may potentially be Upper Pleistocene, and must be investigated further. There is one radiocarbon date from the entire Arabian Peninsula that falls into the classic Upper Palaeolithic time frame. The Soviet Expedition produced a questionable measurement of , ± ,  (Amkirkhanov ) from the basal layer of an exposed open-air section in western Hadramaut. This stratum was correlated with a nearby deposit that had flat cores with parallel removals along the working surface, endscrapers, points, awls, and knives. While Amirkhanov’s Upper Palaeolithic material must be treated cautiously due to the fact that the stratigraphic correlation between the dated sediments and artifact bearing sediments is tentative, it is tempting to draw analogies between the flat blade cores he reported and those from the “Nejd Leptolithic” sites (T and T) presented in Chapter . Whalen and Pease () noted the presence of Upper Palaeolithic forms during their survey of widian draining from the Yemeni Highlands onto the Tihama coastal plain, though these artifacts were found at multicomponent sites at which the dating was determined by the degree of patination and weathering, as well as blank size. 139 One of the most intriguing potential Upper Palaeolithic occurrences is the Faw Well site from the southwestern fringe of the Rub al-‘Khali in Saudi Arabia, where a collection characterized by a prismatic blade/let technology with crested blades, double-backed bladelets [microgravettes], endscrapers, and burins was found (Edens ). Based on these characteristics, the Faw Well assemblage appears similar to late Levantine Upper Palaeolithic or Epipalaeolithic occurrences. There is no other site on the Peninsula with comparable technology or typological forms, nor are there any obvious correlates in East Africa, suggesting a possible ephemeral tie with the Near East during this phase. NEOLITHIC The density of purportedly Neolithic findspots suggests that the onset of the Holocene heralded an explosion of human population in South Arabia, not unlike that observed in the Sahara (Hoelzmann et al. ). Most research has focused on this period, due, in part, to the availability and accessibility of Neolithic archaeological materials. The plethora of data have made it possible for scholars to assemble a detailed chronological sequence for Early and Middle Holocene lithic industries. Examination of lithic technology and subsistence practices suggests there was considerable regional diversity. Tosi () envisions several communities of specialized foragers adapted to either desert steppe or littoral resources, who were geographically distinct but joined by exchange systems. Others have proposed adaptive strategies with a higher degree of mobility, in which there was seasonal movement between the coast and interior (di Mario ; Uerpmann and Uerpmann ). The prevailing chronological sequences are presented below, which are in agreement in regards to the characteristics of each cultural unit, with only slight differences in nomenclature (Table -). Edens and Wilkinson () divided the Holocene lithic industries of South Arabia into four geographically and temporally distinct entities, which they distinguish based on both technological and typological elements: ) a true blade industry with unifacial pedunculates, similar to the Levantine PPNB, found primarily in northwestern Arabia and Qatar, occasionally occurring 140 Table -. Summary of Early and Middle Holocene taxonomic schemes by region.. in other regions, ) the Arabian Bifacial Tradition, which is widespread throughout the entire peninsula, characterized by bifacial stemmed points, foliates, and lanceolates, ) an industry from the eastern Yemeni highlands, described as an “Upland Neolithic Tradition” lacking bifacial tools, and ) an industry found along the Arabian littoral, interpreted as a distinct coastal adaptation, where microliths are present and bifacial elements appear in smaller numbers. The first industry, characterized by prismatic blades with unifacial pedunculates, was initially recognized and termed the Qatar B-group by the Danish Expedition (Kapel ). In his Holocene chronology, Potts () refers to this phase as the Late Prehistoric A, typified by the “Fasad Point.” In Zarins’ () tripartite division of the Neolithic, he uses the term Period . Sites of this category have been found in Qatar, the Nejd Plateau, Wadi Wutayya near Muscat, the Wahiba Sands, and 141 the Ja’alan coast. Only one radiometric date is associated with the Late Prehistoric A—from a Qatar B-group site—placing it in the early th millennium  (Table -). Some publications (e.g. Edens b; Potts ) point out similarities between this industry and coeval PPNB assemblages from Beidha (Mortensen ), Mureybit (Van Zeist and Bakker-Heeres ), and Abu Hureyra (Moore et al. ) in the Near East, which are dated between , and , . They posit a migration of nomadic pastoralists from the Near East into Arabia during the Early Holocene, taking advantage of the ameliorated climatic conditions. Edens and Wilkinson’s second entity is the Arabian Bifacial Tradition. Potts () refers to this unit as Late Prehistoric B, correlated with the A, C, and D-Group sites reported by Kapel () and Zarins () Period . It is noteworthy that Zarins’ and Potts’ proposed Neolithic phases are considered synchronic elements within the Arabian Bifacial Tradition, while Edens and Wilkinson () have categorized the aforementioned blade/pedunculate assemblages as an industry technologically distinct from the ABT. This could be an interesting debate, addressing the questions of cultural continuity and early Holocene peopling of South Arabia, which has not yet received adequate consideration. The ABT/Late Prehistoric B/Period  appears in its “typologically most developed form” (Edens and Wilkinson :) around the Rub al-‘Khali and Ramlat as-Sabatayn Deserts, although it is known from findspots throughout the entire peninsula. There are several facies that comprise this industry, differentiated by the relative frequencies of bifacial elements and technical control over pressure flaking. In general, assemblages are characterized by well-made bifacial points, including barbed and tanged points, lanceolates, and rhomboids. Sites are commonly found in deflation hollows, sometimes consisting of fire-cracked rock, grinding stones, shell ornaments, ochre, ground stone vessels, and/or ostrich eggshell. In some cases, small circular structures are reported. A wide range of radiocarbon dates indicates this industry was relatively long-lived and stable, lasting from approximately , to ,  (Table -), which overlap with the preceding phase. Their typical association with interdunal playas further indicates habitation during the Early/Middle Holocene wet phase. 142 Zarins () sees a final phase within the ABT—Period  defined the presence of trihedral rods. He posits that they “represent flint imitations of bronze/copper points” (ibid.:). Dates for Period  are based on correlates with material from Hayd al-Ghalib in western Hadramaut (Inizan and Ortlieb ) and Habarut  in Mahra, Yemen, which cluster around the /th millennia . The more recently published dates from Level  at Wadi Wutayya  further corroborate this claim (Uerpmann and Uerpmann ). There is a distinct element within the ABT that Uerpmann () has termed the “SaruqFacies,” coeval with the terminal Late Prehistoric B/Period  phases from the interior. Localities of this type are known from littoral sites such as Khor, Saruq, and findspots in Ja’alan. The assemblages are typified by thick, subtriangular bifacial foliates formed by percussion, while pressure flaked tools are rare, if not absent. Radiocarbon dates from mollusk shells at the type site of Saruq place the industry in the mid-th millennium , corroborated by dates from Khor in Qatar, where these distinct bifacial tools are also represented. The third lithic complex proposed by Edens and Wilkinson () is known from an Italian expedition to Yemen that revealed regionally distinct assemblages in the Yemeni highlands that are more or less coeval with the ABT. The data suggest these upland settlements may have been associated with early agricultural communities situated in narrow valleys that received a high amount of rainfall per annum (Wilkinson and Edens ). There are a plethora of sites known from the upstream portion of the Wadi Dhanah, between  and  masl. Two facies have been identified, referred to as Qutran and Thayyilan. Qutran assemblages appear technologically and typologically related to the ABT, with large and small bifacial foliates, pedunculates, trihedral rods, endscrapers, burins, axes, and adzes. The Thayyilan includes some bifacial elements, such as large foliates and trihedral drills, but with a much greater emphasis on flake tools, such as scrapers, denticulates, notches, naturally backed flakes, and perforators. Pedunculates are extremely rare, and heavy equipment such as adzes and axes are completely absent. Both Thayyilan and Qutran industries occur with architecture characterized as small oval or circular huts. Radiocarbon dates from Thayyilan sites range from approximately , to ,  (Wilkinson et al. ; Edens and Wilkinson ). 143 Aceramic shell middens and lithic scatters are prominent along the Arabian littoral, particularly on the Tihama coast that abuts the Red Sea (the fourth industry proposed by Edens and Wilkinson). Sites are often found in dunes adjacent to sibakh, a setting that is comparable to the ABT findspots within dune blowouts in the interior. The shell middens are typically reported  –  km inland from the present shoreline, at the convergence of differing ecological zones including coral reefs, sand littoral, mangrove swamps, tidal creeks, meadows, and wadi bank vegetation. The lithic assemblages from the Arabian littoral exhibit some points reminiscent of the ABT, as well as unique elements such as backed points and microlithic lunates. Also present are net weights, grinding stones, axes, ostrich eggshell, and stone bracelets. These sites are sometimes associated with bead manufacture, indicated by chalcedony microdrills and worked shell debris. The aceramic littoral sites have yielded radiometric dates from the th to the th millennia  (Edens and Wilkinson ). Wilkinson et al. (:) point out similarities with coeval microlithic industries in the Horn of Africa, speculating that “the same connections across the Red Sea could also have been in force much earlier during the Late Stone Age.” In sum, the technology associated with the repopulation of Arabia during the early Holocene consists of prismatic blade cores used to create blanks for retouch into unifacial/partly bifacial pedunculates. Dates for this phase are tenuous at best; the only measurement comes from a single “B-group” site in Qatar, around , . Based on the density of material, the ensuing phase is characterized by an explosion of populations throughout the interior associated with ameliorated conditions, recognized by their well-made, pressure flaked projectile points. By the Middle Holocene, as the pluvial episode began to wane, the density of occupation shifts to the littoral zone and regional lithic facies begin to appear (e.g. Saruq, Thayyilan, Qutran, trihedral rod sites). Chapters  and  have presented a good deal of data from Pleistocene and Holocene lithic findspots in southern Arabia. While surface scatters are ubiquitous throughout the landscape, Arabia suffers from a “lack of relevant sedimentary formations related to prehistoric occupation” (Tosi : ). Unfortunately, advice given by H. Field in the early days of Arabian research has not been heeded, which has led to the present muddle of Pleistocene scholarship: 144 Until a stratified deposit has been excavated within the confines of the Arabian Peninsula south of latitude ° N., all surface finds cannot be assigned to periods comparable to those excavated in the Nile Valley, Israel, Lebanon, north-eastern Iraq and northern Iran. (:) So, there is a poorly defined Pleistocene chronology, in which potential Palaeolithic assemblages are currently classified as Neolithic. Too many researchers, lacking knowledge of East African lithic technology, have consistently expected Arabia to resemble the European and Near Eastern techno-typological trajectories, further exacerbating the situation. One intent of this work is to highlight the potentially older assemblages and present a tentative sequence based on the lithic material described in these last two chapters, as well as new data obtained by COPR. Having presented the Arabian Corridor Migration Model and summarized all of the germane archaeological, genetic, and palaeoenvironmental information, this dissertation will now turn to the findings from the  and  COPR campaigns. The following chapters will describe the geological settings of each site/findspot, sampling strategies, methods used for analyzing the lithic assemblages, and lithic data. 145 Chapter 5 SITE LOCATIONS, GEOLOGICAL SETTINGS, AND SAMPLING STRATEGIES I will get me to the mountain of myrrh, and to the hill of frankincense. —Song of Solomon This chapter describes the geographic location, geological situation, and sampling procedures of the archaeological sites included in this analysis. Our investigations focused upon two regions of Oman: ad-Dakhliyah and Dhofar (Figure -). The ad-Dakhliyah province, situated in north-central Oman, is comprised primarily of a vast alluvial plain stretching from the foot of Jebel al-Akhdar slightly more than  km southward to al-Wusta (the interior) region. The plain is marked by a series of salt domes associated with the underlying evaporitic formations, primarily Cambrian in age. The western extent is desert fringe that gradually slopes toward the Rub al-‘Khali Basin, while  km to the east the opposed border is bounded by the Wahiba Desert. Widian flow southward across the plain and drain into the Haushi-Huqf Depression, which is located in the southeastern corner of the province. The province of Dhofar is situated in the southwestern corner of Oman, abutting the Yemeni border. Moving from south to north, the region is comprised of coastal plain, escarpment, interior plateau, and desert basin geomorphic zones. Dhofar is unbounded to the east, a vast peneplain that trails off into al-Wusta, while the border of Yemen demarcates the western edge. The Arabian Sea coast mantles the southern boundary; in the north, Dhofar descends into the Rub al-‘Khali Basin and eventually encounters the border of Saudi Arabia. During the COPR  field season,  archaeological sites were recorded throughout ad-Dakhliyah and Dhofar. These sites were observed in a variety of geomorphic settings such as rockshelters, caves, inselbergs, salt domes, alluvial terraces, and relict lake basins. Almost every one 146 Figure -. Provinces of Oman where survey was conducted during COPR  campaign (false color landsat image courtesy of Google Earth™). of these findspots is characterized as a surface scatter of lithic debris, although two buried sites were excavated. The arid conditions of South Arabia present a double-edged sword; while lithic scatters are virtually ubiquitous throughout the deflated desert pavements of Oman and there is extremely high archaeological visibility, in most cases these scatters represent a palimpsest of lithic assemblages deposited throughout prehistory. Common features like façonnage reduction and simple unidirectional blade production inevitably appear at most of these sites, indicating the omnipresence of these technological lineages throughout the Pleistocene and Holocene. 147 While so many lithic scatters were recorded during the COPR  season, the  sites described in this chapter were selected for analysis based on three criteria: ) the assemblages all differ to varying degrees, together demonstrating a wide range of technological variability, ) were collected from a variety of geomorphic settings throughout the landscape, and ) indicate minimal admixture. Particular emphasis was placed on this final criterion—by minimizing excess background noise, single-occupation/phase sites are critical for differentiating the variety of characteristics present among South Arabian lithic technologies, especially in the case of Oman, where there are no prior studies relating to the Upper Pleistocene, thus, no pre-existing model of technological development. Ad-Dakhliyah Region Nearly three quarters of ad-Dakhliyah is a vast plain representing the southern edge of the Oman Mountain piedmont. The plain is interlaced by a dense network of seasonally active widian weakly dipping southward into the Haushi-Huqf Depression (Figure -). The bajada landscape Figure -. Aerial view of ad-Dakhliyah alluvial plain (photo courtesy of Yves Guichard). 148 Figure -. Major drainages on the southern ad-Dakhliyah alluvial plain and northern extent of Haushi-Huqf depression, with select archaeological sites discovered during the COPR  campaign (false color landsat image courtesy of Google Earth™). 149 displays little relief, declining from  m in the north to approximately  m in the south (Rogers et al. ). Between the villages of Adam and Haima, the plain is pockmarked by several diapirs (salt structures) that reach up to  m over the surrounding terrain and provide the only vertical relief on an otherwise flat bajada surface. These halokinetic structures, numbering about , are formed as a result of the underlying Ghaba Salt Basin, a phenomenon that is common in sedimentary basins with thick salt sequences. The difference in buoyancy between salt and surrounding geologic beds creates landscape deformation brought on by tectonic events (Loosveld et al. ). During the  COPR campaign, archaeological sites were discovered within some of the halokinetic structures, on low terraces throughout the alluvial plain, and associated with widian draining into the Haushi-Huqf Depression (Figure -). QARAT AL-KIBRIT  (A) One such diapir, Qarat al-Kibrit, is a salt dome of Cambrian age that forms the only vertical relief on the immediate landscape (Figure -). The dome was created by downbuilding, which is the accumulation of sediments around a salt dome with the top of the dome remaining at or near the surface (Figure -). The core is comprised of a disturbed mixture of halite, sulphate (Qarat alKibrit translates as “the outcrop of sulfur”), fetid limestone, and laminated black shale (Rogers et al. ). The structure is approximately one kilometer in diameter and circular in shape. Quaternary Figure -. Qarat al-Kibrit from distance, facing south. 150 sediments are found along the interior slope of the dome, perched about five meters above the bajada surface. The archaeological site Qarat al-Kibrit  is located within these uplifted Quaternary sediments, just a few meters below a gypsum crest that divides the endorheic basin of the salt dome from the outer alluvial system (Figure -). The site is situated beneath massive dolomite blocks that were isolated by selective erosion from the surrounding gypsum-anhydrite rock. Localized drainage gullies incise the surface of the Quaternary sediments, their path controlled by flow around the dolomite boulders. A block of  m was excavated in one meter units just below the dolomite boulder, and a four m test block was placed some ten meters downslope from the main excavation area. All sediments were sieved through five millimeter screen and every artifact over two centimeters was piece plotted. Layers were excavated in natural units. The sequence was excavated through stratigraphic Unit E to a depth of approximately one meter; the depth from there to bedrock is unknown. Five natural units, labeled A through E, are recognized within the sediments of Qarat alKibrit  (Figure -). They are characterized as follows: Unit A. A tabular layer of surface material up to ten centimeters thick. The matrix consists of weakly silty (very fine and fine) sand, an abundance of heterometric stones (maximum dimension of ten centimeters) formed of angular dolomite and rounded/sub-rounded anhydrite fragments. The rocky inclusions have a random orientation and distribution pattern. The sediment color (dry) is described as YR /, or light gray. Phase I archaeological materials were collected from these loose sandy sediments. Unit B. Unit A grades into Unit B, becoming weakly cemented by gypsum. The matrix is silty (very fine and fine) sand, with a decreasing density of dolomite and anhydrite fragments toward the bottom of the layer. The layer, ranging from ten to twenty centimeters in thickness, is moderately compacted with a weakly visible sub-horizontal lamination dipping westward. The color (dry) is similar to Unit A, described as  YR ., or light gray. When moist, the sediments are YR /. There 151 Figure -. Aerial photograph of Qarat al-Kibrit, facing south (photo courtesy of Yves Guichard). Figure -. Schematic profile (western) of local relief at Qarat al-Kibrit. 152 153 Figure -. Northwest stratigraphic section at Qarat al-Kibrit . is a sharp linear lower boundary that slopes southward. Phase I archaeological materials continue into this stratum, although in extremely low density within the lower portion. Unit C. The matrix of Unit C is comprised of silty (very fine and fine) sand, ranging from moderately to highly-cemented. Stones are scarce in this layer, though quite large, reaching a maximum dimension of twenty centimeters. Unit C ranges in thickness from  to  cm. It is YR / when moist. The lower boundary grades into Unit D, marked by the distinct rocky inclusions that characterize the underlying stratum. Phase II chipped stone debris is dispersed randomly throughout this level, with a slightly higher concentration at the bottom, just above Unit D. Unit D. This layer is generally thin, between five and ten centimeters thick. The matrix is silty (very fine and fine) sand mixed with an abundance of small angular pebbles of black dolomite that are between one and five centimeters in maximum dimension. Large anhydrite stones are also prevalent. Sediments within Unit D vary between unconsolidated to highly-cemented; there is a clear lower boundary with Unit E, which abruptly becomes unconsolidated and has a much lower density of rocky inclusions. The Phase III assemblage was recovered almost exclusively from the base of Unit D. Based on the high percentage of flat-lying artifacts, refits, and frequency of yellow radiolarite raw material (unique to Phase III), there appears to have been minimal post-depositional vertical movement of the Phase III assemblage. It should be noted, however, that the slope of Unit D was somewhat steeper than previous strata, and appears to be non-existent toward the eastern extent of the main excavation block; thus, it is likely the artifacts have undergone some degree of movement downslope. Unit E. Unit E has a very low density of rocky inclusions; the sediments are unconsolidated, well-sorted fine sands. The lower boundary of this stratum was not reached; it is greater than fifty centimeters thick. No artifacts were found. Unit A represents the present-day, disturbed trampled surface, partially derived from, and resting conformably on top of Unit B. Units B and C are formed predominantly of aeolian input, mixed with slope waste material formed of elements originating in the local pre-Quaternary bedrock. The sharp linear boundary between B and C, marked by significant differences in cementation and 154 alternate angles of dipping, suggest the top of Unit C was exposed, cemented in gypsum, and possibly truncated prior to the deposition of Unit B. The high concentration of dolomitic clast in Unit D indicates a high-energy environment, probably a brief wet-phase (or event) that caused increased erosion of the pre-Quaternary bedrock and subsequent runoff and deposition of slope waste. Conversely, the thick deposit of well-sorted, homogenous sand in Unit E was probably laid down during a prolonged arid phase. Post-depositional modifications of the sediments include moderate compaction and cementation. The sequence represents a poorly developed soil profile articulated in the A (Unit B) and C (Unit C) horizons, the latter being a weak gypsic horizon. Additionally, there is considerable bioturbation of the sediments in the form of vertical fissures that run through the entire sequence, caused by halokinetic uplift. The depositional processes observed at Qarat al-Kibrit  belong to a former cycle of landscape formation (Angelucci, personal communication). Geomorphic analysis of the locality immediately surrounding the site, illustrated in Figure -, indicates the landscape was deposited in the following sequence (letters correspond with those on figure): (A) The rim around the salt dome, comprised of dolomite and evaporitic rock, has undergone more or less constant uplift since the Cambrian age. (B) The QK sedimentary platform was deposited via the accumulation of aeolian sediments with alternating degrees of slope waste input. (C) The accretion was then incised by a network of channels draining into the dome. Among the artifacts recovered from the Phase I assemblage was a perforated shell bead manufactured from the apex of a Conus sp. shell (Mienis, personal communication). AMS dating of the bead provided a reading of , ±   (Wk-). The gullies (Figure -:C) that incise the platform were active sometime after that. Based on this radiometric date and the stratigraphy at QK, it is estimated that Unit A through Unit E spans the Early/Middle Holocene wet-phase. WADI QILFAH - (A, A, A, A) Wadi Qilfah is one of a series of north-south draining widian on the ad-Dakhliyah peneplain. The complex of findspots is located on the southern fringe of the plain, at the margin where the 155 156 Figure -. Aerial photograph of geomorphic features around site (photo courtesy of Yves Guichard). drainage networks empty into the Haushi-Huqf Basin. Presumably, these wadi channels were active during Quaternary pluvial cycles, feeding a series of inland lakes. Under present climatic conditions, there is a very large sabkha approximately seven kilometers south of the A-A complex (Figure -). The chipped stone material from these scatters is clearly related to the local outcropping of a Tertiary unit rich in flakeable raw material (Figure -). The outcrop is an early Miocene crust comprised of ferruginous and siliceous rocks that formed on the pre-existing surface; thus, various types of silicified material with different aspect and facies can be found in abundance throughout this locality (Rogers et al. ). The composition of the crust, being formed of harder material than the surrounding rocks, tends to form low ridges and occupy positions along the wadi divides. The scatters are located around a series of low, undulating, elongated ridges, some five meters above the wadi bottom—the area is generally flat and very gently dips southward. The immediate surface is covered by a thin veneer of sand, the thickness of which approaches zero cm at the top of the ridges, reaching tens of centimeters in the depressions between the ridges. Below this sand veneer, and associated with the surface scatters of rocks and artifacts, is a lightly indurated and trampled surface formed of slightly compacted, lamellar sands, which overlies a well-developed petrogypsic soil horizon. The surface is not reached by wadi activity; it is only affected by wind deflation and, possibly, moderate surface run-off during major rainfalls. The landscape appears fairly stable, probably dating to the initial Upper Pleistocene, if not earlier (Angelucci, pers. communication). Four scatters were collected from this area (designated A through A). Because the findspots represent in situ workshop activity zones on a landscape that has been more or less stable since the Upper Pleistocene, A and A were sampled systematically in 1⁄4 m units; artifacts were collected from the surface veneer and within the compacted sandy substratum to a depth of approximately five cm. At the A locus, a  m grid comprised of  x  m units was placed over a single high density scatter. A total area of  m was sampled at A; in this case a grid was placed over different areas demonstrating high artifact densities. A and A represent small conscripted scatters, possibly solitary episodes of reduction, and were, thus, collected in toto as single units. Sites 157 Figure -. Rim of sabkha southeast of A-A complex. Figure -. Chert and silicified limestone on surface of A-A complex. 158 159 Figure -. Aerial photograph of A-A site complex (photo courtesy of Yves Guichard). A through A are found within the slight depressions between ridges, while A is a much larger scatter situated at the top of a ridge (Figure -). All four scatters are spatially distinct, although located no more than  m from each another. WADI QILFAH  (A) Wadi Qilfah  is approximately ten kilometers southeast of the Wadi Qilfah - complex, and about five kilometers south of the aforementioned sabkha flat. It is a small, isolated scatter located on the surface of exposed Jurassic bedrock capped by a thin sand veneer and occasional low dunes (Figure -). Surface sediments are poorly mixed and range from fine-grained sands to coarse-gravels, indicating deposit during high-energy surface runoff. The underlying Jurassic bedrock is comprised of alternating beds of calcarenite, marl, and coarse grain stone. Silicification is present throughout the formation, leading to occasional chert veins outcropping from the surface crust (Figure -) (Roger et al. ). The raw material present at Wadi Qilfah  is derived from one such chert outcrop, though no outcrops were observed in the immediate vicinity. The Wadi Qilfah  scatter is comprised of a low density of artifacts distributed over a small area (less than five square meters) and resting exclusively on the surface. Even in the field, it was clear the assemblage represents a single instance of reduction, thus, was collected in toto without a horizontal grid. Dhofar Governorate Following the geomorphic divisions proposed by Zarins (), the Dhofar governorate is divided into four zones: (from south to north) coastal plain, Dhofar Escarpment, Nejd Plateau, and Rub al-‘Khali Desert, visible on the false color landsat map in Figure -. The coastal plain of Dhofar stretches for  km aligned southwest-northeast, and reaches a maximum of about  km in width; it slopes gradually and steadily upward, rising some  m asl at the base of the escarpment. The plain is made up of early Quaternary travertine, ancient terraces, and alluvial fans overlying a Tertiary limestone (Platel et al. ). These coastal deposits are cut by 160 Figure -. A lithic surface scatter in foreground (photo courtesy of Yves Guichard). Figure -. Chert outcropping in vicinity of A (photo courtesy of Yves Guichard). 161 several drainage systems that are active during the khareef season. Salalah, the second largest city in Oman, is situated along the shore in the center of this plain. The city is surrounded by fields of date and coconut palms, banana trees, mangroves, as well as sorghum, millet, indigo, and cotton—littoral South Arabia falls within the Sudano-Zambezian phytogeographic zone, which spans sub-tropical Africa into the western portion of the Indian subcontinent (Takhtajan ). Recent palaeobotanical investigations indicate that vegetation was considerably more dense along the coastal plains and south-facing mountain slopes in antiquity (Radcliffe-Smith ; Sale ; Miller and Morris ; Zarins ). The Dhofar Escarpment rises sharply above the coastal plain, averaging  m in elevation and reaching up to  m at the peak of Jebel Samhan. The hills are formed of uplifted Tertiary limestones, deeply incised by a series of southward (coastal) flowing ancient river channels. Tectonic uplifting of the Arabian plate in the Late Pliocene/Early Pleistocene caused a drop in isostatic sea levels, triggering increased downcutting by the active widian. This erosional activity prompted karstification throughout the limestone plateau, forming an extensive network of early Quaternary caves and massive sinks (Platel et al. ; Hannah and al-Belushi ; Shaw et al. ). Throughout the southern portion of the escarpment the vegetation, fed by seasonal khareef rainfall, consists of trees and dense shrubs, giving way to grassland and eventually barren scrub as one travels inland onto the Nejd Plateau (Miller and Morris ). The Nejd Plateau is a dissected tableland, stretching north from Dhofar (Figure -). The widian that drain north across the Nejd into the Rub al-‘Khali Basin were active throughout the Pleistocene, with at least three distinct terrace systems spanning the last two million years (Zarins ). The southern edge of the Nejd is marked by jagged hills and inselbergs derived from early Tertiary marine strata. The Rus Formation is particularly well developed in this region; an Eocene bed with high-quality brown tabular chert found at the base and small, gray nodular cherts throughout the unit. In the region around Thumrait there is a  km continuous exposure of Rus tabular cherts, occurring as tabular blocks and thin ovoid plaquettes. As one travels north across the Nejd, vertical relief is reduced to a flat, undulating plain comprised of Quaternary gravels overlying Late Tertiary 162 163 Figure -. Major geomorphic zones and drainage systems in the Dhofar Province, with select sites discovered during the COPR  campaign (false color Landsat image courtesy of Google Earth™). limestone beds. The landscape is marked by occasional inselbergs, and Rus Formation cherts still occur, though only in small patches where a chert stratum happens to be exposed on the surface (Figure -). The Nejd gradually descends into a vast basin that houses the largest sand sea in the world— the Rub al-‘Khali. In this desert, dunes reach over  m in height, separated by interdunal deposits that are often made up of late Quaternary lake beds. The sediments that comprise the Rub al-‘Khali dune fields were deposited during a Late Pleistocene hyperarid phase that triggered aeolian reworking of Pleistocene alluvial sediments (McClure ; McClure ; McClure ; Glennie and Singhvi ). WADI AFUR  (T) The site of Wadi Afur  is located along the middle course of the Wadi Afur drainage system in the southern Nejd, one of many north-south oriented widian originating in the Dhofar Mountains and downcutting across the Nejd Plateau. The region around Wadi Afur  is a canyon some  m wide, with limestone walls reaching approximately  m above the channel bottom. Three erosional terraces are observed at , , and  m, seen on the far side of the wadi channel in Figure -. The T findspot lies on the  m terrace that forms the base of a slope leading up two short, bifurcating tributaries. The terrace is wide (> m) and deep (> m), covered with scree Figure -. View of Nejd Plateau, above Wadi Afur, facing west. 164 Figure -. Rus chert exposure in the central Nejd. Mrs. Medici surveys artifacts at a high density palimpsest workshop site called Wadi Mahwis  (SH). sediments derived from the Late Tertiary bedrock. At the head of each tributary are wide and deep rockshelters with low ceilings (Figure -). The floors of both shelters are carpeted in approximately  cm of goat droppings, evidence for recent Beduin occupation. From a small . x . m test-pit, it was determined there is a layer of sediment beneath the dung, presumably comprised of alluvium and slope waste washed into the rockshelter with additional aeolian input. Two non-diagnostic pieces of lithic debitage were recovered from this stratum, and additional chipped stone fragments were observed within erosional gullies that incise the scree sediments just beyond the rockshelter’s dripline. The depth of the sequence was not determined, nor its relationship with the Wadi Afur  lithic assemblage. The lithic material from Wadi Afur  was found in moderate density about  m downslope from the shelter, scattered across the surface of a wide terrace. Nearly  chipped stone pieces were collected within an area of approximately  m. There are two trihedral rod fragments within the assemblage, a tool form diagnostic of the Neolithic that is regionally distinct in southeastern Arabia (Zarins ). This temporal attribution is in agreement with the site’s position on a  m terrace above a wadi that was active during the Early/Middle Holocene wet-phase. Chips are notably low 165 Figure -.  m terrace adjacent to Wadi Afur at site T, facing west. Figure -. T scatter in right foreground around large eboulis blocks, rockshelter at back of terrace, facing east. 166 in frequency, suggesting there has been occasional fluvial activity with enough energy to mobilize low mass debris. Wadi Afur  was collected in order to obtain a comparative sample of Neolithic technology with which to compare Pleistocene industries. Collection was thorough but nonsystematic; there was high archaeological visibility, due to the sharp contrast between chipped stone debris and scree sediments. GHRAIN BLISS (SH) Ghrain Bliss, which translates as “the Little Djinn” in the Mahri language, is a small mesa comprised of a resistant, massive bioclastic limestone that has been isolated by fluvial activity (Figure -); because of its position that gives excellent visibility over the surrounding plain, it currently serves as a hunting location for local Beduin (al-Mahri, pers. communication). The mesa is found in the central Nejd approximately  km west of the Marmul Petroleum Camp, situated on the left bank of Wadi Mahwis; it is a prominent landmark on the otherwise flattened landscape. Its slopes are articulated into a series of thresholds and flats at different heights—terraces were observed at approximately , , and  m above the present-day wadi channel (Figure -). These terraces are covered by very coarse gravels, aeolian sands, and slope waste. The top of the mesa is capped by a gray, thin-bedded nodular bioclastic limestone with tabular chert inclusions (Chevrel et al. ). At least four lithic scatters were observed on the terraces and on top of Ghrain Bliss (SHSH). A Fasad Point, characterized as a lightly retouched flake or blade with unifacial/bifacial pressure-flaking at the base to produce a tang, was noted among the artifacts at the top of the mesa (SH). The SH findspot was discovered on the  m terrace; it is a small, conscripted low-density surface scatter. Based on clearly observable refits at the time of discovery, the site was sampled to obtain an assemblage illustrating a single incident of reduction.  m were systematically collected in  x  m squares. A . x . m test-pit was then excavated following surface collection, revealing three geological units: 167 Figure -. Ghrain Bliss inselberg from distance. Figure -. Site of Ghrain Bliss  (SH) on lower terrace. 168 Unit A. Surface unit formed of loose, yellowish-gray sand, containing three lithic artifacts (probably incidental) and angular limestone fragments modified by thermal alteration. Unit A ranges between two and five cm in thickness. Unit B. Moderately cemented grayish sand with angular limestone fragments displaying a laminar/platy structure. No artifacts were recovered from the ~ cm-thick layer; this compacted unit was probably produced via deflation and evaporitic cementation. Unit C. A degraded rock locally derived from nodular marly limestone. Fissures are in-filled by brownish sand and concentrations of gypsum crystals, and contain no artifacts. BIR KHASFA (SH-SH) The site of Bir Khasfa was first identified by the Harvard Archaeological Survey in Oman, conducted in  and  (Pullar ; Pullar and Jäckli ; Pullar ). A reconnaissance by COPR returned to Bir Khasfa in  to further explore the area (Rose a). The site was recognized based on the rich densities of artifacts scattered across the surface, displaying a wide variety of technologies. The artifacts are found at various elevations on a rock outcrop above the left bank of the Wadi Arah, where the wide, low-energy channel enters a small localized basin, bending from a northeastward to a northward orientation. The region is named for a nearby well (bir) that today serves Beduin tribes throughout the Nejd. The local relief is formed by a crescent-shaped outcrop of Tertiary rock with abundant chert outcropping from within the inner basin, at the margins of an ancient playa lake. The raw material ranges from chunky chert slabs to thin plaquettes derived from the Rus Formation, like most other chert sources on the Nejd Plateau. The undulating surface of the Tertiary bed is covered by a mantle of aeolian sand ranging between  to  cm in depth, below which is a petrogypsic horizon. Artifacts are primarily found on the surface and embedded no greater than five cm within the subsurface aeolian veneer. Even today, the water level is quite high and there appears to be occasional surface runoff in the basin below the site, indicated by the large number of trees and bushes seen flourishing in 169 Figure -. SH Lithic scatter on surface above basin. Figure -. Tertiary crust at Bir Khasfa eroded by alluvial and lacustrine activity. 170 Figure -. Two  x  m test pits were excavated within this basin, revealing an interstratified mix of aeolian and lacustrine sediments. The surface was carpeted with a five to ten centimeter veneer of fine sand, overlying a layer of grayish yellow silty-clay with large gray mottles, fine sub-angular blocky aggregation, and enriched in gypsum (Angelucci, pers. communication). The pit was excavated to  cm; no stones, artifacts, or organic materials were observed in these sediments. A variety of assemblages, each displaying radically different technologies, were observed on the various surfaces of the Bir Khasfa outcrop within an area of roughly five km. Seemingly late artifacts were noted at the highest elevation (about  m above the basin), including a trihedral rod. The greatest density of lithic material appears concentrated below these high terraces, within the inner depression of the Tertiary outcrop, just above the erosional line marking the edge of the basin (Figure -). Two findspots, SH and SH, were collected from this area. The scatters are approximately  m apart, and do not necessarily represent conscripted reduction sites, rather distinct zones of one large homogenous scatter separated by areas of low density. They were chosen based on the frequency of completed tools (in most cases bifacial foliates) within the assemblages. A  x  m grid was sampled from SH, using . x . m units. Artifacts were collected both from the surface and a five cm subsurface stratum of sandy-gypsum. SH is about  m to the southwest, just above a small relict gully. At SH, a  x  grid was placed over the findspot, employing . x . m collection units. Again, artifacts were observed both on the surface and within a shallow subsurface sandy-gypsum carpet. DHANAQR (T) Dhanaqr, named for its proximity to a present-day well of the same name, is located at the confluence of Wadi Ribkhut and Wadi Dhahabun in the northern Nejd, where the widian broaden out and drain across the plain on their way toward the Rub al-‘Khali Basin. In this region, the landscape is primarily mantled by Quaternary sediments composed of wadi alluvium, fans, depositional terraces, khabra, calcareous palaeosols, and travertines, though the latter two geomorphic 171 zones were deposited in the early quaternary, no Upper Pleistocene palaeosols or travertines have ever been discovered in the northern Nejd. There is considerably less vertical relief than in the southern Nejd, though occasional hills and inselbergs mark the vast plain. The site is associated with a Tertiary rock outcrop rising approximately  m above the alluvial plain, comprising the Andhur Member, a geological bed described as yellowish orange shale with thin-bedded whitish bioclastic limestones and green marls (Platel et al. ). The rocky exposure spans roughly five kilometers from east to west and two km from north to south, reaching approximately  m above the wadi channels. Though there is no chert naturally occurring in this geological unit, immediately to the south there is a deflated gravel plain with Rus Formation cherts outcropping in low density on the surface. Raw material near the site consist of Rus gravels transported northward by alluvial activity and redeposited in proximity to the site. It should be noted that the raw material near the site is relatively poor for knapping. The lithic scatter at Dhanaqr is on the interior flank of a small cluster of hills; on a slope that dips gently toward the center of the semi-enclosed hill cluster (Figure -). The location overlooks both widian, and would have presented a tactical hunting advantage by providing both seclusion and elevation over the once-fertile alluvial plain. Chipped stone debris was observed on the surface of the hill slope and in a subsurface layer of loose, unconsolidated sandy gypsum that carpets the hill, ranging in thickness from zero to ten centimeters. A recent nearby P.D.O. rig camp caused minor trampling and disturbance of the site’s surface (visible in the dirt track featured in Figure -), although the high percentage of complete artifacts (. ) suggests this activity had minimal affect on the condition of the lithic assemblage.  m were systematically sampled in  x  meter units. Where present, the sandy-gypsum surface mantle was scraped clean and screened for artifacts. JIBAL ARDIF  (T) Due north of Salalah, just beyond the Dhofar Escarpment, are a series of hills and inselbergs called Jibal Ardif (Figure -). The jagged landscape is mantled by a more or less continuous - 172 Figure -. Dhanaqr (T) lithic scatter in foreground on slope of Tertiary rock outcrop, Wadi Dhahabun in distance, view facing southeast. km-long reg comprised of nodules, plaquettes, and slabs of chert. This Early Tertiary exposure is part of the Rus Formation; at its base is a - m thick brecciated dolomitic limestone with fine-grained brown chert inclusions. The original chert/dolomite bedding has been thoroughly modified by recrystallization and collapse from dissolution of evaporites, giving the present landscape its craggy relief. The hills rising above the chert regs are composed of white laminated, dolomitic, chalky limestone interstratified with three meter-thick strata of coarse dolomitic limestone breccia. This disparity in density leads to uneven erosion of the rock, producing frequent small, low rockshelters within the hills (Platel et al. ). Lithic scatters are ubiquitous on the Jibal Ardif reg surfaces; most of these scatters exhibit a palimpsest of technologies and are probably mixed assemblages. One such scatter, Jibal Ardif , is situated about  km south of Thumrait and five kilometers east of Wadi Dawkhah. The assemblage was initially noted for the high number of bifacial preforms scattered over a relatively small area. This homogenous assemblage is unique among most other scatters on the Jibal Ardif regs, which are typically mixed. The specialized nature of Jibal Ardif  is related to a local outcrop of thin, high-quality chert plaquettes, particularly conducive to the 173 Figure -. Dr. Usik surveys the Jibal Ardif  (T) lithic scatter. Figure -. Fine-grained chert plaquette outcropping in vicinity of T. 174 production of bifacial tools (Figure -)—the site is clearly a workshop for the exploitation of this outcrop. The scatter is on a flat plain between hills; the immediate landscape is lined by a series of small parallel limestone ridges spaced ten meters apart on average (rarely exceeding ten centimeters in height and twenty centimeters in width). The ridges are formed by interstratified beds of tectonically folded limestone strata; subsequent erosional processes have had a greater effect on the softer chalky limestones, leaving prominent ridges comprised of the harder dolomitic stratum. The terrain is covered by a very thin veneer (not exceeding five centimeters in maximum depth) of hard-packed gypsum evaporites, indicating poor drainage flowing from the local basin. Artifacts were recovered on the surface and imbedded in the evaporitic sediments. The scatter is moderate in density, with nearly  pieces collected in an area of  m. The material exhibits minimal weathering, which contrasts the apparent antiquity of the large, well-made hard-hammer bifaces that are characteristic of the late Middle/early Upper Pleistocene. One explanation for this disparity is that the local basin was only recently exposed, though the materials’ position embedded within the gypsum suggests there has been little post-depositional disturbance, at least since the last wet-phase in the Early/Middle Holocene. Because of the pristine nature of the material, the site was collected in . x . m units. AL-HATAB (T) The site of al-Hatab, named for its proximity to the al-Hatab borehole, is located at the southern end of Jibal Ardif; the base of the Dhofar Mountains, just north of the present-day watershed divide. The findspot was discovered within a small tributary in the upper courses of Wadi Dawkhah, on a wide terrace about  m above the current channel. Raw material availability is similar to that at Jibal Ardif , with Rus Formation cherts littering the landscape around the site. In contrast, however, the raw material at al-Hatab occurs as rounded nodules and large slabs rather than as thin plaquettes. 175 The tributary is quite small, approximately  m long and  m wide. There is a small rockshelter perched about five meters above the channel, oriented parallel to the drainage system. A small stone wall has been constructed across the front of the shelter, presumably by recent Beduin occupants. Bedrock is exposed inside; sediments have been scoured clean by erosion and/or human activity. Steep scree slopes flank both sides of the tributary, filling the channel with colluvium. A cm-deep gully incises these sediments, exposing a wide section (Figure -). A plethora of chipped stone artifacts were observed on the left bank of the tributary, both on the surface of the scree and eroding from the section. Three square meters were excavated at al-Hatab to a depth of  cm, yielding just over  artifacts. Few artifacts were horizontally bedded; most were oriented on edge or in vertical position. Excavation was carried out in two arbitrary -cm-thick levels, though there is only one natural unit represented in the sediments (Figure -). The sediments are comprised primarily of coarse-grained, poorly sorted slope waste with some degree of low-energy fluvial input, indicated by small pockets of clayey-silt. The matrix is predominantly silty-gravel with a high density of angular dolomitic limestone, as well as chert inclusions averaging between five and ten centimeters in maximum dimension. Below the colluvium is a sterile layer of cemented gypsum that extends to an unknown depth. Clearly, the artifacts are in secondary position, although whether the material originated from within the rockshelter or hills above the tributary is unknown. The lithics demonstrate extensive evidence of desert varnish, pot-lidding, and shattering, suggesting they were exposed on the surface and underwent thermal fracturing sometime after discard and prior to burial. It should be noted that the arêtes are sharp and there is minimal evidence for rolling, so artifact transportation was over a relatively short distance. Assuming the gully downcutting through the site’s sediments is associated with the Early/Middle Holocene wet-phase, and considering the evidence for the lithics’ prolonged exposure on a hot, dry surface prior to mobilization (presumably during the Late Pleistocene hyperarid phase), the al-Hatab assemblage must have been manufactured sometime prior to ~, . 176 Figure -. Dr. Usik examines artifacts from al-Hatab (T) that were found eroding from an exposed section at the base of the slope. Figure -. Stratigraphic profile of intermixed colluvium and aeolian sediments at al-Hatab (T). 177 Discussion Archaeological sites recorded during the  COPR season come from a wide variety of geomorphic settings. Findspots are common within and around the diapirs that dot the ad-Dakhliyah plain, as well as on alluvial terraces. In most cases, these are just secondary surface scatters on deflated gravel pavements. The preserved Quaternary sediments at Qarat al-Kibrit are unique because the dome was formed via downbuilding, and has not been affected by fluvial activity. It is not surprising to find dense concentrations of lithics throughout the region; high-quality radiolarite pebbles are ubiquitous in the alluvial gravels that mantle the plain, and the channels that interlace the peneplain provided abundant water in antiquity. There were lithic findspots recorded around the northern margins of the Haushi-Huqf Depression; along the widian that drain into the depression. The area further south inside HaushiHuqf was not explored, due to loose sands that were not possible to traverse. Investigations focused primarily along Wadi Qilfah simply because of logistical ease, where archaeological sites were observed in proximity to outcrops of ferruginous/silicified sandstone and limestone (e.g. the Wadi Qilfah - complex). Wadi Qilfah  is not located in the immediate vicinity of flakeable raw material, which is why the low density findspot is an ephemeral site deposited during a single reduction event. Prehistoric inhabitants of Haushi-Huqf were probably drawn to the series of nearby sibakh, these basins were freshwater lakes during pluvial events, and, therefore, attracted considerable biomass. A brief reconnaissance was conducted around the plain south of al-Ghaba, though no sites were located. This is not surprising, given that there is no flakeable raw material and there was minimal surface runoff. This barren plain continues south for several hundred kilometers, eventually leading onto the Nejd Plateau. Archaeological sites are ubiquitous throughout the Nejd, occurring everywhere there is flakeable raw material. Chert is plentiful in this area, due to the high incidence exposed Rus Formation outcrops. The entire plateau north of the Dhofar escarpment provided ideal conditions during episodic pluvials, when the Southwest Indian Ocean Monsoon System intensified and brought considerably increased rainfall. The numerous wadi channels that drain northward from the Dhofar 178 Mountains experienced greatly increased fluvial activity during these episodes, eventually feeding into the Rub al-‘Khali Basin. It is noteworthy that assemblages with diminutive bifacial foliates are concentrated in this northern region, in proximity to the palaeolakes (e.g. Bir Khasfa, SH, and SH). The COPR  survey examined a number of locations throughout the Nejd east of Thumrait. These loci included wadi channels, rockshelters, terraces, inselbergs, and the top of the plateau. The full length of Wadi Dhahabun was investigated: from its source in the Dhofar escarpment to the point at which it dissipates into the northern Nejd basin. There were few assemblages in the southern extent that appeared Palaeolithic, despite abundant raw material availability. Holocene sites, however, are plentiful throughout the southern Nejd, indicated by several sites with diagnostic features. Very little lithic material was observed at the top of the plateau, although admittedly minimal attention was paid to this zone. The material that was recorded appears very early, exhibiting crude radial cores, heavy wind abrasion, and thick desert varnish. Rockshelters hold some promise, though in general these structures are quite small because of the thinly bedded strata. Most of the accessible shelters are washed clean of sediments and date to the Holocene. Earlier shelters are high above the present wadi channel and have been buried by slope waste. There were a handful of shelters that contained preserved deposits with artifacts on the scree; future investigations will focus on sampling these sediments. Finally, the Nejd survey briefly examined the Dhofar escarpment, which is riddled with karstic cavities, as well as travertine along the coast. While there appears to be an abundance of Holocene material, no Palaeolithic artifacts were found. This may be related to the absence of chert south of the Nejd. The following two chapters describe the methods used for lithic analysis and provide a detailed examination of the lithic assemblages sampled for this dissertation. Technological and typological features of each site are described. These data are used to differentiate a variety of industrial complexes. Some of these assemblages fit into the pre-existing Neolithic sequence, while 179 others do not belong to any previously documented industry. The assemblages are classified according to the technique of core reduction, as well as presence/absence of a bifacial component. It will be demonstrated that there are technologies present in central Oman throughout the Middle and Upper Pleistocene, as well as early/Middle Holocene, that suggest technological and typological affinities with every contiguous region: East Africa, India, and the Levant. 180 Chapter 6 METHODS OF ARCHAEOLOGICAL SAMPLING AND LITHIC ANALYSIS The best laid schemes o’ mice an’ men Gang aft a-gley. —Robert Burns, To a Mouse The  season of the Central Oman Pleistocene Research (COPR) program was originally planned to be a systematic survey of the Wadi Arah drainage system in the northeastern region of Dhofar, following J. Pullar’s discovery of Bir Khasfa there in the ’s (Pullar ). A few weeks into the survey, it soon became clear that we could not achieve our research goals in such a limited geographic area. While preserved Upper Pleistocene sediments do exist throughout Oman, locating in situ archaeological materials within such sediments has repeatedly proven elusive. This was the case of Wadi Arah, where only recent Holocene shelters were exposed; more ancient rockshelters are buried under layers of colluvium. Even on wadi terraces, Palaeolithic materials were scarce. The paucity of Palaeolithic surface sites is probably due to the fact that the Rus Formation geological bed, which is the source of the high-quality chert so prevalent throughout the Nejd, has quite a limited exposure in southern Arah (or at least the portions we observed during the survey). Furthermore, it appears that fluvial activity in this zone has been sufficient to either erode or significantly bury potential Pleistocene deposits. Rus Formation cherts occur in high frequency in northern Wadi Arah, and there are myriad lithic workshops sites associated with those exposures. These too, however, did not offer the necessary data to achieve our research goals, because most of these sites represent a palimpsest of lithic assemblages that were impossible to separate from one another. Taking these limitations into account, the project shifted to a targeted reconnaissance survey throughout the Dhofar and ad-Dakhliyah provinces of southern and Central Oman. Over a period of 181  days, we investigated caves, rockshelters, alluvial deposits, gravel pavements, inselbergs, lacustrine overbank deposits, and aeolian dunes in order to locate and collect Palaeolithic materials. This chapter presents the sampling strategies we chose during this targeted reconnaissance, as well as methods used for analyzing the lithic assemblages. Site Sampling Strategy The goals of the fieldwork were threefold: ) to locate Pleistocene deposits, ) to sample sediments with in situ archaeological materials, and ) to locate and collect lithic scatters on ancient surfaces. These goals were intended to obtain a sample of lithic assemblages with which to begin to describe the variety of reduction sequences present in the South Arabian Pleistocene, as well as to locate material in a secure geological context that could be used to build a relative chronological sequence, and, possibly, acquire radiometric dates. The  COPR season was carried out in the ad-Dakhliyah and Dhofar provinces of Oman. Because some of the research was conducted in areas that had never before undergone archaeological investigations, in order to cover as much ground as possible it was necessary to target specific geomorphic zones. Using false-color Landsat imagery and :, scale geologic maps, the survey explored caves, rockshelters, travertines, halokinetic structures (i.e. salt domes), lacustrine deposits, alluvial sediments, and ancient terraces. Work began in the north of ad-Dakhliyah, surveying terraces and inselbergs just south of the Jebel Akhdar mountain range. Lithic findspots were rare in this area, due to the paucity of flakeable raw material. While siliceous deposits are present, they had been thoroughly modified by tectonic activity, and are generally riddled with fracture planes and, therefore, poor for knapping. In response to the paucity of high-quality raw material, the survey shifted its attention southward to the adDakhliyah peneplain and the monoclines that lie at the northern edge of the plain. The relatively flat landscape and high archaeological visibility in central and southern ad-Dakhliyah made it possible to conduct a driving survey throughout this region, targeting findspots associated with raised terraces, salt domes, and sibakh. A plethora of secondary archaeological sites were recorded from low terraces 182 throughout the peneplain. Of the domes scattered across ad-Dakhliyah, Qarat al-Kibrit, Qarat alMilh, Jebel Majayiz, and Qarn al-Alam were systematically explored. In every case, non-diagnostic material was observed on the surface, although only at Qarat al-Kibrit was a buried archaeological sequence located. We conducted brief reconnaissance along the fringe of the Haushi-Huqf Depression, which houses a number of sibakh. The site of Saiwan (Biagi ) had previously been reported from this zone, and is of particular interest to the COPR program, due to its association with the Sibakhan Industry. Unfortunately, loose sands and high winds prevented exploration beyond the fringe of the depression; therefore, we were not able to thoroughly investigate this potentially important area. It was more difficult to explore Dhofar, due to the deeply incised widian that cut the limestone plateau and hinder navigation across the terrain. Our efforts were primarily focused on the region to the north of the Dhofar escarpment—the Nejd Plateau—because of the availability of high-quality chert that is ubiquitous throughout this landscape. The initial phase of the survey included a driving reconnaissance along the entire course of Wadi Dhahabun, from the point of origin at the summit of the Dhofar escarpment, to the northern extent where it widens out onto the relatively flat Nejd Plateau. This was done in order to assess archaeological potential throughout the various portions of the plateau. It soon became apparent that the only preserved Pleistocene material was concentrated in the central and northern reaches of the Nejd, where the flat expanses, occasional inselbergs, and abundant raw material must have offered ideal hunting and living conditions in antiquity. In addition to surveying the central and northern Nejd, we employed a local Beduin guide to take us to caves and rockshelters. After exploring countless rockshelters, we concluded that these held little potential for achieving our goals in the time available, as they tend to be low, shallow, and scoured clean of deep sediments. There are large karstic cavities that riddle the Dhofar escarpment, which may, in fact, have significant Pleistocene deposits. The  reconnaissance survey visited a few such caves in Dhofar, although did not observe any evidence for Palaeolithic occupation. This initial survey was cursory, and, due to logistical constraints, it was not possible to reach many of the more promising cavities. 183 When surface sites were encountered during the survey, preliminary considerations were made to determine the general nature of the lithic material and post-depositional processes that affected the local landscape. Since one of our primary goals was to identify and describe Palaeolithic reduction sequences, it was necessary to collect findspots that were not palimpsest, but comprised of lithic material from a single or minimal phase(s) of occupation, to reduce as much background noise as possible. Therefore, we examined the variety of core types, finished tools, patination, weathering, and geological context as indicators for potential mixing at the site. If we determined the site was relatively homogenous, a grid was established that was comprised of either  x  or . x . m units, depending on the density of artifacts and degree of post-depositional activity. The grid was placed over artifact concentrations in an attempt to identify specific activity zones. The size of the collection area was proportionate to the density of the lithic material, in order to obtain a relative sample that could be used to adequately describe the reduction sequences present at each site (roughly between  and  pieces). All chipped stone material within the grid was collected by unit, and sediments were scraped to a depth of approximately five centimeters to obtain lithics embedded in the surface veneer. In cases where smaller material was present, we sieved sediments using a  mm screen. At the A, A, and A findspots, there were small, confined scatters of less than  artifacts that clearly represented a single instance of reduction; therefore, these localities were collected in toto without a horizontal grid. When lithic material was observed in association with preserved sediments, . x . m testpits were dug to determine if the deposits contained archaeological materials. If subsurface artifacts were identified, further systematic excavation was carried out in  x  m units. Two such buried sites are presented in this dissertation: Qarat al-Kibrit  (A) and al-Hatab (T). At A, the artifacts were more or less in situ, so every piece greater than two centimeters was plotted and the object’s orientation was recorded. All sediments were sieved through a five millimeter mesh screen. The site was excavated in natural units, which were designated Unit A through Unit E (from surface to basal layer respectively). 184 The sediments at T were washed in, and archaeological material seemed to be in secondary context; therefore, in the interest of efficiency, the artifacts were not piece plotted. Although we could not identify any natural divisions in the deposit, the site was excavated in two  cm-thick arbitrary units. All sites were given two names: one initial code in the field and a subsequent designation related to local geographic landmarks (e.g. widian, mountain ranges, water wells, etc…). The initial label consists of a letter representing the local governorate (wilaya) in which the site was found, followed by a number consecutively assigned to each site as they were discovered. The wilayat and corresponding abbreviations are listed in Table -. Table -. Provinces and wilayat of Oman where survey was conducted during COPR  campaign, with corresponding abbreviation. Lithic Analysis The examination of lithic typology and technology in southern Arabia is useful for understanding the geographic origins of its prehistoric inhabitants. The production of stone tools is an integral part of the cultural system of hunter-gatherers; it is an acculturated process; hence, evidence for culturally shared ideas may be observable through particular lithic attributes and the reconstruction of the technological sequence (Wyckoff ). The descriptions of the lithic assemblages presented in this dissertation are divided into typological and technological categories. Typology is significant, as South Arabia exhibits a long tradition of bifacial tool production, which must be carefully considered. Chapters  and  demonstrated that most foliate/point assemblages have been hastily placed within a Holocene framework, under the umbrella of the Arabian Bifacial Tradition, despite the fact that many of 185 these assemblages are undated and cannot be correlated with forms diagnostic of the ABT. Thin, diminutive foliates are not exclusive of the Holocene, they are reported from various industries around the world as early as the Middle Pleistocene, such as Korolevo level a in Transcarpathia (Kulakovska , ) or Galeria Pesada in Portugal (Marks et al. ). During the Upper Pleistocene, foliates appear in the Micoquian (Bosinski ; Soressi ) and Solutrean Industries (Smith ), and are prominent (in varying degrees) throughout the MSA of sub-Saharan Africa. Morphologically similar specimens are common in the early Holocene of North America (Wormington ) and were even collected at Middle Holocene sites in Central Australia (Valoch ). Clearly, these tools are not diagnostic of a single period of time. The lithic analysis presented in this dissertation will focus on describing the production and morphology of bifacial forms, in order to demonstrate that the variety of bifacial industries present in southern Arabia is greater than traditionally perceived. Most publications dealing with Pleistocene/Holocene lithic collections have been restricted to general descriptions of retouched tools. This limitation is not necessarily the fault of the scholar, in many cases the material being presented was initially recovered by non-archaeologists who only collected the most obvious or aesthetically pleasing artifacts—the completed tools. Therefore, most collections have been described purely in terms of general typological elements, there has been little emphasis placed on technological processes. While this approach is useful for identifying the cultural or temporal period from which an assemblage is derived, it ignores the greater part of the lithic assemblage—the debitage. Debitage is essential for shedding light on how tools were made and illuminating past technical processes: “while typologies are static constructs of the archaeologist, technological reconstructions are explicitly concerned with actions, and attempt to understand why and how tools were manufactured” (Monigal :). Although there is no single paradigm to which lithic specialists subscribe, there is general agreement that the goal of their research is to understand the organizational properties of extinct cultural groups that have produced the archaeological record as it presently exists. Through this approach, in addition to investigation of other relevant factors, such as mobility, raw material acquisition, and social activity, a more complete picture of the past can be formed (ibid.). 186 One of the ways this is achieved is by understanding the chaîne opératoire—behavioral chain, which can delineate the technological sequence that transforms a raw material into a cultural product. The primary method used in this study to reconstruct the chaîne opératoire is attribute analysis. Attribute analysis is a tool with which to understand technical products by identifying: ) the type of product, ) its place in the technological process, ) the type of preparation necessary for its removal, and ) the characteristics that result from its removal on the core. Attribute analysis utilizes a set of mutually exclusive categories to describe morphological variation and metric dimensions of artifacts, the goal of which is to deduce the steps undertaken throughout the reduction sequence—raw material selection, core shaping, blank production, core maintenance, tool production, and discard. Establishing a series of criteria for this dissertation proved challenging, as there have been no previous lithic analyses in South Arabia upon which to base this study. There are no universally agreed upon methods, and not enough is known of Palaeolithic technological processes in this region to conduct a specific, targeted study. Thus, a wide array of general categories were selected that, it was hoped, would both be useful for understanding reduction sequences on a basic level, and provide a foundation upon which to base future work (Table -). The attributes used in this study adhere to those defined by Marks (), who revised the attribute types created by Borde (), de Heinzelin (), and Tixier (). These technological descriptions were published in the monographs from the Central Negev project (Marks , , ), and remain one of the first and most detailed treatments of the subject. In this attribute analysis, the assemblage is initially sorted into four categories: debitage, cores, tools, and debris. Debitage is defined as the spalls produced during the reduction of a core, which were neither secondarily modified nor served as blanks for secondary removals. This dissertation will use the word “blank” as a general umbrella term for any spall purposely removed during the knapping process, including flakes, cortical pieces, blades, bladelets, etc… Cores are pieces of raw material from which blanks have been removed. They are typically the byproduct of blank production, though in some cases may have been shaped and modified to serve as implements in their own right. 187 Table -. Qualitative technological observations used in attribute analysis. 188 Tools include any blank whose edges have been modified by direct or indirect percussion and employed for a specific task or purpose. Tool type categories themselves are arbitrary designations constructed by archaeologist in order to classify standard forms that consistently appear in the archaeological record. A tool may also be an unmodified preferential blank removed from a prepared core (e.g. Levallois point). Debris is divided into chips, which are any small lithic fragment under  mm in dimensions, and chunks, which are shattered pieces that exhibit no discernible flaking pattern. These four artifact classes are not mutually exclusive; for instance, a tool manufactured on a blank is not only treated as a tool, but also the blank upon which it is made is described based on its blank attributes. The following describes each attribute that has been recorded among these three general artifact classes (debris is not included). DEBITAGE Blank type. Blank types are critical indicators for the type of reduction sequence and position each individual piece falls within the chain. For this analysis, they were sorted into fourteen types (flake, blade, bladelet, cortical flake, cortical blade, core trimming element, core tablet, éclat de taille, Levallois blank, burin spall, kombewa flakes, débordant flake, débordant blade, and débordant bladelet). Flakes, blades, and bladelets are simply morphological classes: blades are twice as long as they are wide, bladelets are a sub-category of blade having a width of less than or equal to  mm, and the rest fall into the class of flake, being less than twice as long as they are wide. The  mm cutoff for bladelet is an arbitrary designation proposed by Tixier (). Burin spalls are morphologically similar to blades and bladelets, although they result from the specific reduction of a burin or carinated piece. They are distinguished from blades and bladelets by the large platform size relative to the volume of the piece, often with scars visible from the preparation, and have a steep triangular cross-section, approaching laterally-steep edges. Burin spalls are often longitudinally twisted. Cortical flakes and blades fall into the class of primary element, defined as a blank with some degree of cortex remaining on the dorsal face. This attribute is useful for determining the stage of reduction occurring on site—the more cortical flakes, the earlier in the reduction sequence. Since 189 there is another attribute quantifying the percentage of cortex, it was decided that only pieces with  cortex would be sorted into this category. Core trimming elements are pieces the purpose of which appears to have been the preparation or cleaning of the core, whether from cortex, natural impurities, or mistakes (excluding core tablets and débordant pieces). They are recognized not by one specific trait, but a combination of repetitive or standardized characteristics. CTEs often have at least one steep lateral edge, are thick, twisted, and/or asymmetric. Typically they show evidence on the dorsal face of hinge fractures, and may have cortex at the distal and/or lateral edges, as well as remnants of a previous striking platform. Though core tablets and débordant flakes, blades, and bladelets also result from core trimming, they were placed in their own category, given their technological importance. Débordant pieces are characterized by laterally steep cross-sections; in most cases the dorsal surface exhibits part of the core edge, either the platform or unprepared surface edge. These pieces are often associated with the preparation of lateral convexities on a core. Depending on their extent of cortex, débordants may belong to the initial stage of preparation of the core, or a re-preparation stage, following a series of removals from the working surface. Regarding core tablets, A. Marks provides a concise, detailed definition of this class: By-products of core rejuvenation, occurring when the striking platform needs major modification. A core tablet is formed by striking off the platform of a core with a blow perpendicular to the axis of the major flake or blade removals (parallel to the old platform), a centimeter or so below the intersection of the flaked surface and the platform. This removes the old platform and those portions of the scars on the flaked surface just below the platform. (Marks :) Éclats de taille are thinning flakes, which result from reducing the face of a biface, as well as maintaining convexity on the working surface of a prepared core (Borde , Brézillon ). This class is particularly germane to the present study, given the preponderance of bifacial reduction throughout South Arabian prehistory. Éclats de taille are defined by a suite of conditions beginning with a relatively thin piece and an incurvate longitudinal cross-section. Striking platforms are frequently lipped, may have edge grinding, a high-angle, and are faceted or dihedral, which represent 190 remnants of the edge of the biface. Typically, the dorsal scar pattern is radial, bidirectional, or crossed, and exhibits shallow arêtes. The pieces are most often ovoid or trapezoidal in shape, with gradually feathering lateral and distal edges. Though not originally included in the attribute analysis, the presence of Kombewa technology at Qarat al-Kibrit necessitated the addition of this category. Kombewa flakes, named for the locality in western Kenya where they were first recognized (Owen ), are debitage resulting from the reduction of a core-on-flake. They are recognized by evidence of a previous ventral face on the dorsal side of the blank. Classic Kombewa flakes are removed from the proximal end of the core-on-flake, thereby exhibiting an additional bulb of percussion on the dorsal face, though Usik () argues pieces removed from lateral and distal edges may also be placed in this category. For this analysis, Kombewa flakes were described both in terms of their blank attributes, as well as core characteristics. There are entire works dedicated to the recognition and definition of the Levallois method (e.g. Breuil ; Borde ; Kelly ; Bourgon ; Boëda ; Van Peer ; Usik ), although a detailed discussion of Levallois is beyond the scope of this research, given the lack of resolution within the South Arabian data. Because of these constraints, this presentation of Levallois will remain in general terms. Levallois blanks are the flaked product struck from a prepared core, and, hence, are considered both debitage and tool. There are several different approaches for recognizing these blanks; for this analysis it was decided to adopt a strict interpretation following Usik ()— the Levallois method is only the preferential removal of a blank from a convex working surface, in which the striking platform has been prepared by some degree of faceting. Therefore, Levallois blanks were recognized based on a combination of characteristics that consisted primarily of dihedral or faceted platforms, as well as radial, crossed, or convergent scar patterns. There is the possibility for some overlap with éclats de taille, which is not unexpected because, in both cases, both blank types were struck across a flat core/preform perpendicular to the working surface. In order to distinguish these two blank types, Levallois pieces were recognized as somewhat thicker, having a flat longitudinal cross-section, and a relatively low-angle striking platform. It was not necessary to further classify them 191 as flake, point, or blade (see Monigal  for a discussion of Levallois blades), given that this can be independently determined by the “shape” attribute class. Condition. This category describes whether or not the object is complete or broken, and where the break occurs. The five categories are unbroken, proximal (only the proximal end is present), distal (only the distal end is present), medial (the piece is missing both proximal and distal ends), and false-burin. False-burins are longitudinally snapped pieces that break at the time of impact due to flaws in the raw material or excessive force. Though they are morphologically similar to trueburins, these pieces are distinguished by a snapped striking platform; it is assumed the breaks were not intentional (de Heinzelin ; Brézillon ). Platform. Platform preparation is the removal of one or more small flakes at the exterior of a core margin to strengthen the platform for flaking, correct irregularities, or to increase the platform angle, which can result in longer flakes (Whittaker ). Since these goals can be achieved by several different—possibly culturally specific—techniques that produce a variety of platform shapes, nine states were recognized (Figure -). These are divided among unprepared (straight, cortical straight, cortical curved), prepared (dihedral, dihedral half cortex, faceted straight, faceted curved), and unidentifiable (crushed, missing). Borde and Bourgon () proposed using the Index of Faceting (IF)—the ratio of prepared platforms to all identifiable platform type—in order to measure the extent of platform preparation/maintenance within an assemblage. A straight platform is achieved by a single blow across the core platform that leaves a smooth, flat plane from which the blank is detached. A cortical platform is entirely cortical, Figure -. Some platform types recognized in analysis (adapted from Monigal ). 192 naturally weathered, or otherwise unmodified, indicating that the portion of the core from which the blank was removed underwent no previous preparation. Cortical straight and cortical curved are distinguished to help determine the initial shape of the raw material. A dihedral platform indicates the blank was struck at or near the ridge formed by the intersection of two blows across the core platform. Dihedral half cortex indicates the piece was removed at the ridge formed by the intersection of a single blow and the unmodified surface of the core. Faceted platforms exhibit traces of scars from three or more blows across the core platform surface, which may originate from any direction. Two varieties are recognized: faceted straight and faceted curved, which possibly suggest a desired platform morphology at the time the blank was struck. Crushed platforms are created when the previous blow did not succeed in removing a blank, or when the point of impact was too close to the core face, resulting in proximal shatter. Though the platform is not identifiable, these are considered complete pieces. Platforms were classified as missing when the proximal end of the blank is absent due to breakage, or if the piece has been extensively retouched at the proximal end, thereby completely removing the platform. In both cases of crushed or missing, no platform measurements were taken. Edge preparation. Edge preparation is the presence or absence of marginal, nibbling retouch along the exterior ridge of the striking platform where it intersects with the ventral face. This was achieved by grinding or lightly retouching the edge, in order to strengthen the platform by eliminating brittle overhanging remnants from the previous removal (Pelegrin ). In the Near East, edge preparation is a technique employed during the Upper Palaeolithic in the reduction of prismatic blade cores. Alternately, edge preparation is commonly used during bifacial reduction to prepare the platform for the removal of éclats de taille. Lipping. Lipping is characterized as a slight protruding overhang on the bulb of percussion, just below the ridge formed by the intersection of the striking platform and dorsal face. The presence of lipping is often considered an indication of soft hammer percussion, although other variables affect lipping as well, such as the relative hardness of the raw material and angle of the platform (Crabtree 193 ; Tixier et al. ; Ohnuma and Bergman ). For this analysis, the presence/absence of lipping is considered to be a valuable indicator for differentiating varying reduction sequences in the production of bifaces. Edens () identified three stages for the production of “Desert Neolithic” pressure flaked bifaces, ranging from hard hammer rough-outs, to soft hammer thinning, ending with sharpening by pressure flaking; therefore, it is necessary to determine the methods of percussion present in each assemblage, and their associated tool forms to place “Desert Neolithic” thinning flakes in their appropriate stage of reduction. Shape. A deliberately reduced core should produce primarily blanks of a predetermined shape and standardization. Blank shape is a function of several interrelated technical variables, such as the morphology of the removal surface, preceding flake/arête pattern, and shape of the striking platform. In cases of Levallois reduction, there is often a desired product with deliberate core preparation to achieve that result—whether it be triangular, ovoid, rectangular, etc… Alternately, bifacial reduction, while not consciously producing purposely shaped blanks, tends to create a certain set of shapes, based on the nature the thinning face, ranging between ovoid to trapezoidal. Blanks were sorted into five categories in this analysis, including trapezoidal, rectangular, ovoid, triangular, and irregular. Midpoint cross-section. The midpoint section of a blank is the number and configuration of the planes on the dorsal surface between the lateral edges of the piece (Figure -). The cross-section relates to where on the core platform the impact occurred, with the previous blank removals forming guide ridges along which the force travels. The transverse profile is related to the overall thickness of the piece; trapezoidal blanks are typically flatter with more acute edges than blanks with a triangular cross-section. The midpoint cross-section is suggestive of core face morphology—for instance a Levallois core, where a single convex or flat surface is exploited, will typically yield blanks with a Figure -. Midpoint cross-sections (adapted from Monigal ). 194 relatively flat and thin midpoint. Finally, the transverse profile yields information about the number and pattern of dorsal scars (Crabtree ; Owen ). For this analysis, blanks were described as having a triangular, trapezoidal, lateral-steep, convex, flat, or irregular cross-section. Triangular profiles have two major scars forming a central ridge, which indicate the force was applied directly behind a single arête. In trapezoidal profiles, the dorsal surface is formed by three planes that create two central ridges; the impact occurred between these ridges on the core platform. Lateral-steep profiles are triangular, where one side forms a right angle. This suggests that the blank was detached at the edge of the working surface (e.g. débordants). Convex cross-sections do not exhibit any planes, but are formed by the cortical or unmodified surface of the dorsal face. Flat profiles show no significant arêtes, either because of flat cortex or because the dorsal surface is formed by the single scar of a previous removal. Irregular cross-sections have more than three vectors randomly arranged, and are typically the remnant of a weathered, unmodified dorsal surface, or complex scar pattern that does not conform to any other attribute state. Longitudinal profile. This attribute describes the blank profile when viewed from the lateral edge—the curvature of the ventral face. Three states are recognized in this analysis: straight, incurvate, and twisted (Figure -). Presumably, the longitudinal profile reflects the relationship between the axis of removal and the convexity of the flaking surface, as well as being affected by the thickness of the Figure -. Longitudinal profiles (adapted from Monigal ) 195 blank. Because the presence and prominence of the bulb of percussion may obscure the observation, curvature is measured from the base of the bulb, so that it does not touch the ventral plane. This category is useful for differentiating Levallois blanks from biface thinning flakes; it is assumed Levallois blanks are detached from a relatively flat working surface and, therefore, will have a straight longitudinal cross-section, while thinning flakes are, by nature, thin and struck from a smaller, curved flaking surface. Twisted blanks are typically elongated pieces that are usually formed when force is applied obliquely across the core surface. They are also a useful indicator of débordant pieces. Distal termination. The state of the distal termination gives information regarding the way in which a blank was detached: the amount of force applied, how force was applied, and possible errors resulting from these previous two variables. When overpassed, the piece yields information about the morphology of the distal end of the core. The eight distal terminations recognized in this study are feathering, hinge/step, overpassed, cortical, blunt, missing, or retouched. Feathering, cortical, and blunt states are considered normal. In the case of feathering, the blow’s force traveling through the working face of the core ended before it reached the distal extremity, resulting in a thin, tapering, distal termination. Feathering terminations will usually occur if the working surface is convex and force is properly applied. Blunt terminations occur when the blank extends all the way to the distal end of the core, but is not incurvate like an overpassed termination. Cortical ends are a subset of this previous category, indicating the distal end is comprised of an unmodified natural surface, either weathered or cortex (Whittaker ). Hinge or step fractures at the distal termination occur when the outward force of the blow exceeds the downward force. This can happen when the percussor is pulled toward the knapper immediately before the impact, or if the force of the blow is not strong enough. This results in a sharp drop in the velocity, which allows the outward force to expand and roll away from the detaching flake (Cotterell and Kamminga ; Andrefsky ). Hinge and step fractures complicate subsequent removals from the working surface because the extra mass hinders force passing through the core during detachment. Overpassed terminations reflect errors in the magnitude of force that was applied 196 during impact (Ferring ; Cotterell and Kamminga ). The force is so great that the blank runs past the edge of the working surface and removes a portion of the distal extremity. While these errors are problematic for the knapper, they are useful for analysis because they carry information about the preparation (or lack thereof ) of the opposed edge of the working surface. Some blanks lack distal terminations, either due to breakage (classified as missing), or modified by retouch after they were detached from the core (classified as retouched). Axis. This attribute describes whether the longest axis of symmetry (viewed from the dorsal face) coincides with the technological axis of flaking (viewed from the ventral face). In other words, axis describes whether or not the distal extremity of the blank follows the vector originating from the point of percussion. A blank can be classified as on-axis, off-axis, or unknown. Cortex percentage. If cortex is present on the dorsal surface, the amount is quantified in one of four groups: -, -, -, and -. This does not include cortex located on the platform (indicated by the “platform type” variable), or at the distal extremity (indicated by the “distal termination” variable). Along with cortex position, cortex percentage is an important indicator for which point along the reduction sequence a piece may fall. In addition, the presence of cortex at the distal end can be useful for calculating the size of the raw material/core. Cortex position. Cortex position describes the location of cortex on the dorsal surface of a blank. The pattern of core reduction and methods of initial core preparation and decortication can be discerned by this variable. The eight categories used in this analysis include: lateral, distal, proximal, central, lateral-distal, lateral-proximal, proximal-distal, and >. Proximal-distal is particularly useful, as it gives information as to the initial length of the working surface. In cases were cortex was present on the distal, proximal, and lateral extremities, the piece was classified as lateral-distal because this is a better important indicator of raw material/core length. Dorsal scar pattern. The dorsal scar pattern attribute is the position and orientation of dorsal facets that represent the remnants of removals prior to when the blank was detached. They reflect the typology of the core, the number of previous removals, and the disposition of platforms on the core at the time the blank was struck. Fourteen scar patterns are recognized in this analysis: unidirectional, 197 unidirectional-crossed, unidirectional-parallel, convergent, transverse, transverse-parallel bidirectional, radial, crested, natural-crested, lateral-crested, transverse-crested, distal-crested, unknown, and none (Figure -). Unidirectional indicates there are facets on the dorsal face that originate from the same direction as the striking platform. Unidirectional-crossed has one or more dorsal scars originating from the same direction as the striking platform, with an additional scar or scars at an angle approaching perpendicular to this. Unidirectional-parallel exhibits arêtes that are parallel to one another along the technological axis. Convergent is a type of unidirectional reduction where the dorsal scars originate from the sides of the platform and converge inward to a point at or near the Figure -. Some dorsal scar patterns used in analysis (adapted from Monigal ). 198 distal end of the flake. Convergent scars tend to create a triangular blank, and are associated with point core technologies. Transverse scars originate only from the lateral edge(s) of the core, indicating the previous blank(s) were struck perpendicular to the technological axis of the piece. Transverse-parallel follows the same pattern, although with multiple parallel arêtes. These types are associated with core correction, maintenance, or a multiple platform core. Bidirectional patterns have scars originating from both proximal and distal ends of the blank. These can either be the product of bifacial thinning, an opposed platform technology, or may represent convexity maintenance by removing short flakes from the distal supplementary platform. The different types can be distinguished by the degree of invasiveness from the opposed platform scar. Radial scars exhibit previous removals originating from three or more platforms. These types are common in centripetal Levallois reduction, discoids, and bifacial thinning (Ferring ). Crested pieces have scars originating along the central ridge of the blank that are directed transversely toward the lateral edges. In lateral-crested, the scars are positioned on the ridge along a laterally-steep edge, directed inward to the middle of the blank. Transverse-crested has a similar morphology to lateral-crested, though with scars directed laterally toward the edge. Distal-crested has scars at the distal end directed inward toward the center of the blank. These four types are all indicative of core maintenance. The classic crested blade—lame à crête—is a diagnostically significant form associated with Upper Palaeolithic technologies (ibid.). The unknown category includes pieces with scars that are randomly oriented, or with retouch that obscures the flaking surface and prevents proper identification of the scar pattern. Blanks completely covered in cortex or with an unmodified, naturally weathered surface were classified as having no dorsal scar pattern. Raw material. It is critical to factor in raw material type when conducting metric analyses; this variable directly affects knapping properties and blank measurements, in addition to providing a wealth of information regarding raw material procurement behavior. Several types were present in the analyzed assemblages, including chert, radiolarite (jasper), silicified limestone, sandstone, and 199 Figure -. Location of measurements used in analysis (adapted from Monigal ). quartzite. Strictly speaking, flint and chert are recognized as geologically distinct minerals, though for the purposes of this dissertation they are similar enough that the terms will be used interchangeably. Part of the COPR  survey included a detailed analysis of lithic raw material types collected throughout Oman, the results of which are presented in Appendix C. Metric variables. The following metric variables were measured on each artifact: platform width, platform height, length, width, thickness, weight, and platform angle (Figure -). Further indices were derived from these basic measurements, including an index of elongation (IE), index of flattening (IF), blank area (BA), blank volume (BV), relative platform size (RPS), relative platform width (RPW), and index of platform flattening (IPF). Platform width is useful for determining the stage of reduction at the time the blank was removed from the core, and is a function of the morphology and preparation of the core. It was 200 measured as the distance between the lateral edges of the platform, parallel to the axis of midpoint width (Odell ). Platform height is affected by the position on the core where the percussor strikes the platform. It was measured from the point of percussion to the intersection of the platform and flaking surface. Numerous replication experiments have clearly demonstrated that there is a relationship between platform size and the length and thickness of the blank; as the height increases, so does the resultant flake size (Speth ; Pelcin ). Maximum blank length is the distance along the technological axis between the point of percussion to the distal extent of the blank. It represents the minimal length of the working surface of the core, and, depending in the morphology of the blank (e.g. overpassed), may provide the actual length of the core at the time of removal (Monigal ). Maximum blank width was calculated as the widest point along the length of the piece, perpendicular to the technological axis. Width is a direct function of core morphology, and, like length, indicates the minimal flaking surface of the core (ibid.). Blank thickness was measured at an arbitrary point along the technological axis that represents the average thickness of the specimen. It has been suggested that blank thickness is an important predictor of flake size and of its place in the reduction sequence (Speth ). Platform angle is the angle formed at the intersection of the ventral face and the striking platform. It represents the angle at which the blank was struck from the core, and is important for differentiating biface thinning flakes from core reduction blanks—presumably éclats de taille will generally have more obtuse angles than blanks struck from cores. As weight closely correlates with the dimensions of the flake, it is a useful indicator of reduction stages (Magne and Pokotylo ). Weight was taken with an electronic balance to the nearest . g. The aforementioned metric observations were used to calculate several composite variables that are useful for analyzing an assemblage. 201 The index of elongation (IE) is calculated as the quotient of length and width, therefore, the higher the value, the more elongated the piece. Index of flattening (IF) is the quotient of width and thickness. Low values indicate relatively massive pieces, as the thickness of the piece approaches its width. High values are representative of thinner, more gracile specimens (Monigal ). Blank area (BA) is the product of the width and length. It imparts an estimate of the overall dimensions of the blank. Blank volume (BV) is the product of width, length, and thickness. BV is a useful predictor of the stage in the reduction sequence, since the variance of this measure decreases as the reduction trajectory advances (Newcomer ). Relative platform size (RPS) is the blank area divided by the platform area. When a platform is small in respect to the size of the blank, this value will be high, whereas low values indicate a more massive platform relative to blank size. This measure provides a proportional indicator of platform size, and, therefore, can be used to compare across multiple assemblages. Johnson () suggests that RPS may also reflect the consistency and accuracy of the knapper. Relative platform width (RPW) is the quotient of blank width and platform width, thereby providing a relative measure of these two variables. RPW indicates the general narrowness or broadness of the platform; high values reflect narrow platforms relative to blank shape, low values suggest broad platforms relative to blank width. Again, since it is a proportional measure, RPW can be used to compare easily across several assemblages. The index of platform flattening (IPF) is the quotient of platform width divided by platform height. High values indicate that platforms are thin relative to platform breadth; values around  indicate that the platform is rectangular in shape; values approaching  reflect a square platform; values less than  represent a narrow and thick platform. TOOLS Tools include any blank that has been secondarily modified after it was removed from the core, or, in the case of Levallois, the preferential product of a prepared core. 202 Tool type. All tools were first classified into a simplified typology using general tool classes found throughout the Pleistocene and Holocene. Because this analysis is concerned with multiple assemblages that occur over several archaeological phases, it was not possible to adhere to any one particular tool typology; instead, an amalgamation of several typological systems has been used (Table -). Each type of tool is defined by a number of morphological traits involving both the blank and nature of secondary modification. This analysis recognized  different tool types, which, in the grossest sense, can be divided into façonnage (bifacial) and unifacial tools. The concept of façonnage— shaping the face—represents a vastly different approach to producing lithic tools than retouching a blank. In contrast to core tools, which are created by striking a core and retouching the subsequent blank, façonnage reduction is achieved by invasive surface flaking across one or both faces of a plaquette, nodule, or blank. There are six types of façonnage tools recognized: handaxes, foliates, points, bifacial tiles, trihedral rods, and preforms. For the purposes of this analysis, they are distinguished primarily by the method of percussor—handaxes are produced by hard hammer only, bifacial foliates use soft hammer percussion to complete shaping the piece, and bifacial points are made by pressure flaking. Bifacial tiles and trihedral rods are special categories because they are unique type fossils within the Arabian Bifacial Tradition. Bifacial tiles are large pieces typically manufactured on limestone or thin chert plaquettes. They exhibit semi-steep bifacial retouch around the entire margin of the piece. Trihedral rods are small, well-made, pressure flaked tools shaped as thin, elongated points, with the crosssection of an equilateral triangle. Though not yet completed, preforms are viewed as a type of bifacial tool and are quite useful for reconstructing the façonnage reduction strategy. Unifacial tools include three types of scrapers classified as such based on the presence of continuous retouch along at least one entire edge of the blank. The retouch ranges from marginal to invasive. The three scraper categories—endscraper, sidescraper, and miscellaneous scraper—are distinguished by the location of the scraper retouch on the blank. Endscrapers are retouched along one or both narrow ends, i.e. along the edge perpendicular to the morphological axis. Conversely, 203 Table -. Qualitative observations used in typological analysis. 204 sidescrapers have retouch along one or both edges lengthwise, parallel to the morphological axis. Miscellaneous scrapers include those pieces with retouch along more than three edges, often forming ovoid if not completely circular scrapers. Denticulates are defined as any blank with continuous denticulate retouch (though if the retouch covers one entire edge it is considered a scraper with denticulate retouch). Notches exhibit at least one area of retouch forming a concave edge, whether from a single blow or multiple flaking; double notches have two contiguous notches. Two types of perforators, awls and drills, are included among the tools. Awls have a bit ranging from approximately one centimeter to several centimeters in length, formed by a variety of retouch types. Drills are typically smaller and have a much more pronounced bit either comprising the entire piece or exhibiting one or two shoulders at the base. Drills were often formed by pressure flaking and were presumably hafted, in some cases they may just be reworked awls. Burins are defined as such based on the removal of at least one burin spall from a blank. The spall is struck down the edge of the piece, forming a chisel edge. Types of burins are further differentiated by the morphology of the platform from which the spall was struck, as well as the number of spalls. Truncated pieces exhibit at least one narrow edge with continuous retouch forming an angle of °. When the same type of retouch is found on the long edge, it is considered backing. A tranchet has a single blow across the face of a blank that forms a chisel edge. Composite tools are those that exhibit more than one unifacial tool form, in any combination. Condition. See debitage attribute analysis. Blank Type. See debitage attribute analysis. Shape. Most forms follow those already presented in the debitage attribute analysis, with the addition of categories used for bifacial pieces. Since the earliest studies of bifacial handaxes by Boucher de Perthe () and de Mortillet (), scholars have produced a number of classificatory schemes to describe the variety of shapes in which bifaces occur. In this analysis, façonnage tool shapes follow the general divisions proposed by Bordes () and Debenath and Dibble (), where 205 metric variables including length, maximum width, distance from the base to the maximum width, width at midpoint, width at three-quarters, and thickness are used to calculate its shape. The shape name corresponds to an index where elongation is plotted against location of maximum width and the flatness ratio. This analysis includes the following shapes, based on the array of bifacial tools recovered: ovate, cordiform, foliate, fusiform, lanceolate, and pedunculate. Ovate bifaces are those forms that are sub-rounded with convex edges around the entire perimeter. They have an elongation ratio (length:maximum width) between . and .. Cordiform— heart-shaped bifaces—possess a rounded base, convex edges, and a slightly pointed distal extremity. Though Borde () distinguishes between typical and elongated forms, this study includes both types within the one category. Debenath and Dibble () point out that this shape is prevalent within bifaces of the Mousterian of Acheulean Tradition. Foliates, as defined by Ulrix-Closset () are leaf-shaped bifaces with symmetrical, convex edges that are pointed at one or both ends. Lanceolate forms have edges that are straight or very slightly convex and a distal point. Fusiforms are thick bifaces with slight points at both ends in the shape of a ships hull (similar to naviforms, which are distinguished by their relative thinness). Pedunculates, in this study, are limited to bifacial points that exhibit some form of tang at the proximal end of the tool. Midpoint. Like shape, there is overlap with the midpoint cross-sections presented in the debitage attribute analysis category. Additional forms used for bifaces include lenticular, biconvex, plano-convex, trihedral, and spheroid. Lenticular is arbitrarily distinguished from biconvex by the relative thinness of the piece. Retouch Type. Retouch type is a qualitative assessment of the nature and invasiveness of the retouch, defined by Borde () and Tixier et al. (). Fine retouch is nibbling along an edge that does not extend more than a few millimeters from the edge; when it is irregular the retouch is often considered usewear, rather than intentional modification. Invasive retouch has scars that extend past the central axis of the tool, while normal (sometimes called scraper retouch) describes any retouch that falls in the spectrum between fine and invasive (Tixier ). Quina is a specific form of retouch that fractures in steps or scales. It is a more extreme form of scalar retouch distinguished by the depth 206 and angle of the detachments, which imply the use of a dense or sturdy percussor. Quina retouch can be applied to thin blanks, but because of the depth of the removals a thick blank is often used. Brézillon () notes that the application of Quina retouch is a characteristic of the Mousterian industry. Denticulate retouch exhibits consecutive serrations formed by multiple notches. Abrupt follows the aforementioned definition of backing/truncation, with retouch forming an angle of greater than °. Spall retouch was used to classify burins and carinated pieces (though there were no tools fitting the latter category). Parallel and sub-parallel retouch is associated with pressure flaking, and is clearly recognizable by the invasive, ribbon-shaped scarring that often leaves bladelet negatives. Retouch Angle. This category is another qualitative observation regarding the edge angle formed by the retouch. Flat retouch is typical of bifacial thinning, with invasive flakes struck from the blank. Semi-steep are non-invasive edges with an angle between -°, steep are those approaching °, and backing/truncation angles form a right angle perpendicular to the working face of the blank. Retouch Duration. Retouch duration describes the continuity of the retouch; continuous is comprised of multiple contiguous blows, irregular are one or more flakes removed from different areas of the blank that form no recognizable pattern, and single describes the single blow that is often used to form notches. Edge Shape. This category describes the shape of the retouched edge when it is formed by continuous retouch. The five conditions that are recognized include concave, straight, convex, converging, and nosed. Retouched Side. Retouched side describes the face upon which the retouch occurs. Obverse retouch is found on the dorsal face, while inverse retouch is struck across the ventral face. Alternating is contiguous retouch that switches between obverse and inverse, though never at the same position on the edge, in which case it would be classified as bifacial. Retouch Position. Retouched position is used to indicate upon which edge the retouch occurs (in relation to the technological axis of the piece). The positions recognized include: proximal, distal, later, bilateral, full margin, and corner. In some cases the position could not be determined (i.e. amorphous chunks), therefore they were classified as unknown. 207 Figure -. CORES Orientation of core platforms (adapted from Monigal ). The study of individual cores themselves does not provide a significant amount of insight, as each specimen represents an end stage of the reduction strategy. Considered as a continuum, however, the core assemblage offers information regarding nodule shaping, initial preparation, maintenance, and eventual discard. The core attribute list used for this study is comprised of five qualitative categories: core type, surfaces, number of platforms, platform preparation, and scar pattern, as well as raw material and simple metric variables including maximum length, maximum width, thickness, and weight. Viewed from the working surface, the cores were conceptualized with four quadrants in a single plane: proximal, distal, left, and right (Figure -). The proximal end is oriented toward the viewer, and is always the major or last platform used to remove a substantial flake (rather than core maintenance). The distal end (or platform) is opposite the proximal platform, and right and left platforms follow as such. The core back refers to the surface opposite the major working surface when the core is turned over. True discoids have no sides, and fully volumetric, where the working surface exploits the entire volume of the core, have neither back nor sides. Core Type. Assigning core types in the traditional sense was difficult for this study as there is no precedent in South Arabia for detailed core analysis, and the assemblages being studied 208 Figure -. Some core types recognized in analysis (adapted from Monigal ). presumably span several technological phases throughout both the Pleistocene and Holocene. Therefore, it was decided to use broad categories that are purely descriptive, with no temporal implication (i.e. Mousterian disk). As a result, the category of core type is only an assessment of the type of blanks struck from the core. These types include general forms such as flakes and blades, and, when identifiable, scars that reflect a specialized technology, such as Kombewa or Levallois (Figure -). It is necessary to reiterate that, for the purposes of this dissertation, the definition of Levallois is conceptualized in the strictest sense: the preferential removal of a blank from a prepared working 209 surface and striking platform, which exhibits radial, bidirectional, or unidirectional convexity maintenance (following Usik ). Core Surface. Every core must have two basic features: a striking platform and a working surface. In non-volumetric, single-surface reduction strategies (i.e. Levallois tortoise cores), the reduction occurs at the intersection of these two planes. These surfaces are not interchangeable, in other words they cannot increase in size at the expense of the other; therefore blank production capacity is limited to the volume contained in the space that occurs at the intersection of these two surfaces. In contrast, within fully volumetric reduction strategies (i.e. prismatic blade technology) the reduction and preparation of striking platforms and working surfaces can be interchanged (Boëda , ). It has been demonstrated in areas contiguous to South Arabia—Africa and the Near East— that the way in which cores are conceptualized, sensu latu, shifts during the Upper Pleistocene from non-volumetric, single surface cores, to a more volumetric approach to raw material reduction. That is not to say cores with more than one flaking surface are absent in the early Upper Pleistocene, but certainly less frequent and the mode of reduction less standardized. In cases of cores with multiple working surfaces, the surfaces are contiguous but unrelated; such cores are considered as having two or more individual working surfaces. For this study, three states are recognized: one working surface, two working surfaces, or partial/fully-volumetric. Number of Platforms. This variable is simply a count of platforms, including both the primary platform and related or unrelated supplementary platforms, whether the core has one or multiple working surfaces. The platform states are described as single, double, or multiple. Platform Preparation. This category describes the condition of the primary striking platform, which in turn yields information about specific methods of core preparation as well as the position of the core within the reduction sequence. Three states are recognized: natural, cleaned, or faceted. Natural indicates the platform is covered in cortex, cleaned platforms exhibit a single blow across most or the entire striking platform, and faceted indicates two or more fracture planes across the platform. 210 Scar Pattern. The scar pattern variable follows the same states previously described in the debitage attribute analysis section. It refers to the patterns exhibited on the working surface associated with the primary striking platform. STATISTICAL PROCEDURES The lithic analysis incorporates basic measures of central tendency—means, median, and mode. In addition, there is some use of statistics of dispersion, i.e. standard deviation and variance. Since the absolute measure of variation in the data are a function of raw material size and the blank’s position in the reduction sequence, they are not directly comparable across assemblages. The coefficient of variation (CV) compares the relative amount of variation in populations with different means; because it is independent of both mean size and the unit of measurement, it is useful for comparing disparate populations. A lower value of CV indicates greater standardization within the population (Sokal and Rohlf ; Drennan ). Discussion The site sampling strategy and methods of lithic analysis presented in this chapter were tailored for the specific goals of the Central Oman Pleistocene Research program—to describe the variety of lithic reduction sequences present in southern Arabia during the Upper Pleistocene, and begin to place them in a chronological context. There were a number of difficulties in achieving this aim, given the paucity of previous scholarship upon which to base the findings of COPR. Furthermore, much of the territory covered during the survey was terra incognita, providing no preexisting insight into raw material availability, site locations, or palaeoenvironmental conditions. The survey methods used by COPR were intended to cover as much ground as possible, and evaluate certain variables useful for determining an area’s archaeological potential, such as proximity to raw materials and water, site position on the landscape, palaeoenvironmental conditions, and postdepositional processes active within the local ecosystem. The sites chosen for sampling were intended to produce a variability of reduction sequences representative of a wide bracket of time. Though this 211 dissertation is focused on Pleistocene lithic industries, a few Holocene assemblages were purposely selected in order to have data with which to contrast the Palaeolithic material. In regards to the lithic analysis, it was necessary to use a typological system that was as descriptive as possible, while avoiding terms that might carry temporal implications. Since the sites may span from the Middle Pleistocene to the Middle Holocene, a flexible system of classification was employed with several descriptive variables, rather than a single category comprised of preconceived tool types. There is an indefinite number of observable technological attributes within a lithic assemblage; those that have been chosen for this study of technology were aimed at establishing a general description of the reduction sequence(s) present at each site and to begin to identify the characteristics of Middle/Upper Pleistocene lithic industries in Oman. There were a maximum of  basic qualitative and quantitative observations recorded from each blank,  on each tool, and  on each core. Reduction strategies cannot solely be extrapolated based on these morphological attributes, each assemblage is the product of external variables as well: palaeoenvironmental conditions; site size, function, and length of occupation; distance to and quality of raw material; and tool production, use, and discard. The following chapter will analyze the data collected during the COPR  archaeological activities. The analysis will consider the technological and typological variables presented in this chapter, in combination with the available palaeoenvironmental and geological data. By isolating significant technological elements in each assemblage—observable patterns within either individual variables or a suite of characteristics—it is possible to classify each site based on the reduction strategies present, and use that data to examine techno-typological affinities in Africa, India, and the Near East. 212 Chapter 7 ANALYSIS OF LITHIC ASSEMBLAGES You see, I went on with this research just the way it led me. That is the only way I ever heard of research going. I asked a question, devised some method of getting an answer, and got—a fresh question… —H.G. Wells, The Island of Dr. Moreau This chapter examines the lithic assemblages recovered during the COPR archaeological activities. The data from each assemblage are presented seperately, including qualitative and quantitative descriptions of debitage attributes and tool types, following the methodology described in Chapter . From these observations, the reduction strategies at each site will be described. Following the assemblage descriptions, there are inter-site analyses of the various chaînes opératoires. Four different lithic industries, sensu latu, are defined. Techno-typological attributes of these industries are described and placed within a tentative chronological framework. Tables with relevant information are located at the end of this chapter, while the raw lithic data appears in Appendix A. Qarat al-Kibrit  (A), Phase I The lithic assemblages from Qarat al-Kibrit  were divided into three archaeological phases based on the material’s stratigraphic position within distinct geological units. These stratigraphic divisions are corroborated by refits made within layers but not between them, as well as by related features on raw materials within each layer that indicate they are from the same source. Phase I material was recovered from the surface and within a thick carpet of unconsolidated/ semi-cemented sediments, reaching a depth of approximately  cm; consequently, the assemblage is not necessarily associated with a single period of occupation. There is a noticeably higher distribution 213 of artifacts downslope from the dolomite block just north of square D (Figure -). Presumably, this is because water draining into the dome was redirected by the boulder, thereby preventing the lithic material from being washed away. Additionally, human activity at the site may be associated with the presence of the dolomite block itself (e.g., protection from wind and sun, as well as concealment for hunting). Although the location of Qarat al-Kibrit  suggests it may have served as a basecamp, the -:- cortex percentage ratio is ., meaning there was a relatively high percentage of cortical pieces (Table -). So, while the site may have served multiple functions, one of those functions seems to have been early stages of manufacture of lithic artifacts. Figure -. Horizontal distribution of artifacts at Qarat al-Kibrit , Level I. 214 Figure -. Radiolarite cobbles embedded in gypsum just outside of Qarat al-Kibrit salt dome. Artifacts from the Phase I assemblage are derived primarily (.) from reddish-colored, fine-grained radiolarite pebbles that are ubiquitous throughout the alluvial gravels surrounding the salt dome (Figure -); for detailed lithology see LRM - in Appendix C. This is followed by an extremely fine-grained yellow radiolarite that is derived from the same source (), distinguished by its excellent quality and different color (LRM -). Workable limestone, occurring within the dome, comprises . of the assemblage. There are trace amounts of quartzite (.) and a white chert (.), both from unknown, non-local sources. The closest outcrop of quartzite is Jebel al-Akhdar, approximately  km to the north. The aforementioned low cortex percentage ratio (.) of Phase I is indicative of the site’s proximity to radiolarite gravels just outside the salt dome, which provided the predominant source of raw material for the Phase I toolmakers. It is also noteworthy that non-local raw materials are somewhat more frequent in Phase I than the underlying Phases II and III. 215 Figure -. Blade cores from Qarat al-Kibrit , Level I. 216 DEBITAGE AND CORES The Phase I assemblage is comprised of  pieces, of which  are debitage and  cores (Table -). The reduction strategy can generally be described as a crude prismatic blade technology. Twelve of the cores are flake-blade/let, two are flake cores, and one is a blade core. Twelve are single platform cores that exhibit unidirectional-parallel scar patterns on one or more working surfaces, and the remaining three are opposed platform (Figure -:b). Most of the cores () have some degree of edge-grinding at the intersection of the dorsal face and striking platform. Among the identifiable blanks, blades comprise . (Figure -), which is quite prominent (Table -). There are five core trimming elements, four of which are core tablets (Figure -:e), indicating this method was used for rejuvenating striking platforms. Six blanks are éclats de taille, although no bifacial pieces were found in this assemblage. It is possible the presence of these pieces is either the result of some form of turbation bringing artifacts up from underlying layers, or their bifacial tools manufactured or resharpened at the site were carried away. Approximately  of the blanks exhibit some degree of edge grinding (Figure -:a,c), and nearly three quarters have lipping along the bulb of percussion. Furthermore, bulbs tend to be diminutive; all suggesting that most of the flaking was carried out by soft hammer percussion. Regarding platform morphology: . are straight, . are cortical straight/curved, and only . are dihedral or faceted (the remaining . are either crushed or missing). Dorsal scar patterns reflect those of the cores: the most frequent is unidirectional, followed by unidirectional-parallel, unidirectional-crossed, convergent, radial, lateral-crested, and natural-crested, respectively (Table -). While there are three bidirectional cores, none of the blanks exhibit evidence for bidirectional flaking. TOOLS There are just seven tools in the Phase I toolkit. Figure -:a depicts a unifacially retouched composite tool, including burin, sidescraper, and double-notch. -:b is an awl made on a cortical blade with steep obverse retouch, and -:c is a sidescraper made on a blade. Other tools include an atypical endscraper on a flake, a notch and double-notch on a flake and cortical flake respectively, 217 Figure -. Blades (a — d) and core tablet (e) from Qarat al-Kibrit , Level I. and a burin made on a blade. The burin had three burin blows, the first was a flake removed from the platform onto the ventral face, followed by a blow from the proximal end down the lateral edge, and a final blow across the platform—these three removals intersect to form a single burin edge. Although there are only seven tools, the fact that over half of them were manufactured on blades supports the assertion that the production of blade-proportionate blanks was the primary reduction strategy at the site. 218 Figure -. Tools from Qarat al-Kibrit , Level I. Figure -. Perforated shell bead from Qarat al-Kibrit , Level I. 219 In addition to lithic artifacts, a shell bead was found in association with the Phase I assemblage (Figure -). The shell was discovered near the top of geological Unit B, in lightlycemented sands. It is the apex of a Conus sp. shell, which has been sliced off and perforated (Mienis, personal communication). Conus sp. are marine gastropods; thus, the nearest source would be either the Arabian Gulf or Arabian Sea, both approximately  km away. The shell yielded an  date of , ±   (Wk-), providing a terminus ante quem for the Phase I assemblage. The manufacture of perforated shell beads, particularly using Conus sp., was common among coastal sites throughout the Holocene (e.g., Barker ; Tosi and Usai ; Uerpmann and Uerpmann ). The presence of a bead at Qarat al-Kibrit, Level I suggests that either the inhabitants’ mobility strategy included the littoral zone, or there was some degree of exchange with coastal populations. It is possible the Phase I occupation at Qarat al-Kibrit was associated with exploiting halite deposits found within the crater as part of an early th millennium  interaction sphere Figure -. A. al-Mahrooqi points out evidence for halite mining within one of the Qarat alKibrit cave. 220 connecting the coast and hinterland. Even today, halite salt extracted from the cavities inside Qarat al-Kibrit is prized among Omanis for its purity (Figure -); local folklore holds the crater has been a source of salt crystals for medicinal purposes well into the past (al-Mahrooqi, personal communication). Perhaps the Qarat al-Kibrit salt deposits were mined and exported for trade as early as , years ago. Qarat al-Kibrit (A), Phase II Phase II material occurs exclusively within geological Unit C, and is comprised of fewer artifacts than the overlying assemblage (Table -); artifact distribution appears to be one dense scatter concentrated around square D (Figure -). Like Phase I, most of the material is concentrated immediately downslope from the dolomite boulder. By far, the most frequent raw material is red radiolarite (.), followed by scarce amounts of limestone (.), yellow radiolarite (.), white chert (.), and quartzite (.). Raw material procurement was much more localized than in Phase I, as well as there being a different reduction strategy. These observations, in conjunction with the fact that Unit B unconformably overlies the truncated surface of Unit C, indicate that Phase I and Phase II belong to different technological periods and were deposited under different environmental regimes. The dating of these two phases correlates with the palaeoenvironmental record: the Early/Middle Holocene wet-phase ended around , , placing Phase II sometime within the sub-pluvial episode, followed by Phase I right at the onset of greater aridity. Sediment samples were taken from Unit C and Unit D for OSL dating. Results are pending from the Max Planck Institute in Leipzig, Germany. Like Phase I, the -:- cortex percentage ratio (.) suggests that early stages of decortification were carried out on site (Table -). Also similar to Phase I, tools make up a miniscule portion of the overall assemblage, comprising just .. The assemblage from Phase II seems ephemeral—a small lithic workshop, along with some hunting/game-processing activities. 221 Figure -. DEBITAGE AND Horizontal distribution of artifacts from Qarat al-Kibrit , Level II. CORES There were  pieces of debitage and six cores in this assemblage. All six cores exclusively exhibit flake scars, primarily with simple unidirectional scar patterns. One core has multiple platforms and alternating scar patterns; an ad hoc prismatic approach to the core’s volume. Figure - depicts the core with several refit flakes, illustrating this seemingly haphazard reduction strategy. Two of the specimens are cores-on-flakes, and can be categorized as Kombewa, following the broad definition of this category suggested by Usik (). One such Kombewa core from this assemblage was almost completely refit, illustrating the multiple platform, alternating reduction strategy (Figure -). The percentage of blade-proportionate blanks is dramatically less than in the overlying stratum (Table -), another indication of the markedly different core reduction practiced in Phase II. There are ten éclats de taille (Figure -), although the raw material does not match that of the distal point fragment also recovered from Phase II. Scar patterns show a significantly decreased percentage of unidirectional-parallel, with far more unidirectional and unidirectional-crossed (Table -). There were five lateral-crested pieces, in which scars are positioned along the ridge of a débordant blank, 222 Figure -. Level II. Refit multiple platform core with alternating reduction strategy from Qarat al-Kibrit , directed inward toward the middle of the piece. Although these pieces were not classified as typical core trimming elements, they suggest either platform maintenance to rejuvenate a striking platform, or reformation of the convexity around the working surface. Approximately a quarter of the blanks have edge grinding, and two-thirds are lipped. Most of the striking platforms are straight (.), cortical-straight (.), or cortical-curved (.). There are very few dihedral (.) or faceted (.), which would indicate platform shaping. It 223 Figure -. Refit Kombewa core from Qarat al-Kibrit , Level II. 224 Figure -. Biface thinning flake from Qarat al-Kibrit , Level II. is interesting to note that the percentage of mistakes, expressed by hinge/step/overpassed distal terminations, is significantly higher in Phase II than any other assemblage considered in this analysis (Appendix A). This may be further indication of the haphazard, ad hoc core reduction strategy. TOOLS There were three unifacial and two bifacial tools in the Phase II assemblage. The unifacial tools consist of an atypical endscraper on a primary flake (Figure -:a), a denticulate (Figure -:b), and a retouched flake with obverse, marginal retouch. Both bifacial tools are diagnostic of the Arabian Bifacial Tradition. One is the distal fragment of a bifacial point (possibly lanceolate). The piece was made from red radiolarite, and exhibits thinning via both soft hammer percussion, as well as very-fine pressure flaking at the distal extremity (Figure -). Although broken, the shape and retouch of the bifacial point fit into tool Type  of the Rub al-‘Khali Neolithic (Edens ). The second bifacial piece is a “bifacial tile,” a tool form recognized by Kapel () from the Danish Expedition to Qatar. The specimen is made on a large slab of dolomite limestone that outcrops within the salt dome. The piece is flat, about . cm in thickness, nearly  cm long, and  cm at its widest point. The tool is elongated-triangular in shape, and biconvex in cross-section. Retouch consists of semi-steep bifacial scraper retouch along the entire distal margin, forming a convex working edge. According to the work of the Danish Expedition, these tools are diagnostic of the ABT, supporting the correlation of the Phase II assemblage with the Early/Middle Holocene pluvial. 225 Figure -. Unifacial tools (with refits) from Qarat al-Kibrit , Level II. Figure -. Distal fragment of a Type  pressure flaked lanceolate from Qarat al-Kibrit , Level II. 226 Qarat al-Kibrit (A), Phase III Phase III is the lowest archaeological stratum at Qarat al-Kibrit. The material occurs within Unit D, a thin layer of cemented dolomite gravels indicative of an intense, short-term episode of increased rainfall. Like both overlying phases, the artifacts are primarily distributed within a small, conscripted scatter downslope from the dolomite boulder (Figure -). The assemblage is comprised of  pieces, of which  are cores, tools, and debitage, and the remaining  are debris. Of the  chips,  can be classified as éclats de taille, a byproduct of pressure flaking (Figure -:a). This is not surprising, given the presence of a pressure flaked bifacial point found in this stratum, described further below in detail. Yellow radiolarite is by far more prevalent than in the previous phases, comprising nearly  of the assemblage; red radiolarite totals approximately , followed by a small amount of limestone (.). While the red and yellow radiolarite pebbles appear to grade into one another in a single continuum, the yellow specimens are extremely fine-grained and virtually free of inclusions, in general much higher quality. Because of the specialized bifacial reduction strategy apparent in Phase Figure -. Horizontal distribution of artifacts from Qarat al-Kibrit , Level III. 227 Figure -. Biface thinning chip (a) and flakes (b — e) recovered from Qarat al-Kibrit , Level III. III, it is possible the flintknappers consciously chose better raw materials than the somewhat crude reduction strategies seen in Phases I and II. Similar to both of the preceding phases, the -:- cortex percentage ratio is extremely low: . (Table -). So, while the reduction strategies may be different, the procurement strategies remain the same. Phase III toolmakers obtained radiolarite cobbles from the surrounding alluvial plain, and subsequently carried out early stages of decortification at the site. 228 DEBITAGE AND CORES The Phase III assemblage is dominated by façonnage technology; there is almost no evidence for core reduction. Two ambiguous pieces were collected that can be interpreted multiple ways, in light of the related diagnostic toolkit and Holocene date of the site. Initially, they were both classified as Levallois cores (or at least flat, lineal cores), in that they demonstrated bilateral/radial convexity maintenance on both faces and some degree of preparation along the primary striking platform. One such specimen is illustrated in Figure -, showing radial flaking on one face and bilateral thinning on the opposed face (with lateral refit). The final removal was initially interpreted as an overpassed Levallois flake. In light of the Holocene date of the site and associated techno-typological features in the assemblage, the specimens are considered early stage preforms rather than cores. It appears the Figure -. Bifacial preform? (with refits) from Qarat al-Kibrit , Level III. 229 overpassed longitudinal blow removed far too much of the preform, and so the piece was discarded. There was another example of this longitudinal thinning technique observed at T, which also falls within the Early/Middle Holocene. Among the debitage, blades are virtually nonexistent, while éclats de taille comprise a third of all blanks (Table -). This is not surprising, given the number of preforms. At ., the Phase III assemblage has the highest incidence of lipping of all assemblages analyzed. Edge preparation, as well, is relatively frequent on . of blanks. The prevalence of façonnage reduction is exhibited in both striking platform morphology and dorsal scar patterns, where radial scars jump to . (Table -). Unmodified straight, cortical straight, or cortical curved platforms are far less frequent than the preceding phases, accounting for . of blanks, while dihedral and faceted forms increase to .. TOOLS The Phase III toolkit is heavily bifacial. There were seven preforms (including the two aforementioned specimens). Most are from early stages of reduction, and all are on relatively flat radiolarite plaquettes. Figure -:a depicts a very early stage preform, in which only a few exploratory blows were carried out before snapping the piece, while Figure -:b shows a specimen that is somewhat further along in reduction, both sides have undergone some degree of thinning. The only completed tool was recovered on the very last day in the field, while taking sediment samples for OSL dating. The piece is a medial fragment of a pressure flaked, bifacial pedunculate, falling into Type  of Edens () typological scheme (Figure -). The piece is extremely well-made, with parallel ribbon-flaking on both faces, forming a lenticular cross-section. While the proximal and distal ends are snapped off, enough of the butt remains to indicate there was once a tang there. The form and quality of pressure flaking are both indicative of the desert Neolithic. The point is made on yellow radiolarite, although seemingly from a slightly different source than the associated Phase III debitage. 230 Figure -. Early stage bifacial preforms from Qarat al-Kibrit , Level III. Figure -. Medial fragment of pressure flaked pedunculated arrowhead (Type ) from Qarat alKibrit , Level III. 231 Based on these techno-typological characteristics, this archaeological phase is interpreted as an ephemeral hunting camp, in which bifacial points were manufactured and curated. There is no other reduction strategy besides the façonnage production of points; therefore, it seems likely that human activities were fairly limited at the site. Based on the geomorphology of Unit D, which suggests short-term, intense rainfall, it is possible Neolithic hunters were drawn to the salt dome because of water pooling inside, or to exploit animals attracted to the salt deposits. Wadi Qilfah  —  (A, A, A, A) A total of  pieces were recovered from findspots A-A (Table -). While each locality was collected individually and given a different findspot number, following the analysis it was decided that all four scatters exhibit the same or similar variability of typological and technological elements to warrant consideration as a single assemblage; however, it cannot be assumed the assemblage represents a single technological phase. Findspots A-A are single scatters of lithic artifacts probably associated with one or a few reduction events (Figure -), whereas A consists of multiple conscripted loci in which a high Figure -. Horizontal distribution of artifacts at findspot A. 232 Figure -. Horizontal distribution of artifacts at findspot A. density of artifacts are present in at least two separate areas (Figure -). The complex of findspots are clearly linked to the exploitation of locally available chert, silicified limestone, and silicified sandstone that outcrops from an early Miocene crust mantling the landscape. There is little doubt as to the site complex’s function as a workshop, given the blank to core ratio of ., which is the lowest among all eleven assemblages considered in this analysis. This is corroborated by the distribution of the four cortical percentage groups (Table -), which yields a :- cortex percentage ratio of ., also one of the lowest among the analyzed assemblages. The predominant raw material is silicified limestone (.), a medium-quality limestone ranging in color from gray to orange (for comprehensive lithology see LRM samples - and - in Appendix C). This is followed by a course-grained, gray silicified sandstone (.) represented by LRM sample -, and, lastly, a fine-grained brownish-orange chert that is local, albeit rare (.). There are trace elements of radiolarite from nearby alluvial gravels (.), as well as coarse-grained 233 limestone (.). Six pieces, manufactured from silicified limestone, showed a much higher amount of weathering and varnish and may belong to an earlier period of occupation. DEBITAGE AND CORES Attribute analysis of the debitage and cores indicates there were at least four different types of reduction within the assemblage: ) simple, unidirectional production of flake and blade blanks, ) unidirectional-convergent Levallois point production, ) lineal, preferential point production on the narrow elongated face of silicified limestone plaquettes, and ) façonnage technology to create large, flat, limande bifacial implements. It is argued types , , and  all belong to a single continuum of core reduction. All four categories utilize hard hammer percussion exclusively—less than one per cent of blanks exhibit lipping, bulbs of percussion are quite pronounced, and the retouch on the bifacial tools and preforms is typically heavy and invasive. Approximately  of the blanks bear evidence for edge grinding to prepare the platform prior to the removal. Most of the local silicified limestone occurs as thick plaquettes ( –  cm thick); the most common form of core reduction at the site was to utilize the elongated, narrow face of the plaquette to create the working surface of the core. As a result, the production of blade-proportionate blanks (i.e. blades, bladelets, cortical blades, and débordant blades) accounts for . of all measurable debitage; a relatively high percentage. The prevalence of blade production is also indicated by the frequency of cores that exhibit blade/let scar patterns; . of the cores (Table -). The prevalence of débordant pieces struck from the edges of the narrow face of cores, expressed by the high percentage of blanks with a lateral-steep midpoint cross-section (.), provides further evidence for this particular method to rejuvenate the working surface of the core. This reduction strategy is evident in the refit blades depicted in Figure -. That is not to say there is a “true” blade technology present at the site, in the sense of volumetric and recurrent cores. On the contrary, blades were almost always removed from single platform, lineal cores (defined as one working surface on an axis perpendicular to the striking platform). The scar patterns exhibited on both the debitage and the primary working surface of the 234 Figure -. Blade refits from A showing removal of débordants for convexity maintenance. core are predominantly unidirectional, unidirectional-parallel (Figure -), and unidirectionalconvergent (Figure ). Most cores are single platform (.), followed by a few with opposed platform cores (.), with the rare occurrence of multiple platform cores (.). In most cases, the opposed platforms are associated with different working surfaces, only . of all cores can be classified as truly bidirectional. Regarding platform preparation, in nearly every case the surface has been cleaned of cortex, although faceting is rare. There is little evidence (.) for edge grinding at the intersection of the striking platform and dorsal surface, and some of these specimens may, in fact, be caused by aeolian abrasion. 235 Figure -. Unidirectional-parallel blade cores from the A-A complex. In addition to blades from elongated narrow working surface of silicified limestone plaquettes, there is also evidence for the Levallois technique. Fifteen of the  cores recovered from the A-A complex are Levallois, sensu stricto, of which  exhibit a unidirectional-convergent scar pattern on the working surface of the core (Figure -:b). Only two cores bear evidence of radial preparation to maintain surface convexity. In nearly every case, the cores are roughly triangular in 236 Figure -. Unidirectional-convergent blade cores from A-A complex. 237 shape and the working surface is located on the wide, flat surface of the plaquette rather than on the narrow edge (in contrast to the blades). Among the debitage, there were only five blanks (.) that could be classified as Levallois. These specimens include rectangular, triangular, and ovoid shapes. Three of the five Levallois blanks exhibit either dihedral or faceted platforms, one exhibits a straight platform, and the final piece is crushed and, therefore, unidentifiable. Scar patterns on the Levallois blanks are similar to the distribution on the cores;  are converging, while the remaining  are radial. The paucity of Levallois flakes or points may be due to the nature of the site as a workshop, it is likely the desired end-products were carried away from this locality. The third form of reduction at the A-A complex is a strategy that falls between the blade production and Levallois technique. In some cases, flint knappers utilized the narrow working surface of plaquettes to produce unidirectional-convergent points (-:a). In these examples, débordants are removed from the edge of the plaquette to create convexity, followed by a preferential blank bearing the Y-arête that is diagnostic of Levallois points. Both methods represents a lineal approach to the volume of the core, with the preferential removal of a single end-product. Platform preparation varies between cortical, decorticated, dihedral, and faceted. The three modes of core reduction employed at the A-A complex grade into one another, therefore, I argue they belong to a single continuum of core reduction. The fourth strategy is markedly different from the aforementioned methods of core reduction—the façonnage creation of large bifacial implements (Figure -). While éclats de taille only comprise . of all blanks (Table -), . of the toolkit consists of bifaces or bifacial preforms. Similar to the core reduction technology described above, hard hammer was the only method of percussion used to create these bifacial tools, as indicated by the deep scars, pronounced bulbs, and complete absence of lipping. The tools were reduced from large silicified limestone plaquettes and, to a lesser degree, from silicified sandstone nodules; there is not one example of a biface manufactured from a blank. Some of the bifacial preforms and related debitage show scars and breakage patterns similar to the intentional “plunging” technology proposed by Charpentier (). The specimen illustrated 238 Figure -. Bifacial tools from the A-A complex. 239 in Figure -:b exhibits the scar from a large overpassed flake extending across the entire face of the preform. The piece is snapped in half along this scar, indicating that that final overpassed flake caused the piece to break along its weakened lateral axis, due to excessive thinning. The piece was discarded immediately after the break, suggesting it was not intentional but an accident that typically occurs while thinning preforms. Additional “plunging” flakes that have been refit back onto preforms are illustrated in Figures - and -. In both cases, the presumably incomplete preform was discarded following the removal of the overpassed flake, again suggesting these were unintentional mistakes that resulted in the termination of the work on the preform before it was completed (contra Charpentier ). TOOLS The frequency of tool types is presented in Table -. Unifacial tools comprise the majority of the toolkit (.), while bifacial tools account for the remaining .. A brief description of each category is given below: Sidescrapers (). Two of the sidescrapers are manufactured on flakes, two on blades, and one on a débordant piece. The retouch is all semi-steep scraper retouch, four of which form a straight working edge and one convex. In four cases the retouch is obverse, and one has alternating retouch. One specimen is a bilaterally retouched sidescraper. Notch/Double Notch (). There were four notches and one double notch. The single notches are all obverse, three of which are located on the lateral edge and one on the distal extremity. The double notch exhibits alternating retouch and occurs on the lateral edge. Denticulates (). Five denticulates are made on flakes; one is manufactured on a blade. Four exhibit semi-steep retouch and two are steep. Half are retouched on the obverse face, and half on the inverse face. Interestingly, four of the denticulates are made on sandstone, which is significantly higher than the overall frequency of sandstone within the assemblage. It is possible the function of the denticulate was better served by the use of a coarse-grained raw material rather than the finer-grained silicified limestone. 240 Figure -. Early stage bifacial preform with refit from the A-A complex. 241 Figure -. Biface with overpassed flake refit from A-A complex. Burins (). Both burins are dihedral. One is on a large thick flake. The two burins blows are located at the proximal end and originate from the opposed faces of the lateral edges. The other burin is made on a cortical flake, and exhibits two blows, both originating from one lateral edge, which run across the distal extremity. Levallois flakes/points (). See description in Debitage and Cores section. Truncation (). The single example of a truncation was manufactured on a transverse flake. The truncation retouch is found along the dorsal face of the distal end, forming a straight edge. 242 Tranchet (). This piece was made on a large flake and exhibits partial backing along one lateral edge. The backing consists of irregular inverse retouch along the already naturally-backed edge. There was a single tranchet blow struck across the proximal end of the ventral face. Retouched Pieces (). Seventeen retouched pieces were made on flakes (.),  on blades (.), four on cortical flakes (.), one on a débordant (.), one on an éclats de taille (.), and two on chunks (.). In  cases the retouch is marginal, four exhibit regular retouch, and one is invasive. Seventeen specimens have inverse retouch,  obverse, eight alternating, and one bifacial. By far, most retouch occurs on the lateral edge (), followed by distal (), proximal (), bilateral (), and full margin (). Blades and points with continuous, marginal retouch along the lateral edge (Figure -) are considered one defining characteristic of the Sibakhan industry, to which A-A are attributed. Bifaces (). Of the  total bifacial tools, only two are completed, based on symmetry, regularity of the edges, as well as degree and quality of flaking. The first piece has an index of elongation of . and an index of flattening of .; since the IE is greater than . and IF greater than ., the tool is considered flat and elongated. It falls within Zone III of Borde’s bifacial scatterplot (Borde ), within the category of “elongated cordiform” (Figure -:a). The second biface has in IE of . and an IF of ., therefore is also flat and elongated. It falls within Zone IV, and both distal and proximal edges are rounded, thus, it is limande in shape. Both bifaces are biconvex in cross-section, and are manufactured from silicified limestone. They are both slightly assymetrical, and exhibit partial fine retouch on one side, thus can be classified as bifacial scrapers. Bifacial preforms (). The recovered preforms display an even distribution among very early through very late stages of production. The shapes all roughly range between ovate and cordiform; five are biconvex, three plano-convex, and one spheroid in cross-section. Eight are made on silicified limestone and one on sandstone. In over half the cases, the preform was laterally snapped and subsequently discarded. Because certain technological features within the A-A complex are diagnostic of the late Middle/Upper Pleistocene (i.e. unidirectional-convergent Levallois method), this assemblage is 243 Figure -. Blades and points with continuous, marginal, lateral retouch from A-A. critical for addressing techno-typological affinities during this period. The question remains as to the relationship between the core and façonnage strategies. Due to the fact that it is a surface site, it is difficult to determine the degree of homogeneity; however, the patina and degree of abrasion do appear more or less equal on the entire assemblage. 244 One means of addressing the question of homogeneity is to consider the assemblage recovered from Saiwan (Biagi ), which is located  km southeast of the A-A complex. The Saiwan bifaces fall roughly within the same size range as A-A ( –  cm long,  –  cm wide,  –  cm thick), and include similar shapes, such as elongated cordiforms and lageniformes (ibid.:fig. : -, pp. ; fig. :-, pp. ). The blade cores at Saiwan also match A-A; both assemblages exhibit a simple unidirectional reduction strategy on the elongated face of plaquettes, with minimal platform maintenance (ibid.:fig. :-, pp. ). Sidescrapers with continuous, marginal retouch, also similar to A-A, comprise the predominant unifacial tool form at Saiwan (ibid.:fig. :-, pp.). Since these two assemblages exhibit a nearly identical suite of techno-typological features, it is argued that AA, indeed, are homogenous, and these assemblages represent occurrences of the Sibakhan Industry. Wadi Qilfah  (A) Wadi Qilfah  (A) is a very small conscripted lithic scatter some ten kilometers southeast of the A-A complex. The assemblage is comprised of  pieces belonging to a single episode of reduction, based on homogeneity of the raw material and that approximately half of the debitage was refit onto the three cores that were collected. The findspot has a -:- cortex percentage ratio of . (Table -), which is among the highest of all analyzed assemblages, indicating very little decortification was carried out on site. This is not surprising, given that the entire assemblage is derived from a fine-grained brownish-orange chert (same as chert specimens from A-A), which is not available in the immediate vicinity around A. In this light, the A assemblage must fall within a later stage of the reduction sequence, after raw material had been procured and initial decortification was carried out. DEBITAGE AND CORES Attribute analysis and artifact refits indicate the technology was exclusively unidirectionalparallel and bidirectional-parallel blade production. Similar to A-A, convexity is achieved by the removal of débordant flakes and blades at the edge of the working surface (Figure -). This is clearly 245 Figure -. Blade refits from A showing recurrent unidirectional method for achieving convexity. 246 represented by the high percentage of blade-proportionate blanks and débordant pieces (Table -). There is not a single instance of lipping, indicating that the only type of percussion was hard hammer. Striking platforms are predominantly straight (.), followed by cortical/dihedral 1⁄2 cortex (.), with a single instance of faceting (.). The remaining . of platforms are either crushed or missing. This same degree of platform preparation is apparent from the cores as well—in all three cases, cortex has been either completely or partially removed from the platforms, but there is no further modification. Approximately a third of the blanks exhibit some degree of edge grinding at the intersection of the dorsal face and the striking platform. The presence of a core trimming element with a transverse-crested scar pattern indicates that platform maintenance was carried out. Scar patterns on the dorsal face of debitage and working surface of the cores are presented in Tables -. Clearly, the dominant category is unidirectional-parallel. Two of the cores exhibit a lineal, recurrent method of reduction in which parallel-edged blades were removed from a single working surface. The opposed platform core represents a volumetric approach to core reduction; blade scars are found along three of the four faces of the core (Figure -). The working surfaces associated with the opposed platforms are slightly askew to one another, and are not bidirectional in the strictest sense. Figure -. Blade core from the A findspot. 247 TOOLS There were only two tools recovered A—two retouched pieces made on blades (Figure :b). Both specimens have continuous obverse marginal retouch, in one case forming a straight edge; the other piece has a convex edge. Retouched blades fitting this description are consistently found in every Sibakhan assemblage, and are therefore considered a significant characteristic of this industry. Ghrain Bliss (SH) The lithic material recovered from Ghrain Bliss comes from a small, relatively low-density scatter of artifacts on the side of a large inselberg associated with Wadi Mahwis. The scatter is more or less confined to just four square meters, centered on D (Figure -). Raw material types include a fine-grained grayish brown Rus Formation chert that exhibits banding and mottling (.). There is also a semi-translucent, milky gray chert that is moderate-grained and completely free of inclusions (.), and artifacts made on a grayish-blue workable limestone (.). All three raw materials are locally available; there are Rus chert layers interstratified within the inselberg and capping the top. Figure -. Horizontal distribution of artifacts at SH. 248 In contrast to the other assemblages presented thus far, the -:- cortex percentage ratio is relatively high (Table -). This is unexpected, given that chert nodules are ubiquitous at the top of the inselberg just ten meters above SH. Perhaps initial decortification was conducted when the nodules were first selected, and then later stages of flaking were subsequently carried out on site. The location of the site on a low terrace is conducive to hunting game passing through the wadi; according to our Beduin guide the spot is still used today to hunt gazelle (al-Mahri, personal communication). In this light the SH findspot may be an ephemeral hunting camp, although the lithic toolkit lacks any evidence for projectile point manufacture. If the site dates to the historic period, then their absence may be explained by the use of an alternate material (e.g., metal, wood, or bone). DEBITAGE AND CORES There were  pieces collected from SH, including  cores,  tools, and  unretouched blanks (Table -). The general reduction strategy seems ad hoc, all of the cores have multiple working surfaces, unprepared platforms, and exhibit alternating blows across the working surfaces. Furthermore, the cores are manufactured from brittle, low-quality chunks that are riddled with fracture planes. The scar patterns on the working surfaces are predominantly flakes that are unidirectional or unidirectional-crossed (Table -). One core has evidence for blade removals, although this seems more coincidental than intentional. Nearly  of identifiable blanks are flakes, with blades comprising just . of the assemblage (Table -). Like the cores, the predominant dorsal scar patterns are unidirectional and unidirectional-crossed (Table -). Approximately three quarters of the striking platforms are straight or cortical, while only about  have dihedral or faceted platforms. Lipping is almost completely absent and bulbs are pronounced, suggesting the almost exclusive use of hard hammer percussion. The relatively high incidence of chunks from shatter is probably a combination of both this method of percussion and the brittle raw material. 249 There was one piece of debitage that appears to be intrusive, an elongated point (ie=.) with convergent scars, made on fine-grained reddish chert. Given the ubiquity of Nejd Leptolithic scatters throughout the region, as well as the blade industry associated with the early Holocene production of Fasad points, intrusive elements at this locality are likely. TOOLS There were five tools at SH, including three retouched pieces, a notch, and a scraper (Table -). The retouched pieces are all made on flakes and have inverse marginal retouch. The notch is made on a flake as well, and has obverse retouch. The scraper is on a limestone chunk and exhibits converging bilateral retouch. Every feature of the SH assemblage is indicative of a crude, ad hoc technology. The raw material selected is relatively poor and the core reduction haphazard across multiple working surfaces that show an alternating flaking strategy. Retouch on the tools is quite poor, and there are no formal types. The site is interpreted as an ephemeral hunting camp where local, low-quality raw material was exploited. Bir Khasfa (SH – SH) The expansive lithic scatter at Bir Khasfa was first discovered by J. Pullar while surveying for prehistoric sites in the ’s. Pullar () reported a plethora of soft hammer bifacial foliates, leading the COPR project to return to Bir Khasfa during both the  and  seasons to further investigate the site. Two findspots were collected, SH and SH, roughly  meters apart. The areas were chosen because they exhibited the same reduction strategy, albeit early and late stages (respectively). At SH, the distribution of artifacts is moderate, with most pieces concentrated in the northern portion. SH is slightly denser, with a more or less even distribution of artifacts (Figure -). All of the lithics from Bir Khasfa are made from local, medium to high-quality Rus chert that occurs as both nodules and plaquettes. It is fine-grained, with colors ranging from chocolate 250 Figure -. Horizontal distribution of artifacts at SH and SH. brown to light orange (see LRM samples -, -, and - in Appendix C), and varying amounts of patination and/or abrasion. The COPR  collection areas both fall within Pullar’s () site JC-. While she initially reported bifacial foliates in conjunction with pressure flaked arrowheads, we found that the Holocene component was restricted to a small inselberg featured in Figure -, while the findspot with soft hammer foliates is associated with a low terrace about  m south of the inselberg, seen in Figure -. We also found that the patination was considerably different on the diagnostic Holocene forms: a 251 Figure -. Neolithic Type  bifacial lanceolate from outside SH/SH collection areas. Type  bifacial lanceolate (Figure -) and trihedral rod collected from the inselberg. The pressure flaked tools and associated lithic debitage were extremely fresh, and exhibited very light patina ranging from gray to pink, while the foliates had a heavier patina ranging from orange to brown. Neither types are heavily abraded. Both SH and SH are characterized by the same technology—almost exclusively façonnage—although the individual findspots indicate different stages of reduction. In both areas, éclats de taille comprise just over a third of the identifiable blanks. Biface thinning flakes at SH have a cortex percentage distribution as follows: -= and -=. The distribution at SH, on the other hand, is indicative of earlier stages of manufacture: -=., -=., -=., and -=. (Table -). In general, the material from SH is larger and comprised of dense amounts of shatter, resulting from the testing of cobbles. Of the four preforms collected from the area, they are all from early stages of production with only minimal thinning, while SH preforms are predominantly from mid to late stages. Since both findspots are linked to the same chaîne opératoire, they will be considered as one assemblage in the ensuing analysis—a homogenous unit representing different ends of the Khasfian Foliate reduction continuum. 252 DEBITAGE AND CORES The primary mode of reduction at Bir Khasfa is the manufacture of bifacial foliates that can be classified as Eden’s () Type . There is fleeting evidence for core reduction; just two specimens were collected during the combined  and  campaigns. This follows Pullar’s () observation that there was a dearth of cores at the JC- findspot. The first of the two cores was recovered in  in the area around SH. It is an exhausted centripetal Levallois core; one working surface has radial flaking around the entire perimeter, while the opposed face has flat radial retouch (Figure -). The core is derived from the same Rus chert as the foliates and exhibits the same degree of patination. The second core is from findspot SH; an ambiguous fragment of a simple flake core made on a Rus chert nodule. The platform is not present; therefore, not much else can be gleaned from this piece. Among the debitage, slightly over half of the blanks are flakes, followed by a third éclats de taille (Table -). Blades make up just . of the assemblage. Excluding thinning flakes, which have a predetermined suite of attributes (defined in Chapter ), dorsal scar patterns are presented in Table -. The predominance of unidirectional, unidirectional-crossed, and transverse scars are indicative of a simple reduction strategy with minimal evidence for platform and working surface preparation, corroborated by the high incidence of straight (.) and cortical (.) platform types. Only . of the platforms are dihedral or faceted, and a significantly high . are crushed or missing. Both of the core trimming elements have a lateral-steep cross-section; one with transverse-crested scars, the other lateral-crested. There are significant discrepancies among lipped pieces. The non-éclats de taille debitage have a frequency of ., while éclats de taille possess a lipping frequency of .. Between SH and SH, lipping on éclats de taille is . versus . respectively. These numbers suggest the use of both hard and soft hammer percussion at Bir Khasfa. The low percentage of lipped non-éclats de taille, in contrast to the significantly higher percentage of lipped éclats de taille, indicates core reduction was more often carried out using hard hammer, while façonnage reduction employed a mix of the two percussors. The marked discrepancy between lipping on éclats de taille at SH versus 253 Figure -. Radial core from SH, possibly exhausted Levallois with centripetal preparation. SH is another indication that early stage façonnage reduction was carried out at SH, while later stages of work were conducted at SH. Regarding edge grinding, both types of debitage exhibit a low incidence of approximately . TOOLS Of the  specimens, there are approximately  bifacial and  unifacial tools (Table -). It must be pointed out, however, that nine bifacial pieces considered in this assemblage were collected in , under non-systematic conditions. Therefore, there is a bias in the type list toward bifacial pieces; the  specimens were included so as to have an adequate sample of foliates in order to examine the variability within this category. Descriptions of each tool type are presented below: Sidescrapers (). Four sidescrapers are made on flakes, two on cortical flakes, one on a blade, one on an éclat de taille, and one on an unmodified plaquette. Four of them have regular scraper retouch, two have Quina retouch (Figure -), two show more invasive scarring, and one has denticulate retouch. The retouched face is evenly distributed between inverse and obverse. Endscrapers (). These are all atypical endscrapers, classified as such based on continuous scraper retouch along the narrow edge of a blank. Three are on flakes and two are on blades. They all 254 Figure -. Sidescraper from SH with Quina retouch. have obverse, regular scraper retouch. In four cases, the working edges are straight; one piece has a convex edge. Retouched Pieces (). Five retouched pieces are made on éclats de taille, and three on flakes. In seven of the eight cases, the retouch is marginal albeit continuous along the entire edge. Six pieces have inverse retouch, one obverse, and one alternating. Preforms (). In considering the preforms from both SH and SH, there is a fairly even distribution from all stages of reduction. Two preforms are derived from cortical flakes, two on regular flakes, one on a cortical blade, six on plaquettes, while the remaining six are too late in the reduction sequence to determine the initial blank type. An early stage preform manufactured on a flake is illustrated in Figure -; the piece was laterally snapped at the base during thinning, which I argue was the result of a knapping error rather than intentional technology (contra Charpentier 255 Figure -. Bifacial preform from SH manufactured on a blade blank, with lateral break near base. ). The forms falling into the earlier stages of reduction have both plano-convex and biconvex cross-sections, while later forms range from biconvex to lenticular. Khasfian Foliates (). This category is comprised of diminutive bifacial tools with flat, invasive soft hammer percussion retouch. Using Bordian terminology, the shapes include primarily limandes, followed by cordiforms, and discoidal bifaces (Figure -); they are all flat specimens falling within Zone  and Zone  of Bordes’ () classificatory scatterplot. Cross-sections are almost exclusively lenticular. Metric attributes of unbroken foliates (in centimeters) are as follows: mean length=. (standard deviation=.); mean width=. (standard deviation=.); mean thickness=. (standard deviation=.); index of elongation=. (standard deviation .) and index of flattening=. (standard deviation=.). These numbers indicate an assemblage of small, wide and flat bifacial tools, exhibiting a relatively standardized set of dimensions. 256 Figure -. Khasfian Foliates from SH and SH. 257 Wadi Afur  (T) The  artifacts in the T assemblage were collected non-systematically from a low terrace above Wadi Afur (Table -); the low density, diffuse scatter was collected more or less in toto. The site was sampled following the recognition of diagnostic Neolithic forms in the field; therefore, useful for examining Holocene reduction strategies and tool forms. The majority of artifacts (.) are manufactured from typical Rus chert, although there is a wide range in quality of material. Biface thinning flakes are typically on fine-grained chert, while other debitage is on coarse and moderate-grained chert. There were trace amounts of limestone flakes (.). Ten pieces—all debitage and debris—are burned, although there is no evidence for hearths or fire-cracked rock around the scatter. The burned pieces may have derived from the rockshelter above the terrace, which was presumably occupied by the T toolmakers. There is no source of raw material in the immediate vicinity, although beds of Rus chert outcrop fairly regularly throughout the wadi; the nearest source occurs about ten kilometers away. The distance to the source is reflected in the relatively high -:- cortex percentage ratio (Table -). In addition, the bifacial preforms all fall within mid and late stages of reduction, and only . of the striking platforms have cortex (cortical straight, cortical curved, or dihedral 1⁄2 cortex). DEBITAGE AND CORES The core technology at T is comprised of diminutive flake, blade, and bladelet cores. The ten specimens are all volumetric; seven with multiple platforms and three single platform cores. The shapes are typically orthogonal or irregular, and half of them have evidence for edge grinding at the intersection of the working surface and the striking platform. Half of the scar patterns exhibit randomly alternating reduction across multiple working surfaces (Table -), while the remaining half include unidirectional, unidirectional-crossed, and unidirectional-parallel scar patterns on the working face. The distribution of blank types (Table -), shows a predominance of flakes, followed by éclats de taille, and a relatively low frequency of blades and core trimming elements. Nearly  258 of the identifiable blanks have lipping (of which  éclats de taille are lipped). About  of the pieces show edge grinding ( éclats de taille). Dorsal scars are dominated by unidirectional and unidirectional-crossed patterns, a combined . (Table -). The striking platforms are primarily unmodified (.) and crushed/missing (.). The remaining . have some degree of platform modification. The distribution of cortex percentage categories on éclats de taille is as follows: -=., -=., -=., -=. This pattern indicates that nearly all bifacial thinning is from later stages of reduction, which is corroborated by the predominance of mid and late stage preforms. TOOLS There are  tools in the T assemblage (Table -), of which  are unifacial and  bifacial. They are as follows: Sidescrapers (). Two sidescrapers are made on flakes and one on a cortical flake; all have continuous, obverse, regular scraper retouch. Endscrapers (). Six endscrapers are on flakes, two on cortical flakes, and one on an éclat de taille. Seven specimens have regular scraper retouch, while two exhibit pressure flaking. Six working edges are convex, two straight, and one nosed in shape. Like the sidescrapers, all retouch is obverse. Notches/Double Notch (). There were four notches and one double notch. All specimens were made on flakes. The notches all have obverse retouch; the retouch on the double notch is inverse. In one case pressure flaking is used to form the concavity. Burin (). The one piece from this category exhibits a single burin blow on the retouched, truncated, distal end of a flake. Retouched Pieces (). Thirteen retouched pieces are made on flakes, two on blades, two on éclats de taille, and one on a burin spall. Ten tools have nibbling retouch, three irregular scraper retouch, two invasive retouch, two pressure flaking, and one has irregular denticulate retouch. In  cases retouch is obverse, three alternating, and one inverse. 259 Truncation (). This piece, made on a flake, has steep inverse retouch at the distal end, forming an angle of approximately °. There is also irregular scraper retouch along the lateral edge forming a slightly convex working edge. Drill/Awl (). There was one awl and one drill. The awl is made on an éclat de taille and has inverse, converging, bilateral retouch forming an elongated, perforator tip on the blank. The drill is made on a transverse flake and is formed by alternating, steep bilateral retouch that forms a drill bit at the edge of the blank. Preforms (). All of the preforms at T have enough thinning for them to be classified as mid to late stages of reduction. The presence of a diagnostic trihedral rod in the assemblage suggests this may have been the intended product of the preforms. Indeed, most of them are plano-convex in cross-section and lanceolate in shape. In almost every case it is impossible to determine whether the preform is thinned from a blank or plaquette, although there is evidence for both types in the assemblage. Two of the late stage pieces exhibit longitudinal blade scars, which may be a thinning technique to prepare the preform for pressure flaking. Trihedral Rod (). The single tool of this type is broken in half; otherwise, it is a classic example. The piece has the cross-section of an equilateral triangle, with fine parallel ribbon scarring across all three faces. Its presence at T signifies that the site dates to Zarins’ () Neolithic Period , placing it sometime in the th-th millennia . In sum: ) the low ratio of debitage to tools (.); ) evidence for later stage bifacial production; and ) location of the site directly below a rockshelter and on a low terrace above the wadi channel, all indicate that T served as some form of basecamp for Neolithic Period  huntergatherers occupying the southern Nejd. Dhanaqr (T) The T lithic scatter occurs on the interior slope of a Tertiary rock outcrop at the confluence of Wadi Ribkhut and Wadi Dhahabun. The  lithic artifacts recovered from the findspot come from on and within a five centimeter surface carpet of aeolian sands and gypsum. Twenty-six square meters 260 were collected, although the vast majority of material comes from three squares in the southeastern quadrant (Figure -). The predominant raw material is a Rus chert that ranges from extremely fine-grained and high-quality (), to quite poor, riddled with fracture planes and showing a high degree of shatter (.), and everything between (.). There is also a trace amount of limestone artifacts (.). The higher quality material tends to be orangish-yellow in color and glossy, while the brittle chert is a dull brown. Although there is minimal evidence for façonnage reduction, some biface thinning flakes are present among the debitage, and an ambiguous piece that can be interpreted as a preform is included in the toolkit. All of these façonnage pieces are made from the high-quality Rus chert, and suggests preferential treatment of raw material for specialized reduction. There is no immediate source of chert, however, nodules of the material are abundant throughout the wide alluvial plain surrounding T; the nearest source is estimated to be no more than five kilometers away. Still, the -:- cortex percentage ratio is relatively high at . (Table -), indicating initial decortification was probably carried out off-site. Figure -. Horizontal distribution of artifacts at T. 261 DEBITAGE AND CORES The production of blades appears to be one of the primary reduction strategies. Blade/let and flake-blade/let cores comprise . of all recovered cores, while flake cores are just . (Table -). The cores are both lineal and volumetric (multiple working surfaces), exhibiting a range of scar patterns such as: bidirectional (.), unidirectional-parallel (), unidirectional (), alternating (.), and radial (.). Three are classified as Levallois, although it should be emphasized that this is only sensu latu, they are lineal cores and there is evidence for platform and convexity maintenance. There was one Kombewa core, a large éclat de taille that was longitudinally split as a falseburin during reduction. The lateral-steep edge was used as a striking platform for subsequent blanks (Figure -). It is noteworthy that the Kombewa core is derived from high-quality Rus chert, an indication that this material was more intensively exploited. The blade reduction strategy is also apparent in the prominence of blade-proportionate blanks, which represent . of the debitage. The remaining blanks are divided among flakes (.), Levallois (.), and CTE’s (.). The Levallois blanks have faceted platforms; two exhibit bidirectional-parallel scars and the other has convergent scars. Their shapes are ovoid, rectangular, and trapezoidal. Of all assemblages analyzed, T has the fewest number of blanks with simple unidirectional scar patterns; rather there is a wide variety of complex forms (respectively): unidirectional-parallel (Figure -), unidirectional-crossed, convergent, bidirectional, and radial (Table -). This is further indication of the later stages of reduction carried out at the site, and the degree to which raw material was exploited. Regarding the bidirectional method of reduction at T, the distal platforms are only supplementary—short, non-invasive distal removals are used for maintenance of convexity, rather than to obtain substantial blanks for tools. About half of these cores also have supplementary platforms on one or both lateral edges. This method represents a lineal approach to the core volume with a standard technique of convexity maintenance; therefore, the technology is viewed as a Levallois variant. 262 Figure -. Refit Kombewa flake-core from T. Soft hammer percussion may have been utilized at T, as . of the assemblage exhibits lipping. Among the blades, that number rises to .. Edge-grinding is present but rare, about a quarter of the pieces show evidence for abrasion at the intersection of the dorsal face and platform. There is minimal treatment of striking platforms, nearly three quarters are unmodified, and less than ten percent are faceted or dihedral. It is noteworthy that in the few cases of modified platforms, the faceting scars sometimes come from the lateral edge. TOOLS Of the  tools within the T assemblage, all but one is made on a unifacial blank (Table -). Descriptions of each category are presented below: Sidescrapers (). Two of the sidescrapers are made on blades, and one on a cortical flake. One of the blades has heavy Quina retouch, the other two show regular scraper retouch. All retouch is obverse; on one piece it is bilateral. Two of the three pieces are made on blanks with a lateral-steep cross-section, a pattern that is repeated among several of the tool categories. Notches/Double Notches (). There was one notch and three double-notches. The notch is made on a thick blade and has nibbling inverse retouch forming a concavity on one of the lateral edges. The double-notches are made on a flake, blade, and cortical flake. The retouch is all obverse; on one the concavity is formed by invasive flaking, one moderate, and one nibbling retouch. Denticulate (). This piece is actually a composite tool, although the denticulate edge exhibits the 263 Figure -. Unidirectional-parallel blade refits from T. heaviest amount of retouch, thus is classified as such. It is made on a large cortical flake. One lateral edge exhibits continuous semi-steep denticulate retouch, the distal end has a small notch, and there is irregular scraper retouch on the opposite lateral edge. Burins (). Two of the burins are made on flakes, and two on blades. Three of them are simple burins that exhibit a single spall down either the proximal end or lateral edge of the piece. The last piece is a dihedral burin made on the distal end of a transverse blade. Retouched pieces (). Four are made on flakes, three on blades, one on a cortical flake, and on an éclat de taille. The retouching thinning flake is derived from high-quality yellowish-orange Rus chert, and is another example of the intense exploitation of this material. Six of the retouched pieces have a lateral-steep cross-section, upon which retouch is often on the opposed lateral edge. The retouch ranges from irregular scraper retouch to continuous nibbling. In a sense, these lateralsteep retouched pieces can be considered naturally-backed tools. Indeed, the majority of all tools 264 Figure -. Truncation on Kombewa flake from T. () have lateral-steep cross-sections; the prevalence of these pieces may indicate they were chosen intentionally for backing. Levallois flakes (). See description in Debitage and Cores section. Truncated pieces (). Of the two specimens belonging to this category, one is made on a blade with continuous obverse retouch at the distal end forming a truncation, and the other is manufactured from a Kombewa flake (Figure -, also refit in Figure -). Bifacial preform (). The specimen illustrated in Figure - is fairly ambiguous, and, therefore, can only tentatively be placed within this category. It shows flaking on both sides, one has flat invasive scars, and the other face has semi-steep invasive. The piece is snapped in half, which explains why it was discarded prior to completion. The preform is derived from the same fine-grained orangish-yellow Rus chert as the éclats de taille, another indication of the specialized reduction strategy associated with this raw material, and further supporting the classification of this piece as façonnage. Based on the nature of the lithic assemblage and location of the site, T seems to have been some form of basecamp. Its position on an outcrop just above two major watercourses is ideal for hunting and obtaining fresh water. The assemblage is indicative of late stage manufacture, and the function of many of the tools seems to be for game-processing. 265 Figure -. Miscellaneous bifacial implement from T. Jibal Ardif  (T) The scatter at findspot T is one of many on a massive reg surface that extends for nearly  km north to south in the central Nejd. Packages of fine-grained Rus chert are available in dense patches throughout the reg in both nodular and plaquette form. The T artifacts are associated with a dense outcrop of very thin plaquettes; clearly, the site is a workshop for the exploitation of this high-quality raw material. This claim is corroborated by the -:- cortex percentage ratio, which indicates a high amount of cortex; therefore, for primary stages of reduction (Table -). Artifacts were collected from a narrow strip encompassing  m, Figure - shows the distribution of this material is concentrated in a narrow band running through the center of the transect. Every artifact within this assemblage is made from Rus chert, although . exhibit heavier patination and weathering and may be related to an earlier phase of occupation. Mixing is expected, given the position of the scatter on an outcrop of particularly good raw material. The patina on most of the material ranges from pinkish-gray to light brown. In general, arêtes are quite sharp and there is minimal weathering. One reason for the pristine nature of the material may not necessarily be the age 266 of the site, but the position of the artifacts, which were lightly cemented in a shallow surface veneer of gypsum evaporites. There are two distinct reduction strategies at T: the façonnage production of large, flat bifaces, and a flat, lineal core reduction strategy not unlike that from T and T. One of the critical questions regarding the T assemblage is whether or not the core and bifacial reduction strategies are coeval, given the potential for mixing at a surface scatter in proximity to an outcrop of raw material. The patina on the cores is much thicker and there is light/moderate weathering on all of the specimens, while the façonnage material typically has less patina and exhibits fresh arêtes. Furthermore, the patinated cores were all collected within squares A and B at the northern extremity of the site, while the bifacial material is diffuse throughout the entire collection area. Thus, the evidence suggests the two reduction strategies at T are not coeval. The lineal core reduction Figure -. Horizontal distribution of artifacts from T. 267 strategy (henceforth referred to as Phase I) apparently predates the façonnage artifacts (henceforth referred to as Phase II), which are differentiated by their relative degree of patination and weathering. DEBITAGE AND CORES Of the Phase I cores, four are flake cores, one flake-blade/let core, and three opposed platform cores that are loosely defined as Levallois (Figure -), following the same technique of convexity maintenance via the use of a supplementary distal platform (and sometimes lateral) described at T. Scar patterns on the cores’ working surface(s) also follow that of T, the predominant forms are unidirectional-parallel and bidirectional (Table -). There is virtually no evidence for edge grinding on any of the striking platforms. Despite the fact that éclats de taille (Figure -) comprise only a quarter of the assemblage (Table -), most of the other debitage also appear to be associated with Phase II bifacial reduction, although lacking the characteristic attributes of thinning flakes. Nearly all of the debitage have a grayish-pink patina, so appear homogenous. The debitage mean index of flattening (.) is remarkably higher than any other assemblage (Appendix A), indicating a preponderance of particularly flat and thin pieces that are the byproduct of façonnage reduction. Flaking appears to have been carried out via hard hammer percussion, just  of the blanks are lipped ( on éclats de taille). Of the lipped specimens, they all have extremely high-angle striking platforms (>°), which may explain the presence of lipping. Edge-grinding is rare, just . of pieces bear evidence for this method of platform preparation, although the frequency increases to . among éclats de taille (Figure -:b,c). In fact, one preform exhibits extensive grinding along an entire edge (Figure -:a), clearly this technique was employed in façonnage reduction. Platform faceting is one of the characteristic attributes of thinning flakes; therefore, it is not surprising that . of the debitage are dihedral, faceted curved, or faceted straight. Platforms with cortex are also quite frequent (), indicating a high degree of initial plaquette reduction (the platform illustrated in Figure -:c is a typical example of an early stage éclat de taille, in which the cortical edge of the plaquette is still present). 268 Figure - TOOLS Levallois cores from T Phase I. Of the  tools in the T assemblage,  are preforms and one is simply a retouched piece. Clearly, this toolkit is indicative of the near singular activity of making bifacial tools. The one retouched piece has irregular, inverse marginal retouch along the lateral edge of a thinning flake. It may simply be a utilized flake, but the usewear is so heavy as to merit classifying the piece as a tool. The preforms come from every stage of reduction, ranging from tested plaquettes with just a few radial removals across one face (Figure -), to nearly complete forms with flat, invasive 269 Figure - T éclats de taille. flaking across both faces (Figure -:b). The shapes of late stage forms include ovate, limande, and cordiform, and all are biconvex in cross-section with predominantly straight edges. Mean dimensions of the preforms are as follows (in cm): length=., width=., thickness=., and weight= (g). Based on these measurements, the preforms fall somewhere in size between the bifaces at the A-A complex and the SH-SH bifaces (Figure -?). In terms of length and width, T matches Sibakhan dimensions, although the thickness is significantly less 270 Figure - Late stage bifacial preforms from T. 271 Figure -. Early stage preform from T. than the Sibakhan, within the range of Khasfian and Holocene forms. This observation begs the question, to which industry did the T bifacial toolmakers belong? Without completed tools or known correlates from other sites, there is no clear answer to that question, although possibilities are presented at the end of this chapter in the discussion. In sum: Phase I technology at T is only represented by a handful of cores. The reduction strategy is a lineal approach to the core volume, using flat plaquettes for unidirectional, centripetal, and bidirectional flaking. Some specimens exhibit both convexity maintenance and platform faceting, suggesting some form of prepared core technology approaching Levallois. Unfortunately, there are no tools or debitage associated with Phase I. Phase II is exclusively characterized by façonnage technology in which plaquettes were thinned to create large flat bifaces in an array of shapes falling within Zone IV of Borde’s classificatory scheme—ovate, cordiform, and discoid (Borde ). Finished tool types remain unknown and are considered in the discussion section. 272 Al-Hatab (T) Al-Hatab was found in a small gully feeding Wadi Dawkhah, just north of the Dhofar Escarpment. The  lithic artifacts in the assemblage were excavated from just two square meter units, within a half-meter thick deposit of mixed colluvial and aeolian sediments. The loosely compacted sediments were removed in two arbitrary stratigraphic units approximately  cm in thickness; however, it was later determined that the entire deposit represents a homogenous unit of mixed slope waste. It is not known when the T sediments began to accumulate, nor how long they took to form; however, the secondarily deposited artifacts appear to come from the same source and exhibit a homogenous technology, as well as weathering. The lithics have a dark orangish-brown desert varnish, yet virtually no abrasion. In addition, pot-lidding frequently occurs on the artifacts. These observations suggest that the material was subjected to high heat when on the surface, followed by sudden cooling, probably caused by occasional precipitation. Furthermore, arêtes are extremely fresh, indicating the lithics were transported only a very short distance prior to deposition. So, it appears that the artifacts were made and discarded onto the surface during a phase in which the landscape was stable and climatic conditions were xeric. Subsequently, there was active downcutting in the gully that caused the material to become buried in low-energy alluvium and slope waste. Assuming these events occurred within the most recent climatic cycle, then the T artifacts must be contemporary with, or predate the Early/Middle Holocene wet-phase. The frequency of potlidding, caused by extremely xeric conditions, presents evidence in favor of a pre-Holocene date. There is a wide terrace adjacent to the wadi, littered with moderate to high-quality Rus chert. Unlike the T plaquettes, the raw material is available in packages of rounded nodules and angular slabs. While generally free of inclusions, much of the material is brittle and has thermal weathering. Three quarters of the assemblage is made on moderate-quality chert that ranges from gray to beige in color and has banding. Approximately  is a high-quality, glossy orangish-brown material resembling the high-quality material from T. The remaining ten percent is a fine-grained limestone that grades into the moderate-quality banded chert. 273 Although T is located in proximity to a raw material source, the function of the site appears to have been more than simply a workshop. The -:- cortex percentage ratio of . is within the range of other sites near raw material, so a high degree of decortification did take place (Table -). In addition, the debitage:tool ratio is ., indicating there were a relatively high number of completed tools found within the assemblage. It is not surprising that this location served as more than an ephemeral workshop, its proximity to the rockshelter (albeit small), raw material, wadi, and escarpment would have presented an ideal setting for nomadic hunter-gatherers. DEBITAGE AND CORES Twenty cores were recovered, of which  exhibit either blade/let or flake-blade/let scars. Scar patterns on the cores’ working surface(s) are primarily unidirectional-parallel (), followed by unidirectional (), and bidirectional (). One of the bidirectional cores has a similar method of convexity maintenance as T and T Phase II, where the distal platform is used for non-invasive supplementary flaking to create convexity. There is little evidence for treatment of the striking platform on any of the cores beyond simple decortification. Edge-grinding is present, but only on  of the specimens (of which all are either blade/let or flake-blade/let). Blade-proportionate blanks comprise slightly over  of identifiable debitage (Table ), with a mean index of elongation of .. Dorsal scars on blanks, presented in Table -, show a majority of unidirectional (~) and unidirectional-parallel (~) patterns. Only three blanks had bidirectional scars. Most platforms are unmodified, less than  exhibit any evidence of faceting. Edge-grinding mirrors the frequency seen on the cores, about  of blanks have abrasion at the intersection of the dorsal face and striking platform Lipping is present on . of blanks. These debitage characteristics illustrate a method of core reduction at T in which blades (Figure -:a) were struck from simple, single platform lineal cores. The flat cores illustrated in Figure - exhibit parallel and convergent methods within this strategy. In a few cases, supplementary distal platforms were used to maintain convexity on the working surface. There is also evidence for a volumetric reduction strategy, which uses the same unidirectional-parallel flaking 274 strategy across multiple working surfaces. These lineal and volumetric strategies may belong to the same reduction continuum, representing a more thoroughly exploited core. TOOLS All  tools are unifacial, although there was one piece (Figure -:a) that is open to interpretation. It is worked bifacially, thick and plano-convex in cross-section, and exhibits heavily invasive scarring. There is some irregular scraper retouch on one lateral edge. The artifact can be viewed as a core, bifacial piece, scraper, or any combination of said categories. A description of the unifacial tools is presented below: Sidescrapers (). All five sidescrapers are made on flakes. In every case there is obverse regular scraper retouch. Four pieces have convex working edges, and on one the retouch forms a straight edge. There is one scraper with retouch along two-thirds of the margin (Figure -:c). Figure -. Unretouched blade (a) and unifacial tools (b — c) from T. 275 Figure -. Flat, unidirectional-convergent (a) and unidirectional-parallel (b) blade cores from T. Endscrapers (). Two endscrapers are made on flakes and one on a transverse cortical flake. On the flakes, there is regular obverse scraper retouch forming a straight working edge. The cortical flake has steep inverse retouch creating a convex working edge. Miscellaneous Scrapers (). Figure - illustrates the two specimens belonging to this category. The core/biface/scraper, described above, was placed in this category because of its resemblance to the other miscellaneous scraper. The second specimen (Figure -:b) is discoidal in 276 Figure -. Bifacial (a) and unifacial (b) scrapers from T. shape, plano-convex in cross-section, and has semi-steep and steep invasive retouch around the entire margin of the piece. Notches/Double Notches (). There were seven notches and two double notches. Five of these tools were manufactured on flakes, two on chunks, one on a cortical flake, and one on a cortical blade. In six cases the retouch is obverse, the other is inverse. One notch is actually a composite tool with a wide concavity on one edge and irregular retouch on the opposite edge (Figure -:b). Denticulates (). Two were made from flakes, one from a cortical flake, and one from a chunk. In every case, the denticulate retouch is obverse and continuous along the entire lateral edge. 277 Burin (). One specimen of this type is a simple burin exhibiting a single spall removal at the distal end of a débordant blade. The second example falls between a burin and a bladelet core: the working edge of this diminutive piece is formed by perpendicular, intersecting blows. One face is faceted, exhibiting multiple spall removals. Retouched Pieces (). Nine retouched pieces are made from flakes, four from blades, and four from chunks. In eight cases the retouch is simply nibbling, while an equal number exhibit irregular scraper retouch, and one has flat, invasive flaking. Like all of the other tools, almost all of the retouch is obverse; only two pieces have retouch on their ventral faces. Perforators (). There are three awls and one drill in this category. Two of the awls are made on flakes and one on a blade. The drill is made on a bladelet. All of the tools from this category are crude informal types with irregular retouch. It is noteworthy that, while I have argued soft hammer blade production was the primary reduction strategy at T, less than  of tools were manufactured on blade-proportionate blanks. Either the production of elongated blanks was unintentional, or, since one function of the site was a workshop, it is possible other blade-proportionate tools were carried away from the site. Discussion A variety of reduction strategies have been described from the eleven assemblages collected by COPR in  and . The next step of this analysis is to examine the relationship of these technologies to one another and define potential lithic industries. For the purposes of this dissertation, the concept of an industry is only perceived in the sense of a grouping of like assemblages, not in the strict sense of the term as defined by Marks (n.d.). This is due to the absence of relative and/or absolute dates, as well as prohibitive sample sizes (i.e. a limited number of sites and small number of pieces in the assemblages). Assemblages that are grouped together as an industry are considered as such based on qualitative and quantitative techno-typological similarities. In the most general sense, the reduction strategies can be divided into façonnage and core reduction. Types of core reduction include: ) multiple platform, volumetric cores with removals 278 haphazardly across different working surfaces [A.I, A.II, SH, T]; ) simple, unidirectional, recurrent hard hammer blades on the narrow, elongated face of the core [A-A, A]; ) soft and hard hammer, recurrent, unidirectional and bidirectional flake-blade cores using a single, lineal working surface, sometimes grading into a variant of Levallois [T, T.I, T]; ) unidirectional convergent Levallois [A-A]; and ) single and opposed platform, recurrent, volumetric flake-blade cores [A.I, T, T, T]. A common thread running throughout all of these reduction strategies is the paucity of platform preparation. Typically, cortex is removed from the platform and sometimes there is evidence for light nibbling, but, for the most part, actual faceting is rare. By its very meaning, façonnage refers to a strategy of reducing a volume of raw material through invasive bifacial flaking. Therefore, different types of façonnage technologies are articulated by examining the various finished products, percussor(s) (and in which phase of production they were used), attributes of debitage produced, and preform selection (i.e. blank, plaquette, slab, or nodule). Based on these criteria, five different products of façonnage reduction are recognized within the COPR assemblages: ) pressure flaked trihedral rods [T]; ) pressure flaked pedunculates [A.III]; ) pressure flaked points [A.II]; ) soft hammer foliates [SH-SH]; and ) hard hammer, flat handaxes [A-A; T.II]. While traces of bifacial technology were observed at T and T, the samples are far too ambiguous; there are too few data to make any statement regarding bifacial manufacture at these sites. Based on qualitative and quantitative techno-typological observations, how do these assemblages relate to one another? Four classificatory groups are proposed: ) general Holocene; ) Khasfian; ) Nejd Leptolithic; and ) Sibakhan (Table -). NEOLITHIC There are clear stratigraphic and geomorphic indications that A.I, A.II, A.III, and T are all Holocene. In the case of Qarat al-Kibrit , the shell from Unit A provides a terminus ante quem of , ±   for the uppermost stratum. Presumably, the underlying layers are associated with the Early/Middle Holocene sub-pluvial. At T, the site is located on a low terrace just a couple meters 279 above the present channel of Wadi Afur. The T findspot is in southern Dhofar, less than  km from the escarpment, which would mean the channel was fairly active in the Early/Middle Holocene. It is unlikely to predate this wet-phase, or the scatter would have been washed away and/or show greater evidence for alluvial rolling. These rough geomorphic correlations are corroborated by techno-typological similarities, at which point SH can be added to this general “Holocene” grouping. While clearly, there is considerable typological variation among these assemblages, this dissertation is not so concerned with Holocene variability. Also, since units of time in the Upper Pleistocene are on the scale of tens of thousands of years, breaking up Holocene industries into smaller phases would create disproportionate resolution. The Holocene A.I, A.II, SH, and T core technologies are characterized by exclusively volumetric reduction. Lipping is found on approximately half of the blanks. Averaging  – , edge-grinding is also prevalent in these assemblages. The cores tend to be small and ad hoc (Figure -)—they have multiple platforms, multiple working surfaces, exhibit little to no preparation or maintenance, and have simple, alternating working surface scar patterns. There are also volumetric flake and blade cores with opposed platforms and unidirectional/bidirectional-parallel scar patterns. This leptolithic technology is particularly prevalent at A.I and T; and, in both of these assemblages, bladelets account for about  of the blanks, the only two cases where these types of blanks are present in any significant amount. Both assemblages bear other lines of evidence that point to a Middle Holocene date (AMS measurement on shell bead at A.I and diagnostic tool at T). As for associated bifacial technologies, the finished tools often have some amount of pressure flaking. At T, trihedral rods are present, which are diagnostic of Neolithic Period  (Zarins ). The pressure flaked pedunculate in Phase III of Qarat al-Kibrit  is a classic example of a Type  form according to the Rub al-‘Khali Neolithic scheme (Edens ). Although only a distal fragment, the bifacial point from Phase II of Qarat al-Kibrit  seems to be a soft hammer/quasi-pressure flaked lanceolate that fits into Type  of the Rub al-‘Khali Neolithic typology. 280 Table -. Proposed lithic industries. 281 KHASFIAN One of the conclusions reached in Chapters  and  is that there is potentially an unrecognized industry in the interior of southern Arabia characterized by the production of small, soft hammer foliates—Khasfian Foliates. The pieces were made with flat, invasive retouch, thereby giving them a characteristic lenticular cross-section; shapes include sub-triangular forms such as limande, ovoid, and discoidal. Rather than an element of the Rub al-‘Khali Neolithic or Saruq-Facies coastal sites, these tools may be Pleistocene in age. There are morphologically identical tools from Somaliland Stillbay and Somaliland Magosian assemblages in the Horn of Africa (e.g., Clark : pl. :, pl. :-). Connections with the latter industry are not compelling, given the absence of backed microliths in Khasfian assemblages. Taking into account the diminutive size of the foliates, the climatological history of South Arabia, the genetic evidence presented in Chapter , the typological analogies across the Red Sea, and the contrasting degrees of patination between the foliates and Neolithic forms at Bir Khasfa, the assemblage is thought to correlate with the late MSA or early LSA of the Horn, in the range of  —  kya. Holocene Nejd Leptolithic Sibakhan Figure -.  confidence intervals comparing core weights among the three core industries. 282 Bir Khasfa—the Khasfian type site—provides the only sample of Khasfian Foliate material in this analysis. While there are other isolated reports of soft hammer foliates (Smith ; Edens ; Villiers-Petocz ), none come from systematically collected contexts. Therefore, with only one definitive data point, the term industry (sensu stricto) cannot yet be applied. The assemblage is almost exclusively based on façonnage technology; there is virtually no evidence for an associated core technology with the exception of one nondescript core and one discoidal core illustrated in Figure - (perhaps exhausted Levallois). The density of preforms and completed foliates suggests the site was a workshop for the production of these pieces. The two different collection areas represent different stages in the reduction continuum. Early stage preforms are on both plaquettes and large blanks. There was a significant discrepancy in lipping on éclats de taille—. at the early stage locus versus . at the later stage area. It seems hard hammer percussion was used for initial thinning, while soft hammer was employed to complete shaping the piece. This observation is significant when considering the relationship between Khasfian Foliates and the preforms from T.II (discussed below). NEJD LEPTOLITHIC The percentage of blade-proportionate blanks at T and T are . and . respectively. These assemblages seem to represent a widespread blade industry observed at a number of findspots throughout the Nejd Plateau. The term Nejd Leptolithic is used, although in the absence of an in situ type site, this taxonomic designation should not be considered a proper industrial label. Both assemblages exhibit a similar core technology typified by unidirectional-parallel blades removed from the flat working surface of lineal cores. There are also bidirectional cores with supplementary distal platforms (Levallois sensu latu), as well as simple, volumetric flake-blade cores. In both assemblages, the percentage of lipping on blanks is between  – , suggesting some degree of soft hammer percussion. Examination of the  confidence intervals on metrics of unbroken blade blanks at T versus T reveal several significant statistical differences (Figure -). Blank size, represented by 283 T T T T T T Figure -.  confidence intervals comparing complete blade metric variables at T and T. 284 T T T T T T Figure -. 285 cont. weight, blank area, and blank volume, indicate the blades at T are considerably smaller. Based on Independent Samples T-Tests, it seems the index of elongation is nearly identical between the two assemblages (p=.), while the index of flattening is significantly higher at T (p=.). This means the blades are flatter, which explains the discrepancy in weight, area, and volume, all of which are a function of this attribute. The reason for these different dimensions might be explained by the significant difference in relative platform width (p=.). T typically has deep, narrow platforms relative to blank size, whereas T platforms tend to be wider and broader. Consequently, the blades produced at T were thicker than T. Are these statistically significant variations in blade size an argument against placing T and T within the same lithic industry? There is no clear answer to this question, although different raw material constraints, and/or simply percussors of different dimensions might explain the discrepancy. Qualitative observations of techno-typological attributes from both assemblages suggest these sites are, in fact, related. There is a similar distribution of core types and dorsal scar patterns. Furthermore, lipping and edge-grinding are found in similar frequencies. Platform angles on the debitage occur within a similar range. Although the data are scarce, the cores from T.I exhibit a similar technology and may also be tentatively placed within the Nejd Leptolithic. If this is the case, then the Phase II bifacial component at T must post-date the Nejd Leptolithic. SIBAKHAN The Sibakhan is perhaps the best defined industry considered in this analysis, because it is known from multiple findspots, representing different phases of the reduction sequence. The name is derived from the position of these findspots in proximity to sibakh associated with the Haushi-Huqf depression. Presumably, these sibakh would have formed a series of ancient playas during pluvial episodes, or perhaps a single large lake filling the basin. Like the other “industries” defined in this dissertation, the taxonomic designation is tentative in the absence of an in situ type site. The A-A complex, as well as A, are two newly discovered occurrences of the Sibakhan. 286 One of the defining characteristics of Sibakhan assemblages is the production of large, flat bifacial handaxes with semi-sinuous edges and biconvex cross-sections; shapes primarily fall within Borde’s Zone IV (limandes, ovoids, and discoids), as well as some lageniformes, in lower frequency. The bifaces are primarily made from tabular raw material, although Biagi (ibid.) reports rare instances of bifaces made from blanks. The pronounced bulbs and virtual absence of lipped debitage indicate these tools were made exclusively with hard hammer percussion. The other characteristic feature of the Sibakhan is a hard hammer blade technology. The blanks are typically removed from simple, lineal, unidirectional-parallel cores. At the A-A complex, located on a source of large angular slabs, the narrow elongated face of the raw material was used to make blades, also producing a high frequency of débordant pieces. Average striking platform angles are significantly higher on Sibakhan blades than those from the Nejd Leptolithic (p=.), although roughly equivalent with the Holocene blade assemblages included in this analysis (p=.). Other metric analyses indicate the blades are significantly more massive in Sibakhan assemblages than the two other leptolithic techno-complexes, in terms of blank area, blank volume, index of flattening, and weight (Figure -). The recurrent, lineal production of blades at A exhibits the same suite of technological attributes observed within the A-A assemblage, and it is tempting to link the two sites within the same chaîne opératoire. Metric analyses between the two blade assemblages display some statistically significant differences, however, and must be addressed. An Independent Samples T-Test comparing A-A and A blades shows that the range of platform angles among complete blade blanks at A is significantly higher than A-A (p=.). This picture changes somewhat when the index of elongation is filtered to only those blades with an IE between . and ., indicating no significant differences (p=.). Therefore, the discrepancy is attributed simply to a wider range of variability at the A-A workshop complex, as opposed to the very narrow range of higher angle blade blanks from A. Because platform angles naturally increase as the working surface is reduced, these statistics support the assertion that the blade technologies at A-A and A both belong to a single 287 technological-complex; the sites represent both early and later stages of core reduction, respectively. Core weight is significantly lower at A, because they are smaller, more heavily exploited specimens. While very little can be said about the typology at A, it is noteworthy that the only tools present resemble a frequent form within the A-A toolkit, in terms of retouch position (lateral), blank selection (blade), and retouch type (inverse marginal). This adds further support to the notion that both sites collected within Wadi Qilfah belong to the same lithic techno-complex. A second, distinctive type of core reduction was observed only at A-A. The cores are classic unidirectional-convergent Levallois, triangular in shape and, in some cases, with platform faceting. These cores are presumably coeval with the leptolithic and façonnage components, based on minimal differences in abrasion and patina. Discoidal cores were reported from Saiwan, with recurrent centripetal flaking around the entire margin of the piece (Biagi ). Some of the discoids have a single working surface, while others are biconical. Like the A-A complex, all of the artifacts exhibit the same degree of patination and weathering. This analysis raises several questions regarding the Sibakhan. How do these blade, radial, and Levallois core technologies relate to one another? Are they coeval? How old are the assemblages? Regarding the date, examination of landscape morphogenesis around A-A provides few clues. The ridges upon which the scatters are located were formed in the Middle or Upper Pleistocene. Presumably, local occupation in the Haushi-Huqf region was associated with the presence of ancient freshwater lakes, which were active during the Pleistocene pluvial phases reported in Chapter  (OIS , , e, c, a, or ). Techno-typological correlations with material from contiguous regions place the assemblages either within an MP/MSA context or late LP/ESA. Therefore, the Sibakhan may date to either the OIS e pluvial ( –  kya), or sometime between OIS  —  ( —  kya). This temporal attribution must fall within a wide bracket of time and be considered tentative in the absence of absolute dates or a stratified site. Regarding the precise relationship of the various reduction strategies noted at A-A, this question must remain unanswered until a stratified site is finally excavated. Because of potential 288 Holocene Nejd Leptolithic Sibakhan Holocene Nejd Leptolithic Sibakhan Holocene Nejd Leptolithic Sibakhan Figure -.  confidence intervals comparing complete blade metrics in Holocene, Nejd Leptolithic, and Sibakhan Industries. 289 Holocene Nejd Leptolithic Sibakhan Holocene Nejd Leptolithic Sibakhan Figure -. cont. implications regarding modern human emergence out of Africa, future work in the Palaeolithic of southern Arabia should focus on further articulating this industry. T The T assemblage does not necessarily fit into any previously defined category; therefore, it will be treated as a separate techno-typological unit. In the discussion of the Nejd Leptolithic, it was proposed that there were two archaeological phases represented in the T assemblage. The earlier material—Phase I—is comprised of eight cores recovered from two square meters at the 290 northern edge of the collection area that exhibit heavy patina and weathering. The Phase I cores have technological features similar enough to T and T to be tentatively considered part of this industry. The cores are single platform and opposed platform flat cores, some with faceted platforms, that grade into the bidirectional Levallois method characteristic of the Nejd Leptolithic. Regarding the Phase II assemblage from T, the question remains to be answered whether preform manufacture relates to Holocene, Khasfian, Sibakhan, or some other techno-complex with façonnage reduction. Sibakhan can be discarded as a viable option. The Phase II material post-dates the Phase I (Nejd Leptolithic) soft hammer blades. Presumably, the Nejd Leptolithic post-dates the hard hammer blade technology of the Sibakhan Industry, which is posited to date to OIS e or earlier. Therefore, the T preforms must belong to an industry considerably later than the Sibakhan. Figure - presents error bars for metrics of unbroken éclats de taille between T, Holocene, and Khasfian sites. The sites are broken down to individual findspots, primarily to differentiate between early stage activities that took place at SH, versus later stage manufacture at SH. In addition, it was demonstrated that, while T and A are both Holocene bifacial assemblages, T represents a separate, later technological phase. Ironically, most of the significant metric differences occur between thinning debitage from the two Holocene assemblages: relative platform size (p=.), blank volume (p=.), blank area (p=.), index of elongation (p=.), index of flattening (p=.), and platform angle (p=.). In the case of T, material was considerably more diminutive, and pressure flaking much more prevalent than any other assemblage. Aside from this clear difference, there do not appear to be any discernible patterns or affinities among éclats de taille. This may simply be because the method of façonnage flaking precludes any significant metric variations. Turning to the bifacial tools themselves, Figure - presents metric error bars for the completed bifaces as well as preforms, categorized by techno-complex (Sibakhan is included this time to provide a relative comparison). While the Sibakhan and T bifaces are larger than Holocene or Khasfian, T preforms are considerably thinner than Sibakhan forms. This is also apparent in the 291 fact that areas are similar, but there is a difference in volume; indication that the T preforms were too large and too thin to become Sibakhan bifaces. All T metrics fall within the range of Holocene and Khasfian forms, with the exception of biface area. In this case, T preforms have a greater area than Khasfian bifacial tools, although their volumes and thicknesses are comparable. So, while the two bifacial assemblages have the same thickness, T specimens are significantly larger. Taking into account the diminutive size of completed Holocene tools versus the size of Khasfian Foliates, the preforms seems to be on their way to becoming foliates. For example, a relatively large late stage preform that is  cm long and  cm wide was probably meant to be a Khasfian Foliate, ranging  –  cm long and  –  cm wide, rather than a pressure flaked point or trihedral rod that are typically less than five centimeters long and under three centimeters wide. While the assemblages from Bir Khasfa and T both exhibit an exclusively façonnage reduction strategy, at T there is only hard hammer thinning. The reason for this may lie in the discrepancy between lipping at SH and SH, which was determined to result from the use of primarily hard hammer percussion in early stages of reduction, followed by soft hammer for final thinning and shaping. Since T was interpreted to be a workshop for initial thinning and no tools were found, it is possible the usable, late stage preforms were carried off site and completed at a separate location. If this is the case, then it is not surprising that soft hammer flaking was not used at T, since the function of the findspot was only to create preforms for subsequent bifacial tools. Unfortunately, analysis of T technology and typology, in respect to the other façonnage assemblages, has not turned up any significant patterns that can be used to argue for a concrete relationship with any particular industry. Evidence points to Khasfian Foliates, but the association is tenuous and must be further verified. It is equally plausible T may belong to an industry not discussed in the analysis. Again, more work is needed to solve this puzzle. 292 A SH T T.II SH A SH T T.II SH A SH T T.II SH Figure -.  confidence intervals comparing complete thinning flake metrics in bifacial sites. 293 A SH T T.II SH A SH T T.II SH A SH T T.II SH Figure -. 294 cont. Holocene Khasfian Sibakhan T.II Holocene Khasfian Sibakhan T.II Holocene Khasfian Sibakhan T.II Figure -.  confidence intervals comparing complete biface metrics in bifacial industries. 295 Holocene Khasfian Sibakhan T.II Holocene Khasfian Sibakhan T.II Holocene Khasfian Sibakhan T.II Figure -. 296 cont. Summary Sibakhan is considered to be the earliest group of assemblages, indicated by the simple, hard hammer blades struck from lineal cores, manufacture of relatively flat, hard hammer bifacial handaxes, and limited presence of unidirectional-convergent Levallois. Considering this suite of techno-typological features, in conjunction with the palaeoenvironmental record presented in Chapter , as well as the position of A-A on the landscape, Sibakhan assemblages most likely date to one of the pluvial phases preceding OIS . Based on techno-typological affinities with the Mugharan Complex in the Levant and presence of this suite of technologies at the Kapthurin Formation in East Africa (discussed in the next chapter), the most plausible bracket of time is between OIS  and OIS  ( –  kya), although OIS  ( –  kya) should not yet be discounted. Dating of the Nejd Leptolithic is only relative to the other industries defined in this dissertation. The greater presence of Levallois technology and absence of associated bifacial material are taken as indicators that the industry post-dates the Sibakhan. Contrasting degrees of patination at T clearly indicate that the Phase I (Nejd Leptolithic) cores are older than the façonnage artifacts from Phase II. If Phase II is Khasfian, then the Nejd Leptolithic must pre-date the Khasfian. Considering this relative chronological sequence, the Nejd Leptolithic may be associated with subpluvials during OIS e ( – ), OIS c ( –  kya), or OIS a ( –  kya). Unfortunately, while the Khasfian is perhaps the most important techno-typological entity considered in this dissertation because of its posited attribution to OIS  (when mtDNA analyses predict a divergence from the ancestral population), it is also the most poorly understood. Its existence depends upon establishing that it is not coeval with Holocene bifacial technologies, but is a separate, earlier industry. At present, there is no way to absolutely date the assemblage from Bir Khasfa, and no clues to be obtained by examining local landscape morphogenesis. The strongest temporal indication comes from identical bifacial foliates found in late MSA/early LSA assemblages from the Horn of Africa, placing the assemblage between approximately , and , years ago. Assuming the Khasfian does, indeed, post-date the Nejd Leptolithic, then the suggested time 297 frame for this industry is  –  kya. There is overwhelming evidence for an intense pluvial at this time throughout the Rub al-‘Khali (McClure , , ; Schulz and Whitney , ; Lézine et al. ), as well as being the genetically-predicted time frame for the modern human expansion across the Arabian Corridor (Quintana-Murci et al. ; Macaulay et al. ; Thangaraj et al. , ). The final chapter will synthesize these newly defined techno-complexes with the pre-existing chronological scheme in southern Arabia. These data are then applied to the Arabian Corridor Migration model presented in Chapter  and used to examine the variety and development of South Arabian lithic technologies during the Middle and Upper Pleistocene. 298 Table -. A, I A, II A, III A-A A SH SH-SH T T T T debitage  .  .  .  .  .  .  .  .  .  .  . Artifact class by site. tools  .  .  .  .  .  .  .  .  .  .  . debris  .  .  .  .  .  .  .  .  .  .  . total            cores  .  .  .  .  .  .  .  .  .  .  . Table -. A, I A, II A, III A-A A SH SH-SH T T T T            - . . . . . . . . . . . Cortex percentage and ratios by site. -  .  .  .  .  .  .  .  .  .  .  . 299 -  .  .  .  .  .  .  .  .  .  .  . -  .  .  .  .  .  .  .  .  .  .  . Total            -: - ratio . . .  . . . . . . . Table -. A, III Core types by site. T T  .  .  .  .  .  .  .  . flake flake-blade/let blade/let Levallois Kombewa Total A, I A, II  .  .  .  .  .  .  . — — A-A A SH SH- T T  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . Table -. A, II  .  .  .  .         . . . .   . .  .  .  .  .     .  .  . . . .  .   . .  .   . .  . . . .  .   . .  . . . .  .  .  .  .  .  .  .  .  .  .  .  .  . . .     .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . . . . A, III A-A A SH SH-SH . . . . Blank types by site. T  .  .  .  .  .  .  .  .  .  . T  .  .  .  .  .  .  .  .  .  .    .   . .  .     . . . . . .  .  .    . .  .  . .  . T  .  .  .      . . . . . T  .  .  .  .  .  .  .  .  . Flakes regular cortical debordant subtotal A, I     300 Blades blade bladelet cortical burin spall debordant subtotal  .  .  .  .  .  . Levallois Levallois flake Levallois point subtotal Eclats de Taille  CTE misc core tablet subtotal    Total  . Table -. A, I        . . . . . . . Scar patterns exhibited on blanks and working surface(s) of cores. A-A A SH SH-* T T T* T A, II A, III 301  .  .  .  .   .  .  .  .  .  . .  .  .  .  .  .  .  . Debitage unidirectional uni.-crossed uni.-parallel convergent transverse bidirectional radial crested natural-crested lateral-crested transverse-crested distal-crested Total          .    . .  .  .  .  .  .  .  . .  .  .  .  .  .  .  . .  .  .  .  .  .  .  .  .  .  .  .  .  .  . .  .  .  .  .  .  . .  .  .  .  .  . .  .  .  .  .  .  . .  .  .  . .  .  .   .  .  .  .  .  . .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . Cores unidirectional uni.-crossed uni.-parallel convergent bidirectional bi.-parallel radial alternating Total  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . *excludes bifacial thinning flakes Table -. A, II  . . A, III A-A A SH SH- T T Frequency of tool types. T T A, I  .  .  .  .  .  .  .  .  .  .  .  .  . . . . . .  . .  .  .  .  .  .  .  . .  . 302 .  .  .  .  .  .  .  .  .  .  . Unifacial sidescraper endscraper misc scraper notch double notch denticulates burins ret. pieces Lev. flake/point truncation tranchet awl drill composite tool subtotal  .  .  .  .   .  .  .  .   .  .   .   .  .   .  .  .  .  .   .   .  .   .  .  .    .  .  .  . . .  .  .  .  .  .   .  .  .  . .  . Bifacial handaxes foliates points trihedral rod “bifacial tile” preforms subtotal Total  .  .  .  .  .  .  .  .  .  .  . Chapter 8 CONCLUSIONS He is crazed with the spell of far Arabia, They have stolen his wits away —Walter de la Mare, Arabia This dissertation has presented a model of early modern human expansion from the Horn of Africa into southern Arabia during the Upper Pleistocene (Figure -): the same expansion that culminated in the peopling of Australasia. The proposed scenario was articulated by incorporating genetic, palaeoanthropological, palaeoclimatic, and archaeological data to describe the timing and dynamics of this demographic movement. The Arabian Corridor Migration Model The Arabian Corridor Migration Model is based on the premise that hunter-gatherers, adapted to the climate and ecology of contiguous refugia, occasionally expanded their range onto the Arabian Peninsula. The demographic shifts were facilitated by the spread of the savanna phytogeographic zone to Arabia during Upper Pleistocene pluvials. These pluvials were triggered by abrupt fluctuations in the Southwest Indian Ocean Monsoon System. There have been several oscillations in the intensity of the monsoon regime over the last half million years, roughly correlated with glacial/interglacial cycles (Clemens et al. ; Zonneveld et al. ; Reichart et al. ; Petit et al. ; Muzuka ; Lückge et al. ; Ivanova et al. ; Higginson ). The consequences of these episodic intensifications were increased wind and precipitation, as well as a northward shift in the circulation patterns that brought seasonal storms far into the interior of South Arabia (Sarnthein ; Kutzbach ), transforming the largest sand sea in the world into vast grasslands dotted with playa lakes and populated by a variety of ungulates. 303 Figure -. Potential routes of migration out of Africa. As pronounced as these pluvials were, arid periods were equally severe (Anton ; McClure ; Gardner ; Clark ; Glennie and Singhvi ). The sprawling dune fields and expansive gravel pavements that cover the present-day landscape are a testament to the magnitude of hyperarid conditions that prevailed during glacial periods. As global forcing mechanisms pushed the monsoon southward, Arabia’s water supply was cut off, causing inhabitants of the region to leave or die out. Consequently, South Arabia repeatedly experienceed tabula rasa events—periods in which it was wiped clean of hominid occupation. At the onset of each wet phase, populations in neighboring refugia (i.e. East Africa, the Levant, and India) found themselves presented with unoccupied grasslands able to support a considerable biomass—a cornucopia for human groups adapted to hunting on the savanna. It was this perpetual oscillation between habitable and uninhabitable conditions that defined prehistoric occupation in southern Arabia: it acted as a pump by drawing 304 populations in when conditions ameliorated, then forcing them to depart when they environment became desiccated. There have been many studies that document the timing of climatic fluctuations in southern Arabia by analyzing environmental signals such as travertines (Clark ; Burns et al. , ), palaeosols (Anton ; Sanlaville ; Brinkmann and Ghaleb ), lacustrine sediments from the playa lakes and associated fauna (McClure and Swain ; McClure , ; Schulz and Whitney , ; Lézine et al. ), palaeosalinity of the Red Sea, Arabian Sea, and Arabian Gulf (Reichart et al. ; Petit et al. ; Sirocko ; Siddall et al. ; Ivanova et al. ), aeolianites (Glennie and Singhvi ), and landscape morphogenesis (Hötzl and Zötl ; McClure ; Anton ). From these data, there is evidence for at least three pronounced Upper Pleistocene pluvial phases. The first, and most intense, is correlated with OIS e (~ – ~ kya), followed by a sub-pluvial that occurred during OIS a (~ – ~ kya). The OIS a pluvial ended abruptly with the onset of the OIS  hyperarid phase (~ – ~ kya), at which time there was large-scale aeolian deposition throughout the Rub al-‘Khali basin, converting the grasslands back into a vast sea of sand. By the beginning of the next cycle of monsoon intensification around  kya, the massive dunes that had formed in the Rub al-‘Khali during the preceding hyperarid regime had stabilized and created a series of natural interdunal reservoirs. As soon as rainfall increased, this new vertical relief facilitated a chain of over , freshwater lakes strung across the heart of the basin (McClure , , ). The lake period lasted until about  kya, at which time another hyperarid period set in during OIS  ( –  kya). Presumably, the OIS  hyperarid phase triggered another tabula rasa throughout southern Arabia, and set the stage for Early Holocene colonization by expanding Neolithic groups from the Near East, and, possibly, the Horn of Africa. While the possibility of modern human expansion into southern Arabia is not a new concept, until relatively recently restrictive socio-political (as well as geographic) conditions have hindered archaeological exploration of this region. The Arabian Corridor has been excluded from most studies dealing with human origins simply due to the paucity of research projects geared to answer this question. In the last few years, some palaeoanthropologists have begun to consider 305 this route (e.g., Tchernov b; Lahr and Foley ; Cavalli-Sforza ; Rose ; Stringer ; Rose a, b; Petraglia and al-Sharekh ; Field et al. ; Mellars ), as new archaeological and genetic findings suggest the Arabian Corridor was critical in the spread of modern humans. One of the most significant indications of this dispersal route was the recognition of mtDNA haplogroup M (Quintana-Murci et al. ; Forster and Matsumura ; Macaulay et al. ; Thangaraj et al. ). Scholars have found the earliest evidence for this clade among groups living in the Horn of Africa, who supposedly branched from the ancestral population into South Arabia, India, and Southeast Asia between  –  kya. The proposed bracket of time correlates with the OIS  wet phase in southern Arabia. So, according to these analyses, it is feasible that hunter-gatherers from the Horn of Africa began expanding eastward into Arabia (and points beyond) around , years ago. There is no a priori reason, however, to assume that the founder population (basal M) originated in East Africa. Given the overlap in projected dates of divergence in India and Africa, as well as the uncertainty in using mtDNA for timing evolutionary events, it is equally plausible that the haplogroup M lineage originated in India. Middle/Upper Pleistocene Chronology There are not yet archaeological data outside of Africa supporting the Arabian Corridor Migration model. Chapters  and  illustrated this point by presenting the state of research in southern Arabia, which is severely hampered by a dearth of stratified sites. Consequently, there are virtually no absolute or relative dates with which to anchor the myriad lithic scatters strewn about the surface of the land. That is not to say assemblages dating to the Pleistocene have not been recorded. Since the very first expedition conducted by Caton-Thompson in the ’s, archaeologists have written about Palaeolithic findspots throughout the region. Based on the findings of numerous research projects over the last sixty years, it seems there has been intermittent hominid occupation in South Arabia since the late Pliocene. 306 This dissertation is particularly concerned with late Middle/Upper Pleistocene lithic elements. The work of the Central Oman Pleistocene Resarch Program has culminated in the definition of three new industries, sensu latu, that potentially fall between  to  kya: the Sibakhan, the Nejd Leptolithic, and the Khasfian. These new industries will be placed within the framework of the pre-existing body of knowledge from previous archaeological work conducted in South Arabia, and then compared to MP/MSA and UP/LSA lithic industries found the Near East and Horn of Africa. SIBAKHAN The Sibakhan Industry was first collected at the site of Saiwan, located at the northern extent of the Haushi-Huqf Depression, just south of the ad-Dakhliyah alluvial plain (Biagi ). The name Saiwan itself is not Arabic, but a mistranslation of oil well “C-” (as-Sabri, personal communication). To correct this misnomer, the region was renamed Ghanim; Saiwan does not appear on any recent maps of Oman. Since the original name of the type site has been stricken from the record by local authorities, Sibakhan has instead been chosen as a taxonomic designation for this industry. The Sibakhan assemblages have all been found near sibakh (indicative of ancient playas), therefore, a name loosely translated as “people of the lakes” has been chosen. In addition to the type site, the Sibakhan is known from the complex of findspots at A-A, as well as the ephemeral scatter at A. The industry is characterized by large, thin, asymmetrical, biconvex handaxes, which sometimes exhibit marginal retouch on one side. Core technologies include hard hammer unidirectional-parallel blades from flat cores with naturally elongated working surfaces, infrequent unidirectional-convergent Levallois cores with unfaceted platforms, and, most rarely, discoids. Tools with nibbling retouch, often on the lateral edge of blades, are prominent. Because this tool type is present at all three Sibakhan sites, it is tentatively considered a type fossil. There have been reports of potential Sibakhan occurrences throughout several geomorphic zones in southern Arabia. Zarins et al. (, ) and Bulgarelli () mention Upper Acheulean bifaces from surface sites in Saudi Arabia and Yemen, which they classify as “Mousterian of Acheulean 307 Tradition” due to their association with Levallois cores and discoids. The “Group Two” assemblage from Jebel Barakah is comprised of radial cores and the tip of a large biface with flat, invasive, hard hammer retouch. McBrearty () points out similarities between this assemblage and material from the African Middle Stone Age. Inizan and Ortlieb () discovered assemblages in their work around Shabwa, western Hadramaut, that had features characteristic of the Sibakhan. Two occurrences were found on the plateau above Wadi Muqqah; these showed a convergent Levallois technology that included a variation resembling Nubian Type I cores. There were also blade cores at Wadi Muqqah, which, like A-A and Saiwan, were not considered part of a “true” blade technology in the sense of volumetric, prismatic reduction. Platforms are primarily straight, and débordant blades are common. The Wadi Muqqah findspots yielded slightly asymmetrical flat bifaces, generally limande in shape, with marginal retouch on one or multiple edges. Biagi () posited the material from Saiwan was Upper Acheulean, dating to around , years ago. Citing techno-typological correlates in Europe, Inizan and Ortlieb () also speculated the material from Wadi Muqqah fell somewhere around the late Lower Palaeolithic or Middle Palaeolithic. With so little data, dating of the Sibakhan must remain ambiguous, spanning a wide range from approximately  to  kya (between OIS  and OIS ; it is assumed Arabia was uninhabitable during OIS ). Based on techno-typological correlates presented below, the Sibakhan probably falls within the earlier half of this range. Presumably, this industry is associated with a pluvial episode, based on the assumption that hominids could not have survived in southern Arabia during an arid phase; testing this hypothesis will be part of COPR’s future research agenda. Sibakhan connections with the Near East The Levantine Middle Palaeolithic is comprised of three Mousterian entities; to some, these are separate industries (e.g., Meignen ) while others consider them to be related facies (e.g., Copeland ; Jelinek ). At Tabun Cave, all three were discovered within a stratigraphic sequence, and, consequently, are thought of as temporal phases within the Mousterian (Jelinek 308 ). The sequence at Tabun Cave has long been used as the type site for classifying the Levantine Mousterian, and the facies are usually referred to as Tabun D type (Early Levantine Mousterian), Tabun C type (Late Levantine Mousterian), and Tabun B type (Terminal Levantine Mousterian). The Tabun D type Mousterian is characterized by the production of blades and elongated Levallois points struck from unidirectional, and more rarely, bidirectional and centripetal Levallois cores. Dominant tool types include retouched Levallois points, simple sidescrapers, retouched Levallois flakes, as well as Upper Palaeolithic tool types such as burins, awls, endscrapers, and truncations (Hours et al. ; Monigal ). In addition, there are truncated-faceted pieces, which are ubiquitous throughout the Levantine Middle Palaeolithic (Solecki ; Nishiaki ; GorenInbar ; Delagnes ). Tabun C focused on the production of broad ovoid Levallois flakes with faceted striking platforms, which were struck from centripetally prepared Levallois tortoise cores. The toolkit also includes truncated-faceted pieces (Solecki ; Nishiaki ; Goren-Inbar ; Delagnes ) diverse sidescraper types, denticulates, and a very low frequency of retouched Levallois points or flakes. There are few Levallois points or blades, and no Upper Palaeolithic tool types (Hours et al. ; Marks ). Tabun B blanks are characteristically short, broad-based Levallois points with faceted striking platforms (in some cases chapeau de gendarme). They were removed from unidirectional-convergent cores, although radial preparation is also present. Blade production was limited to the byproducts of core preparation. The toolkit is comprised of unretouched and retouched Levallois points, a variety of sidescraper types (Copeland ; Marks ), and truncated-faceted pieces (Solecki ; Nishiaki ; Delagnes ; Goren-Inbar ). Like Tabun C type, Upper Palaeolithic tools do not occur within this entity. Dates for Tabun D generally range between , and , , Tabun C dates between , and , , and Tabun B falls within the , – ,  range (Grün et al. ; Mercier et al. ; Grün and Stringer ). There is a disparity between lithic sequences in the Carmel and surrounding margin. Industries in the arid zones were almost exclusively Tabun D type 309 (with the exception of Tor Faraj and Tor Sabiha) throughout the Middle Palaeolithic and up to the MP/UP transition in the Negev at ,  (Schwarcz et al. ; Marks ; Schwarcz and Rink ), while Tabun C and B assemblages are seldom reported outside of the Mediterranean zone. Assemblages from the Zagros Mountains exhibit a Middle Palaeolithic tradition similar to the Levant (Garrod ; McBurney ; Hole and Flannery ; McBurney ; Hume ; Dibble ; Bewley ; Baumler and Speth ; Dibble and Holdaway ; Solecki and Solecki ; Minzoni-Deroche ; Lindly ). In the most general terms, the technology of this region is dominated by Levallois point production. Also resembling the Levant, truncated-faceted pieces are prominent. These assemblages differ from the Levantine Mousterian in that they have a much greater intensity of reduction and resharpening (Lindly ). Mousterian points with heavily invasive retouch are common, while in the Levant these tools are seldom found. Chronometric dates from the Zagros are scarce; two radiocarbon dates obtained from Shanidar Cave yielded results around , and , , and are considered minimal for the Zagros Mousterian sequence (Solecki ). In terms of inter-regional affinities, the Sibakhan (or any other South Arabian industry for that matter) does not resemble anything from the Middle Palaeolithic in the Near East due to the significant presence of flat handaxes at A-A, Saiwan, and Wadi Muqqah. No bifacial implement has ever been reported from any Levantine or Zagros Mousterian assemblage. Furthermore, truncated-faceted pieces are absent and there are no Levallois tortoise cores in the Sibakhan. While unidirectional-convergent Levallois cores are present at A-A as well as Wadi Muqqah, this technology seems closer to the Nubian rather than the Levantine Mousterian because of the presence of distinct Nubian Type I cores (Guichard and Guichard ; Marks a). If the Sibakhan dates to OIS  – , the industry would be contemporary with the late Lower Palaeolithic/initial Middle Palaeolithic in the Levant. This period includes later Acheulean facies of the Mugharan techno-complex, including Acheulo-Yabrudian, Yabrudian, and Amudian (Hours et al. ; Copeland ; Jelinek ). The dating of the Mugharan is the subject of considerable debate, as the numerous TL and ESR measurements from Tabun Layer E (Acheulo-Yabrudian and Amudian) indicate a span 310 of time around OIS  or OIS  (Mercier et al. ), which is considerably earlier than other dated occurrences from Arida, Hummal b, Umm el-Tlel, Umm Qubeiba, Zuttiyeh, and Yabrud I.; these later sites yielded a range generally between  –  kya (Porat et al. ). Recent work at Qesem Cave corroborate an early date for the Mugharan (or at least the Amudian facies of this complex), producing a range between  and  kya (Barkai et al. ; Gopher et al. ). Taking into account marine data from Adlun and Bezez C (Porat el al. ), the most logical placement for the end of the Mugharan is around mid-OIS  (~ kya). In general terms, Mugharan assemblages are dominated by core technologies, although flat bifacial handaxes are present in low frequencies, sometimes described as “Micoquian” in appearance. Little or no Levallois technique is used, platform faceting is uncommon, and flakes and blades have high-angled platforms. As for the tools, flake blanks are typically chosen for retouch (with the exception of the Amudian), there is a high percentage of sidescrapers, denticulates are rare, and Quina retouch is prevalent. The various industries are distinguished by the presence or absence of bifaces, as well as the predominance of blades. At Tabun XI, excavators noted a gradual increasing and then gradual decreasing of blade production throughout the Acheulo-Yabrudian and subsequent Amudian layers; they speculated this patterning may indicate that the inhabitants of the cave had greater need for elongated blanks during certain periods (Jelinek ). While the Sibakhan does not resemble any single facies within the Mugharan, it bears a number of general Mugharan-like characteristics. The core technology is closest to the Amudian, which is characterized by the production of large hard hammer blades with prominent bulbs (Garrod ; Garrod and Kirkbride ; Jelinek , ; Monigal ). Like the Sibakhan, Amudian blade cores are single platform parallel, single platform convergent, and partially prismatic. Discoids make up a small percentage of cores, although Levantine Mousterian types such as Levallois tortoise and point cores are absent. The Sibakhan and the Amudian have significant numbers of retouched blades with marginal, nibbling retouch along one lateral edge. The bifacial component of the Sibakhan is only similar to the Acheulo-Yabrudian. PreAurignacian and Amudian assemblages have little or no façonnage elements, while the Yabrudian 311 is typified by asymmetrical bifacial scrapers with heavy Quina retouch. Sibakhan bifaces are asymmetrical and sometimes exhibit continuous retouch along one or both edges. One specimen from A-A might be classified as a bifacial scraper, although the scraper retouch is not nearly as heavy as seen in the Yabrudian. Sibakhan connections with the Horn of Africa Unfortunately, there have been far fewer Middle/Upper Pleistocene excavations in the Horn than in the Levant; therefore there is considerably less resolution in both the chronological sequence and definition of the various lithic industries. Absolute dating of the MSA material from sites in the Horn is tenuous because the majority of radiometric measurements are based on obsidian hydration (Michels et al. ; Clark et al. ), a method which has routinely been demonstrated to be inherently flawed (e.g., Ridings , ; Anovitz et al. ; Hull ). More reliable potassiumargon dates were obtained from Gademotta and Kulkuletti (Wendorf and Schild ; Wendorf et al. ; Wendorf et al. ), although the dated material comes from ash lenses within reworked volcaniclastic sediments and colluvium, therefore providing only an indirect, possibly disturbed measurement. Two lithic complexes, sensu latu, are known from the Horn that fall between OIS  and OIS . The earlier group of assemblages is known as the Acheulo-Levalloisian or Abyssinian Fauresmith, characterized by handaxes, cleavers, discoids and some centripetal Levallois core reduction (Clark ; Clark ). Scholars have debated whether this industry represents the terminal Acheulean or beginning of the Middle Stone Age (Brandt ). The assemblages from Lake Ziway (Wendorf and Schild ) and Garba III at Melka Kunture (Bailloud ; Hours ; Chaivallon et al. ) may be roughly coeval with the Sibakhan if a date around OIS  is accepted. At Garba III, excavators described an assemblage with earlier forms such as handaxes and cleavers, as well as Stillbay points and sidescrapers. The authors note platforms are rarely faceted, and there are traces of Levallois technology in the uppermost strata. Unfortunately, the analysis at Melka Kunture is primarily concerned with tools, and there is little 312 mention of core types. Based on the published illustrations, there does not seem to be any significant blade component, unlike Sibakhan or Mugharan occurrences where this is one of the defining characteristics. A sequence of four occurrences was documented at Gademotta and three at Kulkuletti. The oldest date for the sequence comes from a cemented ash layer directly above the lowest archaeological level (Eth--B) at Gademotta. The sample yielded a K-Ar date of , ± ,. A second sample was taken from an ash layer below the lowest strata at Kulkuletti (Eth--), with a K-Ar date of , ± ,. The minimum K-Ar date, associated with the second most recent occupation at Gademotta (Eth--), is , ± , (Wendorf et al. ). Unifacial and bifacial foliates are well-represented within all of the Lake Ziway assemblages, although the excavators note the frequency of these tools decreases through the sequence. “Mousterian” and bifacial points are sometimes derived from Levallois blanks, and subsequently modified via flat, invasive flaking across one or both faces. A particular strategy of resharpening foliate tools was observed—the Prodnik technique—accomplished by striking a lateral “burin spall” from the point tip. This removal effectively set up a new platform from which rejuvenation flakes were subsequently struck (Wendorf and Schild ). The core technology is consistent at all findspots: a predominantly prepared-core reduction strategy with opposed platform cores, discoids, and Levallois cores. In most cases, convexity on the working face of the Levallois core has been achieved by bidirectional or centripetal preparation, while unidirectional-convergent Levallois cores are absent. The blade index steadily increases over time, ranging from . at Eth--B to as high as . at Eth--. In nearly every case, the blades were obtained from Levallois cores with an elongated working surface (ibid.). Although it is outside of the geographic region under consideration, brief mention must be made of the Kapthurin Formation, a series of bedded tuffs interstratified with palaeosols in the Tugen Hills of central Kenya. The well-dated sequence yielded numerous argon-argon dates between approximately  and  kya (Deino and McBrearty ). The findspot of Koimilot Locus  at Kapthurin was interpreted as transitional between the Acheulean and MSA, based on the presence 313 of both Acheulean handaxes and prepared cores. The site, associated with the most recent dates from the formation, produced a unidirectional-parallel and unidirectional-convergent blade assemblage similar to the Sibakhan, as well as the earliest examples of Levallois technology in Africa (Tryon and McBrearty , ; Tryon , , ; Tryon et al. ). There are several sites/findspots in the Horn that likely fall within OIS , most notably: Gorgora (Leakey ), Porc Epic (Teilhard de Chardin ; Breuil ; Breuil et. al. ; Vallois ; Perlès ; Clark and Williamson ; Clark et al. ; Pleurdeau , ), K’One (Kurashina ), and a series of buried sites in the Awash Valley (Yellen et al. ). Gorgora and Porc Epic exhibit the classic characteristics of the Somaliland Stillbay—a significant percentage of bifacial foliates and unifacial points with flat, invasive retouch (bifacial and unifacial points may, in fact, simply represent a continuum of types). Material from Aduma is somewhat different; while classic MSA/Stillbay points are present, blade technology is infrequent. Researchers argue that the “Aduma Industry” exhibits a regional development characterized by diminutive Levallois cores (ibid.). At Porc Epic, points (both unifacial and bifacial) are the dominant form—the frequency is close to  of the entire toolkit (Clark et al. ). In addition to these façonnage elements, there are sidescrapers, denticulates, and backed pieces (including lunates) within the Porc Epic assemblage. A variety of core types were identified at Porc Epic including biconical discoids, centripetal as well as unidirectional-convergent Levallois cores, and partial/fully volumetric blade cores. The blade cores are unidirectional and bidirectional, most often with parallel scar patterns on the working face(s). There are a few cases in which blades were removed from lineal cores with supplementary distal platforms, resembling the reduction strategy from the Nejd Leptolithic. Ridge (crested) blades are reported from the assemblage, indicating core maintenance. Blade-proportionate pieces account for nearly  of the debitage (Pleurdeau ). Radiometric measurements place the assemblage somewhere within the span of  –  kya (Michels et al ), although these measurements are discounted because they were obtained by obsidian hydration dating. 314 There is no significant relationship between Sibakhan and the Somaliland Stillbay assemblages. There are no Stillbay points, backed pieces, centripetal Levallois cores, volumetric blade cores, or ridge blades from the Sibakhan assemblages. While some unidirectional-convergent Levallois technique is present at A-A, it does not occur in nearly the same frequency as in the Somaliland Stillbay (nor is it necessarily coeval with the rest of the Sibakhan). Furthermore, large, flat cordiform and sub-triangular bifacial handaxes are not a component of the MSA in the Horn. The only OIS  assemblages with some degree of technological and typological affinities to the Sibakhan come from K’One, Ethiopia, described as a series of obsidian workshop floors (Kurashina ). K’One is a volcanic caldera containing several buried MSA sites eroding from the badlands landscape. Eleven meters of sediment have been exposed at various points in the erosional landscape above the underlying ignimbrite bedrock. Geologists identified three fining-upward alluvial cycles represented in the sequence, suggesting a situation of prolonged deposition in a swampy zone at the distal end of an alluvial fan. Obsidian and pitchstone flows stretch at least two kilometers along the upper reaches of the caldera rim, about five hundred meters from one of the workshop floors, at its nearest point. Unfortunately, there are no radiometric dates from the site. While the various K’One findspots all clearly belong to the same general lithic tradition, the assemblage with some similarity to the Sibakhan was from Locality  Extension. The predominant core technologies include unidirectional-convergent Levallois cores (ibid.:, fig. -:a-i) and unidirectional-parallel blade cores (ibid.:, fig. -:f ). Like the Sibakhan, discoids are present but rare (.). Just over  of the cores from this assemblage are classified as Nubian Type I. Therefore, the K’One core reduction strategies show the same variety as the Sibakhan, particularly if one includes Inizan and Ortlieb’s () findspots from Wadi Muqqah with the list of Sibakhan occurrences, since Wadi Muqqah also produced evidence for Nubian Type I cores. The bifacial elements from the Sibakhan and K’One are roughly similar as well. K’One bifaces are asymmetrical and range from thin (Kurashina :, fig. -:g) to moderately thin (ibid.:, fig.-:i), and are found in a variety of ovoid and cordiform shapes ranging from approximately six to twelve centimeters in length. The façonnage implements from Locality  315 Extension are manufactured from blanks, and  are partially bifacial. Sibakhan bifaces, on the other hand, are manufactured from plaquettes and are all fully bifacial. This disparity may be attributed to either a difference in technologies, or, simply, variation in the locally available raw material. Since it has markedly different features from other Somaliland Stillbay assemblages, K’One was considered to be unique within the East African MSA: The presence of technologically distinct Nubian cores at K’One distinguishes it from other MSA occurrences from the Horn and eastern Africa which have been broadly termed “Somaliland Stillbay” (Clark :) and “Stillbay Culture” (Anthony ) respectively, but makes it more comparable with the Nubian Mousterian. (Kurashina :) This technological and typological mixing is not surprising, given its location at the northern extent of the Horn, abutting Northeast Africa. There are differences between K’One and the Nubian Mousterian, however, as no classic Levallois points were found within the latter (Guichard and Guichard ; Marks a). If it is related to the Nubian Mousterian, presumably K’One must date to an East African pluvial, when population exchange across the Sahara was possible. Considering all of the possibilities presented above, the Sibakhan appears technologically and typologically closer to the Near East than to Africa. The unidirectional-parallel hard hammer blades that are characteristic of the Sibakhan are not present in Africa during OIS , although may be present in the uppermost levels of Melka Kunture and at Koimilot Locus  of the Kapthurin Formation, which would place it in OIS , if not earlier. There are far more similarities between Sibakhan and Mugharan assemblages. This possibility brings up the question of north-south demographic movement across Arabia during the late Middle Pleistocene, which, in turn, allows for multiple avenues of speculation regarding hominid migration at this time. If one accepts Zeitoun’s () taxonomic designation of the Zuttiyeh skull as Homo sapiens, which is associated with an Acheulo-Yabrudian assemblage, then it is germane to tentatively link the Lower/Middle Palaeolithic technological transition with the introduction of this new species into the Levant. Perhaps archaic Homo sapiens expanded northward through Arabia into the Levantine Corridor, or, it is equally 316 plausible hominids moved from the Levant to southern Arabia. Alternately, the Sibakhan may originate from an Indian source population, an option considered in greater detail at the end of this chapter. NEJD LEPTOLITHIC The Nejd Leptolithic is named as such for its ubiquity in the form of surface scatters throughout the Nejd Plateau in northern Dhofar, although sites fitting the description have been reported several hundred kilometers westward in Mahra and Hadramaut (Amirkhanov ; Zarins, personal communication), well beyond the confines of the Nejd. Recent COPR fieldwork conducted in May of  even discovered several leptolithic findspots in the Maradi Hills of northern Oman, exhibiting a nearly identical range of technologies. The industry is characterized, first and foremost, by the predominance of bladeproportioned pieces, averaging around  of the debitage. Lipping has a frequency between  and  on blanks, indicating the common use of soft hammer percussion. This is not a “true” blade industry in the prismatic sense; the cores are primarily lineal, with unidirectional-parallel or bidirectional removals across the working surface. The distal platforms on the bidirectional cores are supplementary, and seem to be intended for convexity maintenance rather than obtaining usable blanks. Platform faceting is rare (less than ten percent), although it is apparent on some of the cores. The lineal cores that exhibit both convexity maintenance and platform preparation are considered a variant of Levallois. Regarding the toolkit, there is little indication for the manufacture of bifacial implements, likewise, there are no bifacial thinning or shaping pieces among the debitage. A variety of sidescrapers types, denticulates, notches, and retouched pieces comprise the dominant tool categories. At both T and T, ambiguous bifacial implements were recognized (Figure -). In both cases, the tools were plano-convex in cross-section and elongated ovoid in plan view. The specimen at T was classified as a miscellaneous scraper, while the T object was interpreted as a crude preform. 317 Three assemblages were recognized—T, T.I, and T—as Nejd Leptolithic, and a similar blade technology was noted at nearly every palimpsest scatter examined during the COPR  survey. Amirkhanov () recognized potential Nejd Leptolithic assemblages at four stratified and nineteen surface sites in the western Hadramaut. The stratified sites are secondary deposits of artifacts within alluvial aggradations. He describes the technology at these sites as flat cores with parallel removals along the working surface, which is the primary defining characteristic of the Nejd Leptolithic. The associated toolkits are comprised of endscrapers, points, awls, and knives; a different array of tools than those encountered on the Nejd. The Soviet team obtained a sample of datable material from an archaeologically sterile profile that was correlated with the basal layer of one of their leptolithic sites. The material yielded a radiocarbon date of , ± ,  (ibid.). Given the indirect dating of Amirkhanov’s purported Upper Palaeolithic material and the secondary position of the artifacts, these dates are given little weight. Unfortunately, findings from the COPR  campaign did not contribute any significant clues as to the dating of the Nejd Leptolithic. At Jibal Ardif  (T), the Phase I Nejd Leptolithic assemblage possessed a much different patina than the Phase II (Khasfian?) material: Phase I cores ranged from light to deep brown, and had mildly abraded arêtes, while the Phase II artifacts were pinkish-gray with virtually no signs of weathering. Since the site occurs on a deflated reg surface, both assemblages were subject to similar chemical weathering processes; therefore, the markedly different degrees of patina are considered a useful indicator of relative dating. If Phase II is indeed linked to the Khasfian Industry—a link that is only minimally supported at present—then the earlier Nejd Leptolithic industry represented in Phase I is tentatively correlated with OIS  ( –  kya). This proposed span of time rests upon the additional assumption that the Khasfian Industry dates to OIS . Clearly, the basis presented here for dating the Nejd Leptolithic rests more upon speculation than actual data, underscoring the importance of future research to further elucidate this industry. 318 Nejd Leptolithic connections with the Near East There do not appear to be any analogous assemblages in the Near East. While Monigal () describes an uninterrupted leptolithic tradition in the Levant from the late Lower Palaeolithic to the Upper Palaeolithic, none of the industries along this trajectory possess the same or a similar combination of techno-typological features as observed in the Nejd Leptolithic (summarized in the preceding section that discusses potential Near Eastern connections with the Sibakhan Industry). Nejd Leptolithic connections with the Horn If the Nejd Leptolithic dates to OIS , the industry would be coeval with either the Somaliland Stillbay (again, summarized in the Sibakhan section) or Hargeisan, if one accepts the earliest proposed dates for the basal archaeological layer at Midhishi . The Hargeisan Industry was first defined by Clark () in northern Somalia, while more recent excavations at Midhishi  have added greater resolution to this poorly understood techno-typological entity (Brandt ; Brandt and Brook ; Gresham ; Brandt ). Midhishi  cave is situated in the Midhishi Tog Valley of northern Somalia, a spring-fed drainage system that dissects the Somali plateau just northeast of the town of Ceerigaabo. The site is one of three adjacent caves located on the northern side of the valley, about nine meters above the wadi floor. It is a small, round limestone cavity with a shallow stratified sequence. Excavations revealed silty sediments containing eboulis and a dense concentration of artifacts reaching one meter in depth in some portions of the deposit. The bottom of the sequence is brecciated, and contained only MSA artifacts. The excavators describe three “culture-stratigraphic” units within the sequence, exhibiting both MSA and LSA features. The two lower units were classified as typical Somaliland Stillbay, while the uppermost unit was ascribed to the Hargeisan, with its combination of Stillbay features as well as some microliths, endscrapers, and prismatic blade cores. Upon analysis of the three assemblages, it was posited that the entire sequence may be related to the Hargeisan, to some degree: 319 Indeed, all three culture-stratigraphic units, but particularly the upper two, could be viewed as variants of the Hargeisan as defined by Clark…Except for the low frequency of microliths, the upper two culture-stratigraphic units at Midhishi  could easily be interpreted as Hargeisan, and perhaps the low frequency of microliths is within the range of variation for the, as yet, poorly defined industry. In summary, the Midhishi  material best fits Clark’s  culture-history if considered as a “Somaliland Stillbay” industry that exhibits some differences. There appears to be a Hargeisan ‘flavor’ to the material… (Gresham :) Of the , recovered lithics, almost all are manufactured on chert. Points comprise  of all modified artifacts, including Levallois, unifacially retouched, and bifacial points made by façonnage reduction. Scrapers (mostly endscrapers) form slightly over  of the toolkit, and the remaining  comprises burins, awls, backed pieces, and truncated pieces. Cores are unidirectional-convergent Levallois, discoids, bidirectional, and single-platform blade cores. Levallois technology is common in all three levels, accounting for . of cores. Gresham () conducted an analysis of the points from the site, identifying three techniques used in their production: ) “true” Levallois made by preferential, unidirectionalconvergent flaking across a flat working surface, ) non-Levallois with unidirectional recurrent flaking across a flat working surface, and ) non-Levallois with bidirectional recurrent flaking across a flat working surface. He notes that methods  and  resemble the transitional MP/UP technology reported from Boker Tachtit (Marks ; Volkman ). Radiocarbon dates from within the brecciated portion of the deposit produced a date of greater than , , while a sample from the unit above the breccia yielded results of , ±  . A single TL date carried out on material from the lower breccia corroborates the earlier carbon date, indicating a date sometime prior to , . In addition to the radiometric measurements from within the cave, researchers have offered additional evidence for dating the lower brecciated deposit. Sediment samples obtained from tephras in the nearby Golis Mountains were dated to , ± , and , ± , years ; analysis indicates these tephras formed during more humid climatic regimes. The excavators reason that the basal breccia of Midhishi  became cemented during one or more wet phases, and must thus predate 320 the latest date for the tephra—,  (Brandt ; Brandt and Brook ; Gresham ; Brandt ). So, considering possible inter-regional affinities, there is nothing remotely resembling the Nejd Leptolithic in the Levantine Mousterian. While the Somaliland Stillbay has an element of unidirectional-parallel blade technology on flat/partially-volumetric cores (e.g., Porc Epic), Levallois points and Stillbay points are not found in the Nejd Leptolithic; thus, a relationship between the Nejd Leptolithic and the Somaliland Stillbay can be discounted. The Hargeisan, as well, bears minimal resemblance because of its characteristic high frequency of points as well as backed pieces and microliths. There are two other possibilities regarding the development of this industry. It may derive from a blade technology found in the MP/MSA of India, or it may have arisen from the lineal, hard hammer technology of the Sibakhan Industry. The former option is explored later in this chapter. If the latter possibility is true, the Nejd Leptolithic represents an autochthonous development; a period in which the proposed process of hunter-gatherer contraction and expansion into and out of Arabia was disrupted and replaced by regional demographic stability. KHASFIAN The Khasfian Industry is defined by its fossile directeur: the Khasfian Foliate (e.g., Figure ). This tool is a small (mean length=. cm, mean width=. cm) leaf-shaped or ovoid biface, with flat, soft hammer retouch and a lenticular cross-section. They were first recorded by J. Pullar at the site of Bir Khasfa in the northern reaches of Wadi Arah (Pullar ), and have since been reported from surface collections throughout Oman (Smith ; Villiers-Petocz ; Rose a). Upon her initial examination of the material, Pullar recognized potential affinities with artifacts from the Horn, drawing comparisons with the “Somaliland Doian” Industry (Pullar :). Since that time, similar foliates have been recognized within the Rub al-‘Khali assemblages collected by ARAMCO workers and studied by C. Edens (, ). The tools are classified as Type  in the Rub al-‘Khali Neolithic scheme. There are other soft hammer bifacial implements known 321 from Saruq-Facies (D-Group) assemblages along the coast (Kapel ; Uerpmann ; Charpentier ; Uerpmann and Uerpmann ). Some Arabian scholars have posited the Khasfian, D-Group, Saruq-Facies, and Type  foliates all belong to a single industry falling somewhere between the th and th millennia  (e.g., Charpentier ). It should not be assumed, however, that the production of all bifacial foliates in southern Arabia is related to a single time period. Morphologically similar tools are reported from various industries around the world as early as the Middle Pleistocene, such as Korolevo level a in Transcarpathia (Kulakovska , ) or Galeria Pesada in Portugal (Marks et al. ). Foliates appear in the Micoquian (Bosinski ; Soressi ) and Solutrean Industries (Smith ); industries that occur in the European Middle and Upper Palaeolithic, respectively. Nearly identical specimens (albeit fluted) are a characteristic of the early Holocene in North America (Wormington ), and are even known from Middle Holocene sites in Central Australia (Valoch ). After reviewing every documented prehistoric findspot in southern Arabia with bifacial implements, it has been demonstrated that the Type  foliates have never been found with pressure-flaked implements in a secure stratigraphic context. In addition, there is variability among these different types of foliates: the Saruq-Facies materials are typically thick biconvex or plano-convex in cross-section, with subtriangular shapes, while Khasfian Foliates are lenticular in cross-section and exhibit a range of shapes from ovoid to limande. In addition, the related core technologies are markedly different: the Khasfian material is associated with discoidal cores. Finally, non-chipped stone materials that are often found with South Arabian Early/Middle Holocene surface scatters such ground stone, ostrich egg shell, marine shell, ochre, beads, hearths, and structures are absent from the Khasfian localities. Finds at SH-SH (Bir Khasfa) are used to argue for a pre-Holocene date of the Khasfian/ Type  foliates. The massive complex of findspots at Bir Khasfa encompasses a variety of different lithic occurrences, exhibiting the diagnostic characteristics of Neolithic, Nejd Leptolithic, Khasfian, and possibly other undefined groups. Fieldwork conducted by both COPR and Pullar () observed clearly distinct areas of the complex associated with these different technologies. The  campaign systematically sampled an area exclusively with Khasfian Foliates, and found no related 322 Neolithic (or other) material. A diagnostic Type  Neolithic lanceolate and trihedral rod were collected from an inselberg some  m to the south; it is noteworthy that both specimens had no weathering and minimal varnish, while all of the foliates (and associated cores and debitage) have a moderate orangish-brown patina. There was evidence for a technologically diagnostic core technology within the SH-SH assemblage, albeit comprising a miniscule portion of the reduction sequence. One of the specimens from SH is a high-backed discoidal core with a flat working surface, possibly an exhausted centripetal Levallois core, which would suggest MSA/MP affinities. In addition, there was a large sidescraper with classic Quina retouch, also diagnostic of the MSA/MP. The posited dates for the Khasfian Industry are based on the assemblages’ repeated association with interior playas and ameliorated drainage systems on the Nejd and in the Rub al‘Khali. Given the date of the two “lake phases,”  –  kya and  –  kya, it is proposed that the foliates fall within the earlier wet period. This association with OIS  must remain within the realm of speculation in the absence of a datable, stratified site. Khasfian connections with the Near East Like the Nejd Leptolithic, there are no coeval analogous assemblages in the Near East. Levantine technologies are exclusively based on core reduction, while Khasfian is almost entirely façonnage. Near Eastern MP/UP transitional reduction strategies are typified by unipolar or bipolar elongated point cores that evolve from preferential Levallois to recurrent, prismatic, non-Levallois varieties. A type fossil of this phase is the Emireh Point, which is not known from Arabia (Marks ; Volkman ; Clark et al ). The few cores that are associated with the Khasfian are high-backed centripetal Levallois—a technique that is uncommon within most Levantine assemblages, with the exception of Tabun C. In the case of Tabun C, however, there are no bifacial implements, nor debitage to suggest the manufacture of bifacial pieces. 323 Khasfian connections with the Horn OIS  is a problematic and poorly understood bracket of time within the Horn. Radiometric dates from this phase are non-existent, therefore one can only speculate as to what industries might be associated with it, based solely upon relative dating from buried deposits. The most probable candidates are a late stage of the Somaliland Stillbay (what Leakey called “Upper Stillbay”), the Somaliland Magosian, or the Hargeisan. The Somaliland Stillbay and Hargeisan were summarized in the preceding sections dealing with the Nejd Leptolithic and Sibakhan Industries. Somaliland Magosian is known from excavations in the first half of the th century, although there has been no modern work investigating this industry. Material ascribed to the Somaliland Magosian was found in stratified contexts overlying Somaliland Stillbay material at Gure Warbei (Clark ), Bur Yassin (ibid.); and Bur Hakaba (ibid.), all in southern Somalia. Gure Warbei is a rockshelter situated on the eastern slope of Bur Eibe, formed by a massive overhanging granite boulder. The large shelter contained several meters of in situ sediments. Excavations, carried out in , produced a sequence of four distinct industries ranging from early Holocene to Upper Pleistocene. Layer C was posited to represent a transitional phase between Magosian and Doian, while Layer D is typical of the Somaliland Magosian. The Doian/Magosian assemblage (Layer C) overlies a thick layer of sterile, reddish-brown, sandy sediment, which, in turn, overlies laminated yellow sandy silt washed into the shelter from an opening in the back. The Layer D assemblage was recovered just above the sterile sandy layer, and excavators note that there appears to be a major temporal break between C and D, the latter deposited during a considerably wetter climatic regime (ibid.). Layer D yielded diminutive unifacial and bifacial points, endscrapers, sidescrapers, outils écaillés, and a burin. Typical LSA components such as thumbnail scrapers and backed microliths are extremely rare. Cores are also closer to the MSA than LSA; discoids are present and there no bipolar cores. Because of these characteristics, Clark (ibid.) speculated that this basal assemblage may actually be Somaliland Stillbay, or transitional Magosian/Stillbay material. 324 Bur Hakaba, also known as “the Rifle Range Site,” is an open-air stratified deposit located about  km south of Bur Eibe. The site is situated in a small embayment at the north end of a bur. Four industries were recognized at the site, following a sequence similar to that seen at Gure Warbei. Typical LSA artifacts begin to appear in Layer D, characterized as Doian, with microliths, hollow based points, bifacial and trihedral lanceolates and foliates, and blades and bladelets. This layer overlies a hard, compact, calcareous wind-blown sandy stratum (Layer E). Below the sterile aeolian material is Layer F, a gritty black earth containing a dense concentration of workshop debris. Layer F includes bifacial and unifacial points appearing in a variety of subtriangular shapes, sidescrapers, burins (bec-de-flute), and outils écaillés. Unlike Gure Warbei, blades comprise a major element of this assemblage. Straight-backed and arch-backed blade tools are frequent (comprising the majority of the toolkit), as are unidirectional, bidirectional, and multiple platform blade cores. There are also some discoids and bipolar cores. Because of the exceptional preservation, non-lithic artifacts such as ostrich eggshell fragments and worked bone were also recovered. The Layer F assemblage was thought to have been deposited during a wet-phase preceding a period of aridity (ibid.). Khasfian affinities with Africa must be approached with caution, because the former entity is only known from a single findspot; therefore, the full range of techno-typological variability is unknown. Based on the characteristics observed at SH-SH, the material seems closest to the “Upper Stillbay” or even early Somaliland Magosian from Layer D at Gure Warbei. The Khasfian does not have any LSA tool types such as backed pieces, thumbnail scrapers, or microliths. Like Gure Warbei, there are no bipolar cores or blades, only discoids. The predominant Khasfian tool types are sidescrapers and diminutive foliates. One sidescraper collected from SH exhibits classic Quina retouch, another indication of MSA, rather than LSA technology. There are striking similarities between the assemblages from Bir Khasfa and a findspot recently reported from northern Sudan, called Station One (Rose b). Station One is a homogenous surface scatter that was interpreted to be an intrusive element representing a late MSA hunter-gatherer range expansion from East Africa, based on techno-typological characteristics, raw material selection, and comparisons with local Nilotic MP and UP industries. While core reduction 325 is much more prevalent at Station One than at Bir Khasfa, the same suite of technological and typological features is present: high-backed centripetal Levallois cores and discoids (Figure -), as well as Stillbay points (Figure -). Furthermore, the foliates (“Somaliland Stillbay” in Africa, “Khasfian” in Arabia) are found in the same range of shapes (i.e. ovoid, foliate, and limande), and dimensions of these specimens are similar. Given these similarities, it is possible this technotypological package is a representative fingerprint of the genetically predicted groups expanding out of the Horn during OIS . In the case of Station One, the expansion does not appear to have extended further north beyond Nubia. As for Arabia, the Khasfian Industry may mark the first phase in the modern human journey eastward, which culminated in the peopling of Australasia. If this is the case, one expects techno-typological connections with the next region on the southern dispersal route: the Indian subcontinent. Figure -. High-backed centripetal cores from Station One (a,b) and Bir Khasfa (c). 326 Figure -. Khasfian Foliates (b,d,f ) from Bir Khasfa and “Stillbay” points from Station One (a,c,e). 327 A PASSAGE TO INDIA The Indian subcontinent stretches from Pakistan to Nepal and from the Himalayan Plateau to the Indian Ocean, covering over three million square kilometers. Within this area are a range of geomorphic and phytogeographic zones; it is not necessary to summarize archaeological material from all of these diverse regions. For the purposes of this dissertation, only the material from the Thar Desert will be reviewed in detail, due to its location in the northwestern portion of India. This region is closest to Arabia, abutting the eastern side of the Indus Valley, and possibly underwent similar dynamics in facilitating the expansion of modern humans. Findspots are known from the Didwana complex, Budha Pushkar, the Luni Valley complex, and Hokra (James and Petraglia ). The Thar Desert exhibits a palaeoclimatic record mirroring the Rub al-‘Khali, with phases of aridity interspersed with periods of amelioration (Andrews et al. ; Kar et al. ; Deotare et al. ). James and Petraglia () propose a dispersal mechanism nearly identical to that presented in this dissertation—wet phases brought greater resource availability, which permitted populations to expand into the region. Also like southern Arabia, OIS  and OIS  heralded pluvials throughout the desert, punctuated by aridity that began to set in around ,  (Misra ; Deotare et al. ). Indian MP/MSA lithic technology is characterized by prepared and non-prepared core reduction strategies. Levallois cores are prevalent (both centripetal and convergent techniques), as are discoids. Material from the Thar Desert is distinguished by the high frequency of unidirectionalparallel blades (Allchin et al. ). The blades are not derived from prismatic cores, but rather from lineal working surfaces. These flat cores are not systematically reduced; just a few blades and flakes were struck off before discard (ibid.). James and Petraglia () observe that the leptolithic technology may have become more developed during the span of the Middle Palaeolithic. The MP toolkit comprises sidescrapers, denticulates, notches, and burins. While points are prevalent in others parts of the subcontinent, they are absent from Thar Desert assemblages (Misra ; James ). The UP/LSA of the Indian subcontinent is characterized by an increase in blade production as well as variability in assemblage composition. The period began at least as early as , years ago, indicated by dates on loess overlying a UP/LSA assemblage at Site  in Pakistan (Dennell et al. ). 328 James () proposes that this technological phase be termed Late Palaeolithic, choosing a taxonomy that emphasizes the difference between both African and European trajectories. The blanks do not exhibit the same degree of standardization as seen in the Aurignacian and other subsequent industries in Europe. Examination of blade core reduction strategies in the Upper Pleistocene does not reveal any sudden shift at the onset of the Late Palaeolithic; on the contrary, early Late Palaeolithic assemblages possess a comparable leptolithic technology as the preceding Middle Palaeolithic. These assemblages are only distinguished by the frequency of blade blanks, it is not until later in the Late Palaeolithic that prismatic blade cores, geometric microliths, and bladelets emerge (James and Petraglia ). The blade tradition found in the Middle/early Late Palaeolithic of the Thar Desert resembles the Nejd Leptolithic. These two industries have a reduction sequence characterized by blades struck from flat, lineal cores. In addition, working surfaces exhibit unidirectional-parallel and bidirectional scar patterns. The toolkits of both industries comprise a variety of typical Middle Palaeolithic types (e.g., scrapers, denticulates, notches), while points and foliates are absent. This comparison of the two regions is cursory, to say the least. It is meant to intimate potential connections and highlight an avenue for further inquiry, for the the Upper Pleistocene of both regions must undergo considerably more investigation before a vigorous analysis can be undertaken. Summary In the beginning of this dissertation, there was a question as to whether the terms Middle Palaeolithic/Upper Palaeolithic or Middle Stone Age/Late Stone Age should be applied to Arabia. In other words, should the Peninsula adopt the East African or North African/Near Eastern Palaeolithic nomenclature (or something entirely different)? This presentation of Palaeolithic archaeology in southern Arabia has demonstrated affinities with all three contiguous refugia at different times: East Africa, India, and the Levant. Therefore, it is still too early to presuppose one taxonomic scheme over another, since the Pleistocene lithic sequence is an amalgamation of different regional traditions. It is recommended that, pending 329 greater resolution, the Arabian sequence remain demarcated by Lower/Middle/Upper Pleistocene terminology, rather than specific technological phases that might lead one to assume affinities with a single region or tradition. It was argued that the oldest documented industry in this dissertation, the Sibakhan, is most similar to the Mugharan Complex (specifically the Amudian Industry), a late Lower Palaeolithic entity in the Levant with dates spanning between OIS  and OIS . Following the Sibakhan is the Nejd Leptolithic, tentatively associated with OIS . The origins of this industry are ambiguous; similar flat blade cores are known from MSA sites such as Porc Epic, although other elements such as points and foliates, common at Porc Epic, are missing from the Arabian assemblages. An entirely different possibility is that the Khasfian and Nejd Leptolithic are, in fact, coeval facies representing different functional ends of a linked techno-typological continuum. If this were the case, the merged industry would then resemble the Somaliland Stillbay. Alternately, the Nejd Leptolithic has affinities with Middle Palaeolithic/early Late Palaeolithic assemblages in the Thar Desert of northwestern India, suggesting connections between these two regions sometime between OIS  and OIS . The interpretation that the Khasfian and Nejd Leptolithic belong to a single industry is not likely—there are no overlapping techno-typological features. Phase I and II material from T, interpreted as Nejd Leptolithic and Khasfian respectively, exhibit markedly different degrees of patination. According to the hypothetical chronology proposed in this dissertation, the two industries are separated by the OIS  hyperarid phase, a period that would have triggered a tabula rasa event in southern Arabia. In this scenario, a clear break in the Upper Pleistocene Arabian lithic tradition is predicted. If the Khasfian Industry does indeed fall between  –  kya, its widely different technology supports this notion of an abrupt technological shift. Khasfian material is close in both façonnage and core reduction to the “Upper Stillbay” or “Somaliland Magosian,” the latter of which was thought to represent an MSA/LSA transitional phase. 330 Because of their posited dates, The Khasfian and Nejd Leptolithic industries may be associated with the genetically predicted modern human expansions out of Africa and/or India. In both cases, these early humans brought with them a seemingly “MP/MSA” technology and toolkit. Therefore, this stage of the expansion had little to do with systemic behavioral changes that facilitated their colonization of the globe. On the contrary, movement into Arabia was initiated by opportunistic hunter-gatherers following an episodic pulse of their known niche. The Arabian Corridor Migration Model relies heavily upon mitochondrial DNA analyses, in which the divergence of haplogroup M is used to argue for a modern human expansion out of Africa and into Arabia sometime between OIS  and OIS  (Quintana-Murci et al. ; MacaMeyer ; Underhill et al. ; Forster ; Lovell et al. ; Forster and Matsumura ; Macaulay et al. ; Thangaraj et al. ). Based on the posited dates of divergence, there is no a priori reason to presuppose the ancestral population expanded from East Africa rather than the Indian subcontinent. Quintana-Murci et al. () predict a divergence of , ± ,  in East Africa, and , ± ,  in India. Considering this margin of error, it is equally plausible the ancestral population emigrated from the latter region. Further supporting an Indian expansion is the work of Karmin (), which reconstructs the global phylogenetic tree for all major branches of human mitochondrial DNA. Rather than sprouting directly from node L, as the traditional out of Africa model predicts, haplogroup M diverges from basal node M somewhere in eastern Eurasia or India and then branches back into Africa (Figure -). At this point there is insufficient archaeological data to vigorously test either hypothesis. The findings from southern Arabia suggest connections with both East Africa (indicated by the Khasfian Industry), and India (indicated by the Nejd Leptolithic). Either or both may represent footprints of the M expansion. The assumption that there were tabula rasa events during OIS , , and  must also be verified. Depending upon the magnitude of aridity, some groups may have survived in small pockets along the Dhofar Escarpment and/or Yemeni Highlands; another reason exploration of the Dhofar caves should be made a high priority for future research. 331 Figure -. Global phylogenetic tree based on mtDNA haplogroups (adapted from Karmin : p., fig. ) The chronology and demographic scenarios proposed in this dissertation are tentative. This work is intended to put forth a testable model, instigate debate, and generate interest in the Pleistocene of South Arabia. These three newly defined industries still require much articulation 332 since the artifacts do not yet come from in situ stratified contexts. There is much to be done in South Arabia: a variety of undefined lithic industries carpet the landscape. It is only a matter of time until stratified Palaeolithic sites, which have eluded Arabian archaeologists for over half a century, will be unearthed. In ‘Happy Arabia,’ however, the bell has just rung and the door swings ajar, still awaiting those equipped to open it wide. For here lies one of the last of the world’s relatively unknown areas. It is this land of sand and dust, Bedawi and jambiya, buried temple and mud palace, that awaits our return to bring new life and activity. (Phillips :vii-viii) 333 Appendix A LITHIC ATTRIBUTE ANALYSIS FROM COPR 2004 ASSEMBLAGES 334 condition (top); lipping (middle); edge preparation (bottom) 335 artifact class (top); blank type (middle) raw material (bottom) 336 shape (top); transverse cross-section (middle); longitudinal cross-section (bottom) 337 cortex  (top); cortex location (middle); raw material (bottom) 338 striking platform (top); dorsal scar pattern (bottom) 339 distal termination (top); axis (bottom) Appendix B SITES RECORDED DURING 2004 COPR CAMPAIGN 340 341 342 343 Appendix C LITHIC RAW MATERIAL FROM AREAS SURVEYED DURING 2004 COPR CAMPAIGN 344 345 346 347 Appendix D bajada – a wide depositional glacis weakly dipping, formed by coalescent alluvial fans. bir – well. éclat de taille/éclats de taille – biface thinning flake. jebel/jibal – mountain or hill. façonnage – technological term for invasive flaking across one or both faces of a blank. fossile directeur – type fossil. jol – plateau. khabra – sediments consisting of very fine alluvial material deposited in closed or almost closed internal basins; typically they do not exceed a few decimeters in thickness, and consist of interlaminated beige silt and aeolian sand. khareef – rainy season between July and September brought on by the SW Indian Ocean Monsoon System. Mahri – Semitic language spoken by Beduin tribes inhabiting the Nejd Plateau and Mahri province of eastern Yemen. It is closely related to Shahri spoken in the Dhofar Mountains, and Soqotri spoken on the island of Soqotra. reg – form of desert pavement in which surface is deflated, exposed mantle of hard sedimentary rock, often silicified and/or ferruginized strata. Rub al-‘Khali – Empty Quarter. sabkha/sibakh – a salt-flat ordinarily found in proximity to sand dunes. These highly saline areas of sand and/or silt form just above the water-table where the sediments are cemented by evaporites left from seasonal ponds. Shahri – Semitic language spoken by Beduin tribes inhabiting the Dhofar Escarpment in southwestern Oman. It is closely related to Mahri and Soqotri. shuquq – flat interdunal plain, surface is often mottled by silty patches covered with a thin veneer of sand. wadi/widian – seasonal drainage channel. wilaya/wilayat – province, governorate, or region. GLOSSARY 348 Adams, R., P. Parr, M. Ibrahim, and A.S. al-Mughannum. . Preliminary report on the first phase of the Comprehensive Survey Program. Atlal :-. Allchin, B., A. Goudie, and K. Hegde. . The prehistory and palaeogeography of the Great Indian Desert. London: Academic Press. Ambrose, S. . Holocene environments and human adaptations in the Central Rift Valley, Kenya. Ph.D. Dissertation, University of California, Berkeley. Ambrose, S. . Chronology of the Later Stone Age and food production in East Africa. Journal of Archaeological Science :-. Ambrose, S. . Population bottleneck. In Genetics, volume , edited by R. Robinson. New York: Macmillan Reference. Pp. -. Amirkhanov, H.A. . The Acheulean of southern Arabia. Sovetskaia Arkheologiia :-. Amirkhanov, H.A. . The Paleolithic of South Arabia. Moscow: Academy of Sciences. Amirkhanov, H.A. . Research on the Palaeolithic and Neolithic of Hadramaut and Mahra. Arabian Archaeology and Epigraphy :-. Amirkhanov, H.A. . Acheulean sites of southern Arabia. In The Hadramaut, edited by P. Griaznevich. Moscow: Eastern Literatures Board. Andrefsky, W. . Lithics: macroscopic approaches to analysis. Cambridge Manuals in Archaeology. Cambridge: Cambridge University Press. Andrews, J., A. Singhvi, R. Kuhn, P. Dennis, S.K. Tandon, and R. Dhir. . Do stable isotope data from calcrete record Late Pleistocene monsoonal climate variation in the Thar Desert of India? Quaternary Research :-. Anovitz, L.M., J.M. Elam, L.R. Riciputi, and D.R. Cole. . The failure of Obsidian Hydration Dating: sources, implications, and new directions. Journal of Archaeological Science :. Anthony, B. . Prospect Farm. Palaeoecology of Africa :-. Anthony, B. . The Stillbay question. Actes du VIe session du congrès panafricain de préhistoire, Dakar, . Chambéry: Les Imprimeries Réunis de Chaméry. Anthony, B. . The Prospect Industry--a definition. Ph.D. Dissertation, Harvard University, Cambridge. BIBLIOGRAPHY 349 Anton, D. . Aspects of geomorphological evolution: paleosols and dunes in Saudi Arabia. In Quaternary Period in Saudi Arabia. Vol. : sedimentological, hydrogeological, hydrochemical, geomorphological, and climatological investigations of western Saudi Arabia, edited by A.R. Jado and J.G. Zötl. Wien: Springer-Verlag. Bailloud, G. . Les gisements paléolithiques de Melka-Kunturé (Choa). Institut éthiopien d’archéologie, cahier , Addis Ababa. Barkai, R., A. Gopher, S.E. Lauritzen, and A. Frumkin. Uranium series dates from Qesem Cave, Israel, and the end of the Lower Palaeolithic. Nature :-. Barker, D. . Perforated shell and stone objects from Sharm. Arabian Archaeology and Epigraphy :-. Bar-Yosef, O. . Pleistocene connections between Africa and Southwest Asia: an archaeological perspective. African Archaeological Review :-. Bar-Yosef, O. . Middle Palaeolithic human adaptations in the Mediterranean Levant. In The Evolution and Dispersal of Modern Humans. Tokyo: Hokusen-sha Publishing. Bar-Yosef, O. . The contributions of Southwest Asia to the study of the origin of modern humans. In Origins of Anatomically Modern Humans, edited by M. Nitecki and D. Nitecki. New York: Plenum Press. Bar-Yosef, O. . The Middle and early Upper Palaeolithic in Southwest Asia and neighboring regions. In The Geography of Neandertals and Modern Humans in Europe and the Greater Mediterannean, edited by O. Bar-Yosef and D. Pilbeam. Boston: Peabody Museum Bulletin . Pp. -. Baumler, M., and J. Speth. . A Middle Palaeolithic assemblage from Kunji Cave, Iran. In The Paleolithic Prehistory of the Zagro-Taurus. Edited by D.I. Olszewski and H. Dibble. Philadelphia: The University Museum, University of Pennsylvania. Bayer, H.J. . General geology of western Saudi Arabia. In Quaternary Period in Saudi Arabia. Vol. : sedimentological, hydrogeological, hydrochemical, geomorphological, and climatological investigations in central and eastern Saudi Arabia, edited by S.S. Al-Sayari and J.G. Zötl. Wien: Springer-Verlag. de Bayle, D.H. . Première mission de recherches préhistoriques en République Arabe du Yemen. L’Anthropologie :-. Bednarik, R.G. . Seafaring in the Pleistocene. Cambridge Archaeological Journal :-. Bewley, R. . The Cambridge University archaeological expedition to Iran . Excavations in the Zagros Mountains: Houmian, Mir Malas, Barde Spid. Iran :-. Biagi, P. . An Early Palaeolithic site near Saiwan (Sultanate of Oman). Arabian Archaeology and Epigraphy :-. Boëda, E. . Approche technologique de la variabilité de la méthode Levallois: industries de Bagarre et de Corbehm (Pas-de-Calais). Bulletin de l’Association Française pour l’Etude du Quaternaire -:-. 350 Boëda, E. . De la surface au volume: analyse des conceptions de debitage Levallois et laminaire. In Paléolithique moyen recent et Paléolithique supérieur ancien en Europe, edited by C. Farizy. Pp. -. Bohannon, R.G., C.W. Naeser, D.L. Schmidt, and R.A. Zimmermann. . The Timing of Uplift, Volcanism, and Rifting Peripheral to the Red Sea: A Case for Passive Rifting? Borde, F. . L’évolution buissonnante des industries en Europe occidentale. Considérations théoriques sur le Paléolithique ancient et moyen. Bulletin de las Société Préhistorique :. Borde, F. . Typologie du paléolithique ancient et moyen. France: Université de Bordeaux. Borde, F. and M. Bourgon. . Le complexe Moustérien: Moustérien, Levalloisien, et Tayacien. L’Anthropologie :-. Bosinski, G. . Die Mittlepaläolithischen Funde im Westlichen Mittleuropa. Köln-Graz: Böhlau Verlag. Boucher de Perthe, J. de. . Antiquités celtiques et antédiluviennes. Bourgon, M. . Les industries Moustériennes et pré-Moustériennes du Périgord. Archives de l’Institut de Paléontologie Humaine, Mémoire no. . Paris: Masson et Cie Editeurs. Bowen, R., and U. Jux. . Birth of an ocean. In Afro-Arabian Geology: A kinematic view. London: Chapman and Hall. Brandt, S. . Archaeological investigations at Lake Besaka, Ethiopia. In Proceedings of the Eighth Pan-African Congress of Prehistory and Quaternary Studies, edited by R.E. Leakey and B.A. Ogot. Nairobi: The International Louis Leakey Memorial Institute for African Prehistory. Brandt, S.A. . A paleontological reconnaissance of the Somali Democratic Republic. L.S.B. Leakey Foundation News :-. Brandt, S.A. . The Upper Pleistocene and Early Holocene prehistory of the Horn of Africa.The African Archaeological Review :-. Brandt, S.A. . An archaeological survey of the proposed Baardheere Reservoir, upper Juba River, southern Somalia. Nyame Akuma :-. Brandt, S.A., and G.A. Brook. . Recent archaeological and paleonvironmental research in northern Somalia. Current Anthropology :-. Brandt, S.A., and T.H. Gresham. . L’age de la Pierre en Somalie. L’Anthropologie :-. Bräuer, G. . Africa’s place in the evolution of Homo sapiens. In Continuity or Replacement: controversies in Homo sapiens evolution, edited by G. Bräuer and F.H. Smith. Rotterdam: Balkemaand. Pp. -. Bräuer, G., and M.J. Mehlman. . Hominid molars from a Middle Stone Age level at the Mumba Rock Shelter, Tanzania. American Journal of Physical Anthropology :-. 351 Bräuer, G., and K.W. Rimbach. . Late archaic and modern Homo sapiens from Europe, Africa and southwest Asia: craniometric comparisons and phylogenetic implications. Journal of Human Evolution :-. Breuil, H. . Palaeolithic industries from the beginning of the Rissian to the beginning of the Würmian glaciation. Man :-. Breuil, H., and H. Kelley. . Les collections Africaines du Département de Préhistoire Exotique du Musée d’Ethnographie du Trocadéro: documents préhistoriques recueillis par la Mission du Bourg de Bozas, en Abyssinie (-). Journal de la Société des Africanistes :-. Breuil, H., P. Teilhard de Chardin, and P. Wernert. . Le Paleolithique du Harar. L’Anthropologie : -. Brézillon, M. . La denomination des objets de Pierre taillée, matériaux pour un vocabulaire de prehistorians de langue française. rd edition. IVe supplement à Gallia Préhistoire. Paris: Centre National de la Recherche Scientifiuqe. Brinkmann, R. and A.O. Ghaleb. . Late Pleistocene mollisol and cumulic fluvents near Ibb, Yemen Arab Republic. The Wadi al-Jubah Archaeological Project, Volume : environmental research in support of archaeological investigations in the Yemen Arab Republic,  – , edited by M.J. Grolier, R. Brinkmann, and J.A. Blakely. Pp. -. Brooks, A., D.M. Helgren, J.S. Cramer, A. Franklin, W. Hornyak, J.M. Keating, R.G. Klein, W.J. Rink, H. Schwarz, J.N. Leith Smith, K Stewart, N.E. Todd, J. Verniers, and J. Yellen. . Dating and context of three Middle Stone Age sites with bone points in the Upper Semliki Valley, Zaire. Science :-. Brunner, U. . Geography and human settlements in ancient southern Arabia. Arabian Archaeology and Epigraphy :-. Bulgarelli, G.M. . Evidence of Paleolithic industries in Northern Yemen. In Yemen, , Years of Art and Civilisation in Arabia Felix, edited by W. Daum. Innsbruck: Pinguin-Verlag. Burns, S.J., A. Matter, N. Frank, A. Mangini. . Speleothem-based palaeoclimate record from northern Oman. Geology ():-. Burns, S.J., D. Fleitmann, A. Matter, U. Neff, A. Mangini. . Speleothem evidence from Oman for continental pluvial events during interglacial periods. Geology ():-. Buzter, K., F. Brown, and D. Thurber. . Horizontal sediments of the Lower Omo Valley: the Kibish Formation. Quaternaria, Rome XI:-. Cann, R.L., M. Stoneking, and A.C. Wilson. . Mitochondrial DNA and human evolution. Nature :-. Carbonell, E., J.M. Bermúdez de Castro, J.L. Arsuaga, J.C. Diez, A. Rosas, G. Cuenca, R. Sala, M. Mosquera, and X.P. Rodriguez. : Lower Pleistocene hominids and artifacts from Atapuerca TD- (Spain). Science :-. Caton-Thompson, G. . Climate, irrigation, and early man in the Hadhramaut. Geographical Journal ():-. 352 Caton-Thompson, G. . Some Palaeoliths from South Arabia. Proceedings of the Prehistoric Society :-. Caton-Thompson, G. . The evidence of South Arabian palaeoliths in the question of Pleistocene land connections with Africa. Pan-African Congress on Prehistory :-. Cavalli-Sforza, L. . The DNA revolution in population genetics. Trends in Genetics :–. Chapman, R.W. . General information on the Arabian Peninsula. In Quaternary Period in Saudi Arabia. Vol. : sedimentological, hydrogeological, hydrochemical, geomorphological, and climatological investigations in central and eastern Saudi Arabia, edited by S.S. Al-Sayari and J.G. Zötl. Wien: Springer-Verlag. Charpentier, V. . La fouille du campement préhistorique de Ra’s al Junayz , (RJ) - Sultanat D’Oman -. Paléorient ():-. Charpentier, V. . Industries bifacials holocènes d’Arabie orientale, un exemple: Ra’s al-Jinz. Proceedings of the Seminar for Arabian Studies :-. Charpentier, V. . From the Gulf to Hadramawt: fluting and plunging processes in Arabia. In Proceedings of the First International Conference on the Archaeology of the U.A.E., edited by D. Potts, H. al-Naboodah, and P. Hellyer. Pp. -. Charpentier, V., D.E. Angelucci, S. Méry, and J.-F. Saliège. . Autour de la mangrove morte de Suwayh, l’habitat VIe – Ve millénaires de Suwayj SWY-, Sultanat d’Oman. Proceedings of the Seminar for Arabian Studies :-. Charpentier, V. and M.L. Inizan. . Fluting in the old world: the Neolithic projectile points of Arabia. Lithic Technology :-. Chavaillon, J. and J.-L. Boisaubert. . Prospection archeologique dans le Gemu-Gofa et la Basse Vallee de l’Omo. Documents Histoire Civilisation Ethiopienne, RCP , CNRS. Chavaillon, N., and J. Chavaillon. . Gotera: Un site Paléolithique recent d’Éthiopie. Éditions Recherche sur les Civilisations . Paris: A.D.P.F. Chavaillon, J., N. Chavaillon, F. Hours, and M. Piperno. . From the Oldowan to the Middle Stone Age at Melka-Kunturé (Ethiopia). Quaternaria :-. Chevrel, S., A. Berthiaux, J.P. Platel, and J. Roger. . Geological Map of Marmul, explanatory notes. Sultanate of Oman: Ministry of Petroleum and Minerals. Clapp, N. . The Road to Ubar. Boston: Houghton Mifflin Company. Clark, A. . Lakes of the Rub’ al Khali. Aramco (). Clark, G. . World Prehistory: a new outline. Cambridge: Cambridge University Press. Clark, G.A., J.M. Lindly, M. Donaldson, A.N. Garrard, N.R. Coinman, S.K. Fish, and D.I. Olszewski. . Palaeolithic archaeology in the southern Levant: preliminary report of excavations at Middle, Upper, and Epipalaeolithic sites in the Wadi al-Hasa. In The Archaeology of the Wadi al-Hasa, West-Central Jordan, vol. : excavations at Middle, Upper, and Epipalaeolithic Sites, edited by N.R. Coinman. Tempe: Arizona State University. Pp. -. 353 Clark, I. and J.-C. Fontes. . Paleoclimatic reconstruction of northern Oman based on carbonates from hyperalkaline groundwaters. Quaternary Research :-. Clark, J.D. . Short notes on stone age sites at Yavello, southern Abyssinia. Transactions of the Royal Society of South Africa ():-. Clark, J.D. . The Prehistoric Cultures of the Horn of Africa. Cambridge: Cambridge University Press. Clark, J.D. . The cultures of the Middle Paleolithic/Middle Stone Age. In The Cambridge History of Africa, vol. : from the earliest times to c.  BC, edited by J.D. Clark. Cambridge: Cambridge University Press. Clark, J.D. . Middle Stone Age of East Africa and the beginnings of regional identity. Journal of World Prehistory ():-. Clark, J.D. . The origin and spread of modern humans: a broad perspective on the African evidence. In The Human Revolution: Behavioural and Biological Perspectives on the Origins of Modern Humans, edited by P. Mellars and C. Stringer. Edinburgh: Edinburgh University Press. Pp. -. Clark, J.D., B. Asfaw, G. Assefa, J.W.K. Harris, H. Kurashina, R.C. Walter, T.D. White, and M.A.J. Williams. . Palaeoanthropological discoveries in the Middle Awash Valley, Ethiopia. Nature :-. Clark, J.D., Y. Beyene, G. WoldeGabriel, W.K. Hart, and P.R. Renne. . Stratigraphic, chronological and behavioural contexts of Pleistocene Homo sapiens from Middle Awash, Ethiopia. Nature :-. Clark, J.D., and K.D. Williamson. . Recent archaeological research in south-eastern Ethiopia (-): some preliminary results. Annales d’Ethiopie :-. Clark, J.D., K.W. Williamson, M.J. Michels, and C. Marean. . A Middle Stone Age occupation site at Porc Epic Cave, Dira Dawa (East-Central Ethiopia). Part I. African Archaeological Review :-. Clemens, S., W. Prell, D. Murray, G. Shimmield, and G. Weedon. . Forcing mechanisms of the Indian Ocean monsoon. Nature :-. Cleuziou, S., M.-L. Inizan, and B. Marcolongo. . Le peuplement pre-et protohistorique du systeme fluviatile fossile du Jawf-Hadramawt au Yemen. Paleorient :-. Copeland, L. . The Middle and Upper Paleolithic of Lebanon and Syria in the light of recent research. In Problems in prehistory: North Africa and the Levant, edited by F. Wendorf and A.E. Marks. Dallas: Southern Methodist University Press. Pp. -. Copeland, L. . Yabrudian and related industries: the state of research in . In Toward Modern Humans: the Yabrudian and Micoquian, Proceedings of a Congress held at the University of Haifa, November -, , edited by A. Ronen and M. Weinstein-Evron. Oxford: BAR International Series . Pp. -. Cotterell, B. and J. Kamminga. . Mechanics of pre-industrial technology. Cambridge: Cambridge University Press. 354 Crabtree, D.E. . Mesoamerican polyhedral cores and prismatic blades. American Antiquity (): -. Crabtree, D.E. . Flaking stone with wooden implements. Science :-. Day, M.H. . Early Homo sapiens remains from the Omo River region of south-west Ethiopia: Omo human skeletal remains. Nature :-. Debénath, A. and H.L. Dibble. . Handbook of Palaeolithic Typology, vol. : Lower and Middle Paleolithic of Europe. Philadelphia: University Museum, University of Pennsylvania. Deino, A., and S. McBrearty. . Ar/Ar dating of the Kapthurin Formation, Baringo, Kenya. Journal of Human Evolution :-. Delagnes, A. . Eclats a trancature inverse et enlevement posterieurs: reflexions nouvelles dun vieux debat. Bulletin de la Société Préhistorique Français :-. Dennell, R., H. Rendell, M. Halim, and E. Moth. . A ,-year-old open-air Paleolithic site at Riwat, northern Pakistan. Journal of Field Archaeology :-. Deotare, B., M. Kajale, S. Rajaguru, and N. Basavaiah. . Late Quaternary geomorphology, palynology, and magnetic susceptibility of playas in western margin of the Indian Thar Desert. Indian Geophysical Union ():-. Dibble, H. The Mousterian Industry from Bisitun Cave (Iran). Paleorient :-. Dibble, H., and S. Holdoway. The Middle Palaeolithic of Warwasi Rockshelter. In The Paleolithic Prehistory of the Zagro-Taurus. Edited by D.I. Olszewski and H. Dibble. Philadelphia: The University Museum, University of Pennsylvania. Doe, B. . Southern Arabia. New York: McGraw-Hill. Drennan, R.D. . Statistics for Archaeologists: a commonsense approach. New York: Plenum Press. Edens, C. . Towards a definition of the Rub al Khali ‘Neolithic’. Atlal :-. Edens, C. a. Archaeology of the sands and adjacent portions of the Sharqiyah. The Journal of Oman Studies, Special Report :-. Edens, C. b. The Rub al Khali ‘Neolithic’ revisited: the view from Nadqan. In Araby the Blest: Studies in Arabian Archaeology, edited by D. Potts. Copenhagen: Tusculanum Press. Pp. -. Edens, C. . A bladelet industry in southwestern Saudi Arabia. Arabian Archaeology and Epigraphy ():-. Edens, C. and T. Wilkinson. . Southwest Arabia during the Holocene: recent archaeological developments. Journal of World Prehistory (): -. El-Baz, Farouk. . The Arab world and space research: where do we stand? Emirates Lecture :-. Excoffier, L., and A. Langaney. . Origin and differentiation of human mitochondrial DNA. American Journal of Human Genetics :-. 355 Fairbanks, R.C. . A , year glacio-eustatic sea level record: influence of glacial melting rates on the Younger Dryas event and deep ocean circulation. Nature :-. Fernandes, C. A. nd. Mitochondrial DNA phylogeography of the small-spotted genet (Genetta genetta), a pan-African species introduced to Europe. Fernandes, C. A., E.J. Rohling, and M. Siddall. . Afro-Arabian biogeography and the Quaternary Red Sea: the origin of the Arabian populations of the Hamadryas baboon (Papio hamadryas) as a case study. Journal of Biogeography ():-. Ferring, C.R. . Technological Variability and Change in the Late Paleolithic of the Negev. Ph.D. Dissertation, Southern Methodist University, Dallas. Field, H. . New stone age sites in the Arabian Peninsula. Man :-. Field, H. . Ancient and Modern Man in Southwestern Asia. Coral Gables: University of Miami Press. Field, H. . Stone implements from the Rub’ al Khali, Southern Arabia. Man :-. Field, H. a. Stone implements from the Rub’ al Khali. Man. :-. Field, H. b. Carbon- date for a ‘Neolithic’ site in the Rub’ al Khali. Man :. Field, H. . Palaeolithic implements from the Rub’ al Khali. Man :-. Field, J.S., M.D. Petraglia, and M.M. Lahr. . The southern dispersal hypothesis and the South Asian archaeological record: examination of dispersal routes through GIS analysis. Journal of Anthropological Archaeology, in press. Forster, P. . Ice ages and the mitochondrial DNA chronology of human dispersals: a review. Philosophical Transactions of the Royal Society B — Biological Sciences :-. Forster, P., and S. Matsumura. . Did early humans go north or south? Science :-. Garbini, G. . Antichità Yemenite. Annali Instituto Orientale di Napoli :-. Gardner, R.A.M. . Aeolianites and marine deposits of the Wahiba Sands: character and palaeoenvironments. Journal of Oman Studies Special Repot no. , -. Garrod, D.A.E. . The Paleolithic of Kurdistan: excavations in the caves of Zarzi and Hazar Merd. Bulletin of the American School of Prehistoric Research :-. Garrod, D.A.E. . Acheuléo-Jabroudien et “Pré-Aurignacien” de la Grotte du Taboun (Mount Carmel); étude stratigraphique et chronologique. Quaternaria III:-. Garrod, D.A.E., and D. Kirkbride. . Excavation of Abri Zumoffen, a Palaeolithic rockshelter near Adlun in South Lebanon. Bulletin du Musée de Beyrouth :-. Gasse, F., R. Rognon, F.A. Street. . Quaternary history of the Afar and Ethiopian rift lakes. In The Sahara and the Nile, edited by M.A.J. Williams and H. Faure. Rotterdam: A.A. Balkema. Pp. -. 356 Gebel, H.G. . Evidence for Pre-Holocene industries in the greater Hili area. Emirates Natural History Group Bulletin :-. Gillespie, R. . Burnt and unburnt carbon: dating charcoal and burnt bone from the Willandra Lakes, Australia. Radiocarbon :-. Glennie, K.W. . The geological setting of the Wahiba Sands of Oman. Journal of Oman Studies Special Report, no. :-. Glennie, K.W. and A.K. Singhvi. . Event stratigraphy, palaeoenvironment and chronology of SE Arabian deserts. Quaternary Science Reviews :-. Gopher, A., R. Barkai, R. Shimelmitz, M. Khalaily, C. Lemorini, I. Heshkovitz, and M. Stiner. . Qesem Cave: an Amudian site in central Israel. Journal of the Israel Prehistoric Society :. Goren-Inbar, N. . Too small to be true? Reevaluation of cores on flakes in Levantine Mousterian assemblages. Lithic Technology :-. Goren-Inbar, N. . Quneitra: A Mousterian Site on the Golan Heights. Jerusalem: Qedem Monographs of the Institute of Archaeology . Goren-Inbar, N., C.S. Feibel, K. L. Verosub, Y. Melamed, M. E. Kislev, E. Tchernov, and I. Saragusti. . Pleistocene milestones on the out-of-Africa corridor at Gesher Benot Ya’aqov, Israel. Science ():-. Gramly, R.M. . Neolithic flint implement assemblages from Saudi Arabia. Journal of Near Eastern Studies ():-. Graziosi, P. . L’età della pietra in Etiopia e in Somalia. Rivista delle Colonie :-. Graziosi, P. . L’età della pietra in Somalia: risultati di una missione di ricerche paletnologiche nella Somalia Italiana. Firenze: G.C. Sansioni. Graziosi, P. . Missione preistorica Italiana in Somalia (estate ). Revista di Science Preistoriche : -. Graziosi, P. . Figure rupestre schematiche nell’Acchele Gazai (Etiopia). Revista de Scienze Preistoriche :-. Gresham, T. . An Investigation of an Upper Pleistocene Archaeological Site in Northern Somalia. M.A. Thesis, University of Georgia, Athens. Gresham, T., N. Carder, S.A. Brandt, M. Williams, and M. Frank. . Phase II Archaeological investigations in the proposed Baardheere Reservoir, Upper Jubba River, Somalia. Foundation for African Prehistory and Archaeology. Gainseville: University of Florida. Grün, R., C.B. Stringer, and H.R. Schwarcz. . ESR dating from Garrod’s Tabun Cave collection. Journal of Human Evolution :-. Grün, R., and C.B. Stringer. . Tabun revisited: revised ESR chronology and new ESR and Useries analyses of dental material from Tabun C. Journal of Human Evolution ():-. 357 Guichard, J. and G. Guichard. . The Early and Middle Paleolithic of Nubia: a preliminary report. In Contributions to the prehistory of Nubia, edited by F. Wendorf. Dallas: Southern Methodist University Press. Pp. -. Hammer, M.F., A.B. Spurdle, T. Karafet, M.R. Bonner, E.T. Wood, A. Novelletto, P. Malaspina, R.J. Mitchel, S. Horai, T. Jenkins, and S.L. Zegura. . The geographic distribution of human Y chromosomes. Genetics :-. Hanna, S. . Field Guide to the Geology of Oman, vol. : western Hajar mountains and Musandam. Muscat: The Historical Association of Oman. Hanna, S. and M. al-Belushi. . Introduction to the Caves of Oman. Muscat: Sultan Qaboos University. Hansen, T. . Arabia Felix: The Danish expedition of -, translated by J McFarlane and K. McFarlane. New York: Harper and Row. Harrison, D.L. . The Mammals of Arabia. London: Ernest Benn Limited. Harrison, D.L. . Mammals obtained in Dhofar by the  Oman flora and fauna survey. Journal of Oman Studies, Special Reports :-. Harrison, D.L., and J.J. Bates. . The Mammals of Arabia, second edition. Kent: Harrison Zoological Museum. de Heinzelin, J. de Braucourt. . Manuel de typologie des industries lithiques. Brussels: L’Institut Royal des Sciences Naturelles de Belgique. Herodotus. Histories, Books II-IV, translated by A.D. Godley. London: Loeb Classical Library. Higginson, M.J. . A solar (irradiance) trigger for millennial-scale abrupt changes in the southwest monsoon? Paleoceanography ():-. Hoelzmann P., B. Keding, H. Berke, H.J. Kruse HJ, S. Kröpelin. 2001. Environmental change and archaeology: lake evolution and human occupation in the eastern Sahara during the Holocene. Palaeogeography Palaeoclimatology Palaeoecology 169:193-217. Hole, F. and K.V. Flannery. . The prehistory of southwestern Iran: a preliminary report. Proceedings of the Prehistoric Society :-. Horai, S., K. Hayasaka, R. Kondo, K. Tsugane, and N. Takahata. . Recent African origin of modern humans revealed by complete sequences of hominoid mitochondrial DNAs. Proceedings of the National Academy of Science :-. Hötzl, H., and J.G. Zötl. . Climatic changes during the Quaternary period. In Quaternary Period in Saudi Arabia. Vol. : sedimentological, hydrogeological, hydrochemical, geomorphological, and climatological investigations in central and eastern Saudi Arabia, edited by S.S. Al-Sayari and J.G. Zötl. Wien: Springer-Verlag. Hours, F. . Le Middle Stone Age de Melka-Kunturé: résultats acquis et . Documents pour server à l’histoire des civilizations éthiopiennes , Paris. Hours, F., L. Copeland, and O. Aurenche. . Les industries Paléolithiques du Proche-Orient, essai de corrélation. L’Anthropologie 77(3/4):229-280. 358 Hugh, S., and K. Mason. . Western Arabia and the Red Sea. Oxford: Oxford University Press. Hull, K.L. . Reasserting the utility of obsidian hydration dating: a temperature-dependent empirical approach to practical temporal resolution with archaeological obsidians. Journal of Archaeological Science :-. Hume, G. . The Ladizian: an industry of the Asian chopper-chopping tool complex in Iranian Baluchistan. Philadelphia: Dorrance and Co. Ingman, M., H. Kaessmann, S. Paabo, and U. Gyllensten. . Mitochondrial genome variation and the origin of modern humans. Nature :-. Inizan, M.L. . Première mission archéologique Française a Qatar. Paleorient :-. Inizan, M.L. . Premiers resultats des fouilles prehistoriques de la region de Khor. In Mission Archéologique Française a Qatar, edited by J. Tixier. Doha: Dar al-Uloom. Pp. -. Inizan, M.L., and L. Ortlieb. . Prehistoire dans la region de Shabwa au Yemen du sud. Paleorient :-. Isaac, G. . The earliest archaeological traces. In Cambridge History of Africa, vol. : from the earliest times to  B.C., edited by J.D. Clark. Cambridge: Cambridge University Press. Pp -. Ivanova, E., R. Schiebel, A.D. Singh, G. Schmiedl, H.-S. Niebler, C. Hemleben. . Primary production in the Red Sea during the last , years. Palaeogeography, Palaeoclimatology, Palaeoecology :-. James, H.V.A. . Testing theories of modern human origins: the Middle Palaeolithic of South Asia. M.A. Thesis, University of Cambridge, Cambridge. James, H.V.A., and M.D. Petraglia. Modern human origins and the evolution of behavior in the later Pleistocene record of South Asia. Current Anthropology :S-S. Jelinek, A.J. . A preliminary report on some Lower and Middle Palaeolithic industries from the Tabun Cave. In Problems in prehistory: North Africa and the Levant, edited by F. Wendorf and A.E. Marks. Dallas: Southern Methodist University. Pp. -. Jelinek, A.J. . The Middle Palaeolithic in the Southern Levant with Comments on the Appearance of Homo sapiens. In The transition from Lower to Middle Paleolithic and the origin of modern man, edited by A. Ronen. Oxford: BAR International Series S. Pp. -. Jelinek, A.J. . The Amudian in the context of the Mugharan tradition at the Tabun Cave (Mt. Carmel), Israel. In The emergence of modern humans, edited by P. Mellars. Edinburgh: Edinburgh University Press. Pp. -. Jelinek, A.J. . Problems in chronology in the Middle Paleolithic and the first appearance of modern humans in Western Eurasia. In The evolution and dispersal of modern humans in Asia, edited by T. Akazawa, K. Aoki, and T. Kimura. Tokyo: Hokusen-sha Press. Pp. -. Jin, L., P.A. Underhill, V. Doctor, R.W. Davis, P. Shen, L.L. Cavalli-Sforza, and P.J. Oefner. . Distribution of haplotypes from a chromosome  region distinguishes multiple prehistoric human migrations. Proceedings of the National Academy of Science :-. 359 Johnson, L.L. . A technological analysis of an Aguas Verdes quarry workshop. In The Individual in Prehistory: studies in variability in prehistoric technologies, edited by N.J. Hill and R. Gunn. New York: Academic Press. Pp. -. Kapel, H. . Atlas of the Stone-Age Cultures of Qatar. Denmark: Aarhus University Press. Karmin, Monika. . Human Mitochondrial DNA Haplogroup R in India: dissecting the phylogenetic tree of south-Asian specific lineages. M.Sc. Thesis, University of Tartu, Estonia. Kar, A., A. Singhvi, S. Rajaguru, N. Juyal, J. Thomas, D. Banerjee, and R. Dhir. . Reconstruction of the late Quaternary environment of the lower Luni Plains, Thar Desert, India. Journal of Quaternary Science ():-. Ke, Y., B. Su, X. Song, D. Lu, L. Chen, H. Li, C. Qi, S. Marzuki, R. Deka, P. Underhill, C. Xiao, M. Shriver, J. Lell, D. Wallace, S. Wells, M. Seielstad, P. Oefner, D. Zhu, J. Jin, W. Huang, R. Chakraborty, Z. Chen, and L. Jin. . African origin of modern humans in East Asia: a tale of , Y chromosomes. Science :-. Kellogg, K.S., and R.L. Reynolds. . Opening of the Red Sea: constraints from a palaeomagnetic study of the As-Sarat volcanic field, south-western Saudi Arabia. Geophysical Journal of the Royal Astronomical Society ():-. Kelly, H. . Contribution à l’étude de la technique de la taille Levalloisienne. Bulletin de la Société Préhistorique Française ():-. Klein, R. . Why anatomically modern people did not disperse from Africa , years ago. In Neanderthals and Modern Humans in Western Asia, edited by T. Akazawa, K. Aoki, and O. Bar-Yosef. New York: Plenum Press. Pp. -. Kramer, A., T.L. Crummett, and M.H. Wolpoff. . Out of Africa and into the Levant: replacement or admixture in western Asia? Quaternary International :-. Krings, M., A. Halim, H. Salem, K. Bauer, H. Geisert, A. Malek, L. Chaix, C. Simon, D. Welsby, A. Di Rienzo, G. Utermann, A. Sajantila, S. Pääbo, and M. Stoneking. . mtDNA analysis of Nile River Valley populations: a genetic corridor or a barrier to migration? American Journal of Human Genetics :-. Kroon, D., T.N.F. Steens, and S.R. Troelstra. . Onset of monsoonal related upwelling in the western Arabian Sea as revealed by planktonic foraminifers. In Proceedings of the Ocean Drilling Program, Scientific results, vol. , edited by W.L. Prell, N. Niitsuma. Texas: College Station. Pp. -. Kulakovska, L. . The Middle Palaeolithic variations in western Ukraine. In The Middle Paleolithic Variability on the Territory of Ukraine, edited by L. Kulakovska. Kiev: Shliakh. Pp. -. Kulakovska, L. . Le Paleolithique moyen de Korolevo (Ukraine). Acts of the XIVth UISPP Congress. Oxford: BAR International Series . Pp. -. Kurashina, H. . An Examination of Prehistoric Lithic Technology in East-Central Ethiopia. Ph.D. Dissertation, University of California, Berkeley. Kutzbach, J.E. . Monsoon climate of the early Holocene: climate experiment with the Earth’s orbital parameters for  Years Ago. Science :-. 360 Lahr, M., and R. Foley. . Multiple dispersals and modern human origins. Evolutionary Anthropology :-. Lahr, M., and R. Foley. . Towards a theory of modern human origins: geography, demography, and diversity in recent human evolution. Yearbook of Physical Anthropology :-. Lam, Y.M., O.M. Pearson, and C.M. Smith. . Chin morphology and sexual dimorphism in the fossil hominid mandible sample from Klasies River Mouth. American Journal of Physical Anthropology ():-. Lawrence, T.E. . Diary of T.E. Lawrence. England: Corvinus Press. Leakey, L.S.B. . The industries of Gorgora Rockshelter, Lake Tana. Journal of the East Africa and Uganda Natural History Society :-. Léclaire, L. . Late Cretaceous and Cenozoic pelagic deposits palaeoenvironment and paleoceanography of the central western Indian Ocean. In Initial Report of the Deep Sea Drilling Project, vol. , edited by E.S.W. Simpson and R. Schlich. Washington: U.S. Government Printing Office. Pp. -. Lézine, A.-M., J.-P. Saliège, C.R. Wertz, F. Wertz, and M.-L. Inizan. . Holocene lakes from Ramlat as-Sab’atayn (Yemen) illustrate the impact of monsoon activity in southern Arabia. Quaternary Research :-. Lindly, J. . The Zagros Mousterian: a regional perspective. Ph.D. Dissertation, Arizona State University, Flagstaff. Loosveld, R.J.H., A. Bell, and J.J.M. Terken. . The tectonic evolution of interior Oman. GeoArabia ():--. Lovell, A., C. Moreau, V. Yotova, F. Xiao, S. Bourgeois, D. Gehl, J. Bertranpetit, E. Schurr, and D. Labuda. . Ethiopia: between sub-Saharan Africa and western Eurasia. Annals of Human Genetics :-. Lückge, A., H. Doose-Rolinski, A.A. Khan, H. Schulz, and U. von Rad. . Monsoonal variability in the northeastern Arabian Sea during the past , years: geochemical evidence from laminated sediments. Palaeogeography, Palaeoclimatology, Palaeoecology :-. Luis, J.R., D.J. Rowold, M. Regueiro, B. Caeiro, R. Cinnioğlu, C. Roseman, P.A. Underhill, L.L. Cavalli-Sforza, and R.J. Herrera. . The Levant versus the Horn of Africa, evidence for bidirectional corridors of human migrations. American Journal of Human Genetics ():. Luz, B. . Paleoclimatic interpretation of the last , year record of deep-sea cores around the Middle East. In Palaeoclimates, Palaeoenvironments and Human Communities in the Eastern Mediterranean Region in Later Prehistory. Oxford: BAR International . Maca-Meyer, N., A.M. Gonzalez, J.M. Larruga, C. Flores, V.M. Cabrera. . Major genomic mitochondrial lineages delineate early human expansion. BMC Genetics :. 361 Macaulay, V., C. Hill, A. Achilli, C. Rengo, D. Clarke, W. Meehan, J. Blackburn, O. Semino, R. Scozzari, F. Cruciani, A. Taha, N. Shaari, J. Raja, P. Ismail, Z. Zainuddin, W. Goodwin, D. Bulbeck, H.-J. Bandelt, S. Oppenheimer, A. Torroni, and M. Richards. . Single, rapid coastal settlement of Asia revealed by analysis of complete mitochondrial genomes. Science :-. Magne, M.P. and D. Pokotylo. . A pilot study in bifacial lithic reduction sequences. Lithic Technology (-):-. de Maigret, A. . Ricerche archeologiche italiane nella Repubblica Araba Yemenita: notizia di una seconda ricignizione. Oriens Antiquus :-. de Maigret, A. . Archaeological activities in the Yemen Arab Republic, . East and West : -. de Maigret, A. . Archaeological activities in the Yemen Arab Republic, . East and West :. de Maigret, A. . Archaeological activities in the Yemen Arab Republic, . East and West : -. de Maigret, A. . Arabia Felix: an exploration of the archaeological history of Yemen. London: Stacey International. Maillard, C. and G. Soliman. . Hydrography of the Red Sea and exchanges with the Indian Ocean in summer. Oceanologica Acta :-. Maizels, J.K. . Plio-Pleistocene raised channel systems of the western Sharqiya (Wahiba), Oman. In Desert Sediments: Ancient and Modern, Geological Society Special Publication , edited by L. Frostick and I. Reid. Pp. -. di Mario, F. . The western ar Rub’ al-Khali ‘Neolithic’: new data from the Ramlat Sab’atayn (Yemen Arab Republic). Annali dell’Istituto Universitario Orientale di Napoli ():-. Marks, A.E. a. The Mousterian industries of Nubia. In The Prehistory of Nubia, vol. , edited by F. Wendorf. Dallas: Fort Burgwin Research Center and Southern Methodist University Press. Marks, A.E. b. The Khormusan: an Upper Pleistocene industry in Sudanese Nubia. In The Prehistory of Nubia, vol. , edited by F. Wendorf. Dallas: Fort Burgwin Research Center and Southern Methodist University Press. Marks, A.E. . (editor). Prehistory and paleoenvironments of the Central Negev, Israel, volume I, the Avdat/Aqev area, part . Dallas: Southern Methodist University Press. Marks, A.E. . (editor). Prehistory and paleoenvironments in the Central Negev, Israel: volume II, the Avdat/Aqev area, part  and the Har Harif. Dallas: Southern Methodist University Press. Marks, A.E. . (editor). Prehistory and paleoenvironments of the Central Negev, Israel, volume III, the Avdat/Aqev area, part . Dallas: Southern Methodist University Press. Marks, A.E. . The Middle to the Upper Paleolithic transition in the Southern Levant: technological change as an adaptation to increasing mobility. In L’homme de Néandertal v., la mutation, edited by M. Otte. Liège: ERAUL no. . Pp. -. 362 Marks, A.E. . The Middle and Upper Palaeolithic of the Near East and the Nile Valley: the problem of cultural transformations. In The Emergence of Modern Humans: an archaeological perspective, edited by P. Mellars. Edinburgh: Edinburgh University Press. Marks, A.E. and K. Monigal. . Modeling the production of elongated blanks from the Early Levantine Mousterian at Rosh Ein Mor. In The Definition and Interpretation of Levallois Technology, edited by H.L. Dibble and O. Bar-Yosef. Monographs in World Archaeology no. . Madison: Prehistory Press. Pp. -. Marks, A.E., J.-P. Burgal, V.P. Chabai, K. Monigal, P. Goldberg, B Hockett, E. Peman, M Elorza, and C. Mallol. . Le gisement Pléistocene moyen de Galeria Pesada (Estrémadure, Portugal): premiers résultats. Paleo :-. Marks, A.E. nd. Resumé of the definition of an industry: the Khormusan of Sudanese Nubia. Masry, A.H. . Prehistory in Northeastern Arabia: the problem of interregional interaction. Miami: Field Research Projects. Masry, A.H. . Notes on the recent archaeological activities in the Kingdom of Saudi Arabia. Proceedings of the Seminar for Arabian Studies :-. McBrearty, S. . Recent research in western Kenya and its implications for the status of the Sangoan Industry. In Cultural Beginnings, edited by J.D. Clark. Bonn: Dr. Rudolph Habelt GMBH. Pp. -. McBrearty, S. . Lithic artifacts from Abu Dhabi’s western region. Tribulus: Bulletin of the Emirates Natural History Group :-. McBrearty, S. . Earliest stone tools from the Emirate of Abu Dhabi, United Arab Emirates. In Fossil Vertebrates of Arabia, edited by P.J. Whybrow and A. Hill. New Haven: Yale University Press. Pp. -. McBurney, C.B.M. . Preliminary report on stone age reconnaissance in north-eastern Iran. Proceedings of the Prehistoric Society :-. McBurney, C.B.M. . Palaeolithic excavations in the Zagros area. Iran :-. McClure, H.A. . The Arabian Peninsula and Prehistoric Populations. Miami: Field Research Projects. McClure, H.A. . Radiocarbonchronology of Late Quaternary lakes in the Arabian Desert. Nature :-. McClure, H.A. . Ar Rub’ al Khali. In Quaternary Period in Saudi Arabia. Vol. : sedimentological, hydrogeological, hydrochemical, geomorphological, and climatological investigations in central and eastern Saudi Arabia, edited by S.S. Al-Sayari and J.G. Zötl. Wien: Springer-Verlag. McClure, H.A. . Late Quaternary Palaeoenvironments of the Rub’ al Khali. Ph.D. Dissertation, University of London, London. McClure, H.A. . Late Quaternary palaeogeography and landscape evolution of the Rub al Khali. In Araby the Blest, edited by D. Potts. 363 McClure, H.A. . A new Arabian stone tool assemblage and notes on the Aterian Industry of North Africa. Arabian Archaeology and Epigraphy :-. McClure, H.A. . Quaternary climate and landscape of eastern Arabia. In Indian Ocean in Antiquity, edited by J. Reade. London: Kegan Paul International. McClure, H.A. and F.M. Swain. . The fresh water and brackish water fossil Quaternary Ostracoda from the Rub’ al Khali, Saudi Arabia. th African Micropalaeontological Colloquium, Tunis. McCorriston, J. . Early settlement in Hadramawt: preliminary report on prehistoric occupation at Shi’b Munayder. Arabian Archaeology and Epigraphy :-. McCorriston, J. . Holocene paleoecology and prehistory in highland southern Arabia. Palérorient ():-. McDermott, F., R. Grün, C. Stringer, and C.J. Hawkesworth. . Mass-spectrometric U-series dates for Israeli Neanderthal/early modern human sites. Nature :-. McDermott, F., R. Grün, C.T. Williams, and V.K. Din. . New late-Pleistocene UraniumThorium and ESR dates for the Singa hominid (Sudan). Journal of Human Evolution : -. McDougall, I., F.H. Brown, and J.G. Fleagle. . Stratigraphic placement and age of modern humans from Kibish, Ethiopia. Nature :-. McGuire, A.V., and R.G. Bohannon. . Timing of mantle upwelling: evidence for a passive origin of the Red Sea rift. Journal of Geophysical Research (B):-. Mercier, N., H. Valladas, G. Valladas, J.-L. Reyss, A. Jelinek, L. Meignen, and J.-L. Joron. . TL dates of burnt flints from Jelinek’s excavations at Tabun and their implications. Journal of Archaeological Science :-. Mehlman, M.J. . Late Quaternary Archaeological Sequences in Northern Tanzania. Ph.D. dissertation, Department of Anthropology, University of Illinois, Urbana. Mehlman, M.J. . Context for the emergence of modern man in eastern Africa: some new Tanzanian evidence. In Cultural Beginnings, edited by J. D. Clark. Bonn: Dr. Rudolph Habelt GMBH. Pp. –. Meignen, L. . Le Paléolithique moyen du Levant: synthèse. In Préhistoire du Levant II: processus des changements culturels, edited by O. Aurenche, J. Cauvin, and P. Sanlaville. Paléorient ():-. Meignen, L., O. Bar-Yosef, and P. Goldberg. . Les structures de combustions Moustérienees de la grotte de Kébara (Mont Karmel, Israel). In Nature et fonction des foyers préhistorique, mémoires du musée de prehistoire d’Ile de France, edited by M. Olivé and Y. Taborin. Nemours: Edition A.P.R.A.I.F. Mellars, P. . Going East: genetic and archaeological perspectives on the modern human colonization of Eurasia. Science :-. 364 Mercier, N., H. Valladas, O. Bar-Yosef, B. Vandermeersch, C. Stringer, and J.L. Joron. . Thermoluminescence date for the Mousterian burial site of Es-Skhul, Mt. Carmel. Journal of Archaeological Science :-. Mercier, N., H. Valladas, G. Valladas, J.-L. Reyss, A. Jelinek, L. Meignen, and J.-L. Joron. . TL Dates of burnt flints from Jelinek’s excavations at Tabun and their implications. Journal of Archaeological Science :-. Merrick, H.V. . Change in Later Pleistocene Lithic Industries in Eastern Africa. Ph.D. Dissertation, University of California, Berkeley. Merrick, H.V. and F.H. Brown. . Obsidian sources and patterns of source utilization in Kenya and northern Tanzania: some initial findings. African Archaeological Review :-. Merrick, H.V., F.H. Brown., and W.P. Nash.. Use and movement of obsidian in the Early and Middle Stone Ages of Kenya and northern Tanzania. In Society, Culture, and Technology in Africa, MASCA  (supplement):-. Metspalu, M., T. Kivisild, E. Metspalu, J. Parik, G. Hudjashov, K. Kaldma, P. Serk, M. Karmin, D. Behar, M.T.P. Gilbert, P. Endicott, S. Mastana, S.S. Papiha, K. Skorecki, A. Torroni, and R. Villems. . Most of the extant mtDNA boundaries in south and southwest Asia were likely shaped during the initial settlement of Eurasia by anatomically modern humans. BMC Genetics :. Michels, J.W., I.S.T. Tsong, and C. Nelson.. Obsidian dating and East African Archaeology. Science :-. Miller, A.G. and M. Morris. . The Plants of Dhofar. Muscat: Royal Diwan, Government of Oman. Minzoni-Deroche, A. . Middle and Upper Palaeolithic in the Taurus-Zagros region. In The Paleolithic Prehistory of the Zagro-Taurus. Edited by D.I. Olszewski and H. Dibble. Philadelphia: The University Museum, University of Pennsylvania. Misra, V.N. . Evolution of blade element in the stone industries of Rock-Shelter III F-, Bhimbetka. In Indian Archaeology, New Perspectives, edited by R.K. Sharma. Bhopal: Prachya Niketan. Pp. -. Misra, V.N. . Prehistoric human colonization of India. Journal of Bioscience :-. Mithen, S. and M. Reed. . Stepping out: a computer simulation of hominid dispersal from Africa. Journal of Human Evolution :-. Monigal, K. . The Levantine Leptolithic: Blade Technology from the Lower Palaeolithic to the Dawn of the Upper Palaeolithic. Ph.D. Dissertation, Southern Methodist University, Dallas. Moore, A.M.T., G.C. Hillman, and A.J. Legge. . Village on the Euphrates. Oxford: Oxford University Press. Mortensen, P. . A preliminary study of the chipped stone industry from Beidha. Acta Archaeologica :-. Mortillet, G. de. . Le préhistorique: antiquité de l’homme. Paris: Reinwald. 365 Morwood, M.J., P.B. O’Sullivan, F. Aziz, and A. Raza. . Fission track ages of stone tools and fossils on the east Indonesian island of Flores. Nature :-. Moysey, F. . Excavations of a rock shelter at Gorgora, Lake Tana, Ethiopia. Journal of theEast Africa and Uganda Natural History Society -. Murray, S.P., and W. Johns. . Direct observations of seasonal exchanges through the Bab el Mandab Strait. Geophysical Research Letters ():-. Muzuka, A.N. .  ka Organic C record of the monsoon variability on the Oman continental margin, Arabian Sea. Proceedings of the Indian Academy Science :-. Newcomer, M.H. . Some quantitative experiments in handaxe manufacture. World Archaeology ():-. Newton, L.S. and J. Zarins. . Aspects of Bronze Age art of southern Arabia: the pictorial landscape and its relation to economic and socio-political status. Arabian Archaeology and Epigraphy ():-. Niebuhr, C. . Reisebeschreibung nach Arabien und andern umliegenden Ländern, Copenhagen. Nishiaki, Y. . Truncated-faceted flakes from Levantine Mousterian assemblages. University of Tokyo Bulletin :-. Nowak, R.M. . Walker’s Mammals of the World, fifth edition. Baltimore: The Johns Hopkins University Press. O’Connell, J.F. and J. Allen. . Dating the colonization of Sahul (Pleistocene Australia-New Guinea): a review of recent research. Journal of Archaeological Science :-. Odell, G.H. . Experiments in lithic reduction. In Experiments in Lithic Technology, edited by D.S. Amick and R.P. Mauldin. Oxford: BAR International Series no. . Pp. -. Ohnuma, K. and T. Akazawa. . Experimental studies in the determinations of flaking mode. Bulletin of the Institute of Archaeology (London) :-. Overstreet, W.C. . Correspondence by William C. Overstreet. In Contributions to the Anthropology of Saudi Arabia. Coconut Grove: Field Research Projects. Overstreet, W.C. . Contributions to the Prehistory of Saudi Arabia: I. Miami: Field Research Projects. Overstreet, W.C., M.J. Grolier, and M.R. Toplyn. . Geological and archaeological reconnaissance in the Yemen Arab Republic, . In Wadi Al-Jubah Archaeological Project. American Foundation for the Study of Man. Owen, L.R. . Blade and Microblade Technology: selected assemblages from the North American Arctic and the Upper Paleolithic of southwest Germany. Oxford: BAR International Series no. . Owen, W.E. . The Kombewa Culture, Kenya Colony. Man :–. Pappu, R.S. . Acheulian culture in peninsular India: an ecological perspective. New Delhi: D.K. Printworld Ltd. 366 Pandya, A., R.R. Santos, T. Zerjal, V. Ferák, R.J. Mitchell, K. Thangaraj, L. Singh, R. Harding, M.A. Jobling, and C. Tyler-Smith. . Use of Y-chromosomal DNA haplotypes to investigate human population history. In Abstracts of Papers Presented at the Cold Spring Harbor Laboratory, New York, October . Pp. . Parker, A.G., L. Eckersley, M.M. Smith, A.S. Goudie, S. Stokes, S. Ward, K. White, and M.J. Hodson. . Holocene vegetation dynamics in the northeastern Rub al-’Khali, Arabian Peninsula, a phytolith, pollen, and carbon isotope study. Journal of Quaternary Science (): -. Patzelt A. nd. Vitaceae, Rhamnaceae, Salvadoraceae, Balsaminaceae, Oxalidaceae, Rutaceae, Geraniaceae. In Flora of Oman, volume , edited. S.A. Ghazanfar. Payne, S., and A. Garrard. . Camelus from the Upper Pleistocene of Mount Carmel, Israel. Journal of Archaeological Science :-. Payne, J.C. and S. Hawkins. . A surface collection of flints from Habarut in southern Arabia. Man :-. Pelcin, A.W. . The threshold effect of platform width: a reply to Davis and Shea. Journal of Archaeological Science :-. Pelegrin, J. . Technologie lithique: une méthode appliqué à l’étude de deux series du Perigordian ancien. Thèse Doctorat, Université Paris X. Perlès, C. . Reexamen typologique de l’Industrie du Porc-Epic (Ethiopie): les pointes et pieces pointues. L’Anthropologie ():-. Petit, J.R., J. Jouzel, D. Raynaud, N.I. Barkov, J.-M. Barnola, I. Basile, M. Bender, J. Chappellaz, M. Davis, G. Delayque, M. Delmotte, V.M. Kotlyakov, M. Legrand, V.Y. Lipenkov, C. Lorius, L. Pépin, C. Ritz, E. Saltzman, and M. Stievenard. . Climate and atmospheric history of the past , years from the Vostok ice core, Antarctica. Nature :-. Petraglia, M., and A. Alsharekh. . The Middle Palaeolithic of Arabia: implications for modern human origins, behaviour and dispersals. Antiquity ():-. Philby, H. . Rub al’ Khali: an account of exploration in the great south desert of Arabia under the auspices and patronage of His Majesty ‘Abdul ‘Aziz ibn Saud, King of the Hejaz and Nejd and its dependencies. Geographical Journal :-. Philby, H. . Sheba’s Daughter. London: Methuen. Phillips, W. . Qataban and Sheba: exploring the ancient kingdoms on the biblical spice routes of Arabia. New York: Harcourt, Brace and Company. Pirenne, J. . La Grèce et Saba: une nouvelle base pour la chronologie sud-arabe, Paris. Platel, J.P., J. Roger, T.J. Peters, I. Mercolli, J.D. Kramers, and J. Le Métour. . Geological Map of Salalah, explanatory notes. Sultanate of Oman: Ministry of Petroleum and Minerals. Pleurdeau, D. . Le Middle Stone Age de la grotte du Porc-Épic (Dire Dawa, Éthiopie): gestion des matières premières et comportements techniques. L’anthropologie 107:15-48. 367 Pleurdeau, D. . Human technical behavior in the African Middle Stone Age: the lithic assemblage from Porc-Epic Dave (Dire Dawa, Ethiopia). African Archaeological Review (): -). Pliny. Natural History, Books III-VII, translated by H. Rackham. London: Loeb Classical Library. Porat, N., M. Chazan, H. Schwarcz, and L. Kolska Horwitz. . Timing of the Lower to Middle Paleolithic boundary: new dates from the Levant. Journal of Human Evolution ():-. Potts, D. . The Arabian Gulf in Antiquity, vol. : from prehistory to the fall of the Achaemenid Empire. Oxford: Clarendon Press. Potts, D. . The chronology of the archaeological assemblages from the head of the Arabian Gulf to the Arabian Sea, - B.C. In Chronologies in Old World Archaeology, edited by R. Ehrich. Chicago: University of Chicago Press. Pp. -. Potts, D. . The late prehistoric, protohistoric, and early historic periods in eastern Arabia (Ca. - B.C.). Journal of World Prehistory :-. Potts, D. . Mesopotamian Civilization: the material foundation. Ithaca: Cornell University Press. Pullar, J. . Harvard archaeological survey in Oman, : flint sites in Oman. Arabian Seminar : -. Pullar, J. . A selection of aceramic sites in the Sultanate of Oman. Journal of Oman Studies :. Pullar, J. and B. Jäckli. . Some aceramic sites in Oman. Journal of Oman Studies :-. Quintana-Murci, L., O. Semino, H. Bandelt, G. Passarino, K. McElreavey, and A.S. SantachiaraBenerecetti. . Genetic evidence of an early exit of Homo sapiens sapiens from Africa through eastern Africa. Nature Genetics :-. Radcliffe-Smith, A. . The vegetation of Dhofar. The Journal of Oman Studies, Special Report : -. Rak Y., and B. Arensburg. . Kebara  Neanderthal pelvis: first look at a complete inlet. American Journal of Physical Anthropology :–. Ramerini, M. Dutch, Portuguese Colonial History. HTML document, Italy, . Rampino, M. and S. Ambrose. . Volcanic winter in the Garden of Eden: the Toba supereruption and the late Pleistocene human population crash. Geological Society of America, special paper :-. Rampino, M. and S. Self. . Volcanic winter and accelerated glaciation following the Toba supereruption. Nature :-. Rampino, M. and S. Self. . Climate-volcanism feedback and the Toba eruption of ~, years ago. Quaternary Research :-. Reichart, G.J., L.J. Lourens, and W.J. Zachariasse. . Temporal variability in the northern Arabian Sea oxygen minimum zone (OMZ) during the last ,. Paleooceanography :-. 368 Ridings, R. . Obsidian hydration dating: the effects of mean exponential ground temperature and depth of artifact recovery. Journal of Field Archaeology ():-. Ridings, R. . Where in the world does obsidian hydration dating work? American Antiquity : -. Rodgers, D.W., A. Gunatilaka. . Bajada formation by monsoonal erosion of a subaerial forebulge, Sultanate of Oman. Sedimentary Geology :-. Roger, J., J.P. Platel, J. Dubreuilh, and R. Wyns. . Geological Map of Mafraq, explanatory notes. Sultanate of Oman: Ministry of Petroleum and Minerals. Rohling, E.J. and W.J. Zachariasse. . Red Sea outflow during the Last Glacial Maximum. Quaternary International :-. Rose, J.I. . A Reconstruction of Pleistocene Arabia: the state of research and avenues for further inquiry. M.A. Thesis, Boston University, Boston. Rose, J.I. . Survey of prehistoric sites in Mahra, eastern Yemen. Adumatu :-. Rose, J.I. a. The question of Upper Pleistocene connections between East Africa and South Arabia. Current Anthropology ():-. Rose, J.I. b. New evidence for the expansion of an Upper Pleistocene population out of East Africa, from the site of Station One, Northern Sudan. Cambridge Archaeological Journal (): -. Rose, J.I. . The role of the Saharo-Arabian arid belt in the modern human expansion. In From the Mediterranean Basin to the Portuguese Atlantic Shore: Papers in Honor of Anthony E. Marks, Actas do IV Congresso de Arqueologia Peninsular, edited by N. Bicho and P. acker. Faro: Universidade do Algarve. Sale, J.B. . The environment of the mountain region of Dhofar. The Journal of Oman Studies, Special Report :-. Sanlaville, P. . Changements climatiques dans la péninsule Arabique durant le Pléistocène supérieur et l’Holocène. Paleorient :-. Sarnthein, M. . Sediments and history of the postglacial transgression in the Persian Gulf and northwestern Gulf of Oman. Marine Geology :-. Schulz, E. and J. Whitney. . Upper Pleistocene and Holocene lakes in the An Nafud, Saudi Arabia. Hydrobiologia :-. Schulz, E. and J. Whitney. . Upper Pleistocene and Holocene paleoenvironments in the An Nafud, Saudi Arabia. In Current Research in African Earth Sciences, edited by G.M. and H. Schandelmeier. Rotterdam: A.A. Balkema. Schwarcz, H.P., B. Blackwell, P. Golberg, and A.E. Marks. . Uranium series dating of travertine from archaeological sites, Nahal Zin, Israel. Nature ():-. Schwarcz, H.P., R. Grün, B. Vandermeersch, O. Bar-Yosef, and H. Valladas. . ESR dates for the hominid burial site of Qafzeh in Israel. Journal of Human Evolution :-. 369 Schwarcz, H.P. and W.J. Rink. . Progress in ESR and U Series chronology of the Levantine Paleolithic. In Neandertals and Modern Humans in Western Asia, edited by T. Akazawa, K. Aoki, and O. Bar-Yosef. New York: Plenum Press. Pp. -. Schyfsma, E. . Climate. In Quaternary Period in Saudi Arabia. Vol. : sedimentological, hydrogeological, hydrochemical, geomorphological, and climatological investigations in central and eastern Saudi Arabia, edited by S.S. Al-Sayari and J.G. Zötl. Wien: Springer-Verlag. Shaw, B., H. Penney, and F. Tully. . Oman: caves of Dhofar. Caves and Caving :-. Siddall, M., E.J. Rohling, A. Almogi-Labin, C.H. Hemleben, D. Meischner, I. Schmelzer, and D.A. Smeed. . Sea-level fluctuations during the last glacial cycle. Nature :-. Siddall, M., D.A. Smeed, S. Mathiessen, and E.J. Rohling. . Exchange flow between the Red Sea and the Gulf of Aden. The nd Meeting on the Physical Oceanography of Sea Straits, Villefranche. Pp. -. Singer, R., and J. Wymer. . The Middle Stone Age at Klasies River Mouth in South Africa. Chicago: University of Chicago Press. Sirocko, F. . Ups and downs in the Red Sea. Nature :-. Solecki, R. . Prehistory in Shanidar Valley, Northern Iraq. Science ():-. Solecki, R. . Summary report of the Columbia University prehistoric investigations in Lebanon, season . Bulletin du Musée de Beyrouth :-. Smith, F.H. . The role of continuity in modern human origins. In Continuity or Replacement: Controversies in Homo Sapiens Evolution, edited by G. Bräuer and F.H. Smith. Rotterdam: Balkema. Smith, G.H. . New prehistoric sites in Oman. Journal of Oman Studies :-. Smith, P. . Le Solutréen en France. Bordeaux: Delmas. Smith, P. and G. Maranjian. . Two ‘Neolithic’ collections from Saudi Arabia. Man :-. Sokal, R.R. and F.J. Rohlf. . Biometry. nd ed. New York: W.H. Freeman and Company. Solecki, R.S. . Summary report of the Columbia University prehistoric investigations in Lebanon, season . Bulletin du Musée de Beyrouth :-. Solecki, R.S. . Shanidar: the first flower people. New York: Alfred A. Knopf. Solecki, R.S. and R. Solecki. . The pointed tools from the Mousterian occupations of Shanidar Cave, northern Iraq. In The Paleolithic Prehistory of the Zagro-Taurus. Edited by D.I. Olszewski and H. Dibble. Philadelphia: The University Museum, University of Pennsylvania. Soressi, M. . Le Moustérien de tradition acheuléenne du sud-ouest de la France. Discussion sur la signification du faciès à partir de l’étude comparée de quatre sites: Pech-de-l’Azé I, Le Moustier, La Rochette et la Grotte XVI. Ph.D. Dissertation, University of Bordeaux, Bordeaux. Speth, J.D. . Experimental investigations of hard-hammer percussion flaking. Tebiwa :-. 370 Stark, F. . The Southern Gates of Arabia: A Journey in the Hadhramaut. New York: Knopf. Straus, L. . Africa and Iberia in the Pleistocene. Quaternary International :-. Straus, L., and O. Bar-Yosef. . Out of Africa in the Pleistocene: an introduction. Quaternary International :-. Stringer, C. B. . Replacement, continuity and the origin of Homo sapiens. In Continuity or Replacement: Controversies in Homo sapiens Evolution, edited by G. Bräuer and F.H. Smith. Rotterdam: Balkema. Stringer, C.B. . Has Australia backdated the human revolution? Antiquity :-. Stringer, C.B. . Palaeoanthropology: coasting out of Africa. Nature :-. Stringer, C.B., R. Grün, H.P. Schwarcz, and P. Goldberg. . ESR dates for the hominid burial site of Es-Skhul in Israel. Nature :-. Takhtajan, A.L. . Floristic Regions of the World. Berkeley: University of California Press. Tamura, K., and M. Nei. . Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Molecular Biological and Evolution :-. Tchernov, E., a. The Afro-Arabian component in the Levantine mammalian fauna—a short biogeographical review. Israel Journal of Zoology :-. Tchernov, E., b. Biochronology, paleoecology and dispersal events of hominids in the southern Levant. In The Evolution and Dispersal of Modern Humans in Asia, edited by T. Akazawa, K. Aoki, and T. Kimura. Tokyo: Hokusen-Sha. Pp. -. Tchernov, E. c. Eurasian-African biotic exchanges through the Levantine corridor during the Neogene and Quaternary. Courier Forsch.-Institute Senckenberg :-. Tchernov, E. . New comments on the biostratigraphy of the Middle and Upper Pleistocene of the Levant. In Late Quaternary Chronology and Paleoclimates of the Eastern Mediterranean, edited by O. Bar-Yosef and K.S. Kra. Tucson: Radiocarbon. Pp. -. Teilhard de Chardin, P. . Le Paléolithique en Somalie française et en Abyssinie. L’Anthropologie :-. Teilhard de Chardin, P., H. Breuil, and P. Wernert. . Les industries lithiques de Somalie française. L’Anthropologie (). Templeton, A.R. . The ‘Eve Hypothesis’: A genetic critique and reanalysis. American Anthropologist :-. Templeton, A.R. . Testing the out of Africa replacement hypothesis with mitochondrial DNA data. In Conceptual Issues in Modern Human Origins Research, edited by G.A. Clark and C.M. Willermet. New York: Aldine de Gruyter. Pp. -. Thangaraj, K., G. Chaubey, T. Kivisild, A.G. Reddy, V.K. Singh, A. Rasalkar, and L. Singh. . Reconstructing the origin of Andaman Islanders. Science :. 371 Thangaraj, K., L. Singh, A.G. Reddy, U.R. Rao, S.C. Sehgal, P.A. Underhill, M. Pierson, I.G. Frame, and E. Hagelberg. . Genetic affinities of the Andaman Islanders, a vanishing human population. Current Biology :-. Thesiger, W. . Arabian Sands. New York: E.P. Dutton and Company, Inc. Thomas, B. . Arabia Felix: Across the Empty Quarter of Arabia. Oxford: Readers’ Union Limited. Thompson, A. . Origins of Arabia. London: Stacey International. Thunell, R.C., S.M. Locke, and D.F. Williams. . Glacio-eustatic sea-level control on Red Sea salinity. Nature :-. Tishkoff, S.A., A.G. Clark, E. Dietzsch, A. Goldman, H. Soodyall, J.R. Kidd, T. Jenkins, and K.K. Kidd. . Demographic history of modern humans inferred from nuclear genetic haplotype data. In Abstracts of Papers Presented at the Cold Spring Harbor Laboratory, New York, October . Pp. . Tishkoff, S.A., E. Dietzsch, W. Speed, A.J. Pakstis, J.R. Kidd, K. Cheung, and B. Bonné-Tamir. . Global patterns of lineage disequilibrium at the CD locus and modern human origins. Science :-. Tixier, J. . Typologie de l’épipaléolithique du Maghreb. Mémoires du centre de recherches anthropologiques, préhistoriques et ethnographiques, II (Algeria). Paris: Arts et Métiers Graphiques. Tixier, J. . Glossary for the Description of Stone Tools, with specific reference to the Epipaleolithic of the Maghreb. (Translated from the French by M. Newcomer). Newsletter of Lithic Technology: Special Publications No. . Tixier, J., M. Inizan, and H. Roche. . Préhistoire de la Pierre taillée: terminologie et technologie, volume I. Meudon: centre de recherche et d’etudes préhistoriques. Tosi, M. . The emerging picture of prehistoric Arabia. Annual Review of Anthropology :-. Tosi, M. and D. Usai. . Preliminary report on the excavations at Wadi Shab, Area , Sultanate of Oman. Arabian Archaeology and Epigraphy :-. Toth, N. . Behavioral inferences from Early Stone artifact assemblages: an experimental model. Journal of Human Evolution :. Trinkaus, E. . Early modern humans. Annual Reviews of Anthropology :-. Trinkaus, E., F.H. Smith, T.C. Stockton, and L.L. Shackelford. . The human postcranial remains from Mladeč. In Early Modern Humans at the Moravian Gate: The Mladeč Caves and Their Remains, edited by M. Techler-Micola. Vienna: Springer-Verlag. Tryon, C.A. . Middle Pleistocene sites from the “southern” Kapthurin Formation of Kenya. Nyame Akuma :-. Tryon, C.A. . The Acheulian to Middle Stone Age transition: tephrostratigraphic context for archaeological change in the Kapthurin Formation, Kenya. Ph.D. Dissertation, University of Connecticut, Storrs. 372 Tryon, C.A. . “Early” Middle Stone Age lithic technology of the Kapthurin Formation. Current Anthropology ():-. Tryon, C.A. and S. McBrearty. . Tephrostratigraphy and the Acheulian to Middle Stone Age transition in the Kapthurin Formation, Baringo, Kenya. Journal of Human Evolution :. Tryon, C.A. and S. McBrearty. . Tephrostratigraphy of the bedded tuff member (Kapthurin Formation, Kenya) and the nature of archaeological change in the later Middle Pleistocene. Quaternary Research :-. Tryon, C.A., S. McBrearty, and P.-J. Texier. . Levallois lithic technology from the Kapthurin Formation, Kenya: Acheulian Origin and Middle Stone Age diversity. African Archaeological Review ():-. Uuerpmann, H.-P. and M. Uerpmann. . ‘Ubaid pottery in the eastern Gulf – new evidence from Umm al-Qawain (U.A.E). Arabian Archaeology and Epigraphy :-. Uerpmann, H.-P. and M. Uerpmann. . Stone Age Sites and their Natural Environment, the capital area of northern Oman, part III. Wiesbaden: Dr. Ludwig Reichert Verlag. Uerpmann, M. . Structuring the Late Stone Age of southeastern Arabia. Arabian Archaeology and Epigraphy :-. Ulrix-Closset, M. . Le Paléolithique moyen dans le basin mosan en Belgique. Faculté de philosophie et lettres de l’Université de Liège, Publications Exceptionnelles . Liège: Éditions Universa. Underhill, P.A., G. Passarino, A.A. Lin, P. Shen, M.M. Lahr, R.A. Foley, P.J. Oefner, and L.L. CavalliSforza. . The phylogeography of Y chromosome binary haplotypes and the origins of modern human populations. Annals of Human Genetics. :-. Usik, V. . Problems of Kombewa method and some features of non-Levallois reduction strategies of Middle Palaeolithic complex , Kordeva site. Paper Presented at UISPP XIV, Liege. Usik, V. . The variants of Levallois method of Middle Palaeolithic industries of Ukraine. In The Middle Palaeolithic Variability on the Territory of Ukraine, edited by L. Kulakovska. Kiev: Shliakh. Pp. -. Valladas, H., J.L. Joron, G. Valladas, B. Arensburg, O. Bar-Yosef, A. Belfer-Cohen, P. Goldberg, H. Laville, L. Meignen, Y. Rak, E. Tchernov, A.M. Tillier, and B. Vandermeersch. . Thermoluminescence dates for the Neanderthal burial site at Kebara in Israel. Nature : -. Valoch, K. . Stone tool assemblages from Arnhem Land. In Anthropology of the Rembrranga People, A Contribution of the Czechoslovak Anthropos Expedition to Arnhem Land N.T. Australia, edited by J. Jelínek. Czechoslovakia: Anthropos Institute. Pp. -. Van Beek, G., G. Cole, and A. Jamme. . An archaeological reconnaissance in Hadramaut, South Arabia: a preliminary report. Annual Report of the Smithsonian Institution :-. Van Peer, P. . A Model for Studying the Variability of Levallois Technology and its Application to the Middle Paleolithic of Northern Africa. Ph.D. Dissertation, Katholieke Universiteit, Leuven. 373 Van Peer, P., R. Fullagar, S. Stokes, R.M. Bailey, J. Moeyersons, F. Steenhoudt, A. Geerts, T. Vanderbeken, M. De Dapper, and F. Geus. . The Early to Middle Stone Age transition and the emergence of modern human behaviour at site -B-, Sai Island, Sudan. Journal of Human Evolution :-. Van Zeist, W. and J.A.H. Bakker-Heeres. . Archaeobotanical studies in the Levant, : LatePaleolithic Mureybit. Palaeohistoria :-. Vigilant, L., M. Stoneking, H. Harpending, K. Hawkes, and A.C. Wilson. . African populations and the evolution of human mitochondrial DNA. Science :. Villiers-Petocz, L.E. . Some notes on the lithic collections of the Oman Department of Antiquities. Journal of Oman Studies :-. Vogel, J.C. . Radiometric dates for the Middle Stone Age in South Africa. In Humanity from African Naissance to Coming Millennia, edited by P.V. Tobias, M.A. Raath, J. Moggi-Cecchi, and G.A. Doyle. Florence: Firenze University Press. Pp. -. Volkman, P. . Boker Tachtit core reconstructions. In Prehistory and paleoenvironments of the Central Negev, Israel, volume III: the Avdat/Aqev area, part , edited by A.E. Marks. Dallas: Southern Methodist University Press. Pp. -. Wallace, D.C., M.D. Brown, T.G. Schurr, Y.-S. Chen, J.T. Lell, R.I. Sukernik, Y.B. Strakvskaya, and A. Olckers. . Mitochondrial DNA (mtDNA) and Y chromosome variation in human evolution and disease. In Abstracts of Papers Presented at the Cold Spring Harbor Laboratory, New York, October . Pp. . Walter, R., R. Buffler, J.H. Bruggemann, M. Guillaume, S. Berhe, B. Negassi, Y. Libsekal, H. Cheng, R. Edwards, R. von Cosel, D. Neraudeau, and M. Gagnon. . Early human occupation of the Red Sea coast of Eritrea during the Last Interglacial. Nature :-. Watson, E., P. Forster, M. Richards, and H.-J. Bandelt. . Mitochondrial footprints of human expansion in Africa. American Journal of Human Genetics :-. Wendorf, F., R.L. Laury, C.C. Albritton, R. Schild, C.V. Haynes, P. Damon, M. Shafiqullah, and R. Scarborough. . Dates for the Middle Stone Age of East Africa. Science : -. Wendorf, F. and R. Schild. . A Middle Stone Age Sequence from the Central Rift Valley, Ethiopia. Warsaw: Polska Akademia Nauk. Wendorf, F., R. Schild, and A.E. Close. . Egypt During the Last Interglacial: the Middle Paleolithic of Bir Tarfawi and Bir Sahara East. New York: Plenum Press. Wendorf, F., R. Schild, A.E. Close, H. Schwarcz, G. Miller, R. Grün, A. Bluszcz, S. Stokes, L. Morowska, J. Huxtable, J. Lundberg, C.L. Hill, and C. McKinney. . A chronology for the Middle and Late Pleistocene wet episodes in the eastern Sahara. In Late Quaternary Chronology and Paleoclimates of the Eastern Mediterranean, edited by O. Bar-Yosef and R.S. Kra. Tucscon: Radiocarbon. Whalen, N., J. Ali, H. Sindi, and D. Pease. . A Lower Pleistocene site near Shuwayhitiyah in northern Saudi Arabia. Atlal :-. Whalen, N., J. W. Davis, and D. Pease. . Early Pleistocene migrations into Saudi Arabia. Atlal : -. 374 Whalen, N., A. Killick, N. James, G. Morsi, and M. Kamal. . Saudi Arabian archaeological reconnaissance : preliminary report on the western province survey. Atlal :-. Whalen, N. and D. Pease. a. Archaeological survey in southwest Yemen, . Paleorient :. Whalen, N. and D. Pease. b. Variability in developed Oldowan and Acheulean bifaces of Saudi Arabia. Atlal :-. Whalen. N. and K.E. Schatte. . Pleistocene sites in southern Yemen. Arabian Archaeology and Epigraphy :-. Whalen, N., H. Sindi, G. Wahida, and J. Siraj-Ali. . Excavation of Acheulean sites near Saffaqah in al-Dawadmi (/). Atlal :-. Whalen, N., J. Siraj-Ali, W. Davis. . Excavation of Acheulean sites near Saffaqah, Saudi Arabia. Atlal :-. Whalen, N., J. Siraj-Ali, H. Sindi, D. Pease, and M. Badein. . A complex of sites in the JeddahWadi Fatima Area. Atlal :-. Whalen, N., M. Zoboroski, and K. Schubert. . The Lower Palaeolithic in southwestern Oman. Adumatu :-. Whittaker, J.C. . Flintknapping: making and understanding stone tools. Austin: University of Texas Press. Wildman, D.E. . Mammalian Zoogeography of the Arabian Peninsula and Horn of Africa with a Focus on the Cladistic Phylogeography of Hamadryas Baboons. Ph.D. dissertation, New York University, New York. Wildman, D.E., T.J. Bergman, A. al-Aghbari, K.N. Sterner, T.K. Newman, J.E. Phillips-Conroy, C.J. Jolly, and T.R. Disotell. . Mitochondrial DNA evidence for the origin of Hamadryas Baboons. Molecular Phylogenetics and Evolution :-. Wilkinson, T.J. . Holocene environments of the high plateau, Yemen, recent geoarchaeological investigations. Geoarchaeology ():-. Wilkinson, T.J. and C. Edens. . Surveys and excavations in the central highlands of Yemen: results of the Dhamar Survey Project,  and . Arabian Archaeology and Epigraphy : -. Wilkinson, T., M. Gibson, and C. Edens. . The archaeology of the Yemen high plains: a preliminary chronology. Arabian Archaeology and Epigraphy :-. Williams, M.A.J., P.M. Bishop, F.M. Dakin, and R. Gillespie. . Late Quaternary lake levels in southern Afar and the adjacent Ethiopian Rift. Nature :-. Wissmann, H.V. and C. Rathjens. . Vorislamische Altertümer, Hamburg. Wolpoff, M.H., Zhi, W.X., and A. Thorne. . Modern Homo sapiens origins: a general theory of hominid evolution involving the fossil evidence from east Asia. In The Origins of Modern Humans, edited by F. Smith and F. Spencer. New York: Alan R. Liss. Pp. -. 375 Wolpoff, M.H., and R. Caspari. . Race and Human Evolution: a fatal attraction. New York. Simon and Schuster. Wolpoff, M., J. Hawks, D.W. Frayer, and K. Hunley. . Modern human ancestry at the peripheries: a test of the replacement theory. Science :-. Wormington, H.M. . Ancient Man in North America. Denver: Denver Museum of Natural History. Wykoff, D.G. . Refitting and protohistoric knapping behaviour. In Piecing Together the Past: applications of refitting studies in archaeology, edited by J.L. Hofman and J.G. Enloe. Oxford: BAR International Series no. . Pp. -. Yellen, J., A. Brooks, D. Helgren, M. Tappen, S. Ambrose, R. Bonnefille, J. Feathers, G. Goodfriend, K. Ludwig, P. Renne, and K. Stewart. The Archaeology of Aduma Middle Stone Age Sites in the Awash Valley, Ethiopia. PaleoAnthropology :-. Zarins, J. . View from the South: the greater Arabian Peninsula. In Prehistoric Archaeology of Jordan, edited by D. Henry. Oxford: BAR International Series . Pp. -. Zarins, J. . The Land of Incense: archaeological work in the Governorate of Dhofar, Sultanate of Oman, -. Muscat: Sultan Qaboos University Publications. Zarins, J. and H. al-Badr. . Archaeological investigation in the southern Tihama Plain, part II (including Sihi, - and Sharja, -) /. Atlal :-. Zarins, J., M. Ibrahim, D. Potts, and C. Edens. . The preliminary report on the third phase of the Comprehensive Archaeological Survey Program—the central province. Atlal :-. Zarins, J., A. Murad, and K. al-Yish. . Comprehensive Archaeological Survey Program: the second preliminary report on the southwestern province. Atlal :-. Zarins, J., A. Rahbini, and M. Kamal. . Preliminary report on the archaeological survey of the Riyadh area. Atlal :-. Zarins, J., N. Whalen, M. Ibrahim, A. Mursi, and M. Khan. . Comprehensive Archaeological Survey Program: preliminary report on the central and southwestern province survey: . Atlal :-. Zarins, J. and J. Zahrani. . Recent archaeological investigations in the southern Tihama Plain (including Aththar, - and Sihi, -). Atlal :-. Zeitoun, V. . The taxonomical position of the skull of Zuttiyeh. Comptes Rendus de l’Academie des Sciences :-. Zeuner, F.E. . Neolithic sites from the Rub Al-Khali, southern Arabia. Man :-. Zonneveld, K., G. Ganssen, S. Troelstra, G. Versteegh, and H. Visscher. . Mechanisms forcing abrupt fluctuations of the Indian Ocean summer monsoon during the last deglaciation. Quaternary Science Review :-. 376