EARLY IRON AGE SOCIAL AND ECONOMIC ORGANIZATION IN SOWA PAN, BOTSWANA By Adrianne M. Daggett A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Anthropology Doctor of Philosophy ABSTRACT EARLY IRON AGE SOCIAL AND ECONOMIC ORGANIZATION IN SOWA PAN, BOTSWANA By Adrianne M. Daggett The Early Iron Age (ca 200 1000 AD) in Southern Africa was a time of expansion, reorganization, and innovation that laid the foundation for the complex system of inter - group interaction that early European explorers first encountered in the 15 th century and that continues to influence community dynamics today. Across the subcontinent, indigenous hunting - and - gathering communities encountered groups of immigrant farmers and herders from Central and East Africa who brought with them new technologies and new forms of subsistence. Over the centuries, both indigenous and migrant communities experienced demographic shifts, changes in settlement patterns, transitions in economic practices, and cultural and social transformations. Much research for this time period focuses on the changes experienced by hu nter - gatherer communities as they were affected by the presence of encroaching agro - pastoral populations . Another related body of research seeks to understand internal dynamics of agro - pastoral groups over time, particularly in the economic and cultural he artland of Shashe - Limpopo. However, a number of studies have documented hunter - gatherer influence on agro - pastoral community cultural practices. F urthermore, agro - pastoral socioeconomic strategies have been shown to vary regionally to an extent. Here, I as k what external influences acted on agro - pastoral communities, and what interpretations of the Early Iron Age culture(s) as a whole. This project reframes community - level socioeconomic processe s: rather than seeing agropastoralist sites as parts of a predominant and hegemonic sphere of influence, sites are nodes within a cross - continental, multi - scalar network in which multiple avenues of influence social, geographical, and environmental ope rate on all comm unities. In other words, if agro - pastoral communities are recontextualized as one influence among many in a network, how might regional variation in their material culture be explained? R esearchers acknowledge the presence of a multi - scalar network of interaction and exchange incorporating several types of communities (including hunter - gatherers and coastal traders), but most assemblages tend to be analyzed from the point of a local and Iron Age - predominant perspective. Strong preference is ceramics in particular of Iron Age society. On a methodological level, this research asks w hat inquiry that incorpo rates multiple lines of evidence, including non - traditional artifact types, can elucidate about the socioeconomic organization of a geographically and culturally peripheral site. In particular, through the use of a high - resolution site - level spatial datase t, this project seeks to lay the foundation for a more robust interpretation of use of space within sites in Early Iron Age Southern Africa. Excavation of Thabadimasego , survey of parts of its surrounding landscape, and interpretation of the resulting asse mblage formed the basis of the dataset for this project. Thabadimasego is one of several Early Iron Age sites in the Mosu Escarpment area of northeastern Botswana which form a settlement complex that is only beginning to be understood. Overall, the researc h project addresses the role of smaller sites like Thabadimasego within the social and economic exchange network which is so often cited as crucial to the development of socioeconomic complexity in the Southern African Iron Age. Areas such as South Sowa, situated on the fringes of known Early Iron Age settlement distribution, are frequently framed as peripheral, if only implicitly, and in comparison with contemporary The data collected for this project comprise one of the few high - resolution spatial datasets for Early Iron Age sites in Southern Africa that give attention to the site's comprehensive set of material culture components. As such this dataset stands to cont ribute to the ongoing scholarly discussion on the relationship between site organization, socioeconomic organization, and group identity, as well as the interplay between regional economies and supra - regional cultural processes. iv This work is dedicated to my parents, Greg and Sara, for their immeasurable support and love . v ACKNOWLEDGEMENTS I may have written the dissertation , but dozens if not hundreds of people made it possible . I owe them all a debt of gratitude for supporting me, motivating me, and caring for me during the course of my graduate studies. My graduate committee - Professor Emeritus Lawrence Robbins, Professor Ethan Watrall, Professor Emerita Helen Pollard, and Professor Jon Frey have been epic cheerleaders , editors, and butt - kickers , as warranted. The National Science Foundation (Doctoral Dissertation Improvement Grant #1220479) , the MSU Graduate School, and the MSU Alliance of Graduat e Education and the Professoriate funded the research. The Department of Anthropology has also provided help (and funding!) guidance. The following people worked with me directly on the disse rtation research: Professor James Denbow (University of Texas - Austin) and Dr. Carla Klehm (University of Washington St. Louis) gave me pointers for working in Botswana, loaned me equipment, and much more; Dr. Morongwa Mosothwane (University of Botswana) an d Dr. Stefania Merlo (formerly UB, now University of the Witwatersrand) patiently fielded endless questions from me and set me up with a wonderful field crew; Professors Karim Sadr and Thomas Huffman (University of the Witwatersrand) likewise offered advic e , help with ceramics, and logistical support, including a place to stay in Joburg; Professor Jon Carroll weighed in on my spatial methodology; Dr. Sheila Coulson and Miss Sigrid Staurset (University of Oslo) provided advice on lithics and on surviving fie ldwork; Professor Edwin Wilmsen (University of Edinburgh) gave me pointers on working with ceramics and glass; Mr. Abel A. Mabuse and Mr. Phillip Segadika (National Museum of Botswana) provided on - the - ground support, consultation, and logistics help, parti cularly in the post - excavation work; Ms. Mighty Mmolawa and other archaeology staff of the National Museum assisted with labwork; Mr. Obonetse Maoto and Mr. Flex Mashabagole (Mosu office of the National Museum) worked with me and my crew directly on - site and offered lots of good advice ; and finally, Mr. and Mrs. Mike and vi Kerstin Main gave me a place to stay in Gaborone as well as the use of their truck and camping gear (not to mention numerous dinners and conversations) . Friends like Miss Kefilwe Rammutloa and Ms. Sarita Louzolo kept me grounded more than they know. My very skilled full - time field crew included Mr. Mpho Basebi, Miss Milidzani Nthomiwe, and Mr. Thabu Kgosietsile, while my talented and energetic University of Botswana field students included Mr. Joshua Ikhutseng, Miss Patience Mgadla, Miss Anisa Mpolaise (who taught me how to make bread) , Miss (who also came back to work with me in the lab) . Mrs. Leametse Oitutile, a.k.a. Mma Moga, was a w onderful camp attendant and c ook ; her hospitality and unfailing good cheer kept the rest of us sane on many a hot afternoon. Numerous other residents of the village of Mosu helped us sort out crises of water, food, electricity , and very gracious ly show ed u s around. Several people collaborated on the post - excavation data collection and analysis . AMS samples were processed at the University of Arizona AMS Laboratory. Dr. James Feathers at the University of Washington processed the OSL samples and wrote up the results. Professor Shaw Badenhorst (University analysis. Dr. Marilee Wood conducted morphological analysis of the glass trade beads, and Dr. Laure Dussu bieux (Field Museum) guided me through the process of conducting LA - ICP - MS on the beads , under NSF BCS #1321731, as well as on how to interpret the results. Marilee and Laure also co - authored a conference paper and a publication on the glass bead results. Dr. Heather Walder put the bug in my brain to take a closer look at the glass in the first place! Dr. Catrien van Waarden talked me through carbonized grain identification and once very kindly showed me around the stonewalled sites of Francistown. Miss Tsh olo Selepeng, a graduate of the UB Archaeology program, worked with me on the ceramic analysis. I had the assistance of a number of Michigan State University undergraduate students too: Mr. Ian Harrison actually withstood the trial of my company for a whol e month of ceramic identification in Gaborone, and vii Miss Allison Apland, Mr. Josh Schnell, and Miss Taylor Flaherty worked with me on recording ostrich eggshell bead measurements . Mr. Brian Geyer taught me the basics of 3D scanning and rendering and gave se veral helpful theoretical critiques on the interpretation of my results. worth, but a lifetime of loving support and all the reality checks I could ask for. My parents, Sara and Greg, my sisters Jen, Katie, and Shannon, brothers - in - law Bryan and Kevin, nieces and nephew Hannah, Rosie, more grateful than I can say. My writing buddies, co - conspirators and commiserators Dr. Charlotte Cable, Dr. Anna Jefferson, Ms. Rowenn Kalman, and Ms. Cris Chapman have been there for me through thick and thin. for the omission. Truly, this has been a labor of one thousand hands (and one thousand pots of coffee) . viii TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ................................ ............. xi LIST OF FIGURES ................................ ................................ ................................ ................................ .......... xiii Chapter 1 Introduction: site description, research questions, and environmental background ................. 1 1.1 Research questions ................................ ................................ ................................ ............................. 4 1.2 En vironmental background ................................ ................................ ................................ ............... 11 Present day ................................ ................................ ................................ ................................ ......... 11 Climate and physical environment ................................ ................................ ................................ . 11 Fauna ................................ ................................ ................................ ................................ ............... 17 Water ................................ ................................ ................................ ................................ .............. 18 Seasonality ................................ ................................ ................................ ................................ ...... 19 Paleoenvironmental reconstructions ................................ ................................ ................................ . 20 Site description ................................ ................................ ................................ ................................ ... 22 1.3 Chapter overview ................................ ................................ ................................ .............................. 25 Chapter 2 Pertinent archaeological background and theoretical considerations ................................ ...... 28 2.1 Subcontinental framework: Southern African Late Holocene archaeology ................................ ..... 29 Defining the area ................................ ................................ ................................ ................................ 29 Defining t he period ................................ ................................ ................................ ............................. 33 The Later Stone Age - in broad strokes ................................ ................................ ............................... 38 Later Stone Age in Botswana - Regional trends ................................ ................................ .............. 42 Early migrations/ settlements - in broad strokes ................................ ................................ ............... 44 2.2 Regional (Botswana) framework: localized chronologies and socioeconomic trends, etc. .............. 61 2.3 Local archaeological record: the south Sowa ................................ ................................ ...... 65 Prior work in the area ................................ ................................ ................................ ......................... 65 Chapter 3 Field Methods ................................ ................................ ................................ ............................ 72 3.1 Introduction ................................ ................................ ................................ ................................ ...... 72 3.2 Goals for fieldwork ................................ ................................ ................................ ............................ 73 3.3 Hypotheses ................................ ................................ ................................ ................................ ........ 74 3.4 Data Collection (Fieldwork) ................................ ................................ ................................ ............... 75 Phase I - Thabadimasego (main site) ................................ ................................ ................................ .. 75 Ground Truthing ................................ ................................ ................................ .............................. 75 On - site survey ................................ ................................ ................................ ................................ . 75 Test pit survey ................................ ................................ ................................ ................................ . 79 Excavation ................................ ................................ ................................ ................................ ....... 80 Stratified test pits ................................ ................................ ................................ ............................ 81 Radiomet ric data collection ................................ ................................ ................................ ............ 83 Phase II: Thabadimasego periphery ................................ ................................ ................................ .... 83 Test pits ................................ ................................ ................................ ................................ ........... 84 Test unit ................................ ................................ ................................ ................................ .......... 84 In - field feedback based on survey results ................................ ................................ ...................... 86 ix Phase III: Escarpment survey ................................ ................................ ................................ .............. 86 Phase IV: Additional sites ................................ ................................ ................................ .................... 88 Phase V: Visits to previously excavated sites ................................ ................................ ...................... 90 Phase VI: Landscape survey ................................ ................................ ................................ ................ 91 Cl ay deposits ................................ ................................ ................................ ................................ ... 91 Wells, dams and boreholes ................................ ................................ ................................ ............. 92 3.5 Post - excavation procedures ................................ ................................ ................................ .............. 92 Summary of post - excavation activities ................................ ................................ ............................... 92 Goals for post - excavation ................................ ................................ ................................ ................... 93 Cataloging and curation ................................ ................................ ................................ ...................... 93 Cha pter 4 Analytical methods ................................ ................................ ................................ ..................... 96 4.1 Introduction overview of methods ................................ ................................ ................................ 96 4.2 Ceramic analysis ................................ ................................ ................................ ................................ 96 Methods ................................ ................................ ................................ ................................ .............. 97 Results ................................ ................................ ................................ ................................ ................. 98 4.3 Shell bead analysis ................................ ................................ ................................ .......................... 105 Methods ................................ ................................ ................................ ................................ ............ 112 Shell beads results ................................ ................................ ................................ ............................ 112 Summary of findings ................................ ................................ ................................ ......................... 114 4.4 Metal analysis ................................ ................................ ................................ ................................ . 115 Methods ................................ ................................ ................................ ................................ ............ 119 Results ................................ ................................ ................................ ................................ ............... 124 4.5 Faunal analysis ................................ ................................ ................................ ................................ 125 Methods ................................ ................................ ................................ ................................ ............ 128 Results ................................ ................................ ................................ ................................ ............... 129 Discussion ................................ ................................ ................................ ................................ ......... 134 4.6 Radiometric analysis ................................ ................................ ................................ ....................... 135 Ac celerator mass spectrometer (AMS) analysis ................................ ................................ ............... 135 Optically - stimulated Luminescence (OSL) analysis ................................ ................................ ........... 137 Discussion ................................ ................................ ................................ ................................ ......... 140 4.7 Glass beads ................................ ................................ ................................ ................................ ...... 141 Glass bead analysis ................................ ................................ ................................ ........................... 142 Methods ................................ ................................ ................................ ................................ ........ 144 Results ................................ ................................ ................................ ................................ ............... 146 4.7 Macrobotanical identification ................................ ................................ ................................ ......... 152 Discussion ................................ ................................ ................................ ................................ ......... 154 Chapter 5 Spatial analysis ................................ ................................ ................................ ......................... 160 5.1 Spatial information and analysis ................................ ................................ ................................ ..... 160 5.2 Spatial methodology and results ................................ ................................ ................................ .... 160 Spatial analysis methods ................................ ................................ ................................ ................... 163 Histograms ................................ ................................ ................................ ................................ .... 167 ArcMap displays ................................ ................................ ................................ ............................ 169 Looking beyond exploratory data analysis ................................ ................................ ................... 176 Summary of results ................................ ................................ ................................ ........................... 181 Original versus standardized data ................................ ................................ ................................ . 181 Interpretation of activity areas shell beads ................................ ................................ ............... 183 x Site layout ................................ ................................ ................................ ................................ ..... 188 Chapter 6 Conclusions ................................ ................................ ................................ .............................. 193 6.1 Site level: Use of space and diachronic variability ................................ ................................ .......... 193 Site - level analysis ................................ ................................ ................................ .............................. 193 A commen t on the site - level methodology ................................ ................................ ...................... 204 6.2 Intra - site scale: comparing with other sites and engaging with explanatory frameworks ............ 2 05 6.3 Forager - farmer interaction ................................ ................................ ................................ ............. 210 6.4 Conclusion thoughts on identity and theoretical praxis ................................ .............................. 218 Further considerations ................................ ................................ ................................ ...................... 221 APPENDICES ................................ ................................ ................................ ................................ .............. 224 APPENDIX A : MISCELLANEOUS TABLES ................................ ................................ ................................ ..... 225 APPENDIX B : SHELL BEAD DATA ................................ ................................ ................................ ................ 256 APPENDIX C : METAL DATA ................................ ................................ ................................ ........................ 289 APPENDIX D : OSL REPORT ................................ ................................ ................................ ......................... 311 BIBLIOGRAPHY ................................ ................................ ................................ ................................ ...... 319 APPENDIX E : GLASS BEAD ANALYSIS REPORT ................................ ................................ ........................... 321 BIBLIOGRAPHY ................................ ................................ ................................ ................................ ...... 323 APPENDIX F : MATERIAL DISTRIBUTION HISTOGRAMS FOR THE ORIGINAL DATA ................................ .... 332 APPENDIX G : MATERIAL DISTRIBUTION FOR STANDARDIZED DATA ................................ ........................ 368 BIBLIOGRAPHY ................................ ................................ ................................ ................................ .......... 403 xi LIST OF TABLES Table 1 Counts of sherd types ................................ ................................ ................................ ................... 105 Table 2 Zhizo sherds ................................ ................................ ................................ ................................ .. 105 Table 3 List of taxa present in assemblage ................................ ................................ ............................... 130 Table 4 Summary of NISP (Number of individual specimens) present in assemblage ............................ 132 Table 5 AMS Dates from Thabadi masego and Site 12 ................................ ................................ .............. 136 Table 6 OSL results ................................ ................................ ................................ ................................ .... 140 Table 7 Mean values of major and minor oxides from LA - ICP - MS ................................ ........................... 147 Table 8 Series determinations for glass beads from Site 12 and Thabadimasego ................................ ... 151 Table 9 Artifa ct categories for analysis ................................ ................................ ................................ ..... 165 Table 10 Comparison of selected Early Iron Age sites ................................ ................................ .............. 201 Table 11 Pit and Unit Summary ................................ ................................ ................................ ................ 226 Table 12 Flotation inventory ................................ ................................ ................................ ..................... 231 Table 13 Ceramic facies determinations ................................ ................................ ................................ ... 234 Table 14 Finished OES beads ................................ ................................ ................................ .................... 257 Table 15 Finished Achatina beads ................................ ................................ ................................ ............. 263 Table 16 Broken OES beads ................................ ................................ ................................ ...................... 268 Table 17 Broken ACH beads ................................ ................................ ................................ ...................... 271 Table 18 Irregular OES beads ................................ ................................ ................................ .................... 272 Table 19 Irregular ACH beads ................................ ................................ ................................ ................... 277 Table 20 OES blanks and fragments ................................ ................................ ................................ .......... 279 Table 21 ACH blanks and fragments ................................ ................................ ................................ ......... 284 Table 22 River mussel beads ................................ ................................ ................................ ..................... 288 xii Table 23 Ferrous beads ................................ ................................ ................................ ............................. 290 Table 24 Non - ferrous beads ................................ ................................ ................................ ..................... 295 Table 25 Metal fragments, bar and wi re ................................ ................................ ................................ .. 298 Table 26 Slag ................................ ................................ ................................ ................................ ............. 300 Table 27 OSL S amples ................................ ................................ ................................ ............................... 312 Table 28 Radiation ................................ ................................ ................................ ................................ .... 312 Table 29 Dose rates (Gy/ka)* ................................ ................................ ................................ .................... 312 Table 30 Acceptance rates* ................................ ................................ ................................ ...................... 313 Table 31 Equivalent dose (central age model) ................................ ................................ .......................... 313 Table 32 Finite Mixture Model components ................................ ................................ ............................ 314 Table 33 Ages ................................ ................................ ................................ ................................ ............ 314 Table 34 Glass bead compositions (major and minor oxides only) ................................ .......................... 325 Table 35 Mean values (%) of major and minor oxides for Site 12 and 13 beads ................................ ...... 329 Table 36 Elements and oxides included in principal components analysis and their PCA values ............ 330 Table 37 Series determinations for glass beads from Site 12 and Thabadimasego ................................ . 331 xiii LIST OF FIGURE S Figure 1 Location of Sowa Pan in Botswana ................................ ................................ ............................... 13 Figure 2 The South Sowa are a and its known archaeological sites ................................ ............................ 14 Figure 3 Thabadimasego site location ................................ ................................ ................................ ........ 27 Figure 4 Archaeological sites in the South Sowa area ................................ ................................ ................ 30 Figure 5 Areas mentioned in the text ................................ ................................ ................................ ......... 55 Figure 6 Zhizo si tes in the South Sowa area ................................ ................................ ............................... 69 Figure 7 Areas surveyed in 2012 ................................ ................................ ................................ ................. 77 Figure 8 10 - by - 10 - meter grid on Thabadimasego ................................ ................................ ...................... 78 Figure 9 Location of ex cavation units at Thabadimasego ................................ ................................ ........... 82 Figure 10 Location of all units and pits, positive and negative, at Thabadimasego ................................ ... 85 Figure 11 Mosu escarpment survey area ................................ ................................ ................................ .... 89 Figure 12 Summary of pottery types present at Thabadimasego ................................ ............................ 102 Figure 13 Examples of Zhizo pottery from Thabadimasego ................................ ................................ ..... 103 Figure 14 Pottery designated as Ziwa (L) and Eiland (R) from Thabadimasego ................................ ....... 104 Figure 15 OES beads from Thabadimasego ................................ ................................ .............................. 108 Figure 16 Achatina beads from Thabadimasego ................................ ................................ ...................... 109 Figure 17 Diameter ratio for shell beads at Thabadimasego ................................ ................................ .... 116 Figure 18 External diameters of shell beads at Thabadimasego ................................ .............................. 117 Figure 19 Internal diameters of shell beads at Thabadimasego ................................ ............................... 118 Figure 20 Copper beads from Thabadimasego ................................ ................................ ......................... 122 Figure 21 Copper items from Thabadimasego ................................ ................................ .......................... 123 xiv Figure 22 Slag fragment from Thabadimasego ................................ ................................ ......................... 127 Figure 23 NISP percentages ................................ ................................ ................................ ...................... 133 Figure 24 SHCal13 calibration curve in relation to U19 - L4 charcoal sample AMS results ........................ 136 Figure 25 SHCal13 calibration curve in relation to U3 - L2 charcoal sample AMS results .......................... 137 Figure 26 Chibuene (L) and Zhizo (R) glass beads ................................ ................................ ..................... 148 Figure 27 PCA score plot for Zhizo vs Chibuene glass types ................................ ................................ ..... 149 Figure 28 PCA loading plot for Zhizo vs Chibuene glass ................................ ................................ ............ 150 Figure 29 A present - day example of sorghum ................................ ................................ .......................... 154 Figure 30 Probable sorghu m from Thabadimasego ................................ ................................ .................. 155 Figure 31 Screen capture of a portion of the attribute table for the excavation units in ArcGIS ............ 168 Figure 32 Example of histogram displaying frequencies of artifact counts per unit/ pit ......................... 170 Figure 33 Value - weighted plotting of bone (by mass, per pit) at Thabadimasego ................................ .. 172 Figure 34 Stone, mudbrick and ash features at Thabadimasego ................................ .............................. 175 Figure 35 Portion of the stone wall at Kaitshàa ................................ ................................ ........................ 179 Figure 36 Portion of the stone wall at Thabadimasego ................................ ................................ ............ 180 Figure 37 Mater ials with unusually high values ................................ ................................ ........................ 184 Figure 38 Materials with unusually high values plotted with ash and mudbrick features ....................... 1 85 Figure 39 All mat erial types; legend on following page ................................ ................................ ............ 186 Figure 40 Legend for "all material types" density map ................................ ................................ ............ 187 Figure 41 Variation in soil compaction ................................ ................................ ................................ ..... 189 Figure 42 Variation in soil compaction with distribution of key cultural materials ................................ .. 190 Figure 43 Fragment of grooved stone found on Thabadimasego's surface ................................ ............. 192 Figure 44 Thabadi masego surface features ................................ ................................ .............................. 196 Figure 45 Unit 13 west wall profile ................................ ................................ ................................ ........... 197 xv Figure 46 Unit 20 west wall profile ................................ ................................ ................................ ........... 198 Figure 47 Unit 21 west wall profile ................................ ................................ ................................ ........... 199 Figure 48 Cluster location estimates at Thabadimasego ................................ ................................ .......... 200 Figure 49 Site 12 unit profile ................................ ................................ ................................ ..................... 206 Figure 50 Worked metal object proportions (by count) ................................ ................................ ........... 306 Figure 51 Metal items by mass ................................ ................................ ................................ ................. 307 Figure 52 Metal jewelry by mass ................................ ................................ ................................ .............. 308 Figure 53 Metal items by count ................................ ................................ ................................ ................ 309 Figure 54 Metal jewelry by count ................................ ................................ ................................ ............. 310 Figure 55 Radial graphs of each OSL sample ................................ ................................ ............................ 315 Figure 56 Achatina total, count, per pit ................................ ................................ ................................ .... 333 Figure 57 Achatina >50%, count, per pit ................................ ................................ ................................ ... 334 Figure 58 Finished Achatina beads, count, per pit ................................ ................................ .................... 335 Figure 59 Irregular Achatina beads, count, per pit ................................ ................................ ................... 336 Figure 60 Achatina <50% beads, count, per pit ................................ ................................ ........................ 337 Figure 61 Bone, mass, per pit ................................ ................................ ................................ ................... 338 Figure 62 Burnt seed, mass, per pit ................................ ................................ ................................ .......... 339 Figure 63 Charcoal, mass, per pit ................................ ................................ ................................ .............. 340 Figure 64 Dhaka, mass, per pit ................................ ................................ ................................ .................. 341 Figure 65 Ferrous material, mass, per pit ................................ ................................ ................................ . 342 Figure 66 Ferrous beads, count, per pit ................................ ................................ ................................ .... 343 Figure 67 Ferrous beads, mass, per pit ................................ ................................ ................................ ..... 344 Figure 68 Ferrous fragments, mass, per pit ................................ ................................ .............................. 345 Figure 69 Ferrous wire, count, per pit ................................ ................................ ................................ ...... 346 xvi Figure 70 Ferrous wire, mass, per pit ................................ ................................ ................................ ....... 347 Figure 71 Glass beads, count, per pit ................................ ................................ ................................ ........ 348 Figure 72 Metal, mass, per pit ................................ ................................ ................................ .................. 349 Figure 73 Non - ferrous metal, mass, per pit ................................ ................................ .............................. 350 Figure 74 Non - ferrous beads, count, per pit ................................ ................................ ............................ 351 Figure 75 Non - ferrous beads, mass, per pit ................................ ................................ .............................. 352 Figure 76 Non - ferrous wire, count, per pit ................................ ................................ ............................... 353 Figure 77 Non - ferrous wire, mass, per pit ................................ ................................ ................................ 354 Figure 78 OES, count, per pit ................................ ................................ ................................ .................... 355 Figure 79 OES <50%, count, per pit ................................ ................................ ................................ ........... 356 Figure 80 OES beads finished, count, per pit ................................ ................................ ............................ 357 Figure 81 OES beads irregular, count, per pit ................................ ................................ ........................... 358 Figure 82 OES beads >50%, count, per pit ................................ ................................ ................................ 359 Figure 83 OES fragments, count, per pit ................................ ................................ ................................ ... 360 Figure 84 Pottery, mass, per pit ................................ ................................ ................................ ................ 361 Figure 85 Decorated body sherds, mass, per pit ................................ ................................ ...................... 362 Figure 86 Decorated rim sherds, mass, per pit ................................ ................................ ......................... 363 Figure 87 Undecorated body sherds, mass, per pit ................................ ................................ .................. 364 Figure 88 Undecorated rim sherds, mass, per pit ................................ ................................ ..................... 365 Figure 89 Shell items, count, per pit ................................ ................................ ................................ ......... 366 Figure 90 Slag, mass, per pit ................................ ................................ ................................ ..................... 367 Figure 91 Achatina per pit, count, standardized ................................ ................................ ....................... 369 Figure 92 Metal per pit, mass, standardized ................................ ................................ ............................ 370 Figure 93 Shell per pit, count, standardized ................................ ................................ ............................. 371 xvii Figure 94 Non - ferrous items per pit, mass, standardized ................................ ................................ ........ 372 Figure 95 Ferrous items per pit, mass, standardized ................................ ................................ ................ 373 Figure 96 OES per pit, count, standardized ................................ ................................ .............................. 374 Figure 97 Achatina >50% per pit, count, standardized ................................ ................................ ............. 375 Figure 98 Achatina finished per pit, count, standardized ................................ ................................ ......... 376 Figure 99 Achatina irregular per pit, coun t, standardized ................................ ................................ ........ 377 Figure 100 Achatina <50% per pit, count, standardized ................................ ................................ ........... 378 Figure 101 Bone per pit, mass, standardized ................................ ................................ ............................ 379 Figure 102 Burnt seed per pit, mass, standardized ................................ ................................ .................. 380 Figure 103 Charcoal per pit, mass, standardized ................................ ................................ ...................... 381 Figure 104 Ferrous beads per pit, count, standardized ................................ ................................ ............ 382 Figure 105 Ferrous beads per pit, mass, standardized ................................ ................................ ............. 383 Figure 106 Ferrous fragments per pit, mass, standardized ................................ ................................ ...... 384 Figure 107 Ferrous wire per pit, count, standardized ................................ ................................ .............. 385 Figure 108 Ferrous wire per pit, mass, sta ndardized ................................ ................................ ................ 386 Figure 109 Glass per pit, count, standardized ................................ ................................ .......................... 387 Figure 110 Non - ferrous beads per pit, count, standardized ................................ ................................ ..... 388 Figure 111 Non - ferrous beads per pit, mass, standardized ................................ ................................ ...... 389 Figure 112 Non - ferrous wire per pit, count, standardized ................................ ................................ ....... 390 Figure 113 Non - ferrous wire per pit, m ass, standardized ................................ ................................ ........ 391 Figure 114 OES beads >50% per pit, count, standardized ................................ ................................ ........ 392 Figure 115 OES beads finished per pit, count, standardized ................................ ................................ .... 393 Figure 116 OES beads irregular per pit, count, standardized ................................ ................................ ... 394 Figure 117 OES beads <50% per pit, standardized ................................ ................................ ................... 395 xviii Figure 118 OES fragments per pit, count, stand ardized ................................ ................................ ........... 396 Figure 119 Pottery per pit, mass, standardized ................................ ................................ ........................ 397 Figure 120 Decorated body sherds per pit, mass, standardized ................................ .............................. 398 Figure 121 Decorated rim sherds per pit, mass, standardized ................................ ................................ . 399 Figure 122 Undecorated body sherds per pit, mass, standardized ................................ .......................... 400 Figure 123 Undecorated rim she rds per pit, mass, standardized ................................ ............................. 40 1 Figure 124 Slag per pit, mass, standardized ................................ ................................ ............................. 402 1 Chapter 1 Introduction : s ite description, research questions, and environmental background T he last 2000 years witnessed a series of substantial shifts in both the demographic and the economic characteristics of societies in Southern Africa. Prior to then, small and usually mobile communities of hunter - gatherers were the only occupants of the subcontin ent. The ear liest centuries AD , however, saw multiple waves of coloni zation by food - producing migrants incoming from East and Central Africa, thus marking the start of what archaeologists generally characterize as the Iron Age in Southern Africa. As these herding and farming commun ities established themselves first on the coasts and, over time, within the interior of Southern Africa, both the indigenous hunter - gatherers and the underwent transformations in terms of where they settled, how they organized their communities, what they ate, and how they represented themselves as individuals and groups. Furthermore, a series of shifts in the social and economic organization beginning in the late first millennium AD, parti cularly in the Shashe - Limpopo Basin of South Africa, indicate the beginnings of social stratification and political centralization which would come to encompass much of the subcontinent in the second millennium. A considerable portion of Southern African a rchaeological study has been dedicated to understanding these transformations and how they led to the diverse societies encountered by early European explorers in the 15th and 16th centuries AD . We as archaeologists are only beginning to appreciate the com plexity of the social, economic and political systems that for med in the first 1500 years AD in Southern Africa. This project takes a look at some of those systems from the perspective of a specifi c area in northeastern Botswana known as the Makgadikgadi P ans. In particular, an extensive yet little - studied cluster of sites on the s outhern edge of Sowa Pan, the largest of the Makgadikgadi Pans, is the geographic focus of this study. In particular, the material evidence from one small site within South Sowa, called Thabadimasego , is scrutinized. localized adaptation of, region - wide cultural and political systems. For the time period in question - the 2 last half of the first millennium AD , also called the Early Iron Age - archaeologists c onsider the eastern third of Southern Africa (that is, what are now Zimbabwe, s outhern Mozambique, eastern Botswana, eastern South A frica, Lesotho, and Swaziland) to be a coherent geographical unit in terms of its subsistence traditions, ethnic affiliations, and language family. At the same time, it is also widely recognized that multiple streams of migrations, moving into Southern Africa from relatively disparate parts of Central and East Africa, contributed to the series of cultural developments which occurred during the Early Iron Age ( for a map of these migration streams, se e Huffman 2007 :122 ). Likewise , the eastern third of Southern Africa i s home to a wide range of physical environments, from desert - like semi - arid savannahs to montane forests and coastal plateaus. As is discussed in detail in the proceeding chapters, the range of diversity in natural resources and demography alike is someti mes addressed in Southern literature, and sometimes not. Much attention has been given to settlements located in what is often viewed as the cultural center of the Iron Age, like Schroda in the Shashe - Limpopo river confluence (Hanisch, 1980 ; Huffman, 2000 ; Calabrese 2000 ; Van Doornum, 2005 ; Antonites 2014) , and these settlements no doubt made highly influential contributions to socioeconomic processes of th e time. However, numerous smaller and potentially more peripheral sites populate the landscapes surrounding the larger sites (as studies such as van Doornum 2005 and Klehm 2013 c an attest), as well as in less - central areas of Southern Africa. Here an argument is presented for a need to move from broad to more specific localized characterizations of subsistence and social organization, while still paying heed to the way that local studies speak to what as a cultural designation, means for Southern Africa as a whole. The predominant explanatory framework for Early Iron Age lifeways across the whole subcontinent has been developed based on data from one particular region the Shashe - Limpopo Basin of South Africa and partly because of this, variation and change tend t o be explained in simplistic terms of the migration and diffusion of culture - As a result, certain types of material culture, particularly 3 ceramics, have been privileged as objects that signal group identity, behavior, and population mo vement. While shared cultural traditions clearly played a strong role tying together communities throughout the subcontinent - shared or related pottery styles range over hundreds of kilometers, for example, and goods produced in one area can be observed a s trade items in other localities - it must also be recognized that some specific attributes of geography, climate, and community dynamics would likewise be major structuring forces in the way that the set of subsistence practices and sociocultural traditi ons call ed the manifested in any given area of Southern Africa. Much more work needs to be done to interpret regional d evelopments in social and economic behaviors to acknowledge the way these factors came into play. Inferences that Southe rn Africanist scholars make based on extant archaeological datasets can also be problematic. In particular, establishing the socioeconomic identity such as San hunter, Khoe herder, or Bantu farmer of the prehistoric occupants of archaeological sites ha s been a scholarly priority far more than the extant archaeological record has the power to actually answer this question (see, for example, Maggs and Whitelaw 1991 ; Smith et al. 1991; Denbow 1999, 2002; Sinclair 2004 ; Shepherd 2003; A. B. Smith et al. 2008 . Archaeologists have indeed been questioning the prevalence of this line of inquiry for nearly as long as the classic tri - partate scheme of hunter - farmer - herder has been in use within Southern African scholarship (e.g., Lane 1994/5, Walker 1998 ; Meskell 2002 ) . In order to address both of these issues t he data use of EIA lifeways ought to be expanded upon beyond ceramic typologies and high - status objects like metal, and regional case studies which incorporate the full range of variation present in site types and settlemen t patterns must be adopted. The relatively narrow samp le data as well as site types which curren tly inform the existing models provide a weak foundation for theoretical arguments about social organization. Excavating a wider geographic range of sites , using a more robust methodology and with increased attention to the geophyisical and environmental contexts of the remains , is the kind of foundational work 4 that will vastly increase the extant body of data from which Southern Africanist archaeologists de velop thei r models. In other words, archaeologists working in Southern Africa need to spend more time fleshing out our understanding of what can be observed more directly, such as diet, technology, and the taphonomic effects on site formation, before we ca n coherently answer questions about such intangible phenomena as ethnic identity. In light of this, the research presented here does not suggest an alternative model of social and spatial organization per se, but it does make the case for an alternative sy stem of data collection and a more flexible interpretive framework. In this research project, the evidence for subsistence activities, material culture traditions, and spatial organization at Thabadimasego is reviewed in detail and its relevance to underst anding the dynamic between local and regional sociocultural processes of the Early Iron Age is discussed. As will be shown, s mall, peripheral sites such as this one contain enormous potential for providing context - specific feedback to the broad - stroke beha vioral models currently favored in Southern African archaeology. 1.1 Research q uestions Early Iron Age settlements are typically characterized as small Bantu - speaking agricultural communities which were more or less self - sustaining in terms of subsistence but who nevertheless maintained extensive economic and cultural relationships with one another (Mitchell and Whitelaw 2005; Huffman 2007) . These relationships involved, to a greater or lesser degree over the centuries, the exchange of both bulk and luxury items; some of these goods were local products and some, such as glass beads, were acquired via connections to trading networks further abroad across eastern coastal Africa and glass - producing r egions across the Indian Ocean (Gilbert Pwiti 1991; Popelka et al. 2005; Robertshaw et al. 2010) . For the Southern African Early Iron Age, access to and control over the foreign goods exchange network has been cited as a potenti ally important factor in the development of social complexity (Hall 1987; Huffman 2000). My research addresses the role of smaller sites like Thabadimasego within this social and economic network. Areas such as South Sowa, situated on the fringes of known Early Iron Age 5 s ettlement distribution, are frequently seen as peripheral - if only implicitly, and in comparison with which are posited to have widespread influence over their surroun ding landscape ( Denbow 1984; Reid and Segobye 20 00a; Huffman 2000; Calabrese 2007) . The socioeconomic processes that characterize the Early Iron Age have been addressed on a number of scales. Some researchers look at regional patterns of trade or settlement (Denbow 1982; Denbow 1984; van Doornum 2005) , while others look on the level of the site (H. J. Greenfield and Miller 2004; H. Greenfield and van Schalkwyk 2006; Badenhorst 2009) . On the scale of the individual site, (1986, 1990, 2001) (CCP) model has played a major role in informing as social organization. This model emphasizes the role of cattle - keeping, and cattle as bridewealth, as a means of maintaining political authority an d increasing communal longevity, thus placing cattle are the heart of Early Iron Age social, spatial, cultural and economic organization. The original C CP was derived from observations of nineteenth - century Sotho - Tswana and Nguni settlement patterns by Kuper (1982) , though Huffman maintains the claim that this socioeconomic - spatial way of organizing Eastern Bantu - speaking communities stretches as far back as the Early Iron Age. He references in support such sites as Broederstroom and Kwagandaganda ( 5th - 7th c. AD, South Africa) as well as Kgaswe (11th c AD, Botswana). Hoewver, even thou g h t he Central Cattle Pattern addresses dynamics at a site - specific level, and while the spatial component of the model does seem to agree with the layout of multiple archaeological sites dating as far back as the first m illennium AD (including those referenced above), for all that Huffman claims an understanding of an Early Iron Age the model does not incorporate much discussion of netwo rks of exchange and interaction . Its application across many tempora l and geographical contexts of Southern Africa without accounting for regional factors has provoked a number of critiques , suggested alternative interpretations , a s well as calls to reassess the CCP . Lane ( 1995) , for example, raises important theoretical critiques regarding the 6 applicability of ethnography to understandings of prehistory, especially the lack of a direct historical approach in the case of the Early Iron Age. Mitchell and Whitelaw ( 2005) present a number of Early Iron Age sites whose layouts either do or do not fit well with the model. (Badenhorst (2011, 2012) and Sadr and Rodier (2012) both take quantitative approaches to evaluating the faunal component and stone wall layouts, respectively, of Iron Age sites with regard to their ranges of variation. While Badenhorst concludes that the layout attributed as the CCP may in fact have origina ted as a functional means of protecting livestock, Sadr concludes that a range of variation exists for Early Iron Age spatial layouts, of which the CCP is only one. Greenfield and van Schalkwyk (2006) specifically address intra - site layout at Ndondondwane, and theirs is one of the few studies to do so in more than descriptive terms. Their study raise the concerns that taphonomic effects as well as long - term occupation may affect what is interpreted as one condensed site layout. Importantly , Denbow (1979, 1982, 19 84, 1986) has established a framework for addressing dynamics of landscape use and settlement processes in eastern Botswana for the Early and Middle Iron Ages which incorporates inter - site and inter - group processes. In sum, while no one voice predominat es Early Iron Age research, neither has any one model gained as much currency as the Central Cattle Pattern, which remains widely - Collectively, these scholars bring attention to the need to look for alternative investigative methods as well as alternative frameworks for understanding the relationship between site formation, behavior, and worldview. For the most part, scholars taking a position in opposition, or framing an alternative, to the CCP have done so by addressing vario us issues within the framework. S ome, such as Badenhorst (2011, 2012), have considered the components; some, such as Denbow (1982, 1984, 1986) the location, and some, like Mitchell and Whitelaw (2005) the consistency of the pattern). What needs to be addressed further is the question of regional variation within the Early Iron Age. If regions within Southern Africa, through their exchange and shared cultural referents, together formed a cohesive socioeconomic system, then the role 7 of any one site or region within this sphere and its interoperability as a whole ought to be examined. This is particularly true for questions of socioeconomic and cultural organization, replicability of studies which address these questions . The studies cited above demonstrate that consideration of regional factors, variation in assemblage, and differences in scale are all crucial to understanding how and why Early Iron Age ways of life were lived in any one place within Sout hern Africa. This work also underlines the need to do the foundational work of interpreting processes, such as regional subsistence patterns, before something as fluid as social or ethnic identity can be addressed. The work conducted at Thabadimasego and the surrounding landscape has sparked an interest in an area which many archaeologists researching the Southern African Early Iron Age had previously considered marginal ( Reid and Segobye 2000 ; Daggett, Wood, and Dussubieux, in press ). This research, building upon the framework established by previous fieldwork in the area (as detailed in the following sections), asks how settlements in the South Sowa area may be understood when placed in the same comparative framework as other regions of Early Iron Ag e Africa. In so doing, this ongoing investigation also sheds light on the specific social and economic strategies pursued by inhabitants of the South Sowa area, instead of continuing to rely solely on models for these behaviors developed out of work in oth er regions of Southern Africa. More specifically, the data collected for this project are one of the few high - resolution spatial datase ts for Early Iron Age sites in Southern Africa with attention to the site's comprehensive set of material culture compone nts (instead of only the major traditional components of fauna, ceramics and stonewalling. As such , it stands to contribute to the ongoing scholarly discussion on the relationship between site organization and socioeconomic organization, with an additional perspective on the impact of ta phonomic processes (including erosion, deflation, and animal activity) on the spatial patterns used to make these inferences. The purpose of this research is to ga in a comparative understanding of socioeconomic process es in different locales within Southern Africa at the time of early state - level developments, and to 8 understand the influences these locales had on one another. The research begins with the assumption that Sowa Pan agropastoralists represented a frontier of a wi despread trading system, which was controlled to some extent by an incipient stratified state - level society such as Bosutswe ( Denbow 2002; Denbow 1999; Denbow et al. 2008; Reid and Segobye 2000a) or Schroda (Calabrese 2000; Calabrese 2007; Hani sch 1980) . If Sowa Pan was a frontier of this system of exchange, for example, then some key questions arise: 1) What information about tool use, floral and faunal exploitation, and settlement patterning can be discerned at sites on the s outhern margin of Sowa Pan , an area thought to be peripheral to burgeoning polities of the era, and do t hese behaviors change over time? 2) Can the presence of hunter - gatherers be established in the South Sowa area for this time period, and if so, what was the nature of their involvement in agropastoral society - and if not, why? Furthermore, w hat does a comparison of these possible interactions and socioeconomic practices with those of in cipient hierarchical societies, like Schro da say about the nature of the Southern African Iron Age, or rather, about the effects of environment, social agency, or the presence of other communities on what has long been thought to have been a tightly integrated 3) Is change over time discernable in resource utilization, trade volume, a nd spatial organization? If so, what does this say about the changing subsistence goals of the Sowa Pan agricultural and forager communities? 4) Is use of space within known walled sites representative of total activity patterns in the area? Can additional i nformation on subsistence behavior be obtained from investigation of the areas outside these walled sites? Because documentation of cultural material concentrates especially on identifying features like hearths, tool production areas, grain bins, middens, and other foci of economic activity, this research project will provide information not only about diet, storage practices, tool use, and production and use 9 of other goods for trade, but also where within sites and on the broader landscape these activities happened. These activities themselves are important indicators of daily life at South Sowa settlements , and provide a basis of comparison for these settlements with their contemporaries elsewhere. A comprehensive understanding of daily life at less - resear ched places such as South Sowa is a necessary first step to an unde rstanding of how places within Southern Africa related through trade, migration, and/ or political and cultural influence. Likewise, the spatial component of these activities is also signif icant because spatial organization has been a key component of many interpretations of Iron Age settlements, not only regarding intra - site organization and its social implications as w ith the Central Cattle Pattern (Huffman 2001) , but also regarding inter - site patterns of landscape use, especially topography and access to natural resources (see Denbow 1984; Mookodi 2001) . The settlements in this study are already known to cluster along the steep edges of the Mosu Escarpment overlooking Sowa Pan. As the remnant of an ancient lakeshore, the escarpment is a unique geological feature, and therefore site distribution in the area do es not conform geometrically to any curre nt landscape use model for the Southern African Iron Age. However, a GIS visualiz ation of material distribution and patterning across this landscape may indicate regularities. Quantitative records of spatial patterning recorded at and around the site provide a measurable indication of regularity, randomness or clustering with the excav ated material samples. This spatial analysis may perhaps also establish a baseline of comparison for similar analyses at other sites in the South Sowa area in the future. Additionally, the nature of relationships South Sowa had with other communities will be inferred from the types and distribution of items typically understood to be trade goods. Following the expectations laid out in archaeological core - periphery models like those of Hall and Chase - Dunn (1993) and Stein (1998) , trade of bulk, as opposed to luxury, goods indicates different kinds of relationships between peripheries and cores. Furthermore, it may be possible to infer the amount of political or 10 sociocultural control a core has over its periph eries through the identification of spatial differentiation in land use and layout of settlements and restriction of quantities and distribution of luxury goods at peripheral sites. In particular when placed in the context of contemporary sites within both the Mosu Escarpment and other Southern African spheres, g lass trade beads excavated at Thabadimasego offer the potential for tracing movement and economic influence (see, for example, Wood 2005; Wood, Dussubieux, and Robertshaw 2012) . Expected research contributions include an improved understanding of settlement processes in the South Sowa area for both the Later Stone Age ( LSA ) and the Early Iron Age. Later Holocene archaeology in Botswana h as been said to be (Sadr 1997) . Though research programs have been growing since then, most archaeological researc h in the country still takes place in its well - populated areas , near present - day cities, as well as Tsodilo Hills (Walker 1998; Mitchell 2002) . As such, this research project will contribute vital information about an under - resear ched area of the country with a known Iron Age settlement history. Increased archaeological coverage in areas such as South Sowa will provide increased understanding of the settlement processes of the Iron Age within Botswana, whose research programs and m odels are largely inherited from South African schools of thought, but whose distinctions in environment, social history, and other aspects demand understanding in their own right ( Segobye 2005; Mitchell 2004; Mitchell and Whitelaw 2005) . On a broader sc ale, this research project also provide s, through its high - r esoluation dataset, a basis of comparison for location - specific manifestations of Ir on Age settlement processes in Southern Africa, as well as how these locations interacted with and affected one another. This issue draws upon fundamental theoretical argum ents about relationships between material culture, economic traditions, and the maintenance of social identity. Some scholars argue that presence of certain ceramic styles, or type of wall construction, indicates an importation of an entire social, cultura l, and economic package (e.g., Hodder 1982; Kuper 1982; Huffman 1986; Huffman 2001) . Others argue that processes of gr oup or individual agency, environmental change, or other as - yet 11 unaccounted - for factors play roles in restructuring or altering these supposed cultural packages of the Iron Age (e.g., Kinahan 1996; Kinahan 2001; Kusimba 2005; Thackeray 2005; Smith 2005; Humphreys 2007) . Extensive data collection thr ough fieldwork and laboratory analysis will provide the information needed to untangle these complicated processes. In raising the issue of variations in social and economic organization of agropastoral communities, in conjunction with their involvement wi th foragin g communities, this project speak s to the tension that has long existed between this predominant structuralist interpretation of Iron Age sociopolitical organization, such as the Central Cattle Pattern (see Huffman 1986, 2001) and attempts at alt ernati ve explanations (e.g. Lane 19 95; Badenhorst 2009a; Chirikure and Pikirayi 2008) . T he research thereby stands to offer a local perspective on the broader issue of material versus ideological influences on human behavior, one of the fundamental issues within anthropology. 1.2 Environ mental background P resent day Climate and physical environment The Mosu Escarpment consists of a series of bluffs, hills and plateaus stretching east - west for about 50 km just south of Sowa Pan, in northeastern Botswana. For a map of this area, see Figures 1 and 2. The escarpment itself is part of the Thlabala Formation, a rock outcrop of Karoo basalt a nd sandstone dating to about 185, 000 BC (Thomas and Shaw 1991; Ringrose et al. 2009) and are part of the system of strandlines surrounding the entire Makgadikgadi Pans that resulted shore formation from during the P aleo - lake Makgadikgadi phase ( Ringrose et al. 2005) . The bluffs rise about 55 meters from the lowlands to their immediate north at the point where Mosu village sits. They stand at about 990 - 1000 m eters above sea level (Sowa Pan , at its lowest , is 890 m eters above sea level). The slopes tend to be quite steep (about 75 degrees, in some cases) and access to the top of the escarpment must be carefully negotiated via winding footpaths. The escarpment edge, 12 when viewed on a map , has the uneven appearance of a fjord: the ma ny crevasses and rocky protrusions forming the edge would, if stretched into a straight line, be much longer than the extent of the escarpment itself. Archaeological sites tend to occupy t he larg er, flatter protrusions . These are separated as the crow flie s by only one or two kilometers, but the distance grows much longer if one follows the escarpment edge. Implications for both past and present - day navigation of the local terrain abound: one decide whether to drive over the uneven, eroded lowland as close possible, and hope for a reasonable climbing path upwards, or to climb up in one place and cross the footpa ths may exist south of the edge. Regardless of route, an intimate knowledge of the terrain is essential for navigating through the dense vegetation both uphill and down on the plateau. From the unruly fjord - like edges overlooking Sowa Pan, the escarpment slopes generally and gradually downward to the south for several kilometers along a broad sandstone plateau. This plateau has been cited as the best in the area for both a rable and grazing land, as it is free of the salinity of the pan and has soil generated from its base rock (Field 1977; Samuel 1999; Reid and Segobye 2000a; Center for Research 2010) . The plateau a lso hosts populations of grewia ( G rewia bicolor ), mophane ( C olospermum mopane ), and morula ( S clerocarya birrea ). Grewia and morula bear fruits edible to (and valued by) humans and animals alike, while mophane trees are not only good foraging for herds in w inter months; they also are the source of phane worms ( G onimbrasia belina ), a seasonally important protein for local communities. The area referred to as the escarpment base by Samuel (1999), leading from the foot of the escarpment north to the edges of So wa Pan, is in reality more accurately referred to as a gently sloping floodplain, as it shows no more than 10 m change in elevation in the 1 - 5 km distance it spans between the escarpment and pan margins. Like the escarpment, the floodplain, where not clear ed for human settlements, roads or other infrastructure, is covered mostly in acacia and mophane savanna. Occasional stands of real fan palms (also called mokolwane locally; Hyphaene petersiana ) also dot the 13 Figure 1 Location of Sowa Pan in Botswana 14 Figure 2 The South Sowa area and its known archaeological sites 15 area; these highly visible tall trees seem to cluster where water is closer to the surface, su ch as dammed springs and reliable boreholes, and are known to grow well in saline, seasonally flooded soil (Setshogo and Venter 2003) . This species has many potential uses for human and animal populations alike. Real fan palms produce numerous spherical fruits, the nut of which is edible (and, according to at least one safari guide site, is popular with elephants: http://www.botswana - safari.net/botswana_trees.html ). The sap of the tree can also be collected and distilled into a type of alcoholic drink called muchame (although doing so is highly destructive to the tree), w hile the leaves can be used to make baskets ( Setshogo and Venter 2003 ). Substantial portions of the soil surrounding the village of Mosu (especially to the south - west and west along the escarpment, and west/ north - west along the pan edges) consist of exten sive clay deposits. These are well known to the residents of Mosu and several locations are frequently visited by individuals who collect clay for bot h domestic and commercial use. At present, there are no known compositional analyses of these clay beds. W hile, furthermore, it is not yet known whether this clay is used to make cooking vessels in present day, the clay is used by residents of Mosu Village for other purposes, including making decorative ceramics and as house paint (personal observations). With in the village and to its east, the soil tends to be much sandier. Both types of soil are susceptible to the erosion and sedimentation processes resulting from seasonal rainfall (discussed more in detail below) though the sandy soil is less stable. Gullies and dry riverbeds up to about 10 feet deep crisscross the landscape. Those which cross the main road have permanent concrete bridges constructed above them, and a few of the larger gullies that run in and around the village have been paved with cement and stones to facilitate draining during the rainy season. The soil beds themselves, in addition to human and livestock disturbance around Mosu village, are prone to collecting slopewash from the escarpment edges above during the rains (small - scale landslides may also occur where vegetation cover is thin or non existent), so deposits are much deeper below the escarpments than on top, and deposits may also be 16 much more prone to soil - movement disturbance (shifting, settling, redistribution) as well. The margins of Sowa Pan represent a somewhat distinct set of resources from that surrounding the village, floodplain and escarpment. Tree and bush savanna grows more sparsely in the increasingly saline sediment, and cease completely where elevation dips low enough tha t seasonal pan infill threatens to waterlog their roots (with the exception of baobabs, which grow frequently along the pan margins and some of the nearby hills). Instead, robust yellow - brown stands of grass span the edges of the salt flats. Like the grass es elsewhere (in the floodplain and escarpment, which compete at a losing advantage with the trees and shrubs), these deciduous grasses have long roots to reach g r oundwater deposits and while the roots go dormant in the winter, importantly, they don which make them a year - round resource for grazing herds (Silitshena and McLeod 1998) . The surface of Sowa Pan itself is a vast flat duricrust composed primarily of sodium carbonate and sodium chloride evaporite deposits formed from the annual cycle of inun dation and drying (Thomas and Shaw 1991:84). Nothing grows on the pan besides a few expanses of grass at the edges where the duricrust is shallow; beyond that the interior of the pan is featureless save for trails left by vehicles and cattle herds. Visibi lity across the pan is very high; on a clear day (which is most of them) it is possible to see the shadowy outline of Kubu Island on the northeastern horizon from various points along the escarpment at a distance of about 50 km. The pan is, of course, a na tural source of both soda ash and salt. Residents of communities surrounding Sowa Pan have a long history of harvesting and processing the salt of the pan for personal consumption and trade at least as far back as the eighteenth century AD , as documented b y Matshetshe ( 2001) . It is unconfirmed archaeologically if, or to what extent, salt harvesting, processing and trade may have played a role in Early Iron Age economies of South Sowa, b ut the potential has been noted by Reid and Segobye (2000 a ), and Wilmsen et al. (1990) cite historic sources discussing traditional San ownership and control of highly valued salt mines in Sowa Pan. It is a logical possibility that salt was a known and well - exploited res ource long before its use was documented by 17 Europeans. Fauna The South Sowa area hosts a wide variety of wildlife including both predators and historically valuable prey animals. The Makgadikgadi Framework Management Plan ( Center for Research 2010) lists species for the entire Makgadikgadi W etlands System (MWS) in a table . It is important to note, however, that species distribution varies throughout the MWS according to season, carrying capacity, and human intervention as discussed more in det ail below. Map #14, Land Use Plan (Central District Land Use Planning Unit 2000) shows South Sowa as an area of low wildlife biomass for both the wet and dry seasons according to aerial survey conducted in 1995. Notably, however, - 2006 is attributed as 42.6% of the mean total of the national population for the same time period ( Center for Research 2010, Chapter 5 ). Other wildlife found in the South Sowa area include leopards and baboons (both of which favor the high reaches of the escarpment) and lions. No reports of hyenas, jack als, wild dogs or elephants were made in the reports for human - wildlife conflict for either Mosu or Mmatshumo for the years covered by the M akgadikgadi F ramework M anagement P lan ( Center for Research 2010, Chapter 5 ). Additionally, numerous smaller species of fauna inhabit the area such as lizards, scorpions, snakes, venomous centipedes and landsnails of the genus Achatina , whose empty bleached shells can be found frequently throughout the escarpment. The shells of these snails, which only inhabit the wetter parts of Botswana (as well as many other parts of tropical sub - Sub - Saharan Africa), were used during the Early Iron Age to make beads nearly indistinguishable from ostrich eggshell beads; these beads are found at archaeological sites across Botswana and S outh Africa ; as is discussed in Chapter 4, there is a distinct possibility that Achatina specimens were exploited as a food source as has been shown to be the case in historic times 18 elsewhere in sub - Saharan Africa . Water Sowa Pan lies within the Nata catch ment, so that its water inflow comes not from the Boteti River on the west side of the Makgadikgadi, but from the Nata, Semowane, Mosetse, Lephashe and Mosope Rivers ( Central District Land Use Planning Unit 2000, Map # 10; Burrough, Breman, and Dodd 2012) . The water is saline owing to the sediment of the pan, but most human and livestock com munities living around Sowa Pan drink from boreholes which draw naturally - filtered groundwater from aquifers in the sandstone surrounding the pan (Si litshena and McLeod 1998: 43). The recharge of both the pan and the boreholes is dependent upon rainfall, which can vary by up to 35 % from y ear to year (Silitshena and McLeod 1998:38). People living in built - up villages with centralized infrastructure such as Mosu share from publicly - owned water taps, while cattle post occupants must hire a surveyor and dig their own, which even today can be an expensive and timely process. One cattle po st owner consulted by the 2012 fieldwork team reported that it took him three months to dig his we ll, which was about 10 m deep, and that h e had to use dynamite to get through the bedrock . A few natural springs exist in South Sowa area as well, some of whi ch are dammed for year - round livestock watering. Though it may not appear so at first glance, this region is relatively water - secure for Botswana; Silitshena and McLeod (1998:46) list it as lying within the of the water table: not only does the MWS overall occupy the zone with the shallowest water table within Botswana (40 m or less below surface according to Silitshena and McLeod 1998:46), but locall y, the porous sandstone of the escarpment encourages infiltration and the formation of springs and pools. Reid and Segobye (2000) i.e., south of] the rehistoric settlers such as Early Iron Age populations, although no historical or archaeological sources are cited to confirm the presence of these. Additional 19 geomorphological and paleoenvironmental studies may be necessary to confirm the truth of this. S easonality Seasonality in Botswana as a whole is marked by annual wet and dry cycles, and there are basically three major seasons: summer, or hot/wet (November - March); winter, or cool/dry (April - August); and spring, or hot/dry (September - October). Defi nition of the exact demarcation of the seasons will depend on the author, but for a thorough discussion of the factors affecting seasonality in Botswana, see Silitshena and McLeod (1998:31 - 34). In northern Botswana, the annual temperature may range between - 5 and 40 degrees centigrade (Silitshena and McLeod 1998:34). Officially, it is a semi - arid tropical climate. Sources also disagree somewhat on figure s for annual rainfall, as estimates for the entire country may range from 250 - 650 mm per annum (Silitshen a and McLeod 1998: 38) to 300 - 500 mm per annum (Republic of Botswana Map #2). Either way, the Makgadikgadi Wetlands lie within the middle of this range, making its annual rainfall somewhere around 350 - 450 mm per annum. As noted earlier, however, the actual amount per year may deviate from t he mean by as much as 35% . (Silitshena and McLeod 1998:35). Additionally, rain is prone to falling in heavy, short - lived storms, which result in flooding and evaporation; the annual potential evapo - transpiration rate for South Sowa is approximately 49 % ( Central District Land Use Planning Unit 2000, Map # 3). The great majority of this rain falls in one short season from about October through December during the hottest part of the year. This has implications not only for g rowing seasons and crop yields, but for soil and landform stability as well. The escarpment edge is prone to slope wash erosion, putting the sandy clay soils of the lowlands surrounding the pan shores at high risk for flash floods and minor landslides from the runoff. Seasonal streams may fill up the usually dry gullies and ditches which are so common across the landscape, and the usual navigation paths of both people and animals must be renegotiated. In 2012 the Department of Environmental Affairs initiate d a program to attempt to manage the water flow by cutting north - south channels though the mophane savanna from the escarpment to the pan. 20 Up on the promontories of the escarpment, one is much more directly exposed to the elements than in the low - lying are as. Strong winds gust frequently, no landforms are available to block the rain, and few trees large enough to provide shade are able to grow in the thin, rocky soil cover. The soil cover , which at its deepest was 50 centimeters on Thabadimasego , is more pr one to slopewash where vegetation is thin, either on the edges of the bluff or where constant treading has worn it thin. For these reasons, it seems highly unlikely that arable fields were kept on the promontories in prehistory just as they tend not to be at present. To sum up, the features of South Sowa resources suitable for human occupation as well as some which complicate that occupation. The escarpment and Southern plateau host natural springs, edible pl ants such as grewia berry and morula, mophane and acacia for grazing livestock, and arable land for growing crops. The floodplain contains large beds of raw clay, stands of real fan palms which offer multiple uses, and additional grazing land. Sowa Pan and its margins possess additional beds of clay, baobab groves, and of course, salt. Wild animal conflict is historically low relative to other parts of the Makgadikgadi ( as is wildlife biomass overall). As already noted by Reid and Segobye (2000 a ), Silitshen a and McLeod (1998) and Field (1977), the area offers a strong have been appealing to prehistoric communities for these reasons, a nd in the following section human and livestock interaction s with the landscape in the past will be considered. Paleoenvironmental r econstructions Work done in recent years by, e.g., Burrough et al. (2012), Ringrose et al. (2009), and Riedel et al. (2012) on the geomorphology and paleoenvironmental reconstruction for the late Quaternary of Botswana and especially of the middle Kalahari has in many cases used data from Makgadikgadi geolog ical formations. Focus within these research programs has been on clarifying wet/dry periods, lake events, and dating shorelines, which are, of course, interrelated topi cs. Unfortunately, due partly to the scale of 21 the investigations, and partly to the nat ure of the data themselves, very little info about recent (late Holocene) prehistory has been salient from this body of research. Some archaeological research has addressed issues of environment f or the last few thousand years (Huffman 2008 , for example), but are not specific to the Middle Kalahari (central and northern Botswana), so whi le these are generally useful for broad climatic trends for the late Holocene, they apply less precisely to areas such as the Makgadikgadi. Just as in the present day, where at the same time the weather in Cape Town and Harare could be dramatically differe nt, climate for various regions of Southern Africa for prehistory ought to be understood in a regional context (particularly if one is going to interpret that prehistory from a human perspective) because of the high degree of variation in biome and terrain throughout the subcontinent. This is demonstrated by the regional data cited by Thomas and Shaw (2002) . For a detailed description of extant knowledge on past environments of the Middle Ka lahari, see, for example, Thomas and Shaw (1991, 2002). Evidence which does come from the Middle Kalahari largely focuses on the Pleistocene and earlier periods, and is often problematic. Thomas and Shaw (1991) note a lack of agreement on dates for the Hol ocene in particular; what evidence there is comes via the proxies of radioactive isotopes from calcretes and shells. Interpreting the that is the Makgadikgadi geological east at time of publication] of accurate altimetric (Helgren 1984 :299 ) . Much of what is known for Holocene climatic sequences concerns the formation of year - round, permanent lakes and is based upon dated samples taken from the strandlines that now surround the pans as ridges and escarpments. Within the Holocene, there are two known lake events. W ithin the context of developing a climatic sequence for the middle Kalahari for the last 300,000 years, Burrough et al. (2009) document the last Me ga - lake Makgadikgadi event at 8 3 00 BC. Another lake (Lake Thamalakane) may have 22 formed around 500 BC . Thomas and Shaw (1991:177). One final lake event in what are now the nearly permanently - dry salt pans may have occurred around 1000 AD , as indicated by th e distribution of baobabs at Kubu Island (Riedel et. al 2012). This find correlates with other data, also cited by Riedel et al. (2012 :72 ), which suggest that a switch from wetter to drier conditions (analogous to the present ) may have occurred in these ar eas right around 1000 AD Specifically, this study report s that after 1000 AD , a - to sedge - use of fire than to drier conditions . h ese data do not apply precisely to the period with which my research is concern ed (which is more like 850 - 950 AD ), but the possibility that a drying trend began around 1000 AD would mean that the climate during the occupation of Thabadimasego was, in fact, wetter than present day. For the purposes of the archaeological sites of the Mosu Escarpment, it remains safest to rely on the faunal and botanical assemblages (for those and geographically related contemporary sites) as primary lines of evidence. This s eems to be the approach taken by, inter alia , Denbow et al. (200 8) , Badenhorst (2011), and van Zyl et al. (2013) . Opportunities may yet present themselves for geoarchaeological or paleoenvironmental approaches to the Mosu Escarpment as well. Additionally, the ongoing work by Burrough et al. (2012) , Ringrose et al. (2009) , and others represents a substantial leap for ward in our understanding of the Middle Kalahari There is good reason to expect that continued refinement and expansion of paleoenvironmental knowledge will continue in the future as this work continues. S ite description Thabadimasego is one of several Early Iron Age sites located on the Mosu Escarpment in northeastern Botswana which form a settlement complex that is only beginning to be understood (Samuel 1999; Reid and Segobye 2000a; Reid an d Segobye 2000b; Main 200 8; Denbow, Klehm, and Dussubieux, 23 n.d.) . The significance of the site is that it is one of only a small handful so far excavated in the Mosu Escarpment cluster and even across the entire Makgadikgadi Pan region. These sites are still poorly understood, especially in the context of the widespread Early Iron Age system of villages that populated the well - watered parts of the Southern African subcontinent beginning around 200 AD. This area in general has the potential to add substantial information to our understanding of Early Iron Age settlement and trade networks, gain new, or enhance existing, paleoenvironmental data, further develop understanding of how Botswana o) those of other ar eas within Southern Africa, and ask questions about how landscape and environment/ resources shape or are shaped by cultural process. Thabadimasego in particular was chosen after I surveyed the Mosu archaeological area on foot in 2008 as a member of a Nat ional Museum survey team. It is one of the area e sites with comparatively high - density surface scatters and visible remains of a stone wall. Like most sites in the South Sowa area, it sits on one of the most prominent finger - shaped bluffs o f the Mosu escarpment and overlooks Sowa Pan. In that sense it made a good comparison to the other two sites previously excavated in the nearby area, Mosu I and Kaitshàa (as described below). Based on survey records it had the characteristic assemblages of an Early Iron Age site, but had not yet received any research attention, and had a high chance of decent preservation as it was not close to any current cattle posts or the village of Mosu. For these reasons, the site offers an opportunity to investigate the way that cultural and economic processes are reproduced, modified, or negotiated in a region widely considered peripheral for the Early Iron Age. Like Thabadimasego , most sites in the area exist on small promontories along the escarpment edge and are m ost easily accessed in present day by climbing up from the base below . Only a very few archaeological sites can be approached even partway by vehicle, where local N ational Museum staff have worked to establish permanently cleared dirt roads suitable for 4x 4 trucks (and even then the rest of the 24 trail needs to be completed on foot, but usually it is a comparatively easy, if long, climb). From the main body of the escarpment, which from the south (starting roughly at the Orapa - Francistown road) is really just a gradually north - sloping plateau (as described in Samuel 1999) that ends abruptly in the escarpment edge, each promontory is most easily accessed by navigating along the slope edge, which increases the walking distance, but avoids the undulating, low - vis ibility confusion of the thickly - vegetated back area. At present it is not known just how present - day ground cover for this area compares with that of E arly Iron Age occupation or whether settlements in the area would have maintained clear paths throughout this back area. Presently only a few cattle tracks can be found wherever cattle posts are located. The site is fairly sma ll approximately 12 , 0 00 square meters, or 1.2 hectares , whereas by comparison, Kaitshaa and Mosu I, the two other excavated sites in the South Sowa area, are 12.4 and 2.4 hectares, respectively. Thabadimasego occupies the whole of the relatively flat surface of the promontory ( Figure 3 , below ) . Ground cover on the site consists of shallow - rooted grass, dense stands of acacia thornbushes and a few scattered trees. A careful eye will frequently spot a ceramic sherd or ostrich eggshell bead l ying among the grass or in a bare sandy patch where cow wallows have worn the grass away. The site is encompassed by steep slopes on nearly all sides (about 300 degrees) save at its southwest corner. Most of these slopes are too steep or too unstable for humans to climb at present. Towards the south - western portion of the site, the promontory connects to the main body of the Mosu Escarpment vi a a narrow he stone wall remains stretch across this connecting point and gradually disappear into the natural rock formations curving along the west and south slopes. A gap in the itional stone features were identified on the hilltop during the course of fieldwork - both are small ( one to two meters in diameter) stone piles or cairns, one along the southeastern corner of the site near the slope, and the other just north of the most southerly stretch of the remnant stone wall. A few dozen meters southwest of the wall and the neck, additional cultural materials w ere located in a small clearing. T his scatter appears to be an extension of the hilltop site. 25 From the promontories of the Mo su Escarpment, visibility to the north of the low - lying floodplain and Sowa Pan itself is excellent, especially right on the edge of the slopes. Ground cover tends to be quite thick except in areas of very thin soil/ near - to - surface rock protrusions, or (a s at Thabadimasego and other nearby hilltop sites ), where sub - surface cultural deposition has affected soil density. It is not known at present how fully a promontory like the one Thabadimasego occupies would have been cleared of its vegetation during acti ve settlement (for the sake of either visibility of the surrounding landscape, communication with other settlements, or just for living space on the site). However, o ne would assume that it was substantially clearer than at present. At least in present day when much of the landscape is covered in dense thornbush savanna, visibility is comparably poor down on the floodplain (being out on the pans, obviously, as the exception). It can be really tricky to navigate footpaths or even driving paths (not counting the state - built gravel and tarred road that connects to the highway) when one loses sight of the nearest escarpment portion, as the paths wind around to compensate for extremely uneven terrain (produced by seasonal flooding and riverbeds) and large trees t hat people prefer to navigate around instead of cut through when making a path. 1.3 Chapter overview The remainder of the dissertation is organized as follows. Chapter Two reviews the literature relevant for understanding Thabadimasego in its archaeologica l and theoretical contexts, situating this information at a number of scales (regional, local, etc.) that affect the interpretation of this chronology. Chapter Three describes in detail the research design as well as the methods used during fieldwork. Chap ter Four discusses methods and results of the post - excavation analyses, including glass bead typological classification and compositional analysis; shell and metal object classification; and ceramic identification. Chapter Five addresses methods and result s for spatial analysis. Chapter Six approaches the social and spatial interpretations of the results, including the implications of diet, activity, and spatial use (both on the site and on the landscape) as they compare with typical characterizations of Ea rly Iron Age 26 ways of life, and offers some thoughts on further implications for this research. 27 Figure 3 Thabadimasego site location 28 Chapter 2 Pertinent archaeological background and theoretical considerations The South Sowa archaeological record can be framed as part of a series of nested, though not neatly interpolated, spatial scales due to the way work has been conducted in local, national, and subcontinental contexts. In order to be fully understood, it has to be approached in this comprehensive perspective and each scale, with its history of research and predominant trends, considered. Each scale of research has somewhat different research concerns, methodology, terminology and chronological definitions, an d in one form or another they inform each other as is the case in many parts of the world (e.g., Bevan and Conolly 2006; Ridges 2006; Andrews, LaBelle, and Seebach 2008; Lawrence, Bradbury, and Dunford 2012) . The broadest of these is the subcontinen tal scale; the research trends discussed in Chapter One connect and theor etically engage si tes all over Southern Africa and have been the predominant structure of the research paradigm for both the Later Stone Age and the Early Iron Age. Many of these trends need to be placed in the historical context of Southern African archaeological research in order to comprehend fully their current foci and debates. Closer to home, a program of archaeological inquiry specific to Botswana has grown especially in the last four decades with the support of the National Museum and the University of Botswana. It i s helpful not only to consider what more specific trends happen in the archaeological record on a national scale, but also because a good amount of fieldwork within Botswana occurs independently of the academic research framework as it is conducted by CRM firms, and may develop its own typologies and use somewhat different methodologies, as will be discussed below. This kind of work often gets published in literature environmental impact assessments, print ed in small quantities by government or contract firms and available only in hard copy at archives like the National Museum and the National Archives. This work, while very important for documenting the material record of Botswana especially in light of ra pid d evelopment in the better - populated areas of the country, does not always make its way into broader 29 scholarly discourse, and often can only be accessed in hard copy in a few select institutions. Finally, there is the local scale of inquiry of sites on the margins of the Makgadikgadi Pans, and even more locally, those constrained to the south Sowa area itself. Sowa Pan itself can be considered as fact that Iron Age archaeological sites around the pans tend to cluster on its immediate edges, making a unique spatially discrete pattern (Fig ure 4 , below ). Archaeological sites in the S outh Sowa area, particularly those associated with the Early Iron Age, tend to occupy the escarpment edges, and do so in much higher frequency per square kilometer than surrounding areas. For example, according to the Botswana National Register, as of 2008 , 57 sites exist on or near the escarpment within three DSM grid map zones, whereas only 29 archaeological sites exist in the next nearest four DSM grid map zones, according to the Botswana National Register, as of 2008. For these reasons, the following se ctions will consider the archaeological record of the south Sowa area from multiple perspectives, each of which carry their own pertinent theoretical and methodological issues. 2.1 Subcontinental framework: Southern African Late Holocene archaeology Defini ng the area Southern Africa is often defined, for archaeological purposes, as the landmass south of the Kunene and Okavango (Cubango) Rivers to the west, and the Zambezi River to the east. For the purposes of archaeological research and heritage m anagement, this can be a problematic definition for a number of reasons. For example, the Zambezi River cuts through current political boundaries of Zam bia (leaving the majority of that nation outside of the Southern African landmass) and through the middl e of Mozambique; it also does not include Madagascar. However one chooses to define it, countries whose material records most frequently contri bute to a body of knowledge on Southern African archaeology include South Africa, Lesotho, Swaziland, Namibia, Bo tswana, Zimbabwe, and Mozambique. Defining the region for archaeology is a tricky balance between maintaining geographic authenticity for pre - modern 30 Figure 4 Archaeological sites in the South Sowa area 31 purposes, and practicality: permits are needed, political factors are at play, consistency is required by the various heritage agencies, etc. The country in which one conducts archaeological research has a strong effect upon one heritage policies and laws, institutional structure, access to sites, archived materials, and the like), and because how one scholar defines the subcontinent may vary eing included in literature on Southern African archaeology is influenced both by contemporary geopolitical factors, and scholarly preferences for definitions of the geography of the subcontinent. For example, within Botswana, Iron Age sites tend to have much higher visibility and pr eservation than Stone Age sites, and for this reason, as well as the fact that Later Iron Age populations were ancestral to current Tswana and Kalanga residents of the country, Iron Age research tends to gain higher status and visibility than Stone Age research. This is especially true for the Later Stone Age, given its ancestral connection to present - day San communities in Botswana whose status as first peoples is denied by the Botswana government. Research which carries the potential to legitimate San claims of ind igeneity and which can be tied in, if only tangentially, with contemporary self - determination efforts of San peoples, has met in the past wit h resistance by permit - seekers (R . Hitchcock 1996; Schweitzer, Biesele, and Hitchcock 2000) . In addition, the nature and intensity of all types of archaeological research in Botswana is also heavily influenced by the criteria of agencies funding the projects. In other words, some sites are more likely to contain the kinds of data capable of addressing those problems considered high priority by a funding agency (ultimately based on anonymous reviewers and the nature of proposed research topics). For example, Later Stone Age and M iddle St one Age surface scatters without associated fauna will not draw research attention and funding, while stratified sites such as caves and shelters that have good bone preservation, dating contexts, and so forth are magnets for research. Iron Age sites typic ally have good preservation and tend to be highly visible on the landscape, and as such are seen as valid opportunities for answering theoretically significant questions. Another aspect to consider for Southern African archaeology is that South Africa tional 32 research program has overwhelmingly set the tone for the subcontinent as a whole (see, for example, Deacon and Deacon 1999 ; Mitchell 2002; Mitchell and Whitelaw 2005). This has implications for interpretation in other areas - for example, major chronologies are developed from type sites (and related sequences in other places, b ut are instead ad opted wholesale at times. South Africa produces the greatest number of archaeologists working in the subcontinent; it is easily the best - funded country in Southern Africa with longest history of archaeological research, and the most extensive as well as ol dest scholarly infrastructure, so it is the best - represented in the relevant literature. Naturally, a number of reasons exist why a broad unifying framework should exist since Southern Africa is typically defined as a geographically coherent area. First of all, a unifying framework does much to provide a needed comparative and integrative approach to each region. Furthermore, physical and environmental conditions , such as the widespread presence of tsetse , may hav e presented barriers to the entry of domesticated animals into Southern Africa until much later than the rest of the continent (Gifford - Gonzales 2000; Gifford - Gonzalez 2005) . The S outhern Bantu language group includes t hose languages spoken almost exclusively in Lesotho, Swaziland, South Africa and Botswana (Holden 2002) , while the many linguistic as well as genetic studies of San peoples of Southern Africa demonstrat e their very deep history of occupation of the subcontinent (Elphick 1985; Watson et al. 1996; Behar et al. 2008; Stynder 2009; Irish et al. 2014; Morris et al. 2014) . Despite these, there still exists considerable justification to take caution when moving subc ontinent fits in with that of another part). Furthermore, the degree of biodiversity and climatic variation across the subcontinent, as well as the sheer scale of the place - several distinct biomes, from true desert to montane forest, range across an area of approximately 3.8 million square kilometers (Mitchell 2002). Complications of addressing the complex nature of Southern African archaeology with a multi - scalar approach must be acknowledged; for example, doing so may highlight issues such as 33 contradict ion between existing regional and local chronologies, the tendency of some areas to predominate in terms of both data and theory, or a lack of consistency or unification between areas, etc. Despite these problems, the growth of the discipline within Southe rn Africa will not be possible on a global scale without working to resolve our understanding of these issues. Defining the period In chronological terms, the period of concern for this research is the late Holocene, or approximately the last 2000 years. The beginning of the late Holocene is marked by the appearance of domesticates, which marks a significant shift in Southern African subsistence, ecology, and, to some extent, demography as well. The term chapter to reflect the fact that no one culture - historical term encompasses all of the cultural traditions of the last Southern Africa. The late Holocene also includes the last few millennia of the Later Stone Age , which itself began to manifest during the late Pleistocene by at least 20,000 BC ; the I ron Age itself, whose assemblages begin to appear approximately at 200 AD , and the colonial and post - colonial periods. Broadly speaking, therefore, late Holocene archaeology in Southern Africa can be character ized as the study of four separate but related trends: hunter - gatherer, herder, and farmer, and colonial archaeology since the appearance of the first domesticates on the subcontinent. Additionally, in many cases these groups appear as variations across a spectrum of mixed subsistence instead of dis cretely bounded cultural types (A. B. Smith 2001; Kusimba 2005) . Hunting, gathering and fishing i.e., the use of wild food resources - is a common part of the subsistence practices of many agropastoral peoples in Africa during the historic period as well as the prese nt. F or example, Bantu - speaking farmers both in prehistory and in recent centuries hunted both for subsistence and for animal products that could be traded. A number of examples of hunter - gatherer societies acquiring sheep and goats , and in some cases man aging herds , have been recorded in Namibia and Botswana during history (Jill Kinahan 2000; John Kinahan 2001; 34 A. B. Smith and Lee 1997a; Richard B Lee and Hitchcock 2001) . Cattle, sheep, and goats are the primary domesticated fauna which appear earliest in the African archaeological record . S heep and goats are traditionally thought to have spread into northeastern Africa from southwest Asia beginning around 4000 BC (Barker 2006) , while domestication of cattle ( Bos taurus ) in the Eastern Sahara is dated to 8000 BC (Marshall and Hildebrand 2002) . However, there is some evidence that attempts at domesticating ovo - caprids may also have occurred independently in the Central Sahara. An accumulation of dung and remains of Barbary sheep ( Ammotragus lervia ) dating to 7000 BC , along with woven basket fragments containing wild seeds, ha ve been recovered from Uan Afuda, a r ockshelter in southwestern Libya , which suggests intentional penning of the animals and (Barker 2006 : 295 ). By 4000 BC, morphologically domestic sheep, cattle and goats were present , along with evidence for wild seed ha rvesting, at several sites in Libya, Algeria, and Niger, while barley and emmer wheat were fully domesticated along the Nile during the same time frame (Ba rker 2006 : 292 - 299). While Bos taurus appears to have been domesticated f rom local wild cattle , another cattle species , Bos indicus , spread into Africa from Southwest Asia some time after its domestication around 8000 BC, introducing genetic admixture among African cattle populations (Fuller 2006) . The growth of pastoral subsistence in the Sahara has been linked to environmental transition to a drier climate during the Mid - Holocene ; a s desiccation grew in the Sahara, pastoral people s moved southwards. Domestic cattle were present in the Lake Turkana basin of East Africa by about 4,000 - 4,500 years ago (Gifford - Gonzalez 2005) . Why another 2000 years passed before pastoralism spread south of the Zam bezi River is a question still being researched, but at least one hypothesis posits that the widespread existence of tsetse - carrying fl ies in that area acted as a barrier preventing southward migration ( Gifford - G onzales 2005 ). While no domestication events occurred within Southern Africa itself as far as the extant body of 35 evidence shows, about 2,000 years ago (or possibly earlier in the instance of sheep see Henshilwood 1996; Pleurdeau et al. 2012) , a number of domesticated plants and animal species began to appear in Southern African contexts. Caprin es (sheep and goats) dominate the early domesticated record and are present in both rock shelters and open - air contexts at various locations (see below for more detail ). These early remains are associated in some instances with flaked stone tools and potte ry, which taken together has been interpreted as indicative of a mobile pastoralist way of life (A. B. Smith et al. 1991; John Kinahan 1995; Sadr 1998; Reid, Sadr, and Hanson - James 1998) . Around 200 AD, the earliest known evidence for cr op farming alongside animal herding began to manifest at sites occupying the better - watered areas along the Indian Ocean coast in Mozambique, Zimbabwe, and South Africa. These sites include Silver Leaves in the Limpopo Province of South Africa (Klapwijk 1974) , Matola in Mozambique, and Eiland and Mzonjani (Mitchell 2002: 264 - 267 ). Actual organic remains of domesticated plants are scarce in these contexts, but what has been recovered includes pearl millet, finger millet, cowpeas, and sorghum (Jonsson 1998; Barker 2006) in addition to grindstones with long, narrow gro oves perfect for these kinds of grains (T. M. Maggs 1984; Huffman 2007) . Importantly, these domesticated species were in many cases incorporated into a subsistence economy that also involved the exploitation of wild resources such as fish and shellfish (Kiyaga - Mulindwa 1993) , wild game, and wild plants (Jonsson 1998). This should be of no surprise given that comparative contexts worldwide of early/ transitional food - producing economies often u sed domesticates side - by - side with wild species, in some case for thousands of years (Barker 2006 ). There is no simple way to characterize early food - producing trends across the whole of Southern Africa, as the communities they represent co - occurred, inter mingled and at times saw conflict in various contexts. In the drier western two - thirds of Southern Africa (Namibia, the Northern and Western Capes, Namaqualand and the Karoo), where for the most part arable agriculture remains in the present day impossible without borehole irrigation, sheep - and goat - herding communities began to appear in the last centuries BC . These earliest instances of direct evidence for domesticated caprines in Southern African 36 contexts appear not neatly clustered in any one geographi c region, but dispersed across the drier third of the subcontinent; caprine remains have been dated to about 300 BC at Leopard Cave, Namibia (Pleurdeau et al. 2012), 0 AD at Blombos Cave on the Southern Cape of South Africa (Henshilwood 1996) , 60 BC at Spoegrivier on the Western Cape (Vogel et al. 1997) and 100 BC at Toteng I in northwestern Botswana (Robbins et al. 2005) . Page 8 of Pleurdeau et al. ( 2012 ) displays a very useful map of early caprine finds in Southern Africa, which distinguishes between directly - and indirectly - dated finds. In addition to the difficulty narrowing down an earliest point of entry for livestock into Southern Africa, scholars still debate the routes and modes of livestock acquisition for a number of other reasons. One school of thought (A. B. Smith 1990; A . B. Smith 1998; A. B. Smith 2006; A. B. Smith et al. 1991) claims that the introduction of livestock must necessarily have been the product of demic diffusion because of fundamental differences in the social and economic organization of herders and hun ter - gatherers. Smith (1990), for example, argues that hunter - gatherers prevented them from acquiring perso nal property such as livestock, and if they were to do so , it would be on a marginal basis as clients of t he stock - owning herders (Smith 1998). Further work by Smith et al. (1991) and A. Smith et al. (2001) supports this claim in the archaeological record by drawing a distinction between the ceramic and lithic assemblages of pastoral peoples and those of contemporary, spatially related hunter - gatherer communities, even when hunter - gatherer assemblages may contain domesticated faunal remains. A number of interpretive issues arise with thi s perspective, however. Mitchell and Whitelaw (2005), for example, note these scholars historic herding populations (such as private ownership of goods and resources) onto prehistoric herding sites, even where the archaeological data for these prehistoric herding communities, such as their ceramic facies, do not accord well with historic Khoe traditions. Sadr (2003) question s whether hunter - gatherers would have been kept apart from the bodies of knowledge and production that accom pany the herding 37 lifeway for the entire duration of this period of prehistory. Sadr (2008) goes as far as suggesting that archaeologists stop classifying first millennium AD ceramics by their culture - instead focus on their f unctional and behavioral contexts. Just as researchers are working out the full implications of various ceramic, lithic and domesticate assemblage signatures in the west, similarly complicated questions occur for the Kalahari and the better - watered easter n portion of Southern Africa. While, for example, in the eastern third of South Africa, farming settlements rapidly populated riverine and coastal landscapes beginning in the first centuries of the first millennium AD, work by (e.g.) Jolly (1996) and Mitchell et al. (2008) in the highlands of Lesotho and van Doornum (2005) in the Shashe - Limpopo confluence demonstrate the persistent pr esence of hunter - gatherer communities well into a - - dependent cultural horizon. Importantly, the work of these authors further highlights the necessity of understanding the nature of hunter - gatherer presence within a specific Iron Age temporal and geographic context; contra the arguments of the of potential interactions that dissimilar socioeconomic groups may take on. Within agricult ural settlements, particularly for the first millennium AD, the role of wild game as a contributor to subsistence has also been noted, adding to the complexity of the picture of life at the time (see, for example, Denbow et al. 2008, Badenhorst 2009, etc.) . The south - eastern margins of the Kalahari (in what is now Botswana) were populated with agricultural villages by the middle of the first millennium AD, while a few seemingly isolated agro - pastoral settlements appeared elsewhere in places such as the Tsod ilo Hills, the Okavango Delta and Sowa Pan, where water is more readily availabl e than elsewhere in the desert (Robbins et al. 1998 ; Reid and Segobye 2000 a ; van Zyl et al. 2013). As famously documented in the many ethnographies of the Harvard Kalahari group and subsequent scholarly endeavors ( such as Lee and deVore 1968; Howell 2000; Tanaka 1980; Tobias and Biesele 1978; Biesele 1993) , hunter - gatherers collectivel y known as San peoples continued to practice a hunting and foraging lifestyle until the twentieth century in several (often 38 remote) parts of the Kalahari. Situated periodically amongst these communities, however, are also herding communities whose presence in the Kalahari archaeological record, scarce, ill - defined and infrequent though it may be, stretches back to at least 1 0 0 BC at Toteng (Robbins et al. 2005 ) . The occasional find of pierced - lug ware (Sadr and Sampson 2006) or Bambata ware (Huffman 2005) , both of which are potentially associated with pastoral peoples, at sites across the northern Kalahari likewise indicates material and behavioral traditions whose nature has yet to be fully determin ed. That multiple modes of subsistence and production, technological traditions, as well as the social and organizational group dynamics which contained them, overlapped for what appears to be centuries within well - defined spaces implies a set of complex behavioral dynamics that scholars are only beginning fully to understand . Overall, although it encompasses numerous strands of research, late Holocene archaeology in Southern Africa can be said to be the study of communities since the introduction and adop tion of food production strategies, as well as the study of how those strategies affected the dynamics of communities who both practiced food production and/ or hunting - gathering . This includes hunters and herders, hunters who herded, farmers who hunted, a nd any other variation on this theme. As such, research on this time period necessarily incorporates a wide array of literature and theory as well as material culture traditions. Most scholars naturally focus their work on one horizon within the late Holoc ene (such as the Late Iron Age); even so, such a horizon may encompass the use of flaked microliths, hand - forged iron and copper jewelry and tools, as well as European milled goods. Therefore, it is of utmost importance to be both precise and thorough in e valuating material and its context, while at the same time placing it in a broader comparative framework. The Later Stone Age - in broad strokes The Later Stone Age ( LSA; ca. 40,000 BC 0 AD ) in Southern Africa is characterized primarily by a diversification in the array of tools at the disposal of hunter - gatherers, as well as the environments that they exploited. LSA communities are frequently linked as directly ancestral to present - day Khoisan 39 populatio ns, especially those living in northern Namibia and Botswana (see, for example , Lee 1979; Mazel 1989; A. B. Smith and Lee 1997b; Deacon and Deacon 1999; Mitchell 2004; Mitchell 2005) . Speaking very broadly, an LSA assemblage would be likely to include some combination of the following: flaked microlithic tools (typically less than 25 m illi m eters long); bored grooved stones (for polishing beads and straightening shafts); beads and pendants of ostrich eggshell, shell and bone; engraved or decorated shell and bone items; tortoiseshell bowls; polished bone tools such as eye d needles, a wls, linkshafts and arrowheads (J. Deacon 1984; Walker 1998 ; Deacon and Deacon 1999; Mitchell 2002; Mitchell 2005) . These items reflect the increasingly complex and diversified subsistence practices as well as social traditions developed throughout the Later Stone Age as compared to previous eras throughout Southern Africa. In some cases, evidence is found for marine exploitation (shell middens) or fishing (hooks, gorges and sinkers), whereas in conditions of exceptional preservation, organic materials such as plant remains, string, leather and wood; bows and arrows may even be recovered (Deacon 1984; Mitchell 2004). Additionally, within the Later Stone Age, evidence for symbolic behavior, including painted and engraved rock art, and deliberate formal burials, becomes increasingly more frequent as well (S. Hall and B inneman 1987; J. Deacon 1984; Lewis - Williams 2002) . Within the late Holocene (i.e. the last 2000 or so years), ceramic vessels also become a common component in some LSA assemblages. The implications of this - whether hunter - gatherers acquired pottery t hrough trade or, in some cases, developed a pottery industry of their own - has been the subject of study numerous times (see, for example, Sadr and Sampson (2006) for a discussion of the morphological and functional distinctions between - Southern Africa). Importantly, the prevalence of any of these material components at any given site depends not only on the actual usage and depositional history of that site but of the conditions of prese rvation as well. The rock - shelter - dotted landscape of the Western and Southern Capes, for example, lends itself well to the kind of deep, temperature - stable occupation 40 sequences that are likely to yield a much wider array of material culture. Later Stone A ge sites excavated under other conditions in Southern Africa must necessarily be interpreted with this in mind. A number of subdivisions or horizons within the Later Stone Age - at least as it was expressed in South Africa in particular - have been assigned based on shifts within assemblages in lithic and organic technology, lithic raw material types, and dietary evidence. Flaked - stone typologies for Botswana or co - occurrence with domesticate species in the late Holocene, are less well - defined. A complete discussion of each horizon and its predominant trends would be exhaustive and, indeed, has been done elsewhere (see, e.g., Deacon 1984; Walker 1998; Deacon and Deacon 1999; Mitchell 2004; Mitchell 200 5). This review, on the other hand, is primarily concerned with understanding Later Stone Age material in the context of co - occurrence with food producing societies in the late Holocene. This period of hunter - gatherer prehistory (that is, their co - occurre nce with food producers) is typically framed as - gathering groups are nearly always framed in relation to the herding, farming or European colonist communities with whom they inter acted (Smith and Lee 1997 being one exception). The archaeology of inter - group interaction in Southern Africa is concerned with resolving, via archaeological evidence from locations across the subcontinent, the contention over if, and how hunter - gatherers of the Kalahari had experienced prolonged contact with food - anthropologists in the 19th and 20th centuries and what, in fact, the social and cultural implications of this prehistoric contact are. The bo dy of literature stemming from this theoretical dispute has been known Denbow and Wilmsen 1986; Wilmsen and Denbow 1990; Solway and Lee 1990; R B Le e and Guenther 1993; Spielmann and Eder 1994; Sadr 1997; Wilmsen 2003) . As it stands today, the question is no longer whether hunter - gatherers experienced prolonged contact with food producers, but rather, what forms that contact took, what were the bro ader economic, 41 social, and cultural implications of these things for any given community, what is the archaeological visibility of these processes , as well as where such cont act did or did not occurred . As a way of addressing these questions, individual st udies have looked at a numerous array of material and temporal trends for different locales within Southern Africa. For example, some have focused on documenting local chronologies and defining the specific material sequences seen in those areas (e.g., Humphreys 1988; Thorp 2000 ; van Doornum 2005; Mitchell et al. 2008), while others ask whether different food producers versus non - food - producers may be observed in a given time or place (A. B. Smith et al. 1991; Sadr 1997; Tapela 2001; Sadr et al. 2003) . Another important focus has been on the persistence of material or social processes of hunter - gatherers in the presence of growing pressure from outside groups (Bollong and Sampson 1999; Sadr 2002; Sadr 2005) and asking what changes in material culture patterns in the face of long - term contact may indicate for hunter - gatherer group identity and social organization (Phaladi 1991; Wadley 1992; Wadley 1996) . Furthermore, many scholars recognize the need to within late Holocene archaeology, such as categorizing (199 5), Denbow (1999) , Wilmsen (2003), and Reid (2005) , for example, discuss the fallacy of using ethno - linguistic communal identities generated within the colonial era (the simple trichotomy of San, Khoe, and Bantu), to gether with the historically documented linguistic, gendered, political, and ritual traditions drawn from ethnographies of these, to demarcate among what are, they argue, less socioeconomically bounded and more culturally nuanced community dynamics over ti me. Although the praxis of applying these perspectives to the archaeological record remains a challenge, the need to do so is clear. This is an ongoing discussion within the academic community working in Southern Africa and it continues to inspire, and dra w inspiration from, archaeological as well as historical and contemporary social studies (Mitchell et al. 2008; Musonda 2013 ; Forssman 2013) . For a comprehensive critical review of this line of research, the types of evidence it assesses, and the theoretical pe rspectives it encom passes, see Sa dr (1997); Brooks 42 (2002); Mitch ell (2005); and Sylvain (2014) . Later Stone Age in Botswana - Regional trends Later Stone Age assemblages in Botswana contain more or less similar components to those found elsewhere in Southern Africa, although Walker (1994, 1995, 1998) identifies three regional trends within Botswana for the the last 4,000 years, which comprise the clearest and best - represented material record for Botswana at this time. Tool types and attributes vary for the western, north - a nd corresponding LSA communities across the borders in South Africa, Zambia and Zimbabwe, respectively (1998:76). This speaks to the regionalized and interactive nature of socioeconomic traditions even prior to the arrival of food producers in Southern Afr ica - as well as the need to use care when developing or using typologies from region to region. Later Stone Age sites during the late Holocene in the south Sowa area would belong, in all likelihood, to the eastern/ Zimbabwe - related lithic tradition, if t he regionalized generalization may be proved to hold true in the subsequent years since Walker (1994, 1995, 1998) publications. It must be acknowledged, however, that of Zimbabwe and therefore the assemblages may cover a greater timespan and represent adaptations to differing environmental conditions than those observed in eastern Botswana . While LSA sites have been observed in relatively high frequency along the western reaches of the Makgadikgadi Pans and the Boteti River (Helgren 1984; Masundire et al. 1998) , very few stand - alone LSA sites have been observed for the south Sowa area; those that have been documented are surface scatters (see, e.g., Main 2008) . None have been excavated to date. Whether or not these sites represent pre - food - producer occupations or not remains to be clarified, and the technological as well as behavioral implications of the lithic assemblages will need to be explored once this h as happened. Rockshelters (and, therefore, the ir associated deep, well - preserved deposits) are uncommon in 43 Botswana and occur only in the few places such as the Tsodilo Hills where bedrock rises above the sandveld as towering hills. Most Later Stone Age sites, unlike in South Africa, exist as open - air surface scatters which are often disturbed (Walker 1998; Brooks 2002). Walker (1998 ) makes note, however, that even such sites offer opportunities for insights into social and economic adaptations in arid and wetland environments as made manifest in evidence for exchange, style change, and territoriality. This is especially relevant for studies of the late Holocene, when (according to Walker 1998:75) Later Stone Age site frequency increased markedly in Botswana (with the introduction of pottery and the availability of domesticated animal products ) , although the precise reasons for this ar e unclear. The antiquity of livestock herding at sites like Toteng is undisputed (Robbins et al. 2005), and the presence of peoples such as Khoe and Bakgalakgadi, characterized as h istorically traditional herders in histories and ethnographies of the 19th and 20th centuries (e.g., Hitchcock 1978; Cashdan 1987) , has well - documented historical depth. Even so, for the early settlement history of pastoral traditions in Botswana, particularly for the first millennium AD, and the nature of their technological and social processes over the course of the late Holocene, much remains to be understood. Some work has been done towards this end in the interest of parsing out differences between material culture traditions for specific locales within Botswana, with a general emphasis on understanding the changing nature of h unter - gatherer society in Sadr and Sampson 2006). Still, historical examples of hunter - herder (or hunter - farmer, etc.) interaction are understood no t from the perspective of a single locus of interaction, but by characterizing regional patterns of land use, resource access, and social dynamics from a number of documented group interactions over a specified period of time. Therefore, it stands to reaso n that archaeological frameworks for understanding similar forms of interaction will best be approached not by examining material change or resource use at a single site, even over a prolonged time period, but at multiple sites across a landscape in a comp arative context, 44 using multiple lines of evidence. Bollong and Sampson (1999) and Sadr et al. (2003) provide useful examples of this approach. The chronology of early food production by arable farmers and their settlement systems is no less complicated. Wh ile much work has been done to characterize the major attributes of early food - producing settlements in Southern Africa as well as establish a timeline for their introduction to the subcontinent, questions remain about at what scales those attributes accur ately represent a constant and consistent whole environmental contexts. Likewise, the continued presence of multiple modes of subsistence such as hunting and gathering throughout the Iron Age, as discussed in this section, implies another level of social complexity that has yet to be fully incorporated into a comprehensive model for the time period. The following section will address in brief some of the ways that schola rs have addressed the behavioral implications of mixed subsistence in a Early migrations/ settlements - in broad strokes The Early Iron Age (c.a. 200 - 1000 AD) is typically characterized in Southern Africa by the appear ance of new technological complexes, distinct settlement systems, and food production. It is broadly understood as product of demic diffusion from East and Central Africa. The major general characteristics of an Early Iron Age settlement, as described by, e.g., Mitchell (2002 : 259 ) , Phillipson (2005) , or Huffman (2007 :331 - 340 ), include the presence of domesticated plants and animals (primarily sheep, goats, cattle, sorghum, millet, and cowpeas), regional pottery traditions, metal tools and metal production, sedentary villages with permanent structure s such as round huts and grain bins, and distinct patterns of landscape exploitation. These will be discussed in greater detail below. This discussion of the Early Iron Age is largely concerned with contextualizing Thabadimasego (the primary focus of this research) and the rest of the South Sowa archaeological landscape. For a comprehensive review of the archaeology of the Early Iron Age, see e.g. Deacon and Deacon (1999); Mitchell (2002); Mitchell and 45 Whitelaw (2005); Stahl (2005) ; and Huffman (2007). Different regions within Southern Africa (e.g., the Eastern Cape, Kwa - Zulu Natal, Gauteng/ Northern Pro vinces, the Shashe - Limpopo Confluence, Zimbabwe and Southern Mozambique) have their own regional concerns. Here, I am concerned with what is most germane to situating South Sowa and its relevant issues. The term itself is not without probl em and is not uncontested (Mitchell and Whitelaw 2005, in their review of recent Early Iron Age research , refer to early food - producing Age - archaeology in Southern Africa). For example, the forging and use of iron tools, while thought to have been culturally important (Calabrese 2000; Mill er 2002; Chirikure 2007) , was not necessarily an economic mainstay of communities at the time. Iron ore deposits, while not uncommon in Southern Africa, varied highly in terms of quality and distribution (Mitchell 2002:279) and access to or exploitation of any given source of iron ore would likewise have varied. Copper metallurgy is also known from early sites such as Broederstroom ( Denbow 1999); while copper ore deposits , primarily found in Zambia, are far less widespread than iron in Southern Africa ( t hus making copper ore an object of exchange over distances up to 200 k ilometers - see, e . g . , Herbert 1984; Denbow 1990; Denbow 1999; Killick 2009) . Evidence of iron and copper smelting an d smithin g (in the forms of tuy è res, forge bases, slag heaps, etc.) is likewise far less common than the final products themselves. Iron a s well as copper items of jewel ry comprise the majority of metal assemblages from such notable Early Iron Age sites as Divuyu, Nqoma, Broederstroom and Kwagandaganda (Denbow and Miller 2007) , while as Mitchell (2002:276) observes, actual agricultural implements such as hoes, spades, sickles or adzes are rare finds. Chisels, awls, and points are better represented, but metal - working by - products such as slag and bloom fragments are more typical finds (Miller and van der Merwe 1994; Miller 1995; Denbow 1999; Mitchell 2002). Whether the low proportions of utilitarian iron items, and particularly those typically associated with agricultural tasks, can be attributed to a reduction in observable freq uency due to recycling of old tools as raw material for smithing 46 (as described in Miller 2002), taphonomic processes, or is representative of actual frequencies, is a question still under discussion. During the second millennium AD, gold objects (including beads and gold - foil - covered sculptures) became increasingly prevalent as well, particularly in the settlements of the Shashe - Limpopo Basin (Killick 2009 ). In fact, the beginning of exploitation of the gold deposits in south Africa during the 9 th century A D has been explicitly linked to formation of trading relationships with Islamic societies via the Indian Ocean network (Killick 2009), while the control of gold objects within Southern Africa has been cited as a factor in the development of centralized hie rarchical polities beginning in the 11 th century AD (Woodborne, Pienaar, and Tiley - Nel 2009) . The social role of metal - working and metal items for first - millennium settlements is, likewise, worth exploring. Calabrese (2000), Denbow et al. (2008), and Huffman (2000; 2008) , for example, suggest that metal - working may have been associated with ritual power and elite status for early second millennium sites. They predicate this idea b ased on the central placement of many smithing , smelting, and forging areas within settlements, coupled with the higher presence of metal decorative items in elite burials (also centrally placed within settlements). A number of studies have likewise docum ented the ritual significance of iron working for historic Bantu - speaking communities in many parts of sub - Saharan Africa (Chirikure 2007) . While such patterns are far better documented for the early second millennium than for the first, it is worth asking what corollaries explain the frequencies observed in first millennium metal assemblages. Compared to the second millennium especially (Mil ler 2002), metal objects are, on a whole, a very small portion of Early Iron Age assemblages (and in many cases, are not prese nt at sites), even at Nqoma which (according to Miller) has yielded the largest collection of iron artifacts in Southern Africa. F or the reasons stated above, maintaining a focus on this one aspect of the time period encourages not only a slanted, single - component - oriented perspective, but a strongly cultural historical one as well. In the view of the author, an ideal terminology wou ld approach this period of Southern African 47 prehistory with the goal of characterizing its archaeology from a behavioral standpoint instead of an object - oriented one, in order to emphasize the entire set of nuanced processes which shaped this time period. Even so, despite these issues, the fact remains that the term used and unde of arable agricultural settlements in the eastern third of Southern Africa in the first millennium. Apart from the shared material culture attributes, the concept of the Early Iron Age also generally describes the time period before centralized political and economic control took shape in early states such as Mapungubwe (Meyer 2000 ; Huffman 2008) and Toutswe (Denbow 1986; Segobye 1998) . For these acknowledging all the same that it is a problematic one. Although (as discussed earlier in the chapter) it is a problematic interpretation, t he majority of scholars agree that that characteristic components of the Early Iron Age were brought in as a behaviors and technologies with the migration of Bantu - language speakers as modeled (linguistically) by, e.g., Ehret (1982) and Holden (2002) . - sp eaking peoples is currently thought to be in Central Africa in or near present - day Cameroon (Vansi na 1995) . T he ways in which the archaeological record may support or contradict the linguistic model continue to be poo rly understood for this region , thanks to political and economic instability in the area for the last few decades , as well as poor pre servation conditions (Eggert 2005) . B ased on reconstruction of a hypothetical proto - Bantu language, the speakers of this early language are thought to have emerged at least 3000 years ago (Phillipson 2005) . The nature and means by which subsequent migrations occurred from this homeland is still very much debated , altho ugh one prevailing theory is that early emigrants followed the courses of rivers through the equatorial forest (Barker 2006). That Bantu speakers represent a coherent group within Southern Africa appears to be supported by evidence of genetic influx into t he subcontinent around 2000 years ago (Behar et al. 2008; Pickrell et al. 2014) . However, as discussed in detail by, e.g., Phillipson ( 2005 :188 - 192) and Huffman (200 7) streams of ceramic facies across central 48 (Angola), south - central (Zambia), and East Africa (Kenya, Tanzania) and their convergence within Southern Africa suggest that at least three independent population movements occurr ed over the first centuries of the first millennium AD as Called, respectively, the Kalundu, Nope and Kwale branches of the eastern Bantu , the western, central and eastern streams were initially distin guished by Phillipson ( 2005 ) not only by their relatively discrete geographic distribution and distinct ceramic styles, but also by other attributes fundamentally connected to those particular landscapes, such as the exploitation of marine protein instead of livestock in many Kwale settlements. Phillipson streams has never fully been explored in terms of its ability to explain variability in economic or social processes in the Southern African Early Iron Age. In recen t years the same terminology has been re - employed by Huffman (2001, 2005, 2007) to differentiate specifically between the ceramic sequences within each of the three geographic regions. This has had a considerable effect on the way in which the Chifumbaze C omplex and its many component horizons are interpreted as a whole within Southern Africa. The overwhelming emphasis on pottery typology as a way to trace movement and temporal change, while it has established an important chronological framework, has also lent this one line of inquiry greater It is also worth noting that Denbow (1990, 2014) conducted a series of excavations in the Loango coast of present - day Democrat ic Republic of the Congo, adding substantially to the poorly - understood archaeological record of Central Africa, as well as suggesting a potential cultural affiliation between settlements in West - Central Africa and northwestern Botswana. His findings include the abrupt appearance at a few sites of Spaced Curvilinear ware pottery, whose decorative motifs and layouts bear a number of resem blances to those recovered at Early Iron Age settlements in northwestern Botswana as th and 8 th centuries AD, while in Botswana they appear slightly earlier, during the 6 th to 8 th centuries AD at Divuyu in the 49 Tsodilo Hills (Denbow 2014). On the whole, p ottery facies for Early Iron Age Southern Africa are generally well - provenienced thanks to comprehensive work by Huffman (e.g., 2000, 2005, 2007), who drew on decades of primary research at dozens of sites across the subcontinent (including his own) to create a consistent, replicable framework for describing and comparing vessel attributes. As described in these references, Huffman typology for each unique facies compri ses a combination of decorative elements, layout, and vessel generated by tracing the transitions in these attributes over time and place from one fa cies to ano ther in cladistic relationships. Additionally, several of the sites he draws on have been independently dated. As already stated, his work provides a highly valuable framework, especially when conducting research in a relatively unknown area suc h as South Sowa. Often, however, any given facies is also attributed other non - ceramic characteristics and becomes a synecdoche for the entire lifeway of a community. This is especially true for the linguistic groups proposed to have been the makers and us ers of the various ceramic types - the logic goes that a community, speaking a particular language, makes a certain kind of pot (or pots) representative of its group identity, and as linguistic groups/ group identities diversified and branched over time, s o did their material representations (Mitchell and Whitelaw 2005). The constant, unchanging factor across the centuries was the equation of one style of pottery with one group identity and of that group identity with any settlement where that type of potte ry was used (or at least observed archaeologically). This over - simplification of both the kinds of information that may be derived from pottery in an archaeological context, and the potential behavioral inferences that can be drawn from that information ha ve already been challenged numerous times (see, e.g., Mitchell 2002:262 - 264; Pikirayi 2007) . Alternative means of characterizing ceramics have more recently been explored and many of these hold great potential. For example, at University of Pretoria, Ceri Ashley is developing a typology of paste types (Ashley pers. comm. 2013) while Wilmsen et al. (2009) have begun to parse out the relationship 50 between temper sourcing and sherd provenien ce through optical petrography, and a small handful of lipid residue analyses have been conducted for both the Shashe - Limpopo Basin and the Western Cape (Patrick, Koning, and Smith 1985; Copley et al. 2004) . Even so, the morphology - based typology, as it is curren tly understood and employed, continues to form an important part of the Southern African E arly I ron A ge epistemological framework because ceramic facies combined with settlement organization patterns form the archaeological basis for the Central Cattle Pat tern (CCP) model (see, e.g., Huffman 2001). As briefly discussed in the introduction, this model views cattle as the heart of Early Iron Age social, spatial, cultural and economic processes. The original Central Cattle Pattern was derived from observation s of twentieth - century Sotho - Tswana and Nguni settlement patterns by Kuper (1982), though Huffman maintains the claim that this socioeconomic - spatial way of organizing Eastern Bantu - speaking communities stretches as far back as the Early Iron Age. This cla im is predicated on what Huffman sees as the consistent and observable physical manifestation of worldview and economic organization through settlement layout dating as far back as the 5th century in sites such as Broederstroom (Huffman 1990) . Residences organized around a central cattle byre (colloquially known as a kraal), and the presence of smithing facilities, grain bin storage, and elite bur ials all within the central byre form key spatial and structural elements of this model. Social and symbolic elements include a division of space into public/ male (the byre) and private/ female (residential) (Huffman 2001). Since, the logic goes, the sett lement layouts are (presumably) the same or similar to those observed in the twentieth century, so too were the organizing principles behind them. While archaeological sites with similar structural attributes to Kuper term) do, indeed, exist that date as far back as the early horizons of the Early Iron Age, so, too, do numerous sites whose attributes do not fit this model well. Mitchell and Whitelaw (2005:223 - 224) and Badenhorst (2009) discuss several examples of Early Iron Age sites where, for example, metal - working occurred outside of the central byre, women were buried in 51 the central byre, or the cattle byre was not even located in the center of the settlement. Questions remain as to how best to understand these spat ial layouts - as outliers, variations on a theme, or alternatives to a modal expression - as well as what social and ritual significance should be inferred from them. As with the ceramic assemblages, additional lines of inquiry such as faunal analysis (Bad enhorst 2010, 2011; van Zyl et al. 2013) and spatial analysis (Sadr and Rodier 2012) are being pursued in an attempt to understand and explain this variability. While the identification of other components of Early Iron Age sites is hardly a new trend (see , e.g., Denbow 1979; Plug and Voigt 1985; Morris 1992; Plu g and Badenhorst 2001; Steyn and Mosothwane 2004) , the significance of these recent studies lies in that they directly address the concerns raised by critics of the Central Cattle Pattern. The validity of the correlation itself between highly standardiz e d re presentations of recent behavioral/ symbolic patterns, and archaeologically observable material patterns, has also met wi th questioning (e.g., Lane 199 5, 1998, 2005) , especially given that a temporal difference of up to 1500 years lies between the two. Whether a structuralist approach is even the best means by which to understand Early Iron Age processes of economy and political formation is often debated (see, e.g., Mitchell 2002:283 - 284), although the predominance of the Central Cattle Pattern as an explanatory model can undoubtedly be credited with inspiring a substantial quantity of research in the decades since its inception. Huffman (2001, 2004, 2012) has responded to critiques of his model by arguing, for example, that with the detailed ethno - historic record for Bantu - speakers reaching as far back as the 16th century (in a few cases), the direct historical approach can be applied to the archaeological record. Since the occupants of Later Iron Age (ca. 14th - 18th. centuries AD) sites are, he argues, undoubtedly the ancestors to today - speaking populations in Southern Africa, archaeologists can further trace back their lineage through material and structural analogies, including the linkages provided by the evolution of Chifumbaze ceramic facies as well as residential structural patterns. The heart of his argument is that the same physical forms must have the same social and symbolic significance th roughout time. Still, many scholars continue to raise objections to this line of 52 reasoning on an epistemological or theoretical basis (e.g., Greenfield and Miller 2004 ; B adenhorst 2009). While Mitchell (2002) has noted that one probl em with the CCP is that no one has so far managed to formulate a reasonable, widely - applicable alternative to it, calls have been made recently for archaeologically - derived, testable hypotheses with regional foci to explain the distribution of material observed in the Early Iron Age (e.g., Isaacs 2013; Jordaan 2013) . Numerous other scholars focus their work on documenting and clarifying the data regarding particular behavioral aspects or material components of Early Iron Age sites, such as faunal assemblages and their economic significance (Plug and Voigt 1985; Plug and B adenhorst 2001; Badenhorst 2009 , 2011, 2012; van Zyl et al. 2013); microtemporal patterns in site stratigraphy and its taphonomic implications (Greenfield and Miller 2004; F owler and Greenfield 2009); and experimental studies on materials like vitrified dung , which results from accumulated masses of cow dung reaching a critical internal temperature due to intentional burning and transforming into a bubbly, glass - like material (Peter 2001) . Additionally, recent paleoenvironmental st udies of dry - wet cycles during the late Holocene provide an evidence - based framework for making sense of migrations and settlement patterns (J. Smith, Lee - Thorp, and Hall 2007; Russell and Steele 2009) , although their authors are quick to clarify that this is not meant to be a holistic e xplanation for Early Iron Age socioeoconomic phenomena. These studies support the long - held notion that Early Iron Age communities exploited particular landscapes during cycles of optimal rainfall patterns for tropical cereals like sorghum and millet, and that shifts towards greater aridity over the centuries could have influenced migration patterns as well as, in some cases, political strategies to consolidate power via non - agricultural means (an argument also made by Huffman 2008). These lines of inquiry should only serve to bolster our understanding of Early Iron Age socioeconomic processes via their contribution of bodies of evidence by which to evaluate models like the Central Cattle Pattern. However, because the Early Iron Age for Southern Africa cover s nearly a millennium of time, numerous ecosystems and roughly 1,600,000 square kilometers, it is still difficult to put these bodies of data into perspective 53 without some broad - spanning explanatory framework, which is one reason why the CCP remains succes sfully persistent. Still, the CCP lacks robusticity in that it does not encourage comparison between contemporary bodies of data at multiple scales, or of understanding of process over time. As such, one potential alternative which could be fruitful is a r egional focus coupled with a systems perspective (after, e.g., Stein 1998; 1999). This framework would be one way to take a more dynamic look at the same bodies of information. One such regional focus which has long since been established is Denbow 982, 1984, 1986) Toutswe tradition, a three - tiered regional settlement model for eastern Botswana. This model applies to a number of sites occupied between 700 - 1200 AD, situated mainly on hilltops and hardveld outside the Kalahari sands and near seasonal r ivers, including the Tati, the Motloutse and the Shashe. T his general area is located about 300 kilometers northwest of the Shashe - Limpopo Basin region often referred to in Southern African literature as a major locus of Iron Age settlement (e.g., Pikirayi 2001; Calabrese 2007; Huffman 2008b; Figure 5 , belo w ). Using photos taken in aerial surveys over the eastern hardveld of Botswana, Denbow (1979) demonstrated that in this region C e nchrus ciliaris , or buffelgrass, tends to cluster on hilltops formerly used as settlement locations. This grass species grows on vitrified dung, and is an indication that cattle were being kept on these hilltop sites. Using this and the extant archaeologic al record of the area as starting points, Denbow developed the Toutswe model as a cattle - based model of centralized political economy which took advantage of specific features of the landscape, particularly water sources and arable land. He posits a centra lized dynamic of power and economic influence between settlements in the area with three regional capitals, and numerous secondary and tertiary - level sites. D enbow (1984) elaborates on the structuring the exchange of goods and resources, particularly cattl e herds, between sites as fundamental to development of local political centralization. At the time of his earlier writings, Shashe - Limpopo settlements were less well - researched than they are today (Denbow 1984) and therefore a broader supra - regional conte xualization of these processes in terms of their 54 contribution to overall Southern African political and economic trends is still not entirely fleshed out (current work, such as Denbow et al. (2008), tends to focus on second - millennium phases of occupation) . However, Denbow makes the point consistently via his eastern Kalahari data that complex socio - Because t his model incorporates considerations of landscape, environment and other local factors, and also adapts Southern African models of process to a specific region, it makes a useful heuristic device. Additionally, even though when it was developed it wasn specifically framed in opposition to the CCP, the Toutswe model provides an alternative way to frame economic, social and political processes for the early to middle Iron Ages . As such it is a good starting point for asking questions about scale, environme nt and process for the Southern African Early Iron Age. One additional limitation of the Central Cattle Pattern, when used as a catch - all socioeconomic model for the Early Iron Age, is that the CCP operates almost exclusively at the scale of the individual site, as the above discussion of the Toutswe settlement model highlights. The CCP does not discuss processes of exchange and interaction for the Early Iron Age, which limits its explanatory power (as well as spatial logic) for social and political organiz ation of this time period. Various regions within Southern Africa have distinct patterns of landscape and resource exploitation during the first millennium. For example, sites with Mzonjani pottery (an early representation of the Kwale Branch dating to rou ghly 450 - 750 AD) exist for the most part within six kilometers of the Indian Ocean shoreline of KwaZulu - Natal. As described by Mitchell (2002:273 - 274), close proximity to iron ore sources and higher coastal rainfall are cited as prob able reasons why settle rs chose this area. Furthermore, although Mzonjani sites meet the criteria for inclusion within the Chifumbaze Complex in other ways (exploitation of domestic crops; production and use of Kwale pottery; and metallurgy), no evidence has been found among the numerous array of sites in this area for livestock herding; instead many Mzonjani sites contain shell middens consistent with the exploitation of marine 55 Figure 5 Areas mentioned in the text 56 protein sources. This overall picture contrasts strongly with depictions of cattle - dependent economies and, like Denbow (1982, 1984), the Mzonjani settlement system provides a frame for understanding the dynamic of an entire subregion for a substantial period of time withi n the Early Iron Age. Like the Mzonjani sites, other regions within Southern Africa witnessed distinct patterns of land and resource use within the first millennium, particularly the eastern Kalahari (see Denbow 1986; Mitchell 2002:275); this information i s essential when building a comprehensive understanding of Early Iron Age social and economic organization. Regional variation and other explanatory factors do tend to get underplayed overall in subsistence strategies for the Early Iron Age - not only in t erms of which domesticated species were used where and when, and the socioeconomic and environmental contexts thereof, but also additional presence of wild foods and non - dietary resources such as trade goods. Mitchell (2002:275), for example, discusses the presence of differential patterns of livestock utilization within the same ecological zones during the Early Iron Age, providing examples from KwaZulu - Na tal such as Wosi and Ntshekane. This may also be due in some part to a lack of identifiable organic re mains at many sites, particularly for botanical remains. Very few cases exist in which domesticated botanical remains excavated in a Southern African Early Iron Age context have been identified through comparative macrobotanical analysis (or other, micro - s cale techniques) by a qualified archaeobotanist; one such example is Jonsson (1998) Neumann (2005) for further discussion on the complexities of archaeobotanical analyses in sub - Saharan Africa. Currently, no archaeobotanist working in Southern Africa specializes in plant species extant during the Iron Age, domesticated or wild. Additi onally, wild game and wild plant foods make regular appearances in Early Iron Age contexts. At Bosutswe, for example hunted g ame contributed up to 60% of the faunal assemblages for the Early Iron Age (Taukome and Zhizo horizons) period within the western p art of the site (Denbow et al. 2008). Presuming non - differential taphonomic processes or discard behaviors for wild and domesticated 57 remains, we can assume that these proportions represent a behaviorally significant proportion of the overall diet for these settlements. Incorporation of hunted and gathered foods could imply any of a number of things: the presence of hunter - gatherer communities nearby which traded with the settlement (Thorp 2000; Mitchell et al. 2008), incorporation of hunter - gatherers into t he settlement (Wadley 1996; S. Hall and Smith 2000; Sadr 2005) , or the maintenance and use of those bodies of knowledge, and dedication of regular time and resources to so - called - Early Iron Age communities themselves (which is a much more contested line of thinking, but not out of the realm of theoretical possibililty; see, e.g, Odell 2001 ; Schiffer et al. 2001; Eerkens and Lipo 2005) . The same is true for material culture (such as flaked and ground stone tools, bone tools, and shell beads) typically ascribed to hunter - gatherer makers and/ or users. Numerous cases of these types of mater ials deriving from Early Iron Age contexts exist at sites such as Bosutswe (Dubroc 2010) , Schroda, Mosu I (Reid and Segobye 2000b; Tlapela 2001), Thabadimase go , Nqoma and Divuyu (Robbins et al. 1998; Reid 2005), and many others. While the specific contexts of the material must be considered in each case as this may yield additional vital information (e.g., stratigraphic integrity and a secure chronology; mixin g in or clustering LSA Iron Age items or structures, etc.), if provenance of the materials is those materials become impo rtant to consider. For example, specific bodies of knowledge are required to produce and maintain these items, procure the resources needed to make them, and one may assume that tools (lithic and bone points, digging sticks and weights, for example) and no n - utilitarian items (shell beads, etc.) alike have ascribed social and cultural functions and meanings (see, e.g., Weissner 1983) which designate specific uses and users for these items. To what extent these functions and meanings would have been modified or otherwise recontextualized by their placement in Early Iron Age, as opposed - contact Later Stone Age, contexts, is another question worth considering. Early Iron Age settlements have in the past been framed as self - sufficient in terms of subsistence 58 (including domesticates, production of basic technology, etc . ) and political organization (T. M. Maggs 1984) . However, substantial evidence for trade throughout the areas occupied by Chifumbaze settle ments brought to light in recent decades has shown that many settlements across the subcontinent were unquestionably interconnected. Through coastal entr êpots such as Chibuene in Mozambique, settlements of the Early Iron Age across Southern Africa had acce ss to products of the intercontinental trading network throughout the Indian Ocean which connected south Asia, the Middle East, and much of sub - Saharan (and Nilotic) Africa (Popelka et al. 2005; Ro bertshaw et al. 2010; Sinclair, Ekblom, and Wood 2012) . These goods, which primarily included glass beads, are present at coastal sites as well as interior ones, including Matlapeneng and Nqoma in western Bots wana, by the 8th century AD (Denbow 2011; Wood 2011; Wood, Dussubieux, and Robert shaw 2012) glazed ware dating to as early as the 6th century (Sinclair et al. 1993; Mitchell 2002:288). Based on this evidence, multiple scholars argue for the possibility that such foreig n goods, in addition to being items of personal adornment, likely played the role of prestige goods in economies of redistribution and served as one mechanism of political power consolidation towards the end of the first millennium AD (Mitchell et a l. 2002 :288 - 289; Calabrese 2007 ). Items that may have been produced for export to the Indian Ocean trade network include animal products such as c arnivore furs and ivory bangles. T he 9th century Shashe - Limpopo site of Schroda has produced unusually large quantiti es of carnivorous faunal remains as well as ivory - working by - products (Mitchell 2002:289), while 9th century Mosu I, in the South Sowa area of Botswana, has also produced a number of finished and partly - worked ivory bangles (Reid and Segobye 2000 b ). Trade of goods and resources originating from local Southern African contexts also occurred among E arly I ron A ge communities as well. In fact , exchange of resources such as iron and copper ore may have been vital to the maintenance of an Early Iron Age way of life in those regions lacking ore deposits. While iron ore deposits are relatively numerous within the eastern thir d of Southern Africa , the quality 59 and vitality of each source varied (Mitchell 2002:279 ) . The ability to transform ore into usable end - produc ts via smelting and smithing procedures seems likewise to have been variable, based on spotty evidence for furnaces and metallurgical by - products such as bloom and slag (Miller 2002), although it remains unclear whether this was due to uneven distribution of expertise or to variation in chemical attributes of the ore itself. As mentioned previously, c opper o re sites are somewhat rare in Southern Africa (Miller 2002), and copper items appear in Early Iron Age assemblages hundreds of kilometers from the near est accessible ore site. C opper items are relatively infrequent and ap pear only as finished products. However, finds of iron bloom and/ or slag at Divuyu and Nqoma , as evidence of metallurgical activities taking place on those sites, also allude to the tra nsmission of the knowledge and technology for this type of metallurgy (Miller and van der Merwe 1994; Denbow 1999, 2002; Miller and Killick 2004). However, Denbow and Wilmsen (1986) report smithing at Tsodilo but no t smelting. Numerous additional subsisten ce - oriented material has been demonstrated to move between Early Iron Age regions and settlements on a regular basis; these include riverine fauna from the Okavango Delta and Boteti Riv er found at Nqoma (Denbow 1990). These include whole ceramic vessels an d possibly indivi dual ceramic fragments as well (Edwin N. Wilmsen et al. 2009) , marine shells from the Indian ocean coast found at numerous inland sit es (Denbow 1990; Mitchell 2002), and specular hematite from mines in the Tsodilo Hills (Robbins et al. 1998; Wilmsen et al. 2013) . That these items begin to appear as early as the 6th century and consistently make appearances in Early Iron Age assemblages implies regular contact between several regions acro ss the subcontinent, even if the specifics of that contact (e.g., routes, nodes, hubs, technology/ transportation, and the mechanisms and social contexts of exchange itself) have yet to be substantiated. While, therefore, descriptions of typical Southern African Early Iron Age assemblages (e.g., Segobye 1998 ; Mitchell 2002:259; Huffman 2007:335 - 338) usually refer to their elements like ceramic vessels or figurines, metal tools/ jewelry, livestock remains, structural remains from 60 storage or residences, kraal remains, and burials, it is in no way unusual for such assemblages to contain (non - elements may include shell ( ostrich and Achatina ) beads, wild game remains, and lithic and bone tools. items as specular hematite and metal ores. The presence of intersite - contact and exchange has long been acknowledged (e.g., Denbow 1990; Denbow and Wilmsen 1986; Martin Hall 1988; Pwiti 2005; Chami 2006) and its potential importance for political and economic consolidation towards the turn of the millennium has been raised (e.g., Denbow 1984; Huffman 2000, 2008). The moveme nt of people, goods and resources, and information between different subsistence - based communities, among regions within Southern Africa, and between Southern Africa and the broader sphere of first millennium world (especially other places within sub - Sahar an Africa), likewise come as no surprise. Bantu - speaking, metal - using farming communities of the first millennium AD in Southern Africa were themselves, after all, the result of migratory communities from east and central Africa, and ample evidence from ar chaeology as well as genomic studies demonstrate how frequently and widely populations of Africa, the Near East and south Asia in the last few thousand years shared not only goods and information but genetic material as well (Hellenthal et al. 2014; Pickrell et al. 2014) important social and economic roles that they often likely played within Early Iron Age community dynamics, it seems a worthwhile effort to attempt to reframe our general definit ion of Early Iron Age structures and processes to consider how these inter - site dynamics may have systematically affected the patterns of material and spatial organization observed on the scale of a single site. That this has already been done to effect an explanation for late - first - millennium political consolidation and its contribution to the rise of early second - millennium states should not exclude the possibility of doing so for the several centuries of Early Iron Age occupation prior to this consolidat ion as well as those regions of Southern Africa which were, presumably; outside the immediate reach of these effects during the time of 61 consolidation. To be clear, to synthesize so many streams of data across so many places, even for any given point within the first millennium AD, is a rather lofty goal which would obviously necessitate multiple very long - term collaborative, interdisciplinary research projects. However, it is towards such goals that archaeologists ought to aim, if truly our enterprise is to understand behavior and change in a comprehensive and systematic way. On a more pragmatic note, one way that individual researchers can contribute is by approaching regional concerns and processes, but with broader theoretical questions structuring their praxis, in an attempt to understand variability as well as change over time. With these issues in mind, the next section takes a look at some regional concerns within Botswana. 2.2 Regional (Botswana) framework: localized chronologies and socioeconomic tre nds, etc. The Early Iron Age in Botswana deserves some consideration on its o wn both because of the national research trends which shape the bodies of data produced from the archaeological record, and because of certain unique aspects of the regional chron ology, which will be discussed below. Archaeological data collection in Botswana is frequently conducted by both profess ionals and students in training for non - scholarly venues. F or example, cultural resource management (CRM) firms are often hired to conduct archaeological impact assessments for government development projects (Thebe 2011) . Numerous sites around the country are also the subject of fieldwork for University of Botswana field schools and bachelor Samuel 1999) . The data collected from this work often non - peer - reviewed, limited - print media such as government reports, bound copies of manuscripts, field reports, and University of Botswana undergraduate theses. This literature, which for the most part is accessible only in hard copy at institutions such as the University of Botswana library or the National Museum, contains vital detailed information about the national archaeological record which does not always make its way into the more widely - accessible, and therefore more widely - consumed and replicated, peer - reviewed literature on the same archaeological record. For 62 example, numerous Environmental Impact Reports ( E IA s) prepared by CRM firms as part of pre - construction mitigation studies are full of survey and excavation details from projects all across Bo tswana; (presumably) if one knows the author of the report one may ask for a copy. While information from the gray literature may be easily accessib le and commonly referred to on a local (i.e., national, institutional and university) level, at the same time it may not be incorporated on a timely basis into the broader theoretical arguments being made for Southern African Iron Age trends on a whole whi ch draw from data in Botswana. University students and field technicians do eventually gain access to the theoretical i nterpretation of the materials that the y themselves may have worked on , but only after it ademic discourse and brought back to the local arena through now - dated teaching and reference materials. There are happy exceptions to this rule, of course, such as when CRM firms are directed by post - graduates actively engaged with the literature or when university students are actively brought into the theoretical realm by their mentors. For the most part, however, there seems to be a disconnect between those practicing archaeology in Botswana on an everyday basis and those consuming the archaeolog ical da ta for research purposes. This issue is, of course, not isolated to Botswana (see, e.g., Huffman 2012b; Kinahan 2013) . While this is probably inevitable to some extent, the ramifications for what information ends up in scholarly literature are twofold. First, the bodies of data as presented in the scholarly literature become much more restricted, and that data furthermore becomes seen as representative of the country as just the well - represented subset that it is. Second, because the scholarly literature is for the most part published in South African a nd pan - African or international venues, the terminology and classifications used to describe the data are frequently of South African derivation. As discussed above, it is important to have a broad comparative framework, but on the other hand, the possibil ity for local ceramic and/ or lithic sequences, or other material trends, to be subsumed or go altogether unrecognized remains high 63 when such a top - down approach prevails (Wilmsen pers. comm, 2013). As has been discussed to some extent in previous sections of this chapter, these issues are already known and are being addressed by some researchers such as Wilmsen et al. (2009) and Ashley (pers comm, 2013). One aspect of Botswana subcontinental scales is the fact that Botswana is one of a few places in Southern Africa where Kalundu, Kwale and Nkope ceramic facies have all been documented. Naviundu ceramics of the Kal undu tradition are represented at Nqoma and Divuyu (Denbow 1999); Taukome and Zhizo ceramics (Nkope tradition) are found at numerous sites in the eastern part of the country, and early Kwale ceramics may be found at some sites in south - eastern Botswana con temporary with Broederstroom (Mitchell 2002: 264). If one originating from three separa te areas of sub - Saharan Africa (Mitchell 2002:264 - 271; Phillipson 2005 ; Huffman 2007), then the high probability that communities representing these streams interacted during their respective overlapping waves of settlement (e.g., Tsodilo Hills to Lake Ngami, Boteti River to Sowa and Tswapong Hills, etc.) suggests a number of interesting behavioral i mplications. If these ceramic groups (and their subdivisions) do represent distinct language groups (Vansina 1984; Holden 2002) with distinct community identities, then the production of bot h identities and boundaries, the maintenance thereof, as well as the negotiation of goods, resources and information across those boundaries (along with the eventual transformation of those identities into those observed in later eras of the archaeological record), all remain potential ly fruitful avenues of research (Stein 1998; Brooks 2002). The possibility remains that Botswana was a locus of settlement not just for the Early Iron Age in the sense of one coherent, typologically - consistent cultural horizon , but for multiple interconnected societies which also negotiated boundaries and the exchange of resources with indigenous hunting and gathering communities (Walker 1998; Denbow 1999; Reid 2005) as well as pastoralist populations (Cashdan 1985; Reid, Sadr, and Hanson - James 1998; Sadr 200 5) . 64 One further feature of Botswana that is worth discussing, particularly for its northern regions, is the possible existence of a salt trade during the Early Iron Age. The Makgadikgadi Pans are known for their extensive salt de posits (as discussed in Chapter 1). Exploitation by local communities, and particularly of San communities, of these deposits at specific locations within Sowa Pan has been recorded for the 19th and 20th centuries (Matshetshe 2001) , and possibly the 18th century as well (Cashdan 1979 ) . Sowa Pan is credited as the major source of dietary and preservative salt for populations of the Northern Kalahari throughout this time period (Matshetshe 2001), meaning that those involved in its extraction, refinement and transport participated in a highly valued trade network whose extent reached as far as Bulawayo. The decline of this trade network is attributed to the introduction of European - sourced, industrially - produced salt into the area via the growing formal commerce sector sometime around 1965 (Matshetshe 2001). No direct confirmation of a similar process exists for the Early Iron Age in Sowa Pan or the Northern Kalahari generally; however, Denbow (1999) reports that ceramic strainers that may have been used to strain salt have been recovered at both Divuyu and Matlapaneng, with dates ranging from the 7 - 10th centuries. Denbow (2002:356) furthermore argues that salt is a likely option for a trade good, as evidence regular trade in lu xury it such as shale, ostrich eggshell beads, Achatina and mussell shell beads appears at nearly all sites throughout northern Botswana. Both Denbow (1999) and Reid and Segobye (2000a, 2000 b) suggest that Early and Middle Iron Age sites located around Sowa Pan (such as Tora Nju, Thitaba, Lekhubu, Kaitshàa , and Mosu I) were settled deliberately in order to take advantage of the availability of salt deposits and the economic benefits they offered. While this would be difficult to prove per se, it is conceivable that the knowledge and ability to extract and refine salt from the pans would have existed among Early Iron Age populations or their contemporaries, and the resource itself certainly existed in no small amount at the time. Salt explo itation, while intangible in the archaeological record of Sowa Pan, seems like a viable working hypothesis (or part of one at least) for 65 to be learned ab out in what social contexts that exchange may have taken place (see, e.g., S tein 1998; T. D. Hall, Kardulias, and Chase - Dunn 2011) . This same holds true for the cluster of sites along Sowa Pan s outhern margins in particular. Settlement there by Early Iron A ge communities is documented as early as the 9th century AD and contin ued periodically for several centuries (Reid and Segobye 2000 a ), but much remains to be understood about the local settlement system and its role in a wider political economy. The next section will take a detailed look at the chronology of the South Sowa a rea and the history of work in that area with an eye towards relevant theoretical and methodological questions. 2.3 Local archaeological record: the south Sowa Prior work in the area Although stonewalled sites in that area and elsewhere across Sowa Pan may have been included in folklore and oral histories of Botswana, a rchaeological sites in the south Sowa area were first documented by Denbow (1985) for the British Petrole um Soda Ash archaeol ogical impact assessment survey. Denbow (2002:353) makes mention that Kubu Island was offered up by mid - twentieth century South African media as a fabled although there is little else to support this claim a part from the imagination of the popular media. Further site location surveys were conducted by avocati onal crews in the early and mid - 1990s (Campbell and Main 1991 ; Main 2008). These surveys led to the documentation and registration of over 50 archaeological sites along the south and south - eastern margins of Sowa Pan ( see figure 4 in Chapter 1 ). Professor Tom Huffman provided identification of pottery types surface - collected from both surveys (see, e.g., comments in Main 2008) which allowed for assignment of preliminary cultural associations for most of the sites. Common facies included Zhizo and Khami, while t entative additional identifications of Leopard pottery and Later Stone Age lithic material were also made. From this, the interpretation was formed that occupation of south Sowa likely reached as far back as the mid - to late first millennium AD and that additional occupations likewise 66 occurred around the middle of the second millennium, prior to the arrival of Tswana, Kalanga or other historically - documented communities now present in the area. An early date (9th - 10th c. AD) was confirmed by Reid a nd Segobye (2000a, 2000 b) through radiocarbon samples for Mosu I, a site located about 5.5 km west of Mosu village on a portion of the escarpment. Materials recovered through excavation at Mosu I, including glass trade beads, carved ivory bangles, livestoc k and Zhizo ceramics indicate that not only was this settlement a contingent of the Early Iron Age way of life, but also that it participated in the exchange network which connected much of Southern Africa to the Indian Ocean coast at the time . Further excavations at Kaitshàa , a stone - walled escarpment site contemporary to Mosu I located 15 km northeast of Mosu village, provided similar information during the 1990s and again in 2010 when Denbow (pers. comm) returned to find a huge cache of glas s trade beads as well as copper jewelry items at the site ( figure 6, below ). These finds indicated that the south Sowa area, while geographically peripheral to the Iron Age settlement system of Southern Africa overall (as discussed by Reid and Segobye 2000 b), nonetheless maintained important economic connections for the duration of the community Reid and Segobye (2000b) suggest that iron - using agricultural communities closely related to contemporary hierarchical polities such as Toutswe settled along the escarpments to exploit the water sources of the area as well as the plentiful wildlife. Based on the extant distribution of diagnostic ceramic material, they further suggest that occupation of this area by food - producing, ceramic - using pe oples occurred from approximately 900 - 1400 AD, with the peak at around 1000 AD. They, along with Matshetshe (2001) and Denbow (2002), argue that the Sowa Pan settlements would have played a role in a cross - Kalahari trade network as a source of salt and wil d game products, as well as ostrich eggshell beads which they assume were produced by local hunter - gatherer communities. They further suggest that the hunter - gatherer communities would have been incorporated to a significant extent in the trading network b y collecting and providing salt and game in exchange for milk and meat products, and that such 67 movement prompted increased social complexity and the development of new forms of social and economic organization among local hunter - gatherer communities. Flake d lithics and ostrich eggshell beads, trademarks of the LSA, have been found repeatedly in association with ceramics during surface survey (Main 2008), lending some support to this claim, but the lack of excavated LSA sites with well - dated sequences in the immediate area makes it difficult to know yet how hunter - gatherer communities were affected by the presence of agropastoralists in the area. Further survey by Samuel (1999), conducted for his bachelor t the University of Botswana, con firmed the initial observations by Campbell and Main (1991) that the majority of archaeological sites in the south Sowa area are located at or near the escarpment edge. The twenty - one transects of his pedestrian survey, which went from north to south from the Franci stown - Orapa road to the escarpment edge and which covered a total area of 198 square kilometers at intervals of one kilometer, located 53 previously undocumented concentrations of archaeological material. It is worth noting that, although Samuel refers to these concentrations as sites, his criteria for defining an archaeological site differ from other studies in the area. While Main (2008) does list the occasional isolated surface find or surface scatter as a site if the material is unusual in nature (an E arly Stone Age hand - axe, for example), he restrict s his site listing primarily to locations with built stone structures and/ or dense concentrations of cultural material. Samuel (1999), on the other hand, appears to have listed every instance of cultural m aterial encountered on his survey, whether it is one potsherd or multiple stone cairns. As will be discussed in greater detail in later chapters, the nature of distribution of archaeological material throughout the escarpment is such that isolated finds c ommonly occur at the foot of the escarpment as the result of post - depositional processes, particularly seasonal slope wash. Samuel findings, therefore, must be taken with this in mind. However, although the degree to which his survey contributed new loca tions of archaeological sites remains unclear, his study does confirm that cultural 68 material in the south Sowa area concentrates by and large near the escarpment edge, with only a small number of sites or scatters occurring either south of the escarpment a long the plateau (what Samuel calls surveyed on foot as a training exercise prior to beginning the officia significant in terms of (archaeologically visible) landscape use. The majority of documented archaeological sites in south Sowa range from Early, Middle and Later Stone Age surface scatters to Later Iron Age and early historic stonewalled structures. These site s, which may vary in extent from a single isolated surface find to multiple square kilometers, as has been reported in the National Museum site register. The sites also tend to occupy similar parts of the landscape (the escarpment edge in particular) and in many cases, multiple chronological components may be reported for a single site. Site 16 - A1 - 12, for example, which w as foot - surveyed by the author collected during survey confirmed that the site had no visible stone features, but very dense ceramic scatt ers (which likely comprise multiple facies) as well as ostrich eggshell beads and numerous flaked stone items. Some lithics appeared morphologically comparable to typical Middle and Later Stone Age her pieces were amorphous, irregular or unidentifiable. These may have actually been the product of thermal spalling from exposure to natural or human - controlled fire (Staurset, pers. comm. 2013); in some cases, perhaps they were the product of informal or experimental manufacture. Little can be said with any certainty without other similar contexts for comparison. Potential references for comparison include Phaladi 1991; Weedman 1993; and Walk er 1994 ; however, even lithic studies comprise samples from only a few sites around the country. However, importantly, all of these components appeared on the surface of the site, and what this implies for the occupation history of this place - whether it represents a palimpsest, the end - product of displacement of the stone tools, or something else - would need careful study of the site formation processes. 69 Figure 6 Zhizo sites in the South Sowa area 70 For thos e sites which have been reported as containing Early Iron Age components, a number of pertinent issues that could affect the interpretation of any one of them come to mind. First, Early Iron Age sites with other chronological components are common. Fo r som e sites where, for example, both Zhizo (ca. 7th - 9th century) and Khami (ca. 14th - 16th century) components have been reported, confirming provenance should generally be a matter of determining stratigraphic integrity . Doing so may not actually be a s imp le process on hilltop sites with a shallow soil matrix, animal burrowing, and high erosion factors, but with enough subsurface testing, the stratigraphy should be understandable . In other cases, however, particularly those reported by Main (2008) as of determining whether these components represent discrete horizons, some form of coexistence, or whether they even represent distinct components in the first place or should be evaluated by other criteria. Most site descriptions in the National Museum site register (as well as those sites listed in Samuel ed on the register) derive either from small surface photographs of the surface scatters taken during field surveys (Samuel 1999; Main 2008). In many cases, t hen, the cultural components were identified by amateur scholars (while Professor Huffman did identify ceramics collected by the crew consist of represen tative samples from those sites). This issue is not raised to discount the existing site designations out of hand, but as a point of caution. For most of the 20 - plus very small amount of information derived from a body of evidence largely collected by non - experts well over a decade ago which is now inaccessible. Little in - depth information exists in each site report, and the ability to return to each site location for 71 provided location (generally, U niversal T ransverse M ercator, or UTM, coordinates are precise to the nearest 100 meters, which may be a distance larger than the extent of th e site itself) and the density of ground cover along the escarpment and lowland plateau in most areas. Additionally, because the great majority of these sites are identified only by relative means, the precision in determining settlement patterning for the area is currently low. A good deal more research, including the collection of a series of radiometric data, would be needed at multiple sites before their relationship to one another can confidently be established. These sites, which could potentially ran ge in age from approximately the 7th to the 11th centuries AD, have the potential to offer invaluable insight on a unique settlement system in a unique landscape, the density and extent of which in the Botswana archaeological record is only paralleled by t he Toutswe pattern as described by Denbow (1982). Pertinent questions about settlement history for this area include whether settlement occurred in pulses, or was continuous. Whether sites with similar date ranges, such as Mosu I, Kaitshàa , and Thabadimase go , represent separate, co - existing villages, short - lived serial occupations, or Interlinked, differentially functional locales (in the manner of field, kraal and village) is another issue; the uses for each site based on their features and deposits still remain to be thoroughly compared as well. The following chapter will discuss how these issues, particularly those of scale and process, have been brought to bear in methodological design for this research project as well as how prior studies in south Sowa and Botswana generally have aided in developing expectations for the archaeological fieldwork. 72 Chapter 3 Field Methods 3.1 Introduction This chapter describes the methods of data collection used in fieldwork for this research project, as well as the post - excava tion inventory conducted prior to analysis of the assemblage. The categories of information as well as methods used to collect them are described in detail for both fieldwork and the subsequent post - excavation cataloguing process. Additionally the methodol ogical justification for these procedures and their relevance to current research trends is discussed. I also describe the methods of analysis used to make observations about these datasets and why these analytical procedures were used. Methods fall into t wo main categories: data collection and analysis. Fieldwork was conducted over a period of three months from July to October 2012 with a team of three professional field technicians who were recommended through faculty at the University of Botswana. Additi onal help was provided at times by University of Botswana volunteer undergraduate students as well as staff from the local National Museum (NMMAG) office. Several phases of data collection occurred, a few of which depended on feedback from a previous phase to make decisions about coverage or sample size. The fieldwork was broad in scope in no small part due to the fact that so little in the way of archaeological (or other relevant historical, geographical matter) is published about the South Sowa area; ther efore, a substantial portion of the field work ended up being exploratory. In brief (roughly) chronological order, the phases are as follows: I. Thabadimasego survey and excavation Site survey Ground - truthing/ confirmation of site location Thabadimasego site gridding and surface collection Thabadimasego perimeter mapping Test units (3) 73 Thabadimasego subsurface (test pit) survey Unit excavations: purposive units (locations based on subsurface survey) - 17 on hilltop Tape - and - compass stone wall mapping A dditional test pits II. Thabadimasego Periphery survey and excavation Periphery test pits Periphery test unit III. Escarpment survey Surface survey of foot and edge of escarpment between Thabadimasego and site 33 GPS documentation of surface scatters IV. A dditional sites Site 12 perimeter, surface collection and test unit Site 33 perimeter, surface collection, and test unit V. Visits to previously excavated sites Mosu I perimeter mapping Kaitshàa perimeter mapping VI. Landscape survey Visits to clay mines west of village Visits to cattle posts and dammed natural springs 3.2 Goals for fieldwork There were five major goals for this fieldwork. These do not neatly tie into the chronological phases of the data collection, since, as mentioned above, in numerous c ases the work was exploratory and the same kinds of information was often collected at varying scales (for example, distribution of surface scatters across both the site and across the landscape). However, each part of the fieldwork was 74 designed with one o r more of these data collection goals in mind, and as a whole therefore the phases of fieldwork inform the goals quite well. The data collection goals were as follows: 1) Collect information relevant to understanding site formation processes in this partic ular physical landscape and climate, such as stratigraphy, taphonomy, and site integrity. 2) Collect information about surface and sub - surface material distribution (including type, frequency, and co - occurrence) at two scales (intra - site and inter - site) as they inform use of space (and our understanding of how archaeologists define a 3) Collect stratigraphically - secure cultural materials (such as faunal remains, ceramics, trade goods, iron or stone tools) that inform understanding of economic and social processes as elaborated in the research questions posed in Chapter 1 . 4) Collect radiometric samples from relevant depositional contexts to provide an absolute chronological framework against which to compare the results of laboratory analyses. 5) D ocument locally - available natural resources (and geographic features) that might be relevant to understanding economic behaviors and use of space on the landscape (such as tool - making, wild fauna or flora availability, or water availability) for local Earl y Iron Age settlements . 3.3 Hypotheses A number of hypotheses were generated based on the extant body of knowledge for the area as discussed in Chapter 2, as follows: Surface and sub - surface material distribution (type, frequency, co - occurrence) at two scales (intra - site and inter - site) will not be random, nor will they align neatly with the attributes of the Central Cattle Pattern. Type, frequency, co - occurrence of mat erial distribution within and between sites will inform understanding of spatial organization, social behaviors and economic processes as they were expressed in the Early Iron Age. 75 Type, frequency, co - occurrence of material distribution will not align neat ly into clearly - defined ethno - social - historical typologies. Placing the local archaeological record into the context of its geophysical landscape will substantially aff ect interpretation of processes at the site - scale as well as the inter - site scale (as compared to when conducting site - level analysis only). The following section describes in detail the methods used for data collection during these phases of fieldwork. 3. 4 Data Collection (Fieldwork) Phase I - Thabadimasego (main s ite ) Ground Truthing The fieldwork phase (July - October 2012) of research was comprised of surface survey, subsurface (test pit) survey, and excavations. The first few days were spent foot surve ying the hilltops to confirm the locations of the intended sites of excavation, and to verify that surface scatters were in fact confined to the bluff edges and hilltops as reported by previous researchers (or whether a more continuous scatter along the en tire bluff formation might be observed). A map depicting the locations of all foot survey conducted in 2012 can be found below ( Figure 7 ). Once the sites were confirmed, test excavations commenced at the main targeted site (listed in the grant proposal as site 16 - A1 - 31, but confirmed by ground truthing to be site 16 - A1 - 13, which was later named Thabadimasego ). This site was chosen for study because it was listed on the National Museum (NMMAG) site register as one of the few single - component Early Iron Age s ites in the Mosu area, making it a useful unit of analysis for comparison with other contemporary settlements. On - site survey Once the site was confirmed and partially cleared, we established a site datum and grid ( F igure 76 8 ). The site datum was spatially r eferenced to a universal coordinate system with handheld GPS, both north - south and east - west baselines counted out at 10 - meter intervals, and a local coordinate grid was marked with pin flags at 10 - meter intervals across the entirety of the site from the s tone wall (which coincides with the natural - Southern , and eastern extents of cultural material surface distribution (which more or less coincide in most places of the bluff - top with the elevation drop - off where it becomes too steep to walk easily on the slope). Before conducting extensive sub - surface investigations I wanted to gather information about the correlation between surface and sub - surface cultural material, as little information about s ub - surface depositional history is available for sites in this area, sites are primarily identified by a sampling of surface materials collected from prior surveys (e.g., Samuel 1999; Main 2008) and therefore an on - site comparison of the two could be highl y relevant for interpreting the integrity and occupation history of the site. As a measure of this, 100% of the surface material visible was collected in each 10x10 - meter grid square in the western third of the site closest to the wall. The material from e ach grid square was bagged separately and is presently curated in the collections at the National Museum. It may be worth noting that my crew protested at the collection process, saying it was standard practice in Botswana to leave surface material on - site so that future visitors could recognize the loc ation as an archaeological site . To address this concern, we refrained from collecting materials in the eastern two - thirds of the site, which left hundreds of sherds visible to potential visitors. As it was, however, this collection did allow me to gain a picture of the extent to which the site was covered in cultural material as well as the range of that material. Additionally, the perimeter of the entire hilltop site as it is defined by the limit of surface material distribution was walked and track - logged using the handheld GPS unit (this is depicted in Figure 8 as the ). This provided further observations relevant to ground cover, slope and taphonomy. A number of surfac e features were observed on the site, such as the remnants of a stone wall 77 Figure 7 Areas surveyed in 2012 78 Figure 8 10 - by - 10 - meter grid on Thabadimasego 79 marking the connecting point between the hilltop promontory and the rest of the escarpment , as well as two small stone cairns . To record the location of these features, a tape - and - compass survey was conducted using the geotagged site datum and grid as reference point s. In the case of the stone wall remnants, pin flags were placed at short intervals along the center line of the wall along its entire observable extent. For each flagged point, the distance from the last point was measured with a tape and bearings were ta ken with a compass. The width of the wall was measured at these points as well, although it was clear that the entire wall had been very much damaged and knocked down, so the accuracy of these di mensions is probably unreliable (s ee chapter 5 for further di scussion of the walls ) . The starting point of this survey was one of the grid coordinate locations, and the information recorded by the survey is therefore easily reproducible as a georeferenced set of measurements for inclusion in the site GIS. The other stone features on the s ite were recorded in similar fashion, although since they were much smaller, they were measured end - to - end and cross - wise, with at least one flagged point referenced to the grid coordinates. Photographs were taken of each feature as well. Test pit s urvey F ollowing this, my crew and I proceeded to dig small (30x30 - cm) test pits at 10 - meter intervals (generally placed at the grid pin flags) across the entire site wherever cultural material was visible on the surface and at least 10 meters in any direction bey ond this as well. For each test pit, soil color and consistency was recorded, and finds were bagged according to pit. Unfortunately, in retrospect, not sorting out the finds from each pit according to the soil horizon in which they were recovered means tha t I am unable to directly compare this assemblage of finds , in terms of artefactual and depositional associations at the same resolution , with the excavation units and later (as discussed later in this chapter ) , the main go al of this initial test - pit survey was accomplished, which was to get an idea of both the stratigraphy and sub - surface material distribution across the site, as well as to identify any activity areas or soil anomalies for further investigation. 80 Excavation A total of 20 1x1 - meter units were placed on the main hilltop site, with one additional unit placed on the escarpment outside/ southwest of the stone walls (see Figure 9 ). Three of these were placed prior to the test pit survey described above, in order to gather information on stratigraphy near the north and south portions of the stone wall (units 1 and 2) and in the central clearing (unit 3). These units revealed the difference in soil deposition in these areas as well as the high cultural material conten t of the central clearing as found in unit 3. Combining the information on depositional history gathered from units 1 - 3 and the test - pit survey (as described in the section above), an additional 17 1x1 - meter excavation units were placed within the site at locations identified to be purposive sample of the site aimed at gathering the highest - priority information within the given time and resource constraints for completing the fieldwork. Although it appears to be common practice in Southern African archaeology to open one or two large trenches in the central area of an open - air site as the main method of data recovery and as a way to focus on the primary locus of deposition (see, e.g., Denbow et al. 2008; van Waarden, Mosothwane, and Waarden 2 013) , standalone 1x1 - meter units were chosen as a sampling method for this project. This was done in order to be able to cover ground across the entire site while maintaining high informational resolution and the ability to open a bigger trench if neede d (further implications of the different excavation methods will be discussed in Chapter 6). The location of these units was guided by findings in the test pit survey. Due to the process of elimination a substantial portion of the hilltop could safely be i gnored, as it was either disturbed or sterile ground (or both). Activity areas and structural remains were the focus of the excavations. Given the 10 - meter interval coverage of the test pit survey, it seems safe to say that a high proportion of both of the se were located and excavated. As will be discussed in the following section, additional coverage was also given to a few areas deemed of interest with another test pit survey, in order to delimit further the extent of these features. Units were excavated in arbitrary 5 - or 10 - centimeter levels, depending on the context (the great 81 majority of levels were 5 centimeters, but occasionally after one or two low - productivity levels we would revert to a 10 - cm level to finish out the unit). Soil was screened thro ugh 1 - mm mesh and all finds were bagged. Planviews were recorded for levels and at least one wall profile was recorded for each unit. In addition, photographs were taken of each profile and of several level floors. Upon completion of excavations, all units were backfilled and markers of modern disturbance included (i.e., we threw a little garbage that would be easily recognized as modern onto the floor of each unit before backfilling to mark it as an excavated unit). Stratified test pits In order to increase fine - grained coverage of certain portions of the site and to check the extent of previously identified features, five of the 10x10 - meter site grid sections were chosen for an additional 30x30 cm test pit survey. Ten test pit locations were chosen at random within each grid section with a random number generator. Unlike the first round of test pits, these were excavated according to the natural strata with finds from each strata bagged separately. This was done so that information from t hese test pits would be more comparable to the excavation units (in that cultural material would be tied to a particular depth and soil horizon). There seemed to be differentiation in material distribution between the different types of soil strata and I t hought this could perhaps map on to different uses of the site, or phases of site use. A total of 50 test pits were placed in this stratified survey, for a 10% sample of each of the chosen grid sections. As before, pits were dug to sterility and/ or bed ro ck. Depth, color and consistency of each soil horizon was noted. The stratified test pits helped to confirm distribution patterns of material culture across the site. While it could be argued that opening additional excavation units in the vicini ty of buried structural remains would have been a preferable use of time (to describe these remains rather than focus on delimiting potential features across a broader area of the site, at the time my goal was to sample from the t otal extent of all types of features instead of enumerating specific ones. It was 82 Figure 9 Location of excavation units at Thabadimasego 83 felt that these test pits were the most efficient way to do so in the limited time remaining on site. Future work at this or similar sites along the Mosu Escarpment ha s the potential to address the question of high - resolution study of specific structural types using area excavation. Radiometric data collection During the str atified test pit survey and unit excavations, several radiometric samples were collected. Any charcoal flecks or fragments over approximately 5 millimeters long were collected as an aggregate sample per level (or test pit soil horizon). The density and con text of charcoal deposits, including associated features, cultural finds and soil type, was noted in order to inform selection of samples f or AMS processing. C lustered concentrations of charcoal were included in planview and profile recordings as well. Whi le these clusters did not appear to be part of any clearly evident hearth feature, they did frequently co - occur with cultural materials including pottery and beads; this admixture may instead indicate small middens. Additional organic material including c arbonized seed clusters, pottery sherds with charcoal temper, and ostrich eggshell, was recovered during excavations. OES has been determined as a reliable source of dateable organic material; (see Vogel, Visser, and Fuls 2001) , but was not selected for AMS processing. In addition to the charcoal samples, two consolidated bulk sediment samples were collected for OSL analysis as independent co nfirmation of depositional integrity. Both samples were collected following the recommended procedures provided by the University of Washington Luminescence Laboratory ( http://depts.washington.edu/lumlab/about_seds.html ). Both samples were collected from different depths of the same vertical column of a centrally - located unit so that they would be comparable in terms of site use. Phase II: Thabadimasego periphery As descr ibed in Chapter One, the hilltop on which the main site of Thabadimasego rests is a sort of oblong protrusion from the rest of the escarpment (s ee Figure 3 ). The slopes of the escarpment here 84 are steep and full of scree towards the drop - off point where the slope becomes an actual cliff. Also, the vegetation cover (mostly acacia thorn bushes and tall grass) tends to be quite dense on these escarpment margins up to the point where the slope becomes too steep and the soil too thin for anything to grow. The cen tral part of the Thabadimasego hilltop was somewhat of an exception to this vegetational density; there overgrowth only existed in patches. About 80 m south - west of the main hilltop a small surface scatter of ceramic sherds had been o bserved and geotagged during earlier exploration of the area during the same field season. Test pits Using the geotagged location as a second local datum point, my team and I placed 30 more test pits at ten - meter intervals where ground cover and slope allo wed, until test pits repeatedly yielded sterile results (Figure 10) . As with test pits on the main hilltop, all finds from this periphery survey were bagged for recovery, and observations were made for changes in stratigraphy. However, little to no soil co lor changes were evident in this area. Soil composition proved consistently unconsolidated as well (as with the test pits and units excavated on the lower, steeper contours of the main hilltop). As such, these test pits provided little information about de positional history for this area (save, perhaps, that it is disturbed), but the types and quantities of cultural material for each pit were recorded. This perimeter test pit survey did help to confirm that there was significant cultural material presence o utside the walled hilltop (more than what would seem random or the result of colluvial deposits), but also demonstrated the limits of its distribution. Test unit Following completion of the test pit survey, one 1x1 - meter excavation unit was placed near one meter south of the secondary local datum on the site periphery (which turned out to be the general area with the highest concentration of subsurface finds on the periphery). This unit was located in this area for the purpose of gaining deposition al information outside the walls for comparison with the units placed 85 Figure 10 Location of all units and pits, positive and negative, at Thabadimasego 86 on the main hilltop. As with units on the main hilltop, this unit was excavated to sterility in arbitrary 5 - centimeter levels. Planviews for each level and a wall profile were recorded. All recovered finds were bagged. In - field feedback based on survey results The NMMAG site register had listed two sites in thi s general vicinity, so based on the description I decided that these surface scatters were very likely what had been recorded as site 16 - A1 - 31; however, these scatters exist within 100 meters of site 16 - A1 - 13 ( Thabadimasego ), overlap almost completely in k ind with finds from the hilltop, and (as will be discussed later in more detail), additional survey demonstrated that cultural material scatters did not extend further along the escarpment. Given th is evidence, it was decided to append this location and it s material scatters within the definition of site 16 - A1 - 13, although (as will be discussed in later chapters) differences in both stratigraphy and the aforementioned surficial distribution indicate a probable difference in use of space for this area as com pared to the hilltop itself. Phase III: Escarpment survey Included among the research goals for this project is the goal of using high - resolution pedestrian coverage of areas surrounding documented sites in order to ground - truth whether surface cultural ma terial distribution really is limited to specific, north - facing promontories of the Mosu Escarpment as described by Samuel (1999). Samuel - south transects at 1 - kilometer intervals bearing due south from the Mosu Escar pment edge to the Orapa - Francistown road, which enabled him to sample a large portion of this area and its changing landforms for high - visibility archaeological sites such as stone structures. His survey revealed that the great majority of sites were const rained to the escarpment edge; they were not also scattered across the Southern plateau or the from the mid - 1990s which located a few dozen archaeol ogical sites mostly at or near the escarpment 87 edge, but which had been conducted partly by aerial survey and partly on foot by amateur trainees. However, during a 2008 site location survey in the area , our crew observ ed that numerous small artifact scatter s were present at the base of the escarpment, sometimes well away from its north - most promontories. Additionally, the NMMAG site register did list a few sites in the area that a re not located on the north ern - most promontories or isolated hilltops north of the escarpment (but, instead, are located a few hundred mete rs south of the escarpment edge, or at its foot). The intent was therefore to employ a survey method different from Samuel , in order to investigate whether a) the escarpment edge sites as they w ere defined spatially in the site register were accurately described and b) if the material scatters found elsewhere along the escarpment could be explained. The importance of clarifying this point is linked to the understanding of what comprises a site an d what uses of the grea ter landscape can be observed; i .e., if sampling is conducted too narrowly, the potential exists to miss a lot of information. To address these questions, the 2012 field team conducted a surface survey of the escarpment between sites 16 - A1 - 33/16 - 1 - 12 and 16 - A1 - 13. B oth the foot of the escarpment (in the first pass) and its edge (on the return pass) were covered , with 10 - meter intervals separating each of the four surveyors . Survey transects followed t he natural contours of the slopes between the starting and ending points, meaning that the actual distance surveyed was much greater than the bird - eye straight line one would normally measur e between these two points ( Figure 11 ). Ground cover was thornbush forest in the case of both the escarpment foot and its edge above; however, a lack of undergrowth beneath the thornbushes meant that the ground surface itself was highly visible. The goal of this foot survey was to document any surface distribution of cultural material (such as ceramics or lithics as already observed in prior survey and excavation) as well as any natural resources (clay deposits, natural springs, etc.) that would potentially be important to the interpretation of the landscape as a habitable place. The entire path was track - logged with the handheld GPS unit; any cultural finds (and potentially significant natural finds, like clay deposits) were also geotagged. This survey concluded on the plateau of site 16 - A1 - 12, which was marked for further 88 investigation (and whose open, densely - scattered surface was like heaven after pushing through kilometers of thorny brush and finding very little). All in all, a total of approximately 157,000 square meters , covering about five kilometeres in length, was surveyed using this method. Phase IV: Additional sites Three other Early Iron Age sites were listed in the NMMAG site register as located within short walking distance (1 - 2 kilometers) of Thabadimasego . One of these, site 16 - A1 - 31, was determined to be an extension of T habadimasego itself, as already described in the section above. The other two sites, 16 - A1 - 12 and 16 - A1 - 33, are both situated along the escarpment about 2 kilometers northwest as the crow fli es of Thabadimasego . Site 12 is visible as a dense array of pottery sherds, flaked lithics and the occasional ostrich eggshell bead scattered across a broad, open promontory of the escarpment (these findings are briefly addressed in Chapter 6). This promon tory is unique in the area not only for its dense surface scatter but also for its lack of ground cover other than short grasses. Site 33, on sites in t he area) as a surface scatter of pottery sherds located about 100 meters north of the base of the escarpment promontory on which site 16 - A1 - 12 lies. Unlike most of the sites in the South Sowa area, site 16 - A1 - 33 is also situated just south of the local junior secondary school, in a locat ion which experiences daily foot traffic from resident human and domestic animal populations. One of the commonly - used footpaths in this area actually cuts right across the site. These sites were included in fieldwork for a short survey and test e xcavation in order to compare stratigraphy between them and Thabadimasego as well as to collect comparative samples of diagnostic cultural materials such as decorated pottery sherds where possible. At each of these sites, as at Thabadimasego, I mapped its perimeter, using the distribution of the surface scatter as a boundary. The perimeter of each site was recorded as a tracklog on the handheld GPS unit. One 1x1 - meter excavation unit was placed on each site as well, using arbitrary 5 - centimeter levels. At e ach site, the unit was placed 89 Figure 11 Mosu escarpment survey area 90 in a central area near a high density of surface material. In the case of site 16 - A1 - 12, the location of the excavation unit was geotagged wit h the GPS unit. At site 16 - A1 - 33, the location of the unit was recorded in meters north of the permanent NMMAG signpost at the edge of the site, which acted as the local datum. As with units at Thabadimasego , both units were excavated to sterility. Planvie ws for each level and a wall profile were recorded, and all recovered finds were bagged. At site 16 - A1 - 33, test pits at 10 - meter intervals were also placed (prior to excavating the test unit), also using the NMMAG signpost as a starting point and datum. I nvestigating this site and its depositional history, as it occupies a unique position on the floodplain, was of particular interest. As with prior test pit surveys, finds were bagged for recovery, and observations were made for changes in stratigraphy. It soon became clear that this site contained a palimpsest of soil deposits that was in all likelihood highly disturbed; the soil horizons were inconsistent across the expanse of the site and shifted frequently below the surface as well. It was decided, based on these findings and based on site 33 that site 33 was probably the product of colluvial deposition from site 12. Phase V: Visits to previously excavated sites So that I could gain a better understanding of the layout and geophysical setting of Thabadimasego as it compares to previously - researched sites in the South Sowa area, the 2012 field team visited both Kaitshàa (Denbow et al ., 2015 ) and Mosu I (Reid and Segobye 2000a, 2000 b) over the course of the field season. Kaitshàa is about 15 kilometers northwest of Thabadimasego on a relatively large promontory directly overlooking Sowa Pan, while Mosu I is further west on one of a series of ridges set a few kilometers back from the pan. In this area, the Mosu Escarpment begins to blend into the surrounding topography. At both sites, the perimeter ( as defined by the surface scatter extent ) was mapped using the handheld GPS to create a tracklog of each perimeter. At Kaitshàa , as at Thabadimasego , the surface scatter more or less coincided with the contours of the hilltop and decreased in frequency where the contours abruptly increased in steepness at the hill ie on an isolated 91 promontory of the escarpment like those two sites, and the ground cover there is very dense as well; for these reasons navigating the site and distinguishing its limits proved problematic. M ultiple visible surface scatters were geotagged at Mosu I to accommodate this uncertainty. At each site, photographs were taken of surface finds (which were not collected) as well as of the viewshed. Phase VI: Landscape survey The final phase of fieldwork involved conducting searches further afield (re lative to the archaeological sites in question) within the South Sowa area for natural resource locations and observations of their present - day uses. The purpose of this was to identify what, if any, natural resources were available in the area that may ha ve been useful to an Early Iron Age community (in addition to those, such as the location of salt and water, which had previously been reported in l iterature). This information contributes to an understanding of Early Iron Age sites uth Sowa landscape. Clay d eposits One of the field team members, after talking with some female residents of the vil lage, shared the observation to that the source of the brightly - colored and elaborately - decorated house village was, in fact , clay deposits found throughout the area . Our neighbor in the village, who was also employed as our camp attendant and cook at the time, agreed to take us around to visit some of these deposits, which are actively mined by village residents and considered a unique and proprietary resource, on the condition that we not disclose their specific locations. This walking survey was conducted over the course of one d ay. While visiting the mines (which are in fact shallow pits excavated from the surface), the fiel d team was allowed to collect small samples of the various clays, which range in hue from green to white to purple to yellow to the more usual reds and browns. The field team also talked to several residents of Mosu to ask about how the clay was used (and were granted permission to photograph a few houses painted with the clay). We learned that the clays are used, in additio n to house paints, as material to make pieces of art that are sold by women 92 entrepreneurs . We were also told that one woman, who reside s in Letlhakane and is related to people in Mosu, uses the clay actually to make fired ceramics, but I was never able to verify this. Wells, dams and b oreholes During the last few days of our field season, the field team visited briefly with families on two cattle posts to learn about their wells. We also visited two of the dammed natural springs in the area and talked to people about how they are used today. Though brief, these visits yielded the following observations: Dammed springs are used by livesto ck, not people, except in times of severe drought. Locations for wells are identified by contracting surveyors, or by observing surface water extents. Wells are dug in dongas (riverbeds) where water is closer to the surface. The wells are dug by hand, usi ng pickaxes and shovels, and can take up to 4 months to dig in a relatively shallow water table (about 20 feet down to water). Dynamite may be brought in to breach some of the rock. Water is then pumped up to cattle troughs using a generator. The water is not salty (or relatively low in salt) and fairly plentiful; multiple wells can be placed on the same riverbed. The wells must be regularly maintained to keep them free from sediment build - up. The mouth of the well is lined with rocks and plaster to keep th e soil from washing away during rains. The cattle posts we visited were west of Mosu, south of the Mmatshumo road about halfway between the road and the foot of the escarpment, on a sloped area with low ridges. Riverbeds themselves tend to be around 5 mete rs deep, and occur near large rocky outcrops. 3.5 Post - excavation p rocedures Summary of post - excavation activities After the conclusion of the field season, I spent several weeks (October - November 2012) at the archaeology lab of the National Museum in Gaborone on post - excavation tasks, including cataloging, photography, and sample preparation for export. These activi ties will be described in further detail below. 93 During the spring of 2013, while back at Michigan State University, I collected additional morphological data from the shell, glass, metal and lithic artifacts I had exported. I also submitted three charcoal samples to the University of Arizona AMS lab for dating: two from units at Thabadimasego and one from site 16 - A1 - 12. In July 2013 I returned to the National Museum in Gaborone to conduct flotation, complete final data collection on shell and lithic assembl ages, and prepare the faunal remains for export to the University of Pretoria for further analysis. Arrangements were also made during this time for parts of the assemblage, including glass, faunal and ceramic items, to be analyzed by a number of consultan ts. In fall 2013 the glass bead assemblage was sent to Marilee Wood for facies identification, while in May 2014, LA - ICP - MS analysis was conducted on the beads by myself and Laure Dussubieux at the Field Museum in Chicago . Further arrangements are being ma de for the completion of the ceramic analysis. Conversion and and continued until late 2014. Goals for post - excavation The goals for post - excavation were as follows: Prepare materials for long - term storage in the NMMAG collections Inventory/ catalog of all recovered finds Photograph samples of finds Prepare samples for analysis to answer research questions Cataloging and curation During fieldwork, a running inventory of recovered finds was kept throughout all survey and excavation procedures. Each batch of finds from a unit level, test pit or [other provenience] was recorded in the inventory, called the lot book, and assigned its own uni que lot number. The lot book recorded the provenience, types of finds, date of recovery, and excavators for each batch. Lot numbers were also recorded on artifact find bags and data collection forms where appropriate, for cross - referencing. 94 Inventory durin g the post - field phase, therefore, primarily involved rectifying this existing record and collecting new data fields for several find types to increase the specificity of the record. During the fall of 2012 and summer of 2013, the following activities took place: Washing select finds, including shell beads, bone, and the remainder of the ceramics. Many of the ceramics had already been washed while still in Mosu by the crew to fulfill their Saturday half - day work requirement (which is considered a standard p art of the work week for a field crew in Botswana). Sorting level bags into their constituent item bags (i.e., one level bag containing all finds from a level would be sorted into multiple bags each containing one of the following: glass, ceramic, metal; e tc. Usually these smaller bags were grouped together into a larger general bag for safekeeping. Counting and weighing finds Refined description of find categories (i.e., for a find of the catalog would record numb er of fragments, number of whole beads, number of broken beads, etc.) Photographing ceramics, including all decorated sherds and a sample of undecorated sherds. The profiles of all decorated sherds were photographed as well during a return trip in 2014. Se lecting and preparing radiometric samples Procuring export permits for selected materials S orting, description and photography of botanical remains P rocessing of soil samples using a small handheld flotation system Although the cataloging process itself was fairly extensive and recorded much descriptive information that is useful for answering research questions about diet and economy at Thabadimasego , several categories of artifacts were chosen for follow - up analysi s based on their perceived importance to 95 the Early Iron Age way of life. The following chapter covers in detail the analytical methods which were used for the artifacts which were subject to further inquiry. 96 Chapter 4 Analytical methods 4.1 Introduction overvie w of methods Following the completion of fieldwork and the post - excavation cataloging process, a number of follow - up evaluations were conducted on several components of the artifact assemblage recovered from Thabadimasego . The work on Thabadimasego blage was on many occasions a collaborative one. Some of the work I was not qualified to do myself, such as faunal identification, and some of it was done side - by - side with specialists and research assistants. Where this is the case for a given analytical method, provided. I believe it is a strength of this research project to have offered the opportunity to develop so many working partnerships with a mul titude of talented people. While material was also collected during fieldwork at sites 12 and 33 (the sites adjacent to Thabadimasego ), it was not included in subsequent analyses (except for one charcoal sample from site 12 submitted for AMS dating, discu ssed below). The single 1 - by - 1 - meter unit excavated at each of these sites detail with the work planned for Thabadimasego . However, s ites 12 and 33, and the ir cultural materials, should be revisited at some future date (and hopefully in some future excavations). Generally, analysis was done categorically according to the class of artifact the ceramics, the faunal remains, the glass beads, etc., were all gro uped together by their respective material and worked on in batches. Hence, the remainder of this chapter reads as a series of individual reports. The last section of the chapter will offer a brief summary of the resu lts gained from each analysis. Due to i ts level of detail, t he spatial analysis (including a consideration of spatial cluster ing of specific artifact types) has been placed in the following chapter. 4.2 Ceramic analysis Archaeologists working in Southern Africa generally accept ceramic facies as an indication of 97 group identity , i.e. of culture groups that are limited to a certain temporal and geographic distributions (Huffman 2005, 2007; Sadr and Sampson 2006), a lthough there is some argument over the nature of what social or ethnic information was intended to be communicated (Pikirayi 2007). As such, identification of chronology. In Southern Africa, standard methods for identifying ceramic facies include determining the shape/ profile of the vessel, and the type and placement of the decorat ions . These major attributes together make for different facies, many of which are accepted as diagnostic across Southern Africa. Due to a recent rise in functio nal and compositional analyses of ceramics, such as the optical petrographic study of ceramic tempers by Wilmsen et al. (2009) , lipid residue analysis by Collins (2013), and functional analysis of temper and paste by Ashley (2013 pers. comm.) , the current model for characterizing ceramics has the potential to change considerably. Due to the limitations of time and resources of this study, however, the standard technique of identifying facies has been used. Here, the ceramics are considered as an indicator of cultural type to serve as a diagnostic marker of comparison with other contemporary sites. Methods In order to identify the ceramic facies present in the assemblage recovered at Thabadimasego as closely to the standard method as possible, the following information was collected for decorated rims and decorated body sherds: Impression type (e.g., comb - stamping; linear incision) Motif (e.g., multiple horizontal lines of comb - stamping) Layout (position of motif relative to the rim, e.g., upper rim; neck; etc.) Presence/ absence of burnishing or paint This data was collected by the researcher and two research assistants. Ian Harrison, an undergraduate student at Michigan State University who received training in pottery identification prior 98 to participatin g in this study, and Tsholo Selepeng, a graduate of the University of Botswana who completed from the Kaitshàa surface collection was also examined. Although the ideal characterization of Southern African pottery, according to Huffman Handboo k to the Iron Age , includes reconstruction of vessels to assess their shapes, sizes, and possible functions in addition to examining decoration motifs and layouts, in practice vessel reconstruction is not always possible. In this case, it proved to be beyond the means of the study to attempt a full refit study. Quantitative data , such as sherd thickness and vessel size estimates from rim curvature, was also not collecte d. Although this information too is of interest for the assemblage, it was not considered critical information for this study because it is not typically incorporated into standard facies determination techniques . The spatial component of the pottery assem blage that is, where and in what contexts pottery clusters at Thabadimasego , will be discussed in Chapter 5, which covers all the spatial analysis. A facies designation was assigned to each sherd based on the combination of the attributes listed above, u sing the images and descriptions of standard facies published in Huffman (2007). Additionally, the face and profile of each rim sherd, a s well as a small selection of unusual body sherds, were photographed. A sample of sherds bearing various motifs were al so drawn to highlight further the range of decorations present in the assemblage. Each sherd was labeled with its provenience information with permanent ink. Additionally, most sherds were washed, in particular to reveal decoration details and/ or presence of burning, paint, etc. However, a small sample of undecorated body sherds was set aside unwashed for future potential lipids analysis. Results A total of 277 decorated rim and body sherds from Thabadimasego were examined, along with another 20 decorated rim sherds from Kaitshàa sherds from the Thabadimasego assemblage, all of which had previously been inventoried and 99 photographed, were missing from the collection. Despite several searches, these sherd s could not be located. However, a tentative fac i es designation was assigned to them based on the photographs taken in 2013. Another 11 sherds had been stored in the incorrect unit/ level bag (as indicated by the photographs) and could not be reconciled wi th their correct provenience. These sherds were rejected from the sample. For an overview of quantities of types of sherds present in the assemblage , see Tables 1 and 2 and Figure 12. Of the 297 sherds examined from both sites, the great majority appear to be fragments of Zhizo - style vessel s ( Figure 13 ). The Zhizo facies has been documented at dozens of late first - millennium sites in eastern Botswana, western Zimbabwe, and the Shashe - Limpopo river basin of South Africa. This facies is characterized by the p resence of single or multiple bands of comb - stamping, bounded by either horizontal linear incisions or additional lines of comb - stamping, along the lower rim and neck of a vessel (Huffman 2007:145). Graphite burnishing or red paint may also be present occa sionally, as was seen in this assemblage. A firm designation of facies, based not only on impression type and a complete motif but on layout as well, could only be made for 12 of the sherds. The remainder were classified as (or probable ot her) given the incompleteness of the decorative motif (in the case of rim sherds) or due to an unclear position relative to the rim (for body sherds). How sherds show a strong similarity in both impression and motif to the e stablished Zhizo facies as compared to any of the Handbook , so even a designation of is made with reasonable confidence. Seven of the Thabadimasego sherds (six body sherds and one rim sherd) proved to be unid entifiable based on the low integrity of the decorations present. Another two sherds possessed motif and impression types atypical of the Zhizo facies. The best guess for these are Ziwa and Eiland respectively. The geographic and temporal ranges of Eiland - and Ziwa - style pottery differ from that of Zhizo - 100 style pottery. Ziwa pottery is associated with very early settlements (roughly 300 - 550 AD) in central and Southern Zimbabwe (Huffman 2007:137). Eiland pottery has been documented in south - eastern Botswana a nd north - eastern South Africa for around 1000 - 1300 AD (Huffman 2007:227). Zhizo - type pottery, on the other hand, is primarily documented in eastern and south - eastern Botswana, south - western Zimbabwe, and South Africa near the confluence of the Shashe and Limpopo Rivers, for the period between 750 - 1050 AD (Huffman 2007:145).The Eiland - style rim sherd in particular deserves some attention for its unusual thinness (as well as its unexpected presence in an otherwise uniform assemblage). While the rest of th e decorated pottery sherds were about one to two centimeters in thickness, this sherd is about three millimeters thick. Its presence as an isolate raises a few questions about both exchange and site formation ( Figure 14 ). The predominance of Zhizo - type sherds at Thabadimasego is consistent with both the AMS dates obtained from the sit e (see Table 5 , this chapter) and with contemporary, regionally contiguous sites in eastern Botswana, including Bosutswe lement, Mosu I, as well as Kaitshàa (Reid and Segobye 2000; Denbow et al. 2008; Denbow et al. 2015). A Zhizo assemblage could indicate cultural ties to the east (rather than to the west, with the Kalundu Tradition and with the Tsodilo and Okavango sites. T he Kaitshàa sherds, although far fewer in number (and from a less reliable surface provenience), had a similarly consistent range of decoration types. About half of the sherds examined were designated as (Huffman distinguishes these as separ ate facies, but some disagree - Denbow, personal communication) with the remainder designated either as Zhizo (6) or unknown (3). What little has been published about Kaitshàa indicates that its phases of occupation were longer than those at Thabadimasego - its deposits are much deeper and contain greater array of materials, and are associated with a wider range of dates (Denbow et al., in press). Despite the scarcity of definitive information of this type for sites in the South Sowa area, the consistencie s in pottery types (as well as radiocarbon dates) support the idea of a related group of settlements occupying the area during the late first millennium AD. 101 Overall, the pottery from Thabadimasego is quite consistent in style. The few sherds that are inconsistent with the chronological or geographical ranges of Zhizo pottery could benefit from further analysis. Their presence could indicate trade activities, re - occupation of the site at a later date, or some kind of intrusive element. Since, howeve r, they represent such a small percentage of the assemblage, their presence does not much influence the picture of Thabadimasego as a late - first - millennium AD occupation with a Zhizo cultural affiliation. 102 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000 Pottery, Undec Body, mass (g) Pottery, Dec Body, mass (g) Pottery, Undec Rim, mass (g) Pottery, Dec Rim, mass (g) Aggregate mass (g) Types of pottery present at Thabadimasego Figure 12 Summary of pottery types present at Thabadimasego 103 Figure 13 Examples of Zhizo pottery from Thabadimasego 104 Figure 14 Pottery designated as Ziwa (L) and Eiland (R) from Thabadimasego 105 Table 1 Counts of sherd types Decorated body sherds Decorated rim sherds Total Thabadimasego units 216 19 235 Thabadimasego STTPs 37 5 42 Kaitshàa 0 20 20 Total 253 44 297 Table 2 Zhizo sherds Zhizo Probable Zhizo Other Unknown DB ( Thabadimasego ) 4 242 1 6 DR ( Thabadimasego ) 8 14 1 1 Thabadimasego total 12 256 2 7 Kaitshàa 6 11* 0 3 * or Leokwe 4.3 Shell bead analysis Ostrich eggshell (OES) beads are typically associated with hunter - gatherer production and use for both archaeological and contemporary populations. OES beads appear in assemblages in both East and Southern Africa as early as the Middle Stone Age and are a common find in Later Stone Age assemblages across S outhern Africa (Walker 1998; Deacon and Deacon 1999:107 - 127; Mitchell 2004). The oldest known OES bead In Botswana is a broken blank from Later Stone Age deposits at the Tsodilo Hills White Paintings Shelter. The broken blank was directly dated by AMS to 2 4,510±300 BC (Robbins 1999:11 - 16). Historical and present - day San communities of Botswana and Namibia have likewise been known to make and wear OES beads strung as necklaces, headdresses and other forms of adornment ( L ee 1979 ; Howell 2000 ). Today, shell bead ornaments are primarily worn in special circumstances as part of to tourists. Examples of some OES and Achatina beads can be s een in Figures 15 and 16, below . 106 Sh ell beads continue to make regular appearances in LSA assemblages during the later Holocene after the introduction of food production, and are a component of many herder and Early Iron Age assemblages as well (Smith et al. 1991; Tapela 2001; Dubroc 2010). This has given rise to the question of what their presence indicates in the context of herding and farming tradition. Jacobson (1987) created a provisional typology of OES beads based on bead diameter for several sites dating w ithin the last 3,000 years in central Namibia. He distinguishes three types of beads: one, the smallest, associated with pre - herding LSA assemblages; another associated with LSA assemblages contemporary with herding; and a third, the largest, associated wi th herder assemblages themselves. Jacobsen draws the further conclusion that style differentiation of bead adornments between the communities may have been a form of persistent social identity differentiation as the two communities coexisted over the centu ries (1987:56). A. B. Smith et al. (1991) draw a similar conclusion for hunter and herder signature assemblages for several sites on the South - Western Cape of South Africa. They also note that size is a mong the distinguishing characteristics of the OES beads; hunter - gatherer sites have beads with small diameters while herder s ites have large beads. Their study, which comprises a series of sites spanning the last three millennia before European settlement, yielded a picture similar to Jacobsen groups with the maintenance of separate cultural identities: hu nters and herders who existed side by side in the south - herders would have maintained separate social and ethnic identities during prehistory, despite their co - occupat ion. This is based on the notion that a herding way of life involves socioeconomic factors such as individually - (A. B. Sm ith 1990) . The bead - size and subsistence - identity line of inquiry was picked up by Tapela (2001) for OES bead assemblages in Botswana for both LSA and Iron Age contexts. Helpfully, the article reviews much relevant 107 literature on OES bead production and presence/f requency in various contexts for the last 2000 years. Tapela (2001: 62) d escribes the following general trends: Southern Africa during the last 2000 years, ostrich eggshell bead sizes were a marker of ethnic and economic identity (southwestern Cape and Namibia). In some parts bead sizes may have increased in time as fashions changed (Namibia). Elsewhere, ostrich eggshell beads seem to have been exclusively produced by Late Stone Age hunters and trade d to Iron Age herders (Natal). In other parts ostrich eggshell beads may have been important trade items produced by both hunters and herders (Botswana). In some places bead sizes changed but the significance and meaning of the change remains enigmatic (Zi Mogapelwa, a site in northwestern Botswana near Lake Ngami. As Robbins et al. (2009) state, 2618±43 BC . This large bead substantially pre - dates the arrival of domesticated livestock in Southern Tapela examined 819 beads were examined from three L ater S tone A ge sites and 5 Iron Age sites, including Kaitshàa and Mosu I in the South Sowa area. He drew a similar conclusion to those of Jacobsen (1987) and Smith et al. (1991), seeing a statistically signific ant distinction in bead diameter between L ater S tone A ge These results are not universally recognized, however. In another study of several sites on the Geelbek Dunes in Namibia, Kandel and Conard (2005) found a less - robust trend towards smaller OES beads associated with earlier sites, but also that the tool and faunal assemblages associated with later sites (and their larger beads) obfuscate who was using the beads: at t he time of their publication, no domesticated fauna was found in association with the larger beads, and lithic tool assemblages presented similar types and frequencies as at smaller sites. While overall the authors concluded that Jacobsen model could not be viewed as universal, Kandel and Conard maintain the fundamental connection 108 Figure 15 OES beads from Thabadimasego 109 Figure 16 Achatina beads from Thabadimasego 110 between hunter - gatherers and small - beads, and the appearance of larger beads with settlement of the area by pastoralists (2005:1720). Finally, Dubroc (2010) situated the shell bead assemblage from Bosutswe, an Iron Age site, within this discourse. Bosutswe is a hilltop site in eastern Botswana whose occupation spans roughly a millennium from 700 - 1700 AD; this period of ti me includes both early and later manifestations of the Iron Age (Denbow et al. 2008). Like Tapela (2001), this study is concerned with addressing the question of whether bead diameter correlates with other material types significantly enough to make interp retations about the economic practices, and thereby the ethnic identities, of the beads 2010:48). Importantly, Dubroc examines not only the ostrich eggshell beads and fragments, but other species of shell as well including river m ussel and Achatina (a genus of land snail native to many parts of sub - Saharan Africa, including Sowa Pan - watered parts of Botswana). Dubroc finds that every excavation level contained a variety of sizes of OES beads, with "no clear patterning in time or provenance," (2010:48) despite the fact that Bosutswe - dated soil horizons indicate an undisturbed depositional history (Denbow et al. 2008). Furthermore, several units on the site contained beads at different sta ges of production, also in a range of sizes, indicating that beads of all size were produced on the site itself. Dubroc (2010:48 - 49) makes the further observation that: e the most uniform and polished. This suggests that they were worn for a long period of time. I would suggest that these beads were curated and worn by multiple generations before being lost to the soil. For example, a present - day analog would be that in s ome Southern African Bantu groups mothers Southern Africa fail to acknowledge the role of many artifacts as heirlooms for future generations. One possible reason could be that curated items would test the utility of tight - knit stylistic 111 chronologies based on bead size. Therefore, explanations concerning the ethnic or cultural makeup of matter - of - fact. The situation becomes even less clear when one factors in the presence of long distance trade and the possibility of a multi - . Achatina species are found widely in Southern Africa. The empty shells dot the Mosu escarpment today, and are sometimes picked up by local residents for use as decorations on house exteriors (personal observation). It is unclear how much of the shell bead assemblag es studied by, e.g., Tapela (2001) may in fact be Achatina or in Tapela (2001) and Kandel and Conard (2005). To judge from both the Bosutswe and the Thabadimasego shel l assemblages, Achatina may not be uncommon (at least in Botswana), and other species of shell such as river mussel may be included as well. van Zyl et al. (2013) also makes mention of the presence of Achatina remains among faunal assemblage in the E arly Iron Age levels at Xaro I, a site on the Okavango Delta in northern Botswana, and notes that they may have been both exploited as a food source as well a s a known source of bead material. Achatina species in Ghana are, in fact, farmed today and sold as a protein source ( A. Logan, personal communication, 2015). Whether different species of shell served different roles or functions in Southern African Early Iron Age contexts, however, remains unclear until further studies address these issues. Shell components of Iron Age assemblages in general are underreported in that, while shell finds may be included in the initial catalogs and site reports from an excav ation, they tend to be excluded from the broader behavioral and cultural interpretations. This is evident from the major survey literature such as Mitchell Southern African Archaeology ; Mitchell and Whitelaw Southern mo (2007) Handbook to the Iron Age . Despite this, it seems apparent that the frequency with which they do occur in the Early Iron Age (as reported in Dubroc 2010) merit s their consideration as social or economic 112 objects for Early Iron Age communities in their own right, rather than just evidence of hunter - gather presence. Methods In light of this, data were collected from the Thabadimasego shell beads following a similar procedure to the one described in Dubroc (2010). Each bead or fragment was identified as to taxa (ostrich, Achatina , mussel, or other) using a number of defining attributes such as color and texture. Bead diameters were mea sured to the nearest tenth of a millimeter (although previous studies used whole millimeters, it was felt that the very small size of the beads (many were 3mm or less in total diameter) warranted a smaller unit of analysis. Unworked shell fragments were me asured as well; as these were generally rectangular or triangular in shape, the maximum length and width of each fragment were recorded. Beads were marked as whole or broken, and also characterized by their condition, as either blanks (bead - shaped with sha rp edges and no drill hole), jagged (sharp edges with a drill hole), rough (edges smoothed but still irregularly shaped), or finished (edges smooth and circular). Finally, the overall condition of each bead or fragment, whether burned, chipped, worn or oth erwise, was noted. For the full resul ts of the data collection, see Appendix B . Shell beads results Following classification of the shell beads, exploratory data analysis (in the sense of Tukey 1977) was performed to illuminate trends in the data. Based on the results of prior studies (e.g., Jacobsen 1987; Tapela 2001; Dubroc 2010), particular attention was paid to the internal and external diameters of each bead. These two measurements were converted into a ratio for each specimen so as to compare with oth er attributes such as degree of completion and burning. Boxplots were produced for bead ratios ( by completion, burning, and material type ) in order to display the distribution of data as well as identify outliers. Scatter plots were likewise produced for t hese combinations of variables. Only a few trends were clearly visible in these data displays: primarily, that two overlapping but distinct distributions of bead 113 diameter ratios existed for ostrich and Achatina beads. It also became clear that degree of co mpletion (whole or broken) required further subdivision as to whether broken beads had more or less than 50% of the bead present; less than 50 % present meant that bead diameter ratios could not be measured since the total width of the bead fragment was a c hord of the circumference but not the bead The ratio of external to internal diameter was calculated for each bead because of the sizable differences in internal diameter lengths for beads of the same external diameter: beads with larger i nternal diameters have a much lower volume (i.e., they contain substantially less raw material) than those with smaller internal diameters. Presumably these beads were worked longer than the small - internal - diameter beads, and they could be strung on a wide r variety of thread/ fibers. It is not known at this stage whether these distinctions are behaviorally or socially significant for Later Stone Age or Early Iron Age peoples. However, as they have the potential to inform variations in both production and us e of the beads, these data were collected. Regardless of which metric (external diameter only, or external - to - internal ratio) is used, the differences between ostrich and Achatina shell beads remains statistically significant. For the sake of using the m ost conservative measure, diameter ratio is used in this study for discussion of results. For further analysis, the following specimens were excluded from t he sample: 1) shell beads from S ites 12 and 33 (so as to narrow focus to a single, well - re presented site); 2) beads with less than 50% diameter present; 3) irregular or jagged - edged (unfinished) beads; 4) unshaped fragments; and 5) beads of the representative). For the remaining sample (which comprised ostrich and Achatina shell beads, burnt or unburnt; either whole or greater than 50% present, from Thabadimasego ), the specimens were subject first to tests of equal variances and then a student t - test (assuming either equal or unequal variances, depending on the results of the first test) in the following combinations of variables: Material type - diameter ratio 114 Material type - degree of bead completion Material type - burning present Diameter ratio - burning present ( ostrich shell beads only) Diameter ratio - burning present ( Achatina shell beads only) Diameter ratio - degree of bead completion (ostrich shell beads only) Diameter ratio - degree of bead completion ( Achatina shell beads only) The tests of equal vari ance as well as the t - tests initially used only whole beads, but were subsequently re - run to include both whole beads and those with greater than 50% present. No difference s w ere found in the results between these two groups, therefore the results discussed below rely on the larger sample (including the broken, greater - than - 50% beads). Summary of findings Significant results (p < .05) from the t - tests were obtained for material ty pe vs. diameter ratio, and for diameter ratio vs. burning present (for Achatina shell beads only). Degree of completion did not make a difference: whole and broken beads for both OES and Achatina shell had similar distributions. Burned beads had a differen t distribution for Achatina , but not for ostrich eggshell. Considering that only 47 burnt Achatina beads were present in the sample, it i s not easy to conclude that this differentiation means anything ; further comparison with other as semblages would be nee ded. On the whole, material type clearly accounted for the greatest amount of variation. Achatina shell bead ratios have a smaller mean as well as a smaller range of distribution than do ostrich eggshell bead ratios, although a histogram of their distribut ions demonstrates considerable overlap. The bead size distributions for each type of shell was much more distinct when plotted either for internal or external diameters only, but regardless of which metric was used, Achatina shell beads showed a continuous distribution of sizes, while ostrich eggshell beads showed a somewhat bimodal distribution. It remains difficult to contextualize these findings within comparable Southern African 115 assemblages, since only one oth er study (Dubroc 2010) addresses the presenc e of different species of shell among the beads. In terms of bead size, Dubroc follows Tapela (2001) in distinguishing between - ones (greater than 6 mm in diameter). The ostrich eggshell beads at Bosutswe fall i nto both of these categories with a fairly continuous distribution in diameters (2010:34), which speaks to the fairly arbitrary division between sizes. Dubroc unfortunately does not go into detail about the range of diameters for Achatina shell beads. As f or shell items that are not finished beads, the quantities of bead blanks and unworked fragments appear to be much lower than in the Central Precinct of Bosutswe, where these numbered in the hundreds (Dubroc 2010:38). The Central Precinct at Bosutswe is, f urthermore, centuries younger than Thabadimasego according to its radiocarbon determinations (Denbow et al. 2008), so it does not make for a close comparison. On their own, the ostrich eggshell blanks and fragments of Thabadimasego do not appear to indicat e any kind of long - term bead manufacturing activity. Their presence in small and scattered quantities could, however, be indicative of occasional bead production occurring as an itinerant activity on the site. One further observation can be made about the shell bead assemblage. Whereas, at Bosutswe further to the south, Achatina beads would have been imports (Dubroc 2010:45), the ready availability of empty Achatina shells dotting the Mosu landscape speaks to the local availability of this species for occupants of Thabadimasego . 4.4 Metal analysis Metal objects and their fabrication, as hallmarks of culture types, have been studied from a number of angles since the inception of the presently - accepted Southern African chronological sequence (Phillipson 2005). Today, the analysis of metal artifacts may include chemical and micro - morphological studies (Miller and van der Merwe 1994; Miller 2001; Miller 2002; Miller and Killick 2004; Killick 2006; Killick 2009) , examinations of the social contexts of metallurgy such as craft specialization 116 Figure 17 Diameter ratio for shell beads at Thabadimasego 117 Figure 18 External diameters of shell beads at Thabadimasego 118 Figure 19 Internal diameters of shell beads at Thabadimasego 119 (Chirikure 2007) and its role in ritual and political processes (Calabrese 2000 ; Huffman 2001; Chirikure 2004), and comparative analyses of the spatial dimension of metal production (H. J. Greenfield and Miller 2004) . For a review of general trends in the development of metal use in sub - Saharan Africa, see Childs and Herbert (2005 ) and Killick (2006, 2009). Metal items overall comprise a relatively small component of most Sout hern African assemblages, with decorative objects (beads, bracelets, earrings, etc.) representing by far the most frequent types of metal finds (Denbow and Miller 2007) . Agricultural im plements such as hoes or adzes, or other tools like spears are far less common, although the comparative contexts in which these categories of metal objects have been recovered may be a useful avenue of research in the future. Metal production in archaeolo gical contexts has been associated widely across sub - Saharan Africa with social and ritual symbolism concerning both fertility and danger (Mitchell and Whitelaw 2005; Chirikure 2004; Chirikure 2007) . Given the lack of d irect historical connection, it may not be productive to try to define specific associations of symbolic meaning or social function for the Early Iron Age . Yet it can still be recognized that metal production would have had spatial, social, functional and economic dimensions as well as technical aspects, such as ore sources and production techniques (Killick 2009). M ethods The goal of analysis of the metal objects from Thabadimasego was fairly straightforward. Items were to be ident ified according to metal content and sorted by form, so that the assemblage could be compared in type and variety to other documented metal assemblages of the Early Iron Age. While a great degree more work can be conducted on this assemblage in the future ( for example, X - ray fluor escence can identify chemical makeup, and thin - sectioning and micrography can identify smithing and smelting techniques), it is sufficient for the goals of this research project to describe and characterize the styles of metal artifacts. Doing so contributes to the overall picture of economic behavior at Thabadimasego and helps pla ce the site in its broader cultural framework. 120 Two hundred and sixty metal objects from the Thabadimasego deposits were examined ( Figure s 20 and 21 ). This metal assemblage includes both ferrous and non - ferrous (presumably iron and copper, respectively) obj ects, and both worked items (such as beads and wire) and by - products like slag. No formal metallographic or chemical analyses were conducted on this assemblage to verify the contents or structure of the objects; a distinction between ferrous and non - ferrou s was made based on macroscopic attributes like color, texture, and corrosion, as well as presence or absence of magnetism. Objects were tallied by both count and mass (in grams). High degrees of corrosion in some of the items (especially the amorphous iro n fragments) as well as the friability of the slag necessitated that these items were better quantified by mass, in lump sums by lot (rather than by individual pieces). These items were too easily fragmented both in situ and during analysis for a per - objec t count to be meaningful. However, for the sake of comparison, counts and mass were recorded for all classes of metal objects. The general shape and color of each object were des cribed and the presence or absence of magnetism and corrosion were also noted. For the worked objects (beads, wire, rods/ bars), length, width and height (in millimeters) were also recorded . Consistent trends in shape recurred among both the iron and copper beads, s o in order to describe their morphology adequately, additional measurements (such as thickness and gap opening) were collected for these items. These shape categories included horseshoe, c - shape, butted circle, and other types which are characterized fully in Appendix C . Since the availability of iron and copper items in the Early Iron Age depended on different ore sources as well as a set of different production techniques (Miller 1995), taking a look at the relative proportions of each metal present in t he Thabadimasego assemblage is of some interest, especially in how they compare to those of other contemporaneous sites. In general, iron was better represented than copper in terms of both mass and object counts. The total mass of copper (or at least non - ferrous) slag outnumbered iron (magnetic) slag by almost 3:1, the only instance in which copper represented a greater aggregate sum than iron. However, this observation should not be granted too much importance, since 121 as a waste by - product, slag can vary g reatly in its makeup and non - magnetic slag may full well be a by - product of iron ore reduction (Ironbridge Gorge Museums Trust 2009) . The overall quantity of both ferrous and no n - ferrous slag is furthermore very small in comparison to the quantities recovered at sites which are thought to have been loci of actual metal - working: the Thabadimasego slag can be measured in tens of grams, while slag mass at, e.g., Divuyu or Ndondondwa ne is counted in dozens or hundreds of kilograms (Miller and Killick 2004). No other corollary evidence of metal - working, such as furnace remains, was likewise recovered at Thabadimasego and so what little slag was there presents something of a mystery. Fo r these reasons, the slag has been considered separately from the worked metal items (which identify as iron and copper, by their color and texture, much more readily). The total quantity, either in terms of mass or count, of worked items (beads and wire/ bars fragments) is not particularly high. 76 beads and 21 pieces of wire or bar were recovered, totaling 27.5 grams and 24.3 grams respectively. Another 49 pieces (by best approximation of the highly fragmented objects) of amorphously - shaped iron, some of which may have been wire or bar prior to corrosive degradation, totaled 15.1 grams. The shape of these gives no indication of their function however (or even whether they were end products or simply by - products with high iron content). Among the 7 6 beads, seven distinct bead shapes were identified. Miller (2002) and Denbow & Miller articles describe and provide illustrations for a wide range of metal too ls and jewelry recovered from Bosutswe, Divuyu, Nqoma, Broederstroom, and Mapungubwe, among other places. One of the bead shapes identified in the Thabadimas - shaped in the tables) may in fact consist of wrapped - wire nec klace or bracelet fragments, but without more intact versions of these objects with which to compare it remains difficult to tell. Four of the seven bead shapes were present in both the iron and shape appears only in ir flat horseshoe shapes are only present in copper. Each bead is quite small and the average 122 Figure 20 Copper beads from Thabadimasego 123 Figure 21 Copper items from Thabadimasego 124 mass per bead (both iron and copper) is about .35 grams. Iron beads generally outnumber copper ones, as with the rest of the metal assemblage, with a ratio of about 3:2 (for both count and mass) inclusive of all bead shape s. The wire and bar fragments present a somewhat different picture. In terms of count, iron exceeds copper by six to one. However, in aggregate mass the two metals are nearly equal. Most of the iron wire/ bar pieces are quite small (averaging about 16 mm l ong by 5 mm wide), while the great majority of the copper wire bulk consists of one well - preserved coil. Results Overall, the types of metal present at Thabadimasego - iron and copper - and their worked forms, including beads or bangle fragments, wire, an d flat hammered bar fragments, are very similar in kind to other metal assemblages of the Early Iron Age, including Bosutswe and Nqoma. The flattene d copper link - shaped beads (F igure 20 above) bear a striking resemblance to similar objects from Broederstroom, a site whose occupation may be as early as the fourth century AD (Miller 2002:1087). However, the copper chain from Broederstroom is of questionable provenience and trusive from a much later occupation of the Mason, 1986:143 in Miller 2002). The disparity in occupation dates between the two sites is therefore not of much concern. The thicker and more rounded cuff - style beads from Thabadimasego are comparable i n form to those from Bosutswe, a site whose early occupation phase was contemporaneous with Thabadimasego (Miller 2002:1 088; Denbow and Miller 2007:281 ). Miller Bosutswe include iron and copper as well as bronze a nd alloy beads. The Thabadimasego beads of this style were sorted as both iron and copper, but further metallurgical analysis may indicate the presence of a wider array of metal types. The presence of minute fragments of metal slag (some of which contains ferrous material, some of which apparently does not) is difficult to explain ( Figure 22 ). A few mere grams of slag cannot readily be explained in functional terms, especially without any further evidence of metal - working activities (like 125 tuyere remains or a forge base) in the area. Thabadimasego appears to be unusual in this regard, and some kind of ritual purpose may be behind the presence of the slag (Gavin Whitelaw, personal communication, 2013). Given the well - documented associations between spiritual potency and metal - working for Bantu - speaking communities in Central and Southern Africa during the historic era (see, e.g., Killick 2004), this is not outside the realm of possibility. However, this issue remains better left to future research. It would be of interest to explore whether other sites as small as Thabadimasego that may have served some kind of peripheral supporting role to actual villages also include similar assemblages. 4. 5 Faunal analysis Faunal analysis comprises a critical part of the interpretation of Early Iron Age lifeways, as it may provide information on not only overall diet but herd management strategies in the case of livestock (Kinahan 199 5; Denbow et al. 2008), insight into the ratio of wild to domestic animal exploitation (Atwood 2005), and in some cases even social and political organization (Huffman 2001; Badenhorst 2009). As discussed in chapter 2, domestic animals were introduced to Southern Africa by 0 AD as can be seen by the faunal remains found at several sites, including in northern Botswana at Toteng, as well as at Leopard Cave in Namibia (Robbins et al. 2005; Plerdeau et al. 2012). However, what the presence (or absence) of various species of fauna a ctually implies for the community exploiting them may vary depending upon the context. For example, it is highly debatable whether or not the fundamental symbolic and social importance of cattle as described in Kuper Southern Bantu Cattle Pattern may be interpreted for - documented existence among Nguni communities (Badenhorst 2010). The relative importance of cattle specifically in an E arly Iron Age eco nomy has also been debated over the years. Huffman (2001:30) argues that cattle are under - represented in the faunal record: Moreover, if there was one cow, the re had to have been at least 100 in the area in order for the 126 herds to reproduce ( Huffman 2001 ). Clearly, it is not possible to determine herd size or infer cultural importance directly from faunal remains alone. Ethnographically - derived settlement models presence of cattle in first - millennium sites based on proxy evidence like dung (Huffman 2007:8; Huffman and Schoeman 2011) , and the circular arrangement of structures at villages akin to Kuper social, as well as economic, significance of cattle throughout all periods of the Iron Age in Southern Africa. M. Hall (1986) offered an early critique of the ated to cattle - keeping, pointing out the absence of any cattle remains in the earliest agro - pastoral settlements of Southern Africa as well as the high ratio of marine resources found at coastal settlements for during this time. In more recent years, Baden horst (2009) challenges the use of dung and dung - based phytoliths as proxy evidence for cattle specifically, pointing out that sheep dung in fact decomposes more quickly than that of ca ttle. Badenhorst (2011) has developed based on the num ber of individual specimens present ( NISP ) proportional to that of caprines in order to document the changing frequency of catt le at Iron Age sites over time, and has shown that Early Iron Age sites generally had a significantly lower ratio of cattle to ca prines. However, this study does not appear to account for, nor comment on the importance of the presence or absence of, wild faunal remains. The presence of wild fauna at Early Iron Age sites in many parts of Southern Africa may be seen as an important di etary component (Gilbert Pwiti 1996; Sadr and Plug 2001; van Zyl et al. 2013) . Denbow et al. (2008:469), for example, report that upwards of 60% of the total faunal assemblage for the Zhizo horizons of the Western Precinct at Bosutswe was comprised of hunted game. Van Zyl et al. (2013), in their analysis of faunal remains from Xaro, a cluster of sites near the Okavango Delta whose occupation sequence spans the last two millennia, find a significant wild animal presence among ceramic and livestock 127 remains. Included in their findings are Achatina and mussel shells . T he authors note that both species could have been exploited as a protein resource, but also make mention of the use of Achatina shells in bead production (2013:56). Their conclusions overall suggest that, at least for open - air sites in the Okavango area, assemblages do not group easily into discrete forager and food - producer categories. The presence of carnivore or non - hunted animal remains (such as leopards), too, may indicate the exploitation of these animals for their fur, ivory or other luxury items produced for trade, as is the case at Schroda (Plug 2000) , although these animals may have been a food source as well. On the other hand, Badenhorst (2012:266) ascribes ri tual significance to smaller quantities of wild fauna. For the Thabadimasego faunal assemblage, goals for data collection included taxon identification Figure 22 Slag fragment from Thabadimasego 128 as well as evidence for dietary exploitation of those species (such as butchering or herd management). T he entire assemblage of faunal remains was provided to Lu - Marie Fraser, a graduate student at the University of Pretoria working under the supervision of Professor Shaw Badenhorst, for species identification. Unfortunately, due to time constraints, Ms. Fra ser was unable to provide a summary of evidence for butchering. However, this information will be available in her forthcoming thesis (Fraser 2015, personal communication). Methods Ms. Fraser kindly provided a summary of the methods she used for data coll ection on the faunal assemblage, which is reproduced here: manual I will utilise and then the specifics of the manual and what each aspect that can be identified regarding fauna can tell us. The manual I will follow is a manual that Driver (1991) set up for the description of vertebrate remains for the Crow Canyon Archaeological Center and which Badenhorst then adapted in 2009 (hereafter called 2009 adaption). I wi remains from the four Iron Age sites as I found it to be the most complete manual and as mentioned before, the 1991 manual was compiled to add a degree of standardisation and a reduction in errors to zooarchaeology (Driver 1991). describe the faunal remains. The two main categories will be identifiable and non identifiable specimens. Refitting will be attempted where possible, on both non ide ntifiable specimens and identifiable specimens, where refitting was not already done, unless the specimens are too fragmented . As suggested by Driver (1991), comparative collections, published guides or keys and measurement systems will form part of the me thodology. The comparative collection at the Ditsong National Museum of Natural History (Archaeozoology and Large Mammal Section) will be utilised. The measurement systems will follow Von 129 den Driesch (1976) and Peters (1985 86), which will also serve as pu blished and illustrated guides, along with any other guides that may need to be consulted. All relevant specimens will also be digitally photographed using scales to accurately document the specimens. sible by species (or family/group), part (what portion of the element is available), sex, length, measurement, and side using a code system. Fusion, breakage pattern, any modification (fashioned into an artefact or tool), taphonomy (carnivore and rodent gn aw marks, trampled, digested, cut and chop marks), burn intensity (for example black, grey, localised), pathological condition and age will be indicated where possible on both identifiable and non - identifiable spe cimens using a code system . These categorie s can help with the identification of hunting, butchering and cooking techniques. These will be explained more in detail with the recording of identifiable specimens below. - . As mentioned previously, Voigt (1983) used categories in addition to the above mentioned three, such as fragments are part of identifiable specimens (example in Badenhorst & Plug 2004/2005:11; Badenhorst & Plug 2011:78; Driver 2005; Driver 2011:25:28) and will be documented according to size/taxon and any other identifiable characteristics where possible. Results No clear pattern was evident for either type or distribution of the faunal remains according to the preliminary information gathered from the assemblage prior to sending it to Pretoria for identification. A good deal of the assemblage is highly fragmented into small pieces. Very few of the bones appeared burned and none were articulated in situ. Larger bone fragment s often occurred in a context alongside 130 ceramics or shell beads; however (as Chapter 5 discusses), small quantities of bone are also scattered fairly widely across the site and no clear pattern of distribution is apparent among them. Bone fragments were, i n fact, among the most ubiquitous type of material encountered on the site (second to undecorated Following is the list of species present in the Thabadima sego preliminary report. Table 3 List of taxa present in assemblage Taxon Common name Type Homo sapiens sapiens Human Human Bos taurus Cattle Domesticated cf Bos taurus Cattle Ovis aries Sheep cf Ovis aries Sheep Ovis/Capra Sheep/ goat cf Ovis/Capra Sheep/ goat Viverridae sp. indet Genet or civet (species indeterminate) Wild Loxodonta africana African bush elephant Equus quagga Zebra cf Potamochoerus porcus Warthog Raphicerus campestris Steenbok Aepyceros melampus Impala Syncerus caffer Cape buffalo Struthio camelus Ostrich Achatina sp. Land snail cf Achatina sp. Land snail cf Geochelone pardalis Tortoise Bufo/Rana Frog or toad Insectivora sp. indet Insect ivore (species indeterminate) Carnivora medium Carnivore Rodent small Rodent Rodent medium Rodent Lagomorpha Hare Aves small Bird Aves small - medium Bird Aves medium Bird Tortoise Tortoise Reptile small Reptile Sauria Lizard 131 Taxon Common name Type Terrestrial gastropod Snail Mollusc Mollusc Bovid I Cloven - footed, ruminant mammal Unknown Bovid II Cloven - footed, ruminant mammal Bovid II - non domestic Cloven - footed, ruminant mammal Bovid III Cloven - footed, ruminant mammal Bovid III - non domestic Cloven - footed, ruminant mammal Bovid III/IV Cloven - footed, ruminant mammal Mammal small Mammal Mammal medium Mammal Mammal large Mammal Table 3 (c o n ) 132 Table 4 Summary of NISP (Number of individual specimens) present in assemblage Taxon NISP observed Domestic Bovids 36 Wild Bovids and Equids 16 Indeterminate Bov I 18 Indeterminate Bov II 229 Indeterminate Bov III 162 Indeterminate Bov II - Non Domestic 1 Indeterminate Bov III - Non Domestic 1 Indeterminate Bov III/IV 6 Indeterminate Mammals 371 Carnivores 1 Birds 7 Rodents 164 Reptile 3 Human 2 Various other small animals (lagomorph, tortoise, Achatina etc.) 92 Total NISP 1109 133 Domestic (36) 3% Wild (285) 26% Indeterminate (786) 71% Human (2) 0% Total NISP observed at Thabadimasego Figure 23 NISP percentages 134 Discussion The presence of human remains among the faunal assemblage was a complete surprise. After further inquiry, it was made clear that the human remains were represented only by two teeth (Fraser 2014, personal communication). Apart from the human remains, little that is out of the ordinary appears in the assemblage. A combination of wild and domesticated bovids (cattle, sheep, goats, antelope) was to be expected based on reports from contemporary faunal assemblages from around Botswana and South Africa (Denbow 1999; Denbow et al. 2008; van Zyl 2013). Species diversity for wild animals is, in present day, low for South Sowa due to the extensive use of land in the area for arable f arming and grazing pasture . Hundreds of kilometers of fences have been built to control cattle herd movements as well as erosion; these too have seriously affected wild herd migrations in the area ( Department of Environmental Affairs and Centre for Applied Research 2010) . However, the Makgadikgadi Pans overall today continues to host populations of many of the taxa enumerated in the faunal analysis, including zebra, land snails, warthogs, impala, and a wide variety of birds and rodents. Elephants, on the ot her hand, are more likely to be encountered further to the north near the Chobe Game Reserve. Various taxa of small ve rtebrates, such as the rodents and possibly some of the birds, may be intrusive remains from post - occupation depositional processes and ro dent burrows. Their presence in the assemblage should not be taken as indicative of the exploitation of these taxa for dietary purposes, assemblage at Xaro, it is also possible that Achatina (land snail) and other gastropods and molluscs were used as a food source in addition to the manufacture of beads out of their shells. Achatina in particular has been cited as a commercially farmable food for present - day c ommunities in West Africa (Hardouin 1995) . Given the small space that Thabadimasego occupies, and given the shallowness of deposits on the site, the presence of over 1,000 individual specimens seems like a relatively large amount. As is discussed 135 in the spatial analysis section, excavations at the site produced evidence for no more than one or two mudbrick structures. The site is therefore too small to be a village, where the accumulation of such a number of faunal remains could be expected. Unfortunately, details about the specific provenience (unit/ pit and stratum) of any particular taxa are not yet known 4.6 Radiometric analysis Accelerator mass s pectrometer (AMS) analysis Out of the several charcoal samples recovered during excavations, three were chosen to submit to the NSF - Arizona AMS Laboratory at the University of Arizona. Unfortunately, the funds available for AMS dating w ere quite limited, so priority was given to the samples that, taken together in context, would provide the most comprehensive overview on the deposi tional history of the sites excavated. Two of the samples, 16 - A1 - 13 - U3 - L2 and 16 - A1 - 13 - U19 - L4, were chosen from central units in the main deposition area of Thabadimasego . Unit 3 Level 2 contained part of the compacted soil horizon found just below the sur face on the central area of Thabadimasego and was also associated with a cluster of ostrich eggshell beads and faunal remains. Unit 19 Level 4 contained an expanse of collapsed mud - brick structure wall as well as one of the largest charcoal concentrations at the site. The third sample, 16 - A1 - 12 - U1 - L2, was collected from the compacted soil horizon in the test unit at Site 12, which also contained a number of ceramic and bone fragments. This sample was submitted in the interest of having a comp arative date from an analogous soil horizon at the E arly I ron Age site nearest to Thabadimasego . Upon receipt of the uncalibrated dates, OxCal ( https://c14.arch.ox.ac.uk/oxcal.html ) was used to calibrate the dates using the Southern Hemisphere 13 calibration curve (Hogg et al. 2013) . 136 Table 5 AMS Dates from Thabadimasego and Site 12 Sample Median 16 - A1 - 13 U3 - L2 (AA101289) 860 - 970 AD 774 - 985 AD 897 AD 16 - A1 - 13 U19 - L4 (AA101290) 864 - 971 AD 775 - 985 AD 906 AD 16 - A1 - 12 U1 - L2 (AA101288) 900 - 1014 AD 892 - 1020 AD 967 AD (Arizona sample numbers in parentheses) Figure 24 SHCal13 calibration curve in relation to U19 - L4 charcoal sample AMS results 137 Optically - stimulated Luminescence (OSL) analysis Two OSL samples were collected from two separate soil horizons at Thabadimasego , as detailed in Chapter 3. The soil horizons from which the OSL samples were taken are similar to those from which the charcoal samples used for AMS dating came from. One AMS and one OSL sample each came from the highly compacted upper soil horizon, and one of each type also came from the less - compacted horizon further down (see Figure s 44 and 45, Chapter 6, for an example of this stratigraphy) . The OSL samples we re sent to the University of Washington Luminescence Laboratory and processed by Dr. James Figure 25 SHCal13 calibration curve in relation to U3 - L2 charcoal sample AMS results 138 Feathers (Feathers 2003)(Feathers 2003) (Feathers 2003) (Feathers 2003)(Feathers 2003)(Feathers 2003)(Feathers 2003)(Feathers 2003)(Feathers 2003)(Feathers 2003)(Feathers 2003) . His full report, including the tables and figure s mentioned below , is attached as Appendix D. The following description of the methods used in OSL analysis is cited from this report (Feathers 2014) : Sample material is removed from the collection container, leaving asid e any portions (such as the ends of tubes) that may have been exposed to light. The latter may be used for dose rate information. From the unexposed portions, about ¼ is set aside as a voucher (material that can be used at a latter date if necessary). I f separate samples for measuring moisture content have not been collected, the voucher can be used for this. For moisture the sample is simply weighed wet, and then dried for several days at 50°C before weighing again. The wet minus the dry weight divide d by the dry weight gives the percent moisture by weight. abundant silt or clay, the sample is wet sieved through a 90 µm screen. Otherwise it is sieved dr y. The greater than 90µm fraction is treated with HCl and H 2 O 2 , rinsed three times with water and dried. It is then dry sieved to retrieve the 180 - 212 µm fraction (or any other fraction deemed appropriate). This fraction is etched for 40 minutes in HF an d then rinsed with water, HCl and water again. After drying, it is passed through the 180µm screen to remove any degraded feldspar. The material caught in the screen is density separated using a lithium metatungstate solution of 2.67 specific gravity. in the appendix [see full report for details]. Table 2 gives the relevant concentrations as determined from alpha counting (U and Th) and flame photometry (K). T he beta dose rate calculated from these concentrations is compared with that measured directly by beta counting, and this is also given in Table 2. There are no significant differences that might be caused, for example, by U - Th disequilibrium in the U 139 dec ay chain. Moisture content was estimated at 6 ± 3 %, which is more than the measured values of about 1%, which, however, do not reflect seasonal change. Table 3 gives the estimated dose rates, which are similar for both samples. ured on single - grain quartz using a green laser. Equivalent dose was determined as described in the appendix [see full report for details]. Table 4 gives the number of grains measured, the number rejected using the criteria given in the appendix, and the number accepted. The samples showed relatively high luminescence sensitivity for quartz. The acceptance rate for signals from which an equivalent dose could be measured averaged 13%. A high number of zero - aged grains in UW2850 suggests some contaminat ion from the surface. acceptance criteria. The central tendency of the derived/administered (~16 Gy) ratio, from the central age model, is 0.96 ± 0.02, which i - dispersion of the ratio distribution is 7.5%, which is a measure of intrinsic variation due to machine and sample factors and whic h can be taken as the minimum over - dispersion expected for a single - aged sample. A value of 10% was taken as typical for a single - age sample when evaluating age distributions. berts 2012) and the over - dispersion for each sample. The over - dispersion is quite high for both samples. A finite mixture model was applied to divide the sample into single - value components (see appendix). These are shown in Table 6. The samples appea r quite mixed. Given the mode of deposition, it is unlikely the samples are partially bleached. The younger components probably represent downward movement of grains exposed at the surface in these shallow deposits. Also unlikely to be correct are the o ld components in UW2851, particularly the fifth one, which would give a late Pleistocene age. Radial graphs are shown in Figure 1. For UW2850 most values seem to cluster around the third component. This probably best represents the 140 depositional age. For UW2851, even discounting the larger and smaller values, there is still substantial scatter. Short of any better information, the central age model, represented by the red line with a value similar to the third component, may present the best estimate, although arguments could be made for a younger age. same age as UW2850, which would suggest rapid deposition of the deposit. The older age, from the central age m odel, separates the layers in time and also agrees with a radiocarbon date (Daggett 2014, Table 6 OSL results Sample Age (ka) Error (%) basis Calendar (years AD) UW2850 0.65 ± 0.04 6.9 Largest component 1370 ± 40 UW2851 1.27 ± 0.18 14.3 Central age model 740 ± 180 UW2851 0.61 ± 0.04 6.8 Largest component 1410 ± 40 Discussion The OSL data appear to support two different sedimentary ages, one roughly consistent with a Zhizo occupation (740 AD ± 180) and one dating to the middle of the second millennium (1370 - 1410 ± 40) . The more recent date is associated with the very compact soil which covered much of the central area of the site from about 5 - 15 centimeters below surface. The earlier date is associat ed with the soil horizon just below the compacted one, which extended about 15 - 25 centimeters below surface, depending on the unit. The levels excavated within this deeper horizon also produced the greatest quantities of cultural material overall. This ser stratigraphic integrity. A more comprehensive OSL study of the site (sampling from multiple locations on the site with a similar columnating sampling procedure) would give a more thorough picture of site formation proc esses. However, from these results, it is possible to suggest that the compacted upper soil horizon, as differentiated from the less - compact soil matrix below it, may be a result of taphonomic effects instead 141 of a separate cultural horizon. There is little to no change in either color or soil makeup between horizons, and the types of cultural material from each horizon are of the same styles. In short, there is no a priori reason to believe that a second extensive prehistoric occupation occurred at Thabadim asego . However, cultural material found within the upper soil horizon and on the surface could be considered not to be in situ . As the tables for the AMS and OSL results indicate, there is reasonable overlap between the chronological ranges provided for ea ch radiometric type for the samples taken from the deeper soil horizon (U19 - L4 and UW2851). The margin of error for the OSL date gives it a range of 560 920, and the 2 - sigma margin for the AMS date gives it a range of 775 - 985. That the OSL and AMS dates for the upper soil horizon are not in agreement the charcoal sample is contemporary with the earlier horizon lends further support to the notion that this upper horizon is in some way disturbed. 4.7 Glass beads Glass beads are a non - utilitarian good im ported into sub - Saharan Africa from manufacturing sources in the Near East and South Asia since the middle of the first millennium AD (Wood et al. 2012) , as time (Pwiti 1991; Chami 2006) . As discussed in Wood (2011), beads changed in both glass composition and appearance over the centuries, which reflects shifting patterns of trade between African communities and the various sources of glass production. The glass beads recovered from late first - millennium (Early Iron Age) and prehistoric second millennium (Middle and Late Iron Ages) contexts within Southern Africa have been organized into a typology by Wood (2005, 2011, 2012) primarily by their morphological attributes such as color, shape , diameter and opacity. This glass bead series demonstrates a robust correlation with the existing chronologic al sequence in Southern Africa ( which is derived primarily from indigenous ceramics; both the ceramics and the beads are supported by radiocarbon ) , so that certain types of beads can be associated with a certain time period with reasonable confidence (Robertshaw et al. 2010). 142 Glass bead analysis During excavations in 2012, approximately 40 glass beads wer e recovered from Thabadimasego . An additiona l handful of glass beads was recovered from a test unit at site 16 - A1 - 12, located on the nearest escarpment protrusion to the west of Thabadimasego . Inclusion of glass trade beads among Early Iron Age assemblages is a well - documented phenomenon across a nu mber of regions of Southern Africa, including coastal trading depots such as Chibuene (Wood et al. 2012); the regionally - organized settlement systems of the Shashe - Limpopo Basin and eastern Botswana (Robertshaw et al. 2010), and the village of Nqoma to the west in the Tsodilo Hills (Denbow 1999). The regions where the glass used to make the beads was produced can be determined through comparative compositional studies of glass objects from sub - Saharan Africa, the Middle East and South Asia. In recent years a number of compositional analysis studies have been conducted on glass beads from Southern African contexts, relying for the most part on laser ablation - inductively coupled plasma - mass spectrometry (LA - ICP - MS) to determine elemental compositions of the b eads (e.g., Dussubieux, Robertshaw, and Glascock 2009; Robertshaw et al. 2010; Wood, Dussubieux, and Robertshaw 2012; Denbow, Klehm, and Dussubieux 2015) . These studies have provided independent confirmation of the morphological seriation developed by Wood. For further information about LA - ICP - MS and its applications to arch aeological glass, see; Gratuze, Blet - Lemarquand, and Barrandon (2001) , Dussubieux et al. (2009), and Gratuze (2013a, 2013b) . A number of ancient glass compositional types that have been identified by researchers appear frequently in Southern African bead assemblages, especially ones linked to glass production in the Mesopotamian region and South Asia (for full descriptions of their typology; see Robertshaw et al. 2010 and Wood et al. 2012). In addition to the information the beads can provide about trade connections between Southern African communities and the rest of the ancient world, they also have the potential to inform understanding of social organization and economic influence within Southern African political 143 spheres. As a non - utilitarian exchange good, the beads are interpreted as a luxury good and status item (Denbow 200 2; Robertshaw et al. 2010; Wood 2011). Their presence at Early Iron Age sites, and the mechanisms by which they were redistributed between Early Iron Age communities, is therefore of considerable interest as they provide insight into the political economic dynamics of the time. Within Southern Africa, numerous studies on the morphology and chemical content of beads have resulted in a chronological series that is unique to the subcontinent. As Wood over the centuries glass beads form distinct groups in terms of both chemical composition and attributes such as color, shape, diameter and opacity. These differences reflect shifting patterns of trade between African communities and glass and bead producers, as well as various Indian O cean merchants who carry on the trade (Robertshaw et al. 2010). The glass bead series also demonstrate a robust correlation with the existing chronological sequence in Southern Africa. Identification of beads with a particular series therefore provides an independent diagnostic element for Southern African sites. For a detailed explanation of each bead series and its morphological characteristics, see Robertshaw et al (2010); Wood (2011), and Wood et al. (2012). Wood overlap well with the compositional types determined by Robertshaw et al. (2010) and Wood et al. (2012). Robertshaw et al. (2010) in particular compared the Southern African glass with bead assemblages from across the continent, providing an independent ch eck on the significance of types observed within Southern Africa; this also demonstrated that the kinds of beads traded to Southern African communities were traded elsewhere. This makes LA - ICP - MS a useful (and inexpensive) tool for verifying bead types, pa rticularly when samples are heavily corroded and many of their morphological attributes are unobservable. For earlier bead series such as Zhizo beads, devitrification and corrosion are common problems due to the glass recipe used at the time (Robertshaw et al 2010), and while in some cases devitrification is even too extensive to provide reliable elemental signatures, LA - ICP - MS techniques which use single - point sampling, as opposed to raster 144 sampling, have been shown to be more effective at limiting the imp act of corrosion on the results (Dussubieux et al. 2009). Methods The morphological classification of the Thabadimasego and site 12 beads was conducted by Marilee Wood, using the procedures described in Wood (2011). A total of 49 beads and one amorphous gl ass fragment were examined; 45 of the 50 specimens were recovered from Thabadimasego units and stratified test pits, while 5 beads were recovered from the test unit on site 16 - A1 - 12. All beads were given a unique identifying number based on method of manuf acture, shape, end treatment, diameter, length, glass diaphaneity and color (including Munsell number), and glass type. Additional information, such as glass quality and condition, were also noted. Dr. Wood as Appendix E . Laure Du ssubieux (Field Museum Department of Anthropology) wrote the following part for a coauthored paper on the beads (which is forthcoming ): Thabadimasego were subjected to laser - ablation inductively - coup led - plasma mass spectrometry (LA - ICP - MS). The analyses were carried out at the Field Museum of Natural History in Chicago, USA, with a Bruker Inductively Coupled Plasma - Mass Spectrometer (ICP - MS) connected to a New Wave UP213 laser for direct introductio n of solid samples. The analytical menu consisted of 12 oxides and 44 trace elements commonly found in archaeological glass. Two different series of standards were used to measure major, minor and trace elements. The first series of external standards are standard reference materials (SRM) manufactured by the National Institute of Standards: SRM 610 and SRM 612. Both of these standards are soda - lime - silica glass doped with trace elements in the range of 500 ppm (SRM 610) and 50 ppm (SRM 612). Certified valu es are available for a very limited number of elements. Concentrations from Pearce et al. (1997) are used for the other elements.The second series of standards were manufactured by Corning. Glass B and D are glasses that match compositions of ancient glass (Brill, 1999, vol. 2, p. 544). The isotope Si29 was used for internal 145 standardization due to its relative abundance. In order to obtain absolute concentrations for the analysed elements, the concentration of the internal standard has to be known. Concentr ations for major elements, including silica, are calculated assuming that the sum of their concentrations in weight percent in glass is equal to 100% (Gratuze, 1999). co nsidered for the calculation of concentrations. A homogeneous glass composition for the beads was assumed based on the prior studies of Southern African assemblages. The detection limits range from 10 ppb to 1 ppm for most of the elements. Accuracy ranges from 5 to 10 % depending on the elements and their concentrations. A more detailed account of the performances of this technique can be found in Dussubieux et al. (2009). A total of 50 glass samples were processed (49 beads and bead fragments, and one amor phous glass fragment). Four samples were too corroded to be provide usable data. Following completion of the data collection, the reduced compositions for each sample were calculated by normalizing the seven major and minor oxides (SiO 2 , Na 2 O, Al 2 O 3 , MgO, K 2 O, CaO and Fe 2 O 3 ) to 100%. This process isolates the main components of the glasses, removes most of the compositional effects of Once the standardize d compo sition ratios were determined for every sample, prior studies were referred to as comparative samples. The results of the Thabadimasego analysis were compared to figure s from the Southern African assemblages in Robertshaw et al. (2010) as well as the assem blage from Chibuene, a coastal site in Mozambique (as discussed in Wood et al. 2012). Ancient glass compositions from Moretti and Hreglich (2013) were also referred to for cross - reference. The rations of the major and minor oxides were compared to these assemblages to determine the type( s) of glass (which, for this time period, are generally either plant - ash glass or mineral - soda glass). Trace element ratios were compared to determine the sub - types of plant - ash glass present. Major and minor oxides are as follows: 146 SiO 2 (Silicon dioxide; silica) Na 2 O (Sodium oxide; soda) MgO (Magnesium oxide; magnesia) Al 2 O 3 (Aluminum oxide; alumina) K 2 O (Potassium oxide) CaO (Calcium oxide; lime) Fe 2 O 3 (Ferric oxide) Following Robertshaw et al (2010), the values for each of these oxides were subject to ex ploratory data analysis. The minimum, maximum, mean, and standard deviation was calculated for each oxide. Bivariate scatter plots to look for significant correlations between oxides and elements, as per Wood et al. (2012) Results Wood ysis had determined that, based on glass quality, bead shape and color, 42 of the beads most likely belonged to the Zhizo series. Three of these were too corroded to allow determination of color. Thirty - six are light cobalt blue (the most common color foun d in this series), one is yellow and two are an unusual bluish - green. The remaining six beads, 2 a light greyish cobalt blue and 4 a greyish blue - green, were tentatively assigned to the Chibuene se ries ( Figure 26 ). Most of the beads are corroded to varying degrees, a condition that is often found with the Zhizo series due to the composition of the glass from which the beads are made and the conditions in which they were buried. The LA - ICP - es submitted for LA - ICP - MS (three of the beads and the lone glass fragment) produced Na 2 O signatures well below 10% (the standard threshold for normal, uncorroded soda - lime - silica glass, per Robertshaw et al. 2010:1902). One further bead was rejected from the sample due to its very low silica content. The remaining 45 were assigned to a known subgroup based on ratios of major and minor oxides as well as trace elements. All 45 147 beads were determined to be made of plant - ash glass as opposed to mineral - soda gla ss, based on their concentrations of magnesia (MgO). MgO levels in plant ash glasses are usually above1.5%. Below this level, mineral soda (e.g. natron) is assumed to be used. In the Thabadimasego beads MgO levels are always above the 1.5%. The mean conce ntrations for constituent oxides fall within the ranges described for Zhizo beads by Robertshaw et al. (2010) with high magnesia and lime concentrations, and a very low concentration of alumina (see Tabl e 7). Table 7 Mean values of major and minor oxides from LA - ICP - MS Bead Series SiO 2 Na 2 O Al 2 O 3 K 2 Fe 2 O 3 Zhizo 69.62 13.15 4.31 3.26 3.23 5.5 0.94 K2 64.51 16.22 0.43 11.85 3.34 2.34 1.3 K2 GR 61.05 14.36 0.37 16.63 3.39 2.85 1.35 Indo - Pacific 63.08 14.75 0.59 13 3.46 2.85 2.27 Islamic 63.21 13.71 4.83 6.05 3.91 6.63 1.66 Map Oblate 61.88 13.47 5.8 7.67 3.47 6.66 1.04 Zimbabwe 60.98 15.81 4.33 6.71 3.74 6.94 1.48 Khami 61.4 18.66 1.21 9.81 2.82 3.39 2.7 Thabadimasego & Site 12 65.69% 14.67% 3.46% 3.26% 4.21% 4.68% 1.25% In order to differentiate the Chibuene series from the Zhizo series beads, principal components analysis (PCA) was conducted for the Thabadimasego and site 12 datasets using the menu of oxides and elements which overlapped with other datasets from Southern Africa (Robertshaw et al. 2010; Wood et al. 2012) used for comparison, and excluding those known to be coloring and opacify ing agents, such as co balt . The PCA distinguished two clear groups of beads, which show strong agreement with the parameters for v - Na - Ca 1 (Zhizo) and v - Na - Ca 3 (Chibuene) beads in both the Robertshaw and Wood datasets. Based on these results, fourteen of the 45 glass beads can be placed within the Chibuene (v - Na - 3) series as described in Wood et al. (2012), while the remaining 31 beads fit within the Zhizo, or v - Na - 1 serie s ( Figure s 27 and 28 ). Of the 1 4 Chibuene beads, 12 were from Thabadimasego and 2 were from the neighboring site 12. The remaining beads b elong to the Zhizo se ries ( Table 8 ). 148 Figure 26 Chibuene (L) and Zhizo (R) glass beads 149 Figure 27 PCA score plot for Zhizo vs Chibuene glass types 150 Figure 28 PCA loading plot for Zhizo vs Chibuene glass 151 Table 8 Series determinations for glass beads from Site 12 and Thabadimasego Site # Zhizo Chibuene Undetermined 12 Lot 373 Lot 375B Lot 375A Lot 377 Lot 379 13 Lot 159 Lot 181 Lot 23 Lot 162 - 3 Lot 201 Lot 36 Lot 163 Lot 232 Lot 146 Lot166 Lot259 Lot 232B Lot 172 Lot 281 Lot 224B Lot 205 Lot 295 Lot 21 Lot 328 Lot 222A Lot 37C1 Lot 222B Lot 37C2 Lot 223 Lot 37C3 Lot 224A Lot 37F Lot 227A Lot 45 Lot 227B Lot 227C Lot 227D Lot 229 Lot 259B Lot 26 Lot 261 Lot 27 Lot 284 Lot 330 Lot 346 Lot 37A Lot 37B Lot 37D Lot 37G Lot 37H 152 4. 7 Macrobotanical identification Positive identification of macrobotanical remains in Southern African Iron Age contexts remains problematic. Very few direct studies of macrobotanical remains in Iron Age contexts have been conducted for Southern African sites; Jonsson - gatherer and farming sites is one notable exception. A handful of other sites have had small quantities of seeds, husks or nuts identified, although it is unclear by what means the identifications were conducted; see Mitchell (2002:274). Evidence for cultivation of plant species at Southern African sites typically comes by way of proxy indicators. The most common of these are grinding stones with grooves suited to sorghum and millet seeds (Mitchell 2002:273), mud - brick structural remains interpreted as grain bin foundations (Mitchell 2002:274; Huffman 2007:335), and ethnobotanical observations of present - day use of wild and domesticated plant species among subsistence communities in Southern Africa (Jonsson 1998; Mitchell 2002:274). An argument can also be made that communities practi cing plant cultivation showed a preference for alluvial and colluvial soils near major rivers or (in earlier cases) marine coastlines (see, e.g., Maggs 1984; Pwiti 1996); however, Mitchell (2002:273) and Mitchell and Whitelaw (2005:222,223) point out that site location choices also conform with the availability of other critical resources such as iron ore or shellfish, and therefore that soil types on their own are not a reliable indicator of agricultural behaviors. It is widely recognized in scholarly research on the transition to agriculture that proxy evidence such as grinding stones and storage facilities do not necessarily imply the existence of cultivated crops (e.g., Piperno et al. 2004; Nadel et al. 2012) . In several regions of the world, the use of grinding stones for cereal preparation and construc tion of food storage facilities has been demonstrated to predate domestication event s for plant species (Barker 2006 :74 - 76). While current models for the appearance of agriculture in Southern Africa posit migration of communities already practicing a well - developed form of crop cultivation who carried both the technology and plant species with them, there is still substantial 153 evidence for the continued consumption of wild plant species among f ood - producing communities (Maggs 1984; Denbow 1986; Jonsson 1998). It may be useful to take the perspective that, without direct evidence, crop cultivation and animal husbandry did not necessarily go hand - in - hand, even where other material indicators of th e Chifumbaze cultural the Chifumbaze cultural type, see Phillipson 1977). Sites representing the earliest phases of the Chifumbaze complex do not, in fact, contain many instances of preserved botan ical remains at all (Huffman 2007:338), although microbotanical studies of pollen and phytoliths may change this understanding in the future. In light of this concern, care was taken both during excavations and in post - excavation activities to watch for bo tanical remains. Several clusters of carbonized macrobotanical remains were in fact recovered from a number of units at Thabadimasego ; the use of 1x1 - mm mesh to screen excavated soil was helpful in this instance (as with other small finds such as glass bea ds). Additionally, soil samples were collected regularly throughout excavations, following a - 67) and were later processed using water flotation . Although Pearsall (2000:75) recommends collecting at least 10 lit ers of soil for each sample for optimal recovery of botanical and faunal remains, this quantity proved not to be feasible for this project. The small size of the features in question necessitated collection of much smaller (1 - 2 liters) samples, an issue al so encountered by Jonsson (1998:52). Because water and drainage facilities were both very limited at the NMMAG archaeology lab, flotation was conducted at a small scale following similar lines as the manual bucket - and - scoop technique as described in Pea rsall (2000 :35 - 37). After undergoing flotation, the heavy and light fractions from each sample were dried and stored separately. The heavy fraction contents (typically larger bone fragments, small ceramic sherds and the occasional bead) were inventoried and added to the general assemblage catalog. The light fraction for each sample was weighed and included in the catalog; however no further work was done apart from packaging them for storage. The intent had been to contract with an archaeobotanical speci alist to work 154 with these samples for the purpose of taxon identification, but as was learned, the only archaeobotanist in Southern Africa currently working with Iron Age remains is presently conducting her Master research and therefore unavailable . The carbonized botanical samples recovered during excavation were photographed using a Dino - Lite digital microscope at 30 - 35x magnification in order that they might be identified through comparison with other images, or ac tual extant botanical samples. Discussion No formal analysis of the botanical remains could be conducted. While there are some archaeobotanists who do work with Southern African prehistoric remains, they tend to specialize in Pleistocene and early Holocene that is, Middle and Later St one Age environments. Hope remains that in the future an archaeobotanist will be available to identify seeds and nuts from Iron Age contexts. In an effort to provide some description of the botanical remains recovered at Thabadimasego , a number of archaeologists who, while not trained as archaeobotanists, have nevertheless dedicated their long careers to researching Iron Age lifeways were asked to examine the Dino - Lite photographs of the remains. While this in no way constitutes a formal determination of type, the general agreement was that the carbonized seed clusters appeared most like s orghum ( S orghum bicolor ). Images of the Thabadimasego seeds and a n example of present - day sorghu m (source: user Sahaquiel9102/ Wikimedia Figure 29 A present - day example of sorghum 155 Figure 30 Probable sorghum from Thabadimasego 156 Commons, http://commons.wikimedia.org/wiki/File:Millo_ - _Sorghum_bicolor_03.jpg, accessed 2/21/2015) are included here for reference . 4.8 D iscussion of results as a whole Ceramics Most ceramics are of Zhizo type (or very similar type) A very few stylistic oddities Very little apparent refit (based on casual observation) Shell Two different (but overlapping) ranges of variation for diameters Achatina and ostrich shell beads Significant difference in diameter between burnt/ unburnt for Achatina shell beads No apparent difference in diameter between b urnt/ unburnt for ostrich eggshell beads Unworked fragments and bead blanks occur in quantities too small to associate with a bead Metal Two types of metal present (based on outward appearance, magnetic characteristics, and compariso n with other Early Iron Age assemblages); copper and iron Quantity of metal items overall is low relative to descriptions those from larger contemporary settlements (Nqoma, Bosutswe) No metal tools present; pieces of jewelry, wire and bar fragments compris e the entire assemblage Copper and iron beads/ bangle fragments have similar morphology to beads/ bangle fragments from a number of sites (Nqoma, Bosutswe, Broederstroom) Slag occurs in unusually low quantities Faunal 157 Over 1000 individual specimens counted in assemblage Wild bovids, domesticated bovids, other mammals and vertebrates, two human teeth Further work needed for clearer picture of ratios of types of bovids Radiographic AMS dates indicate a median time frame of late ninth/ very early tenth century (consistent with Mosu I) OSL indicates two dates, one 6 th 9 th century and one 13 th - 14 th c. Stratigraphic context of both types of samples indicate probable disturbance in upper soil horizon Apart from later OSL dates, results are generally in agreement with timeframe of a Zhizo - era occupation Glass Morphological and elemental analyses both indicate two facies of beads present: Chibuene and Zhizo Botanical No formal analysis possible Preliminary designation of seed clusters as domesticated possibly sor ghum Taken collectively, the artifact assemblage from Thabadimasego as characterized in the sections above is largely reminiscent of a late - first - millennium AD occupation whose cultural ties lie primarily to the regions east and south of the Makgadikgadi P ans. While the margins of error for radiocarbon and luminescence samples from the site provide a potential time frame of occupation as wide as the sixth through tenth centuries AD, the likely scenario for occupation falls towards the latter half of this ra nge as is consistent with the radiocarbon dates acquired from Mosu I (Reid and Segobye 2000 a ) and Kaitshàa (Denbow et al. 2015 ). The possibility does remain that Thabadimasego was occupied earlier than the ninth 158 century as well, particularly in light of several Chibuene - style glass beads present among the assemblage. Chibuene glass beads are relatively rare finds in Southern African sites (or at very l east, are only beginning to be recognized as a separate facies) and are thought to date to the sixth and seventh centuries AD based on their initial discovery at the eponymous site in Mozambique (Wood et al. 2012). Given, however, the potential for durable luxury items such as glass beads to be curated, saved, traded, and handed down over generations, it is also possible that beads of such an early provenience only made their way to Thabadimasego long after their initial arrival in Southern Africa. The mate rial remains from Thabadimasego by and large provide an impression that subsistence activities typical for the Early Iron Age occurred on the site. Cattle and sheep/goats, as well as wild game animals such as zebra and steenbok populate the faunal assembla ge. Carbonized seed remains, such as might be produced at a cooking fire, are found in substantial quantities along with numerous multi - gram flecks of charcoal. Decorated and undecorated pottery, much of it burned, abounds across the site, most of it in a highly fragm ented state. Shell beads, present in Southern African assemblages since the Middle Stone Age, also occur frequently. Glass beads traded inland from coastal depots as well as copper and iron decorative objects occur in moderate quantities. The p resence of both types of items may support the notion that occupants of Thabadimasego engaged in a network where highly valuable goods from the South Sowa area (such as, perhaps, i vory; see Reid and Segobye 2000b ) were exchanged for these luxury goods. Whi le some of the more specific questions posed by this research project about the exploitation of certain resources (such as wild versus domesticated plants and animals) must remain unanswered until more work can be done on the assemblage, it would be easy t o characterize Thabadimasego as a small, low - ranking settlement that perhaps acted as a subordinate in a Zhizo - era settlement system where a site like Kaitshàa acted as the higher - status larger village. Generally speaking, Early Iron Age sites that contain some structural remains, regardless of their overall size, seem to be characterized as some type of 159 residential settlement (see, for example, Mitchell 2002: 279 - 283 ). In the following chapters, however, the argument will be presented as to why Thabadimaseg o should not be presumed to have been a residential settlement, and why the site itself and its presence in the surrounding landscape may instead lend support towards an understanding of Early Iron Age settlement processes in more regionally variable terms . 160 Chapter 5 Spatial analysis 5.1 Spatial information and analysis Spatial attributes of a n archaeological site represent more than just the background of the material objects themselves. Rather, these attributes both constrain and are influenced by the activities which occurred on the site. Space is both functional and symbolic, and the manipulation or organization of space is, as such , an important component of any anthropological study (David and Kramer 2001; Branton 2009) . Spatial variables and spatial attributes are too often overlooked within Southern African Iron Age research as significant components of past communities d uncritically and are not contextualized by other factors like climate, terrain or social influence; such is the case with the Central Cattle Pattern when decoupled from its direct historical connection and its original locale (Lane 199 5). Badenhorst (200 9) , for example, points out the irregularities among features (such as hut shape and village layout) sites at which the Central Cattle Pattern has bee n used as an interpretive model. He suggests that placement of these features are primarily functional, no t ideological, and that similar settlement layouts occur in other parts of Africa outside the Eastern Bantu/ Chifumbaze frame of reference. Regional settlement patterns are also increasingly recognized; in addition to Denbow - known (1984) Toutswe mod el for eastern Botswana, other scholars have developed local settlement sequences for the Shashe - Limpopo Valley (Meyer 2000; van Doornum 2005) and the Eastern Cape (Feely and Bell - Cross 2011) . 5.2 Spatial methodology and results Critical studies of the use of space in Iron Age contexts are beginning to tease out some of the variables which may b e understood from a comparative perspective that accounts for site formation processes, variation in site function, and phys approach which on borrows details from a specific ethnographic analogy. In their work at Ndondondwane, for example, Greenfield and Miller (2004) and Fowler an d Greenfield (2009) demonstrate by examining 161 microstratigraphic changes as well as by sequencing shifts in metallurgical activities is the result of reoccupations over time , albeit within a single cultural horizon . Fowler and Greenfield (2009) relate stratigraphy to radiocarbon and ceramic sequences at Ndondondwane to document changes in site use over time . Greenfield and Miller (2004) classify metal - w orking residues according to type, frequency, and spatial arrangement in order to discern trends in metallurgy over time at Ndondondwane. To address the need for a critical assessment of spatial variables within the scope of this project, the use of space was examined primarily at the site - level scale. In addition, in Chapter 6 the site will be discussed in the context of its landscape. While other spatial scales, such as the regional interaction of Zhizo - era sites, are also significant for the understandi ng of socioeconomic process, it was felt that the intra - site and inter - site (landscape) scales were most informative for the research questions stated in Chapter 2, particularly as the factors affecting these scales have not yet received much attention for the South Sowa area. The overall goal for this spatial analysis is not to produce a specific model which reframes socioeconomic dynamics of E arly Iron Age communities in an alternative to the Central Cattle Pattern , but rather to understand the underlyi ng principles that dictate how space was used within the site, to the extent that this is possible using only the artifact distribution (as opposed to, for example, using ethnographic or ethnoarchaeological information as well ). In other words, how are cul tural materials organized at the site are certain kinds of artifacts more likely to co - occur with other types of artifacts, or more likely to occur in one part of the site as opposed to another? Also, what other identifiable factors - taphonomic, geologi cal, and historical - influenced the distribution of cultural material ? M y future research goals include building a database that can provide a comparative perspective on the variability already observed in site organization among Early Iron Age landscape s. Attributes considered on the intra - site scale include geophysical (soil horizons, terrain, elevation, etc.); taphonomic (erosion and bioturbation ) , recent disturbances made by people ; and behavioral (types of features, 162 relative placement, cultural mater ial distribution relative to non - portable features, etc.). Landscape attributes considered include the natural and cultural boundaries imposed on sites; and geophysical factors such as land forma tions relative to site location . This project on its own will not accomplish a full accounting of these data, but will merely lay the groundwork for it. No a priori assumptions about artifact distribution at Thabadimasego were made prior to mapping the site (apart from the usual law of superpositioning as it relate s to stratigraphic integrity). The first null hypothesis is that all cultural materials are randomly distributed across the space of the site and that no cultural factors were involved in the placement of these materials. This is highly unlikely to be a va lid hypothesis (especially since some clustering was observed in the field). Therefore, a secondary null hypothesis is that none of the types of cultural materials (pottery, glass, structural remains, etc.) have any more influence than the others in the wa y the site is structured. In other words, no clusters of any kind of artifacts influence the clustering of any other kind of artifact. If this null hypothesis were to be supported, no conclusions can be drawn about the behaviors that influenced the site organization, or about the socioeconomic context in which these behaviors were enacted. It must be remembered too that non - random distributions do not automatically translate into direct evidence of a set of intentional behaviors causing the artifacts eposition - we cannot immediately rule out clusters forming from, for example, taphonomic effects such as slopewash displacing lighter artifacts (small beads, charcoal, etc.) and leaving in place heavier ones (such as pot sherds and structural remains). Ev patterning provides a starting point for further investigation wherein non - behavioral factors may be explained. Natural factors may have played a role in site layout a t Thabadimasego . Because the edges of the hilltop are erosion - prone (particularly on the Southern side), it makes sense that structures would be a role. For example, exposure to wind incoming from Sowa Pan could have affected location choice for 163 structures and middens. The spatial informati on collected during the 2012 excavation season was generally accurate to the arbitrary unit level (most often 5 c m below datum; occasionally 10 - cm when a unit seemed especially sterile). In mapping artifact d istributions, then, it would be possible not only to represent their X and Y coordinates (northing and easting coordinates of the unit datum from the site grid), but their Z coordinate as well (depth below surface or datum). Artifact distributions could be generated for each soil horizon in this way, which could be informative as to vertical distribution and depositional processes. However, for the purposes of this project at this time, the aggregate of each pit or unit (combining all its levels/ horizons) was taken as the unit of analysis instead. This was done primarily because the working hypothesis for this project (based on uniformity of the assemblage) is that the site represents a single, fairly short - term occupation, and so all cultural materials, re gardless of their present - day context, were deposited in what can be considered a single event (such as a temporary camp ) . As discussed in the previous chapter, the presence of artifacts in the upper horizon of soil is likely attributable to post - depositio nal m ovement caused by bioturbation and soil displacement. It will probably be of interest in the future to revisit this dataset with an eye to differentiating the vertical distribution, but that remain s a problem for another time. Spatial analysis methods The majority of the spatial analysis was completed using the software suite ArcGIS 10.2. Among GIS software , ArcGIS is somewhat unusual in that its workflow is shared between several programs, all of which are part of the software suite. Two of these prog rams, ArcCatalog and ArcMap , performed the bulk of the operations for this project . ArcCatalog acts as a file and metadata manager, while ArcMap hosts the ability to edit the data stored in the files, and to create maps (which in reality are simply compila tions of the types of files created in ArcCatalog and stored in a separate database). The starting point for the spatial analysis was to convert the existing artifact inventory (stored in 164 Excel spreadsheets) into a table compatible with ArcGIS, and to geo reference this information. This on its own was quite time consuming as it necessitated standardization and synthesis of the multiple formats by which the catalog had been compiled over the last two years (all of them were in MS Excel, but different termin ology was used for the same type of artifact at different points, listed inventories in different ways, and in some cases artifacts were even classified somewhat differently at different stages. For example, while the ceramic identification in the 2012 pos t - excavation initially began by sorting out the burned, undecorated body sherds from the unburned, undecorated body sherds (and had additional categories for, e.g . , unburned decorated rim sherds, etc.), later these two were lumped together in a single cate gory as the distinction was not always clear and it seemed unnecessary to introduce that level of uncertainty into the dataset. A listing of each art ifact category can be found in Table 9 . Before work in ArcGIS could begin, it was necessary to address th e fact that, of the 218 pits and units excavated at Thabadimasego by the 2012 field team , 21 of these were 1 - by - 1 - meter units which comprised a much larger volume than the other 197 test pits (whose surface area dimensions averaged about 40 - cm - by - 40 - cm). F or example, a 1 - by - 1 unit excavat ed to 50 cm would comprise .5 m 3 total volume, whereas a 40 - by - 40 - cm pit excavated to 50 cm would comprise only .08 m 3 total volume. A unit and a pit, even if excavated to the same depth, would therefore each represent very disparate units of analysis. To the knowledge of the author, this is not a common practice in post - excavation spatial analysis, at least within Southern African archaeology; however, upon review of the methods employed in this chapter with Dr. Sarah Hessi on, a statistician with Michigan State University, it was agreed that in theory the standardi zation of artifact counts per unit volume would, in fact, provide a more quantitatively meaningful picture of the data. Such a disparity in the size of each area sampled meant that counts of artifacts for pits and units were not directly comparable. However, since cultural material was generally well - dispersed throughout any given unit or pit (i.e., material did not tend to cluster in any portion of the 1 - by - 1, but was usually 165 spread throughout it), this problem could be addressed fairly easily by converting each pit into a standard size. A copy of the original artifact inventory for all 218 units and pits was made where the artifact counts for each pit were standardize - by - 1 - meter surface area. This was done by calculating the original surface area for each pit and dividing the count of each artifact type by the resulting fraction. For example, if pit A had dimensio ns of 35 - by - 35 - cm, its surface area is 0.1225 m. If the same pit contained 150g of standardize - by - 1 would be about 1225 g. While, given that at some larger scale clustering is known to occur on the s ite (hence the entire reason for the spatial analysis) and this correction is therefore not a perfect approximation of standard sample volumes, it was felt that using a standardize d count of artifacts would address the more immediate problem of unit compar ability. However, the original, non - standardize d artifact counts were also retained as a separate file, so that, ultimately, spatial analysis could be performed on both and a measure of sensitivity could be gained for the standardize d data. Table 9 Artifact categories for analysis Shell OES beads, finished (count) OES beads, irregular (count) OES beads, <50% present (count) OES fragments (count) OES all types aggregate (count) ACH beads, finished (count) ACH beads, irregular (count) ACH beads, <50% present (count) ACH all types aggregate (count) Shell all types aggregate (count) Pottery Pottery, all, mass (g) Pottery, Undec Body, mass (g) Pottery, Dec Body, mass (g) Pottery, Undec Rim, mass (g) Pottery, Dec Rim, mass (g) Glass Glass beads, count Metal Ferrous beads, count Ferrous beads, mass (g) Non - ferrous beads, count 166 Non - ferrous beads, mass (g) Ferrous fragments, mass (g) Ferrous wire/ bar, count Ferrous wire/ bar, mass (g) Non - ferrous wire/ bar, count Non - ferrous wire/ bar, mass (g) Ferrous all, mass Non ferrous all, mass Metal all, mass Other Slag, mass (g) Dhaka, mass (g) Burnt seed, mass (g) Bone, mass (g) Charcoal, mass (g) Once the data were organized, cleaned and converted into ArcGIS geodatabase tables (one each for the original and the standardize d artifact counts), these were each appended to a point feature clas s which contain the location of each excavation unit and test pit each represented as X,Y points. The resulting files contained information (stored as attributes of the points) about the quantities of each type of artifact (glass beads, metal beads, variou s types of pottery, etc.) in each unit and pit in a spatial ly - referenced, mappable format ( F igure 31 ) . One such feature class was created for the original, unaltered data and another for the standardize d data. The resulting feature classes, along with back ground shape files such as hill contours and site perimeter, were incorporated into an ArcMap map file for further analysis. ArcGIS 10.2 offers a wide array of options when it comes to structuring and displaying spatial data. Each separate shapefile or fea ture class makes up a separate layer in an ArcMap map file (much like, in graphics editing software such as Photoshop, different elements of the image can be sorted into different layers, which can then be individually turned on or off and edited on their own). The properties of each of these layers can be changed depending on what kind of information one wishes to display in their map. In this case, the point feature class containing all 218 units and pits (and their corresponding artifact Table 9 (co n 167 counts) first ha d to be used to generate subordinate shapefiles for each category of artifact, so that each of these artifact categories could be a separate layer (and therefore visualized distinctly) in the map. This is done with the following: a applied to the point feature class to select each point containing any quantity of a given type of artifact, and the resulting selected points are exported to their own shapefi le under an appropriate name. This is repeated for each artifact category. Hence, the 35 categories of artifact included as attributes for each of the 218 units and pits resulted in 35 separate shapefiles which could then be modified and displayed separate ly. This process was done both for the original and the standardize d datasets. Histograms In order to get a quick overview of the distribution of each kind of artifact on the site, a series of histograms was also produced for each artifact category for bo th the original and standardized datasets. The histograms depict the frequency with which any given quantity of the given artifact type appears in each unit or pit. The histograms demonstrate that, for any kind of artifact, regardless of whether using the original or standardized datasets, the distribution is non - normal and skews towards zero. These histograms can be viewed in Appendix F; see Figure 32 below for an example. In general, the shapes of the distributions are very similar for standardized and or iginal data. Possible exceptions to this include ferrous beads (mass), finished OES beads, aggregate Achatina , and charcoal. There turned out to be inconsistency in the binning used for the histograms, either between artifact types or between original and standardized, because I was aiming to represent the distributions in as fine - ifferences between types (and also between original and standardized, when standardi zation created much larger quantities instances of it on the whole site ), but it does for undecorated body sherds, because doing so eliminate d false zeros. 168 Figure 31 Screen capture of a portion of the attribute table for the excavation units in ArcGIS 169 Although generally, increasing t its detail. ArcMap displays The display of an ArcMap shapefile for any artifact category can be modified to show something very similar. Attribute values of a given feature (in t his case, quantities of artifacts per point) can be symbolized in a number of different manners in ArcMap, by varying the size, shape and color of the feature. Here, the values of the attribute in question were represented with symbols of graduated size an d color, where each symbol represented a range of values (e.g., 0, 1 - 2, 3 - 5, etc.). The map then displays the locations of larger quantities of a given artifact type by showing a larger symbol, and smaller quantities with a smaller symbol, and so on (for an example, see Figure 33). Since the goal for spatial analysis is to understand not only what material is clustering where, but how clusters of different materials may relate to one another - in an effort to look for an underlying structure to the site - then these displays can be l ayered and, in a sort of georeferenced multivariate scatter plot, used to compare various combinations of artifact distributions across the site (though it is not truly a scatter plot, because the X and Y axes represent the location coordinates of each poi nt, not the values of the variables in question). Given that there are 35 different categories of artifacts, there are hundreds of potential multivariate combinations. However, certain kinds of cultural material are thought to hold more weight in the organ ization of any Early Iron Age site. The arrangement of mudbrick structures in particular - such as huts or grain storage bins - is the primary focus of most representations of site layout (Mitchell 2002, Huffman 2001). Burials, cattle kraals, and metal - wor king forges, if present, are also important. Essentially, these are the features of a site, or at least a homestead or village site, which are the are assumed to form around and in relation to these features. These features are given primacy because they represent the major cultural and economic elements of Early Iron Age subsistence patterns - 170 Figure 32 Example of histogram displaying frequencies of artifact counts per unit/ pit 171 permanent, sedentary living spaces; cattle - keeping and crop production; creation and use of metal items (which also is thought to hold significant spiritual power, hence the importance - see, e.g., Miller and Killick 2004). Materials in the archaeological record that indicate the remains of such features, such as stone structures, dhaka (mudbrick), and cattle dung, can therefore be given primacy when comparing the distributions of different artifact types at a site. It can be expected that these materials would contribute to the site - in other words, that they would be the independent variables in terms of spatial analysis and that other artifacts wo uld be, in e ssence, the dependent variable In short, what is already known about Early Iron Age socioeconomic organization provides a reason to prioritize a few variables in the search for underlying structure of a site that the extensive subsurface reconnaissance at Thabadimasego recovered no remains of a burial, forge or smelting furnace, but that stone structures, mudbrick structural remains as well as what may be ash, dung, or both are all present on the si te. Figure 34 shows Thabadimasego used to generate polygon features representing the general areas in which ash and mudbrick are fo und on the site. Since ash appeared in small, dispersed clumps throughout an area several meters in diameter (as indicated by the few pits ash was found in), the original point features for ash were given a 5 - meter buffer to produce the resulting polygon. Mudbrick required somewhat different treatment. Small chunks of mudbrick (amounting to no more than a couple hundred grams) appeared in several units and pits on the site, while actual structural remains - fragments of walls or floors - only appe ared in a few units in the center of the site (specifically, units 3 and 5, which are contiguous, and units 19 and 20, which are set 5 meters apart). The remains in these units added up to over 20 kilograms of mudbrick per unit. In addition, unit 16 contai ned about 2 kg of dhaka (mudbrick), which is still almost an order of magnitude greater than any other unit or pit. At the time of excavation, there was unfortunately not enough time to place an extended block of excavation 172 Figure 33 Value - weighted plotting of bone (by mass, per pit) at Thabadimasego 173 units in this central area so that the full extent and shape of the walls/ floors could be established (this would be very much of interest for future work on the site). The remains definitely do not extend more than 10 meters in any direction from the units in which they were found, as the ten - meter interval test pit survey coverage established. Therefore, as a rough estimation of the extent of the significant mudbrick features on the site, a 5 - meter buffer was used for units 3, 5, 19 & 20 (5 meters being the maximum distance between any of these units), and a 1 - meter buffer for the smaller feature in unit 16. The resulting polygon representations, which togeth er with the stone structures and elevation contou rs comprise the for the remaining spatial analysis. For future work - it would be informative to excavate a several - meter block of units where these wall/ floor re mains were found to see if any specific structural organization can be identified, or whether it really is just a few collapsed surfaces strewn about. However, the presence of multiple subtypes of the same material - e.g., shell beads at different stages of production, or iron it ems of different forms does allow the opportunity to examine the distribution of these materials at varying levels of specificity. For example, the distribution of any and all metal present on the site may be viewed in comparison with that of, say, only copper items, or only iron beads (both of these being subtypes of the higher - order category looking to determine the appropriate scale of analysis for materials. Are any and all metals treated the same in their placem ent on the site, or are copper items located differently than iron ones? Are metal beads located differently from wire or slag? That is to say, are they treated as part of the same group or as different groups? Without knowing a priori the significance the occupants of Thabadimasego (or any other Early Iron Age site) may have given to any one category of material as these categories are defined by the archaeologist, the use of spatial displays to examine differentiation (or lack thereof) among organization of material types can provide one window of insight into the behaviors which produced the archaeological record we now excavate. The distribution of artifacts over the site may also be influenced by whether or not they were 174 intentionally discarded. Intenti onal discard of certain types of material goods, such as broken pottery or butchered bone for example, would be liable to occur repeatedly in the same locations. Unintentional loss of other kinds of goods, such as glass, shell and metal beads lost one or t wo at a time from a broken necklace, would, on the other hand, be more scattered. To some extent this appears to be true for Thabadimasego . Small decorative items like the beads tend to occur in low concentrations across large portions of the site. On its own this could say something about the occupation of the site how long would it take for nearly 40 glass beads, for example, to accumulate through accidental discard one by one? However, the presence of the cluster of ostrich and Achatina shell beads nea r the west wall remnant, many of which are burned or broken, suggests that these items too were discarded intentionally at some point and separately from other materials. The stone wall itself, it should be noted, had little in the way of cultural materi als directly associated with it, based on the excavation units and test pits placed adjacent to it. The chronology of hardveld, western Zimbabwe and the Shash e - Limpopo Basin, van Waarden (n.d.) associates the origins of the Mambo ceramic facies in Huffman 2007), even while she notes that sites along Sowa Pan, including Kaitshàa , have occupations which date prior to this. Unlike Kaitshàa, however, Thabadimasego has not produced evidence of a multi - phase occupation which extends into the second millennium AD (Denbow et al. 2015). There is, in other words, no reason to assume that the stone walling at Thabadimasego is associated with anything other than its Zhizo - era occupation. There is little further that can be said about the stone wall at this time, beyond des cribing its physical attributes. Its current height averages about 50 centimeters tall; its width roughly two meters. This reflects the severe degree to which it has been subject to damage by livestock and human over the last millennium. Its two segments, 175 Figure 34 Stone, mudbrick and ash features at Thabadimasego 176 out from the opening. The wall appears to have been constructed out of local m aterials (probably calcrete), given that its structure gradually gives way to what looks like a natural outcrop of the same type 30 - 50 cm in length an d appeared uniformly weathered ; they are distinctly larger, rounder, and less evenly coursed than classic Zimbabwe - type stone walls . T he stone wall remains at Kaits hàa ( Figure 35 ) are sufficiently intact to appear comparable to the P - style walling, an earl y form of stone wall construction that appears in the Shashe - Limpopo Basin, at Great Zimbabwe (in association with its early phases), and eastern Botswana (van Waarden 2010) . The wall at Thabadimasego , on the other hand, has been destroyed too much to ass ess its construction style ( Figure 36 ). Looking beyond exploratory data analysis The following section discusses some additional work that was conducted on the quantified spatial dataset in an attempt to understand how higher - order spatial statistical techniques operated on the data and what insights might be gained from using spatial statistics to evaluate a fine - grained data set like the one collected at Thabadimasego . In Southern Africa, the use of spatial metrics to evaluate site formation and site organization on as small a scale as has been done here is relatively uncommon (for one example, see Greenfield and van Schalkwyk 2006). The use of higher order spatial statistics, such as those discussed below, is unprecedented for Southern African Early Iron Age archaeology as far as the author is aware. As such, the proceeding discussion represents an exploration of the applicability of such methods to site - level archaeological data more so than an analysis in itself. Some of the problems and challenges related to attempting to employ spatial statistics on the dataset are discussed as well. The work di scussed in this section was conducted under the tutelage of Sarah Hession, a research specialist at the Center for Statistical Training and Consulting at Michigan State University. However, any errors with either the methods or the data are mine alone. Hig her - order spatial statistics can shed additional light on the clustering of material culture in a 177 site by highlighting not only where those clusters lie, but which clusters are in fact statistically significant (I.e. that they would not be expected to appe ar in a normal random distribution). Hotspot analysis is one such technique (Lawson 2010 ; . Whereas spatial cluster analysis is a technique to measure the degree of clustering and/ or dispersion of point lo cations across a landscape, hotspot analysis takes measure of the degree of differences between values for a given attribute associated with the points (for example, quantities of pottery found in test pits). This technique can be used to identify location s which contain an unexpectedly high quantity of something (a hotspot) as well as an unexpectedly low quantity (a coldspot). However, a lot of assumptions about the organization of the data being subjected to these tests need to be sorted out before the re sults ar e meaningful (basically meaning that other exploratory tests need to be run first to evaluate the data, and the data need to be examined at different scales, to identify appropriate parameters). Unfortunately, completing this assessment proved to b e outside the reach of this research project for the time being; what follows is a brief summary of what techniques were explored and how they may inform spatial organization of an archaeological site. Hotspot analysis may be conducted on a dataset in ArcG IS 10.2 using the Getis - Ord Gi* function. The purpose of this function is, as described on the ESRI help files, ven a set of weighted data points, the Getis - Ord Gi* statistic identifies those clusters of points with values higher in magnitude than you m ight expect to find by random chance. After selecting hotspot analysis for its ability to evaluate quantitative attributes of specific locations, a series of Global Moran the dataset prior to the hotspot analysis itself, in order to identify relevant neighborhood matrices. In essence, a neighborhood is an area, defined by a given distance, around the point of comparison where one would reasonably expect co - occurring material to have some kind of functional relationship or, in other words, to share space for some reason other than random distribution (Getis 2008) . The measures of distance employed for defining neighborhood matrices were the K nearest neighbor function and the fixed distance band. The K nearest neighbor function defines a neighborhood as a user - defined given 178 number of nearest points. If, for exampl e, the parameters are set to look for the four closest points, then the function will include whatever four points are closest to the point in question - no matter their absolute distance from it - as the neighborhood. Conversely, the fixed distance band m easure relies on an absolute radial distance around the point as the neighborhood. In other words, it includes anything within a given distance (e.g., 10 meters) of that point. When working with prehistoric archaeological data, there sometimes is no clear - cut answer to what inherently makes sense, within the parameters of the dataset, as the outer limits of a spatial cluster. An archaeologist is an outsider attempting to make sense of remains without being aware of the choices and logic employed by the com munity at the time of the site reference, archaeologists must use what means and assumptions they can to look for structure in the data. In this case, doing so translates to playing around with a range of dista nces to find out how the data respond to them. For several artifact types (bone, undecorated body sherds, ostrich eggshell, glass) the global Moran the data is normally distributed. A positive index v alue indicates a tendency towards clustering, while a negative value indicates a tendency towards dispersion, with larger absolute values indicating stronger tendencies (for more information, see .The series of index values generated by inputting a variety of distances were then plotted those distance values in a scatterplot. The r esulting graph presents an indicator of whether clustering in that kind of artifact drops off gradually as first law), or whether any more sudden drop - of fs occur (which would be a good indicator of a neighborhood distances to plug in to the Getis - Ord Gi* function. These distances would then be tailored 179 Figure 35 Portion of the stone wall at Kaitshàa 180 Figure 36 Portion of the stone wall at Thabadimasego 181 to the parameters of the specific data in question, and the resulting hotspots identified would have much more meaning in the context of the site. As mentioned before, this approach was abandoned for the time being. After running the Moran I several times, it became apparent that the 35 different categories of a rtifact types employed in the spatial dataset would require a number of different neighborhood definitions, some of which were probably too small to be useful for the parameters of the data as sampled at Thabadimasego . For example, test pits were spaced at 10 meter intervals, and in some areas of the site, no additional units and pits were placed, so the minimum distance between points in those areas of the site is 10 meters. However, it is already known that charcoal, for instance, clusters on a much small er scale than 10 meters - more like 10 centimeters. While a one - meter scale could be useful in identifying clusters of charcoal, a one - meter scale - grained. Furthermore, other more ubiqui tous materials such as bone and ceramic likely cluster at a much larger scale, where using a 10 - meter distance as a neighborhood definition could be perfectly appropriate. In future research, it would be of considerable interest to pursue this line of inqu iry. It would be of particular interest to perform a comparative study over multiple sites with a range of scales. Ultimately it was decided that, for the purposes of this research project, these issues were too much of a challenge. This is particularly th e case when, as will be discussed in the following section, non - metric visual displays of material culture distribution proved sufficient for building a basic understanding of site layout at Thabadimasego . Summary of results Original versus standardize d da ta Overall, standardizing the counts of artifact types to a standard measure of volume proved to be an effective way to compensate for inconsistency in sampling procedure. While the standardize d data did appear to yield smoother distributions for the most part (there are fewer large breaks between values), the ranges of values and the shape of their distributions are, for the most part, the same o r similar to the 182 original data ( these may be seen in Appendix E) . There are a few exceptions to this though, wit h dramatically overrepresented values for certain artifact types. This m ay point to either errors in calculations of the standardize d data, or the presence of clustering at the scale of the individual pit for those particular artifact types (i.e., if locat ions of excavations happened to hit on a small dense area of something that wouldn finished OES beads, broken OES beads, decorated rim sherds, copper beads, and iron fragmen ts. Increases beyond the original range occurred for slag and undecorated body sherds as well, but not to the same degree - the standardize d data in these cases could well be representative of what could be found in a 1 - x - 1 - meter unit based on values that came out of other units. These findings underpin the importance of dataset one is working with. The presence of unexpectedly high values for certa in artifact types is, however, precisely what this analysis aimed to identify. When those artifact types that yielded unexpectedly large standardize d values - finished OES beads, broken OES beads, decorated rim sherds, copper beads, and iron fragments - ar e isolated in ArcMap, their distribution can be seen as a series of relatively dense clusters, mostly in the western half of the site ( Figure 37 ). When mudbrick and ash features are added back into the map display, the arrangement of the above artifact typ es becomes even more apparent ( F igure 38 ) . Furthermore, when additional artifact types are added to the map, those same locations tend to display higher density of those artifacts as well. This holds true whether using the original or standardize d datasets (but it was easier to pick out from the standardize d data tha nks to the exaggerated values) (F igure 39) . Additional, albeit slightly more dispersed, concentrations of material occurred within and just adjacent to the mudbrick remains, as well as in an ar ea several meters wide outside of the wall. More than likely, based on both the mudbrick remains and concentrations of other materials, the materials 183 excavated at Thabadimasego represent the remains of two structures (or possibly two parts of one structure ) - unit 3/5 and 19/20 respectively. Going back to the argument over what constitutes materials in terms of site organization, the mudbrick and ash clusters can be imagined as the sort of functional center of the site, and other materials under stood by their placement relative to these features. Adding yet another layer of information - the differentiation in soil matrices across the site - provides a further insight as to the site xists overall within the site in terms of soil horizons. The great majority of soil on the site is consistently the same color, texture, and composition - a dry, light brown silty sand with a minor clay component. What does vary is the degree to which the soil is compacted in the first 10 - 15 centimeters below the surface. Figure 40 shows the variation in soil compaction within the top soil horizon across the site, while Figure 41 shows the same information but including the distribution of key cultural ma terials. As the Figure shows, most of the densest concentrations of cultural material fall within (in the case of the mudbrick cluster) or just adjacent to the compacted soil horizon. Interpretation of activity areas shell beads As tempting as it would be to invoke the presence of the shell bead remains as evidence of a bead manufacturing workshop i.e., for shell bead production as a sustained and high - volume activity at Thabadimasego the spatial distribution of the shell on the site does not conclus ively support this idea. Shell was certainly worked on the site, but w on the site depends to some extent on how one interprets the density of the materials in question. For instance, production of a flaked stone tool, through the process of reduction from the original cobble, often leaves behind a highly concentrated scatter of lithic debitage that becomes an easily identified feature in the archaeological record. Makers of shell beads may not have left behind such a visible footprint. Schapera (1930) , one of the earliest ethnographers of San peoples, says the following about shell 184 Figure 37 Materials with unusually high values 185 Figure 38 Materials with unusually high values plotted with ash and mudbrick features 186 Figure 39 All material types; legend on following page 187 Figure 40 Legend for "all material types" density map bead production: 188 stone or iron borer. They are then threaded on to a strip of sinew and the rough edges chipped off with a horn. Soft bark fibre is next twisted between the beads, making the chain very taut, and the edges are Dubroc (2010) menti ons that, by way of evidence for shell bead production at Bosutswe, nearly 300 unfinished OES beads are found in the Lose (1200 - 1400 AD) levels in the central area of the site. He latively small area, unlike what has been observed for similar materials at Thabadimasego. Dubroc also notes the presence of part of the site. One broken grooved stone was recovered from Thabadimasego (Figure 42), but it is not of the softer calcrete native to the escarpment, and therefore was imported from another location. Sadr and Fauvelle - Aymar (2006) summarize the way that groove shapes and depths found at sites across sub - Saharan Africa may indicate a very wide range of different uses (grinding particular grains or pigments, and sharpening bone and stone tools, among others); therefore it seems hasty to assume that the grooved Site layout Overall, it would be entirely possible to interpret the layout as a c oncentric semi - circular ring extending west from the mudbrick cluster with the ash feature sitting in front of it. The majority of material clusters not associated with mudbrick appear to occupy the western half of the site in a loosely defined zone arci ng from the western wall remnant to just north of the Southern wall remnant. Numerous low - density artifact scatters occur among and outside of this area as well; this could be interpreted as the se of its occupation. Alternatively, the argument can also be made that no overarching structure exists to the site apart from the fact that material generally concentrates in its center (this effect can also be argued to be taphonomic, which will be discussed further in the next chapter). Some of the material clusters primarily 189 Figure 41 Variation in soil compaction 190 Figure 42 Variation in soil compaction with distribution of key cultural materials 191 feature one or two artifact types - the shell bead cluster near the west wall, for example, and the two pottery concentrations on the north and south edges of the central area. Without extended quantitative analysis, and especially without additional data collected at a similar resolution from other Early Iron Age sites, the placement of material clusters themselves can easily be construed as randomly placed middens within a given (the central area of the site). In the end, either interpreta tion is ultimately an arbitrary imposition of perceived structure, and a prioritization of some categories of information over others. The data can support either argument and, perhaps both interpretations should stand as competing hypotheses until further data is collected. Additional excavations at Thabadimasego - particularly in the central area where the mudbrick features lie - as well as new excavations at neighboring sites may change the picture substantially. Likewise, whether these interpretations c an be substantiated by independent statistical assessment is a mystery that will have to wait for another time (and maybe a more comprehensive dataset). The following chapter will delve into these matters as they speak to archaeological understandings of space and form beyond the immediate boundaries of this one site. Much can be said in the way of methodological comparisons between fieldwork at Thabadimasego and at other Early Iron Age sites, as well as of the interpretive ramifications different methodol ogies produce. 192 Figure 43 Fragment of grooved stone found on Thabadimasego's surface 193 Chapter 6 Conclusions In order to approach a synthetic understanding of the analytical results, it seems appropriate to return to the research questions posed in Chapter 1. What can the results of analysis discussed in Chapters 4 and 5 offer in terms of understanding diachronic variability in South Sowa, of the use of space both within walled sites and further abroad on the landscape, and of the forager - farmer relationship in the South Sowa area? This chapter will address each of these issues in turn. 6.1 Site level: Use of space and d iachronic variability Site - level analysis To recap very briefly the conclusions drawn in the previous two chapter s, Thabadimasego is a small hilltop site which was at one point walled off from the rest of the escarpment, although the spread of cultural material extends beyond the wall onto the escarpment. As previously mentioned , t he site contains a good diversity of material, from glass and shell beads to copper and iron jewelry to extensive bone and ceramic scatters. Small amounts of fragile material such as carbonized seeds, red ochre, and slag were also recovered. One or two structures were located in the center o f the site. No forges or burials were observed (it is possible that the two stone cairns located on the site could contain burials, but as the cairns were left undisturbed for the sake of their preservation, this remains unknown) . As discussed in chapters 4 and 5, the collective body of evidence recovered from Thabadimasego points to a single and likely short - term (i.e. single generation or less) occ upation of the site. The stratigraphy observed on the hilltop was fairly simple. In the center of the hill, a compacted horizon of light brown loamy sand overlay an uncompacted horizon of the same soil type, which gave way to large calcrete cobbles and eventually solid bedrock. Elsewhere (including on the adjoining escarpment), the soil w as uncompacted from the surface down to bedrock. Most cultural material occurred in the uncompacted layer, including in the center of the site. Very few features (hearths or ash pits, for example ) were visible in the soil matrix and there did not appear to be any evidence of this type of feature overlaying one 194 another. Radiometric samples likewise point to a single date (around 850 - 900 AD) for the deposits in the uncompacted horizon, while the compacted horizon in the center of the site may be somewhat youn ger. Cultural material also does not exhibit any clear differences among the soil horizons, apart from increasing in density below the compacted horizon in the center of the site (although it is possible that a formal spatial analysis incorporating depth below surface as a variable could produce more specific insights on this). Worked materials, such as pottery, glass, and metal, whose styles and methods of production could indicate a difference in time period or population (or both), likewise maintain con sistency throughout the depths of the deposits at Thabadimasego . This is true even for the glass beads - Zhizo and Chibuene beads, which are made from chemically distinct glass recipes and date to different periods, are both found at varying depths on the site. This, in fact, adds to the suggestion that the older Chibuene beads may have been curated and handed down over generations since their original point of entry onto the African continent. In other words, the organization of the site does not follow any known model for Early Iron Age village layouts, and given the fairly small area occupied by the site as well as its small quantity of structures, it does not follow that Thabadimasego was in fact a long - term residential occupation. This should be clear when comparing a map of Thabadimasego Bosutswe by Denbow et al. (2008:460). While dozens of grain bin foundations, stone clusters, and (see Denbow et al. 2008 for a map) , only a handful of these are present at Thabadimasego . Bosutswe also hosts a number of vitrified dung exposures, the likes of which only form under conditions of long - term, continuous accumulation. Stratigraphic profiles of both sites likewise demonstrate a similar difference in complexity: whereas multiple overlapping mudbrick (dhaka/ daga) floors and ash deposits are evident in the Central Precinct excavation profile from Bosutswe (Denbow et al. 2008:46), such features were rare finds at Thabadimasego , did not overlap at all, and only occurred in one soil horizon. 195 The same is true when comparing Thabadimasego with other contemporary sites around Southern Africa. Ndondondwane, an 8th - century single - occupation site in Sout h Africa is arguably the most similar to Thabadimasego in terms of site size and occupation length (although their environmental settings differ considerably, as detailed in the table below). Ndondondwane is half the size of Thabadimasego , yet includes a substantially greater degree of complexity in its features. Greenfield and , central and peripheral, which include among its features several middens, a charcoal preparation area, and a burial. The sites whose locations make them the most comparable to Thabadimasego - Mosu I and Kaitshàa, both located on the escarpment edges of the Sowa area - do not unfortunately have sufficient published data on their layout to characterize them accur ately. Thabadimasego clearly does not follow the Central Cattle Pattern; its few structures do not circle a kraal and its cultural material distribution gives no indication of any gendered or class - based divisions es not conform to the site layouts described by Denbow (1984) for any class of the Toutswe system, which in fact are quite similar to the CCP. Apart from these two, no other intra - site spatial model has been developed for Botswana to which the site t could be compared. There do seem to be discrete concentrations of cultural material, as with Ndondondwane, but these by and large appear to be middens of discarded materials instead of activity areas per se. Thabadimasego also contains no known burials (and given how shallow its sedimentary deposits are, is unlikely to contain any at all), and no apparent kraal around which its very few structures might sit. Overall Thabadimasego appears considerably less complex than many multi - generation village sites of its time. It may instead be - one occupied for a brief period of time that depicts a way of life for a very specific time and place. Given its anomalous, low - complexity features, it should also be characterized as some thing other than a long - term residential site. If Thabadimasego does not fit the Central Cattle Pattern spatially, it follows that the site cannot 196 Figure 44 Thabadimasego surface features 197 Figure 45 Unit 13 west wall profile 198 Figure 46 Unit 20 west wall profile 199 Figure 47 Unit 21 west wall profile 200 Figure 48 Cluster location estimates at Thabadimasego 201 Table 10 Comparison of selected Early Iron Age sites Site name Site area (m 2 ) Site area (hectares) Time period for site Features observed on site Site organization Components Location Environment Sources Thabadimasego 12000 1.2 9th - 10th c Small ceramic and shell bead middens, structures, stone walls, ash or dung Central structures, peripheral middens Single - phase occupation (Zhizo) South Sowa (Botswana) Mosu Escarpment/ Sowa Pan GPS tracklog, 2012 Kaitshàa 124340 12.4 7th - 10th c Glass bead cache, stone wall (other unknown); six small mounds unknown Multi - phase occupation (Zhizo; Leopard's Kopje) South Sowa (Botswana) Escarpment/ Sowa Pan GPS tracklog, 2012; Denbow et al. In press Mosu I 24000 2.4 10th c Ash, dung, ivory bangle cache unknown Likely single - phase South Sowa (Botswana) Mosu Escarpment/ Sowa Pan Reid and Segobye 2000a, b Bosutswe 32000 3.2 8th - 18th c O ver 200 stone features, inc . grain bin foundations, stone granary platforms, and 7 semi - circular stone walls on the top of the hill; additional 2 precincts: central and western Multi - phase occupation (Zhizo; Toutswe) Eastern Kalahari (Botswana) hardveld, hilltop Denbow et al. 2008 (estimate from site map) 202 Site name Site area (m 2 ) Site area (hectares) Time period for site Features observed on site Site organization Components Location Environment Sources gra in bins and other stone features at the base of the hill Schroda 150000 15 900 - 1000 AD clay figurines; houses encircling a large cattle enclosure Central Cattle Pattern Multi - phase occupation (Zhizo; LK/Leokwe) Shashe - Limpopo Valley (South Africa) plateau in valley van Schalkwyk and Hanisch 2002; Calabrese 2007 Ndondondwane 5000 0.5 750 AD Infant burial, charcoal prep, midden, domestic household complexes 2 zones: central and peripheral Single - phase occupation ( Ndondondwane) Thukela Valley (South Africa) river valley Greenfield and van Schalkwyk 2006; Fowler and Greenfield 2009 Table 10 (co n 203 be expected to conform to the social expectations laid out by th is model. Instead, some limited conclusions about social processes at the site may be drawn from the specifics of its material record. Activities conducted at Thabadimasego clearly included some subsistence activities, such as processing of animal remains an d cooking, and likely included small - scale shell bead manufacture as well. Hunting and herding both contributed to the diet of the site not yet clear, and which was considered more valuable is even less clear. There is no reason as yet to assume that cattle were especially important for occupants of Thabadimasego , as compared to any other protein source. There is reason, given the demonstrated importance of cattle elsewhere and at later dates, to kee p an eye out for things that would point to cattle being important, but assuming it a priori puts the cart before the horse in terms of interpreting data. Likewise, there is not sufficient evidence that livestock were actually kept on the site. The one exp anse of residue that may be ash or dung is, quite frankly, very small, and does not constitute its own horizon; the material occurs in small chunks rather than an expansive patch. Because of this, it is more likely to be discarded ash than dung. Overall, i n fact, there is not sufficient evidence that people actually established a permanent residence at the site. There are so few structures (possibly only one), and little in the way to suggest storage and hearth areas as might be expected from the layouts at Ndondondwane and Bosutswe. Evidence for behavior other than subsistence activities is, however, present at Thabadimasego . Clearly, the site further abroad. They decorated the ir ceramics in a style shared with other sites of the region, and they acquired metal and glass decorative objects through exchange networks. In addition, other evidence for non - economic activities is present at Thabadimasego . The tiny fragments of slag an d red ochre present on the site are not explainable by any known economic practice for the time period within Southern Africa, although the possibility exists that the slag fragments are the result of on - site small - scale iron tool repair; such a pattern ex ists for copper items in Mesoamerica ( H. Pollard, pers. comm. 2015) . A ritual purpose for 204 these items has also been suggested ( G. Whitelaw, personal communication), but it is not clear what that purpose would be. Little is known about ritual practices during the first millennium AD for this part of Botswana in the first place. Undoubtedly, these oddities will become explained in time with further research both in South Sowa and elsewhere in the country. A comment on the site - level methodology The inten sity of the sampling strategy employed (as described in Chapter 3 ) is, to my knowledge, unusual among archaeological investigations within Southern Africa. However, the use of a gridded approach incorporating both surface and sub - surface sampling was, for this project, inspired by the cultural resource management (CRM) data collection techniques I became familiar with during work with CRM firms in the Midwestern United States. Placing small test pits at regular intervals over a substantial area, even in an area where cultural material is already known to exist, is a technique commonly used in CRM both to identify features within a site as well as define the boundaries of the site. At Thabadimasego , a site whose extent was already roughly defined by the scatt er of pottery sherds clearly visible on its surface, demonstrated the utility of this technique inasmuch as i within the site, allowed for the development of an overall site layout model, and furthermore demonstrate d the continued extent of the site onto the escarpment beyond the stonewalled hilltop. In addition, the documentation of stratigraphic profiles in the test pits, at hundreds of points across the site, provided substantial information by which to make infor med interpretations regarding site formation processes. It should be noted, however, that such an intensive sampling strategy was successful in part thanks to the shallow deposits at Thabadimasego . Conducting a comprehensive gridded test pit survey over th e entire site would have proven much more difficult if the soil deposits were substantially deeper, especially given how rocky the soil matrix was (as this disallowed the possibility of employing a soil augur for testing). Even so, this technique is entire ly adaptable to a wide variety of circumstances. The coverage 205 may be modified to an interval larger than 10 meters, for example, or additional data collection techniques such as magnetometry and ground - penetrating radar may be employed in the service of ge nerating a site - wide layout model. Given the importance placed in Southern African Iron Age archaeology on site organization and spatial layouts, it must be said that comprehensive survey techniques, while time - consuming, are invaluable methods of collecti ng data that will help resolve some of the most pressing theoretical issues that the discipline faces today. 6.2 Intra - site scale: comparing with other sites and engaging with explanatory frameworks Thabadimasego on its own cannot, unfortunately, provide m uch if any insight into change over time in economic practices or material consumption in the South Sowa area. A s a period in the late first millennium AD, the site is very valuable, but it cannot say much about change over time or larger - scale variation. By way of comparison, Site 16 - A1 - 12 sits on the next closest escarpment hilltop to the west of Thabadimasego surface is dense with flaked stone and what (on casual inspection) appears to be multiple facies of pottery. While the single excavation unit placed on Site 16 - A1 - 12 bore witness to a very similar stratig raphic record as Thabad imasego ( Figure 48 ) , future work at this site may provide valuable information pertinent to interpreting the settlement history at South Sowa, particularly for shallow deposits containing multiple occupations. It remains a certainty as well that area excavations at numerous sites throughout the escarpment will need to occur before its Early Iron Age settlement history is understood well enough to characterize in detail. Thabadimasego and Kaitshàa , for example, present some interesting comparisons and contrasts - in particular the differences in their size and depth of deposits. However, it remains to be seen how other Early Iron Age sites will compare in terms of lengths of occupation, ranges of cultural material and structures, and how these individual depositional records sp eak to one another. Accordingly, this section takes a step back from the layout of the site itself in order to consider how site - level analysis informs an understanding of the broader context of the South Sowa 206 landscape. Any archaeological site is best understood in the context of its local physical environment, surrounding site record, and the known settlement/ use history of the area. No site makes sense without knowing something about the ter rain, climate, natural resources, and social history of the place it occupies. What constitutes the local environment for any given settlement, however, depends in part on how the settlers of that site perceive and use the space around them. For example, a hunter - gatherer whose range of mobility may extend dozens of kilometers in the normal course of a day may define in a way very differently from a farmer who spends the majority of his time on or near a homestead and Figure 49 Site 12 unit profile 207 the surrounding fields , or from transhumant pastoralists, whose concerns include access to water sources, grazing pasture, and protection of herds from predators . We can no t be certain how occupants of the South Sowa landscape in the late first millennium AD viewed the area or what they considered local, long - distance, or otherwise. What can be measured, however, is the location of currently known archaeological sites in relation to one another and an estimation of their function, in context of the settlement pattern and physical environment. Furthermore, much has already been written about Early Iron Age settlement patterns for Southern Africa and about typical activity patterns for agro - pastoral communities in general. In the present - day Mosu area, people ty pically spend their time in one of three places: the village itself, their cattle posts, or their farm fields. They will also head out to the bush to collect firewood, raw clay, mophane worms, and other economically important resources on a regular basis. This system of land use is fairly typical for rural Botswana (Silitshena and McLeod 1998). People tend to move around from one of these places to another fairly frequently; walking several kilometers in a day to get from place to place is nothing unusual. Likewise, herds of goats and cattle will cover several kilometers over the course of their grazing, since herds range more or less freely in Mosu (as, again, in much of Botswana) , even given the presence of leopards in the area . The space that the Mosu com munity occupies therefore ranges far beyond the boundaries of the village itself. It can be assumed that an Early Iron Age community would be no different in this regard. The archaeological literature has addressed this point to some extent already. As ha s been discussed in Chapters 1 and 2, Huffman Botswana offers one example of a hierarchic al system of villages participating in an integrated socioeconomic network. In this model, smaller, subordinate settlements situated in lowlands near water and arable land serve as providers of resources to larger, dominant elite hilltop sites. The larger settlements in turn maintained economic and social stability through regulation of herding strategies as 208 well as prestige goods. The villages in this system occupied a shared landscape across which each settlement utilized a different but interrelated set of resources. In other words, the Toutswe region would have comprised an integrated and self - contained socioeconomic entity not only because the settlements in that area existed in the same physical environment, and shared the same language and subsistence practices, but also because of their mutual interdependence. These communities presumably acted within the same set of rules about who uses what, who makes and who consumes what. The Mosu area can be thought of in similar terms, albeit on a smaller scale . The site files at the Botswana National Museum indicate that over a dozen Zhizo - era sites exist within 35 kilometers (although most of them are much closer than that) of Mosu village, most of which lie on or near the edge of the escarpment. Most of these sites have been identified as Zhizo only via a small sample of material collected from their surfaces, and a great deal of additional work will be needed to verify both the chronology of each site as well as their respective extents. Another couple of doz en sites containing either an unidentified Iron Age component or a mix of Early/ Later Iron Age components (again, as indicated by small - scale surface collection) also exist in the same area. Nothing definitive can be said as yet as to the nature of Early Iron Age settlement in the Mosu area - whether multiple self - sufficient villages co - existed, or a hierarchical settlement pattern was in play wherein one or two large elite villages extended control over numerous smaller satellite residential areas, or oth erwise. But the argument can still be made, given the density of contemporaneous sites in the area, that this locale did in fact comprise some sort of integrated settlement system. Consistencies in radiocarbon date ranges as well as material finds (Zhizo g lass trade beads in particular) among the three sites so far excavated in the area further support this premise (Reid and Segobye 2000a, b). The forthcoming publications by Denbow et al. ( 2015 ) featuring results of excavations at Kaitshàa will no doubt prove highly informative towards these issues. Although the material scatter documented on the escarpment outside the walls of Thabadimasego could be construed as a separate archaeological site (and it in fact is listed as a separate 209 one in the national register), it doesn better explained as a continuation of the hilltop occupation for the following reasons: there is no evidence of structures or activity areas per se, jus t fairly low - density scatters of material (and also because the two co - occur so closely together on the escarpment. The survey conducted in 2012 south of the edge of the escarpment as well as along the bluff edges, supports the observations made by Samuel (1999) and Main (2008) that Iron Age sites cluster along the escarpment edges. Of course, further survey work could always provide a different perspective, but for now it stands as a reasonable, evidence - based assessment of site distribution in the area. I t is important to bring attention to the fact that the observations made in this chapter, and indeed the questions that gave rise to this research, build upon a number of other bodies of research. In particular, Denbow tlement system around Toutswemogala, and especially the data - driven approach taken in his work, were foundational for the research conducted here publicati ons address organization at both the site scale and the landscape/ regional scale. Whereas his focus is placed more on the regional scale (most of his fieldwork at the time was aerial and foot survey), in this case the focus flips and takes a closer look a t the site level (much like Denbow 2002, 2008 has done since for Bosutswe). Denbow (1984) also calls for a multivariate approach for evaluating sites (as opposed to, for example, focusing only on establishing ceramic chronologies). He also points out the n eed for tatements parallel (as well as pre cede ) the discussion made in previous chapters of this dissertation. The concerns raised over 30 years ago by Denbow continue to be valid, especially as the frontiers of archaeological knowledge base expand to incorporate heretofore understudied areas like South Sowa. There is a continued need to examine critically the assumptions that 210 underpin archaeological interpretations of a social as well as economic nature; there is also a continued need to specify and account for co ntext - not only the immediate physical context of a site but its broader regional connections and environmental constraints. Even so, the Toutswe model cannot be applied directly for the South Sowa area. With the few sites for which a substantial body of evidence exists, a range of sizes is observable ( Kaitshàa is big, Thabadimasego is quite small, etc ; see Table 6 - 1 ); however, the sites are almost all situated on escarpment edges or on isolated hilltops. In other words, there does not appear to be the dif ferentiation in location choice in South Sowa between small, potentially subordinate sites and larger villages as has been observed in the Toutswe region. Neither is there, at least for Thabadimasego , a site layout indicative of a residence with a kraal, t herefore this begs the question of whether the South Sowa settlement system incorporated non - residential, speci al - purpose sites (similar to Binford 's 1980 logistic mobility model for f oragers, wherein certain locations on the landscape are utilized as either cache points or vantage points for hunting ) . The function of these sites is not yet established; they may have been temporary camps, procuration sites, or otherwise. Other possib ilities include that they functioned as lookout sites, were temporary seasonal camps, or even s plinter groups forming new villages. It seems clear, however, that Thabadimasego may have fit in to the South Sowa settlement scheme as one of these outpost or s pecialist sites. Considerably more fieldwork will need to be conducted in the area before a comprehensive idea can be had of either site - level organization practices or of settlement dynamics on a landscape scale , and further analysis may indicate that a n ew model of settlement will be warranted . 6.3 Forager - farmer interaction Because Thabadimasego provides a short - term, that contributed to the site, it is hard - pressed as a body of evidence to provide information about change over time in either material culture or social dynamics for the South Sowa area. As a snapshot though, it offers some insight to interpreting overlapping culture - 211 Interpretations of inter action between foraging and farming communities in Early Iron Age Southern Africa often rest upon the assumptions that 1) users and makers of non - luxury/ non - exotic goods are from the same community (Huffman 2 001) ; and 2) that specific socioeconomic traditions, passed down through generations as learned adaptations, are exclusive to specific communities ( Smith 1990, 1998). For many (though not all) researchers in Southern African archaeology, interaction be tween the three socioeconomic groupings of the later Holocene - that is, hunter - gatherers, herders, and farmers - is measured in the archaeological record by the presence of one type of material culture set - e.g. pottery and sheep bones - in a c ontext usually associated with another type - e.g., rock shelters (e.g., Denbow and Wilmsen 1990; Smith and Lee 1997; Smith 2001; Denbow 2002; Sadr 2002). Processes of interaction are usually typified as assimilation, displacement, coexistence/ symbiosis a nd/ or trade, or more than one of these over a period of centuries. The presence of iron tools in rock shelters or other typical hunter - gatherer sites is usually interpreted to mean trade between farmers and foragers, or assimilation of foragers into a far ming economy, depending on the proportion of Iron Age material present (e.g., Denbow 2002) . On the other hand, the presence of stone tools or ostrich eggshell beads in walled farmer settlements is generally interpreted as either evidence of trade or the pr esence of hunter - gatherer residents in the farming village (Smith 2001; Sadr 2005; Mitchell et al. 2008). In other words, similar kinds of assemblages are interpreted differently depending on the geographical location in which they are found; group identit y and ethnicity are assumed to be strongly tied to settlement locations as well as cultural material traditions. Acquisition of new technology and cultural practices is generally assumed to occur passively within hunter - gatherer communities only; the assum ption is implicit that agency in, and control over, change belonged solely to farmers. For example, the presence of ceramics, iron or livestock in hunter - gatherer sites is often taken to represent the passive incorporation of new traditions of knowledge in to hunter - gatherer communities (e.g., Sadr 2002). Underlying these distinctions is the supposition that some technology and subsistence activities, such as ceramic production or reliance on 212 animal products, are easily learned and diffused, while others, e. g. livestock herding or lithic tool production (see Smith 1990, 1998), are both the product of specialists and products restricted within certain sociocultural and cultural/ethnic boundaries. In other words, although material exchange may flow both ways, t he acquisition of new skills and knowledge is generally assumed to flow one way. Clearly, some assumptions about agency and power underpin this view. The roles of agency and choice in hunter - gatherer behavior are discussed in some case studies. F or exampl e , Thorp (2000) details the cooperative exchange relationship maintained by hunter - gatherer communities in South Africa Likewise, Mitchell et al. (2008) demonstrate that hunter - gath erers adopted cattle - and sheep - herding during the Early Iron Age in Lesotho. In general, though, agricultural communities are positioned as superior in both social and political power; they are both larger and stronger entities with whom peripheral hunter - gatherer communities must deal with (or be dealt with). Even these models rest upon problematic assumptions, however. As Hammond - Tooke (2000) writes, classifications of material culture in terms of supposed ethnic affiliations is - that is to say, if they grouped in discrete and non - overlapping assemblages - there would be little reason to question the social identity associated with the artifacts. However, this is not the case. As several researchers working in different parts of the Kalah ari have established, hunter - gatherer communities over the last 1500 years have adopted technology and acquired goods from herding and farming communities, but the reasons for doing so and the actual course by which these transferences occurred differed gr eatly in each case. At Cho/ana, Namibia, for example, Smith and Lee (1997) examine a long sequence of hunter - gatherer occupations, recording the presence of pastoralist ceramics and beads dating to the late first millennium AD. They argue that these items were acquired through trade for hxaro exchange, that evidence for sustained direct contact with outside food - producing groups did not occur in the area (contra 213 Denbow and Wilmsen 1986) , and that evidence for socioeconomic transformations does not appear in the archaeological record until the late nineteenth and twentieth centuries. Oral histories recorded from some of the area assemblages from rock shelters in south - east ern Botswana suggest that assimilation did occur in each case after contact with farmers, but that the way each happened was very different. Archaeology from deep in the Kalahari suggests that hunting and gathering remained a way of life till recently, but by the 1960s, it was no longer the sole source of subsistence as previously claimed (Smith and Lee 1997; Brooks 2002) . Smith (2001) also compares case studies from the Middle an d Late Iron Ages in northeastern Namibia and southeastern Botswana. He found that San of Namibia were in contact with Kavango farmers during the Late Iron Age, but only to a limited extent for trading purposes, while, according to changes in the material c ulture record, San in rock shelters around Gaborone were gradually encapsulated by the farming economy. Even open - air sites (as opposed to rockshelters) do not sort easily; van Zyl et al. (2013 :54 pen - air sites in the Okavango area cannot be g rouped into convenient categories such as farme r, pastoral or hunter - gatherer" ( cf. Sadr, 1997). This appears to be true for the South Sowa area too; even though certain aspects of the sites and their assemblages - namely, the pottery and the metal - bear strong connections to a la rger Early Iron Age system, this may not actually indicate a specific form of socioeconomic organization. There are, unfortunately, far fewer studies which evaluate changes in Iron Age assemblages in reference to the adoption of h unter - gatherer technology or socioeconomic behaviors. By and large, these focus on the acquisition of San ritual practices, not of material technology or economic behavior, by Bantu - speaking farming communities. Van der Ryst, Lombard, and Biemond (2004) write that appropriation of San ideology and ritual practices by farming communities was relatively common during the Later Iron Age. Hammond - Tooke's (1998, 1999) work on historic Nguni communities corroborates this, and suggests further that ideological borrowing was very specific, at least in the case of the Nguni. C ertain practices 214 such as trance dancing were adopted and recontextu alized for particular Nguni rituals; San ideology was not adopted wholesale (Hammond - Tooke 1998:14). Physical spaces with cosmological significance were likewise appropriated from their original San creators by Bantu - speakers in the Middle and Later Iron A ges, in south - eastern Botswana as well as north - eastern South Africa (S. Hall and Smith 200 0 ; van der Ryst et al. 2004). In these cases, even though knowledge moved from hunter - gatherer to food - producer populations (unlike what is suggested by the studies of hunter - gatherer sites discussed above), it would appear that th e power dynamic is no different, such that farming communities still maintain the controlling share of transactional power by co - opting San knowledge and places. The overall impression one is left with is of a subtle but uncomfortable tension between the evidence as excavated, and the manner in which it is framed and interpreted ( for example see Wilmsen, Dubow, and Sharp 1994; Wilmsen 2002, 2009) . Evidence clearly shows that behavior and material culture differ over time as well as by reg ion - particularly among hunter - gatherers living near herders and farmers, but also among the food - producing societies. Because of this, a number o f scholars (Hall 1984; Lane 1995 ; Denbow 1999; Sadr 2008) have questioned the utility of characterizing assem blages in terms of socioeconomic identity, much less ethnicity. At the same time, for lack of another well - developed frame of reference, archaeologists working in the Southern African Early Iron Age feel pressed to rely on those selfsame culture - historical tropes to build our arguments anyway. Theory alone, it is clear, will not be capable of resolving this tension, nor of clarifying specific regional chronologies. These issues must be addressed on both epistemological and methodological levels. The assembl age from Thabadimasego provides an opportunity to parse out some of these issues. As has been clearly demonstrated in this research, Thabadimasego is an open - air, stone - walled site located on a hilltop (a common choice of terrain for Iron Age sites in eastern Botswana). This single - occupation site contains materials classically associated with iron - tool - using agropastoralists (metal and glass jewelr y; domesticated animal remains; Zhizo pottery; permanent structures although not as many 215 structures as expected on a typical village site ) as well as some material classically associated with stone - tool - using hunter - gatherers (ostrich eggshell beads; wil d animal remains). Wild animals are, in fact, a common component of Early Iron Age faunal assemblages as well, while shell beads (both ostrich and Achatina ) are likewise frequently reported in Iron Age deposits, but often factored out of behavioral interpr etations of those sites. Like Bosutswe (Dubroc 2010), Thabadimasego contains evidence that both Achatina and ostrich eggshell beads were not only used but also produced on - site. Hundreds of shell beads and unworked fragments, in various conditions and stag es of production, were recovered at Thabadimasego . The shell beads also comprised one of the few clearly clustered midden areas on the site, and as such cannot be ignored in an interpretation of the site from what Achatina beads tend to cluster along the smaller end of the size spectrum. All sizes of beads are found concurrently in the deposits; there is no stratigraphic differentiation by size. The shell bead assemblage does not therefore fit the hypothesis originally proposed by Jacobsen (1987) that smaller beads were produced by hunter - gatherers and larger ones by food - producing communities. The tradition of shell bead production, of course, reaches back tens of thousands of years before the arrival of Bantus - speaking Iron Age populations in Southern Africa. Although the bead production tradition has furthermore been associated in ethnography with San - spe aking communities in Namibia, Botswana, and South Africa (Lee 1979; Tanaka 1980), there needs to be a consideration that these beads were both made and used on a regular basis by people who farmed and used metal. It does not make sense to infer the presenc e of a - gatherer ancestors made these beads too. It makes a lot more sense to infer that whoever occupied Thabadimasego , regardless of what languag e they spoke or what ethnic or social affiliation they identified as, was capable of learning that set of skills. 216 Likewise, hunting as well as herding contributed to the faunal assemblage at Thabadimasego , and presumably the diet of the site well. The presence of wild game in proportions of up to 60% is, of course, already documented for several sites occupied during the Early Iron Age in Botswana, including Bosutswe, Divuyu, Nqoma, and Xaro (Denbow 1999; van Zyl 2013). The number of individu al specimens (NISP) identified at Thabadimasego totaled 11 09 (see Figure 23 in Ch apter 4). Ho wever, it is too early to make a conclusive assessment of the proportions of game to stock at Thabadimasego . Fraser (2014) identification of the faunal assemblag e, while ongoing at the time of writing, has indicated that the great majority of the assemblage - 71%, or 786 NISP - consists of indeterminate mammals. Of these, only slightly more than half (n = 415) could be identified as bovids, meaning that while they were likely consumed for meat, it is still unclear what kind of animals these remains represent. Of the individual specimens that could be identified more precisely, wild animal remains (n = 285) - including game animals as well as birds, reptiles, rodent s, and carnivores - greatly outnumbered domestic stock (n= 36) by a factor of almost eight to one. However, of these wild fauna, only 18 were identified as large game (i.e. bovids and equids). Over half (n = 164) were rodents and quite possibly intrusive t o the deposits. 92 specimens were identified as small animals (such as hares and tortoises), one as a carnivore, and seven as birds of indeterminate species. The small animals, birds and even the carnivore may have also been food sources for the people of Thabadimasego , but without further information it is too early to tell, and too early to make direct comparisons with faunal assemblages from other Early Iron Age sites. It remains to be seen how the ratios of wild to domestic, and food to non - food, fauna species may change if a greater number of specimens can be assigned specific taxa. All that can be said with any confidence at this juncture is that people at Thabadimasego hunted, herded, and probably set traps and snares as well. Although paleoenvironmen tal reconstructions for the South Sowa area (particularly for the Holocene) are not yet robust, it may be assumed based on the taxonomic similarity between faunal remains recovered from Thabadimasego and those species known to populate the Makgadkikgadi Pans during the wet season 217 that South Sowa would have been an attractive hunting ground in the first millennium AD when the pan was filled with water. However, based on the growing number of sites for the first millennium AD in Botswana whose assemblages indicate exploitation of an admixture of wild and domestic food species, it does stand to reason that the evidence does not support the framework of the Early Iron Age (at least for Botswana) as a predominantly farming, and particularly cattle - centric, ti me period. At the very least, the food sources for Early Iron Age should be taken on a site - by - site, or perhaps region - by - region, basis. To return to the issue of inferring identity from material remains, the faunal assemblage in particular raises the que stion of whether or not the practice of distinguishing hunter - gatherers from herder - farmers is even a fruitful exercise for the study of Early Iron Age Botswana at this point, at least for the South Sowa area where knowledge of the archaeological record is so incomplete. This is particularly true given the absolute paucity of formal analyses for Early Iron Age botanical assemblages; there is simply too little extant data to be able to make comprehensive inferences about agricultural/ plant consumption prac tices in general for this time period. There is little doubt that cultural and economic traditions of totally different origins, derived from separate ancestral populations, contributed to the system of settlement and patterns of behavior collectively term ed the Early Iron Age. There is also compelling genetic evidence (e.g. Mitchell 2010; Schuster et al. 2010) that, in some cases, these populations and the communities they formed remained distinct up to a point. At the same time, hunting, herding, and farm ing all contributed to the diet of people at Thabadimasego as well as at other Early Iron Age sites. It appears as if a range of subsistence practices were in use in Botswana for the first 1000 years AD, and clear distinctions between ancestrally San and a n cestrally Bantu populations can not be made on the basis of their subsistence alone, even for open - air sites that include pottery (van Zyl et al. 2013). That pottery deriving from Bantu traditions (Urewe or Kalundu) is found in rockshelters alongside flake d stone tools (for an example, see Sadr 2002) for this same time period further underscores that boundaries of material usage were permeable. For example, Robbins (P ersonal communication , 2015) has found Divuyu 218 potsherds and flaked stone artifacts undernea th of a mine tailing at Tsodil o Hills. The assemblage from Thabadimasego bolsters the growing evidence for a range of region - based diversity in economic practice and social organization during the Early Iron Age in Southern Africa. As stated above, t here i s no a priori reason to assume that, for example, ostrich eggshell beads recovered from a deposit dating to about 900 AD were created by people who identified as hunter - gatherers both socially and economically. Instead, there is good cause to consider who the makers and users were on a case - by - case basis taking the specific contexts of the finds into consideration, rather than generalizing from broad theoretical assumptions developed out of ethnographic studies. It would be more productive to generate a ser ies of regional case studies focused on characterizing the full complement of a site features and what those features indicate about behavior - much like Schiffer (1995) and Schiffer et al. (2001) famously argue , and as Denbow (1982, 1984) likewise demonstrates for the landscape around Toutswemogala in east - central Botswana. As opposed to depending on models that prioritize normative socioeconomic identity, allowing the exc avated evidence to both set the foundation for - and provide feedback to - dynamic theoretical models will develop a stronger understanding of the range of variability present among Early Iron Age societies. The above discussion should in no way be taken as a suggestion to start with a blank slate, however. The existence of those normative models based on a culture - historical framework, against which the Thabadimasego assemblage may be compared, make it possible in the first place to recommend a modified a pproach. As is the case with any discipline, archaeology is the process of contributing to a continuous feedback loop of data and interpretation. As Denbow (1984) stated, there must exist a continuing interchange of information from the local scale to the regional to theoretical level and back again, and again. 6.4 Conclusion thoughts on identity and theoretical praxis Processes and behavior cannot be directly observed in the archaeological record; they cannot be 219 excavated or recorded. Instead, researche rs must make theory - based (but also, one hopes, data - informed) bridging arguments about the behavior that created the material record (Binford 1978) . We must always be clear about the differences between data and interpretation; too often interpretations are later construed as immovable facts or data in themselves. This leads to a weakening of our understanding of Sout hern African prehistory overall. The material record is raw, unprocessed data; the social and behavioral models, on the other hand, are highly processed interpretations and are therefore several steps removed from the archaeological record itself. We must always return to the extant data when we evaluate models in light of new data. We cannot rely solely on published interpretations as our basis for evaluating new data - we need to understand how the data compare directly, or as directly as is possible in a ny given circumstance. We have to, in short, be painstaking in setting forth just what our data consist of, and on what scale or scales they were measured, as well as what biases or shortcomings may have contributed to the creation of the dataset as it exi sts. At Thabadimasego , despite similarities in styles of material culture - particularly ceramics and metal - there really is not much to suggest that a specific set of non - subsistence - based social behaviors in general can be inferred. There is evidence, v ia the stylistic similarities of certain materials as well as the exotic trade goods, that connections existed; however, there is no real evidence as to what those connections actually meant to the people who participated in them. Furthermore, absolutely z ero indication exists of gendered behavior on this site (excluding any arbitrarily - imposed 19th - or 20th - century gender roles, as with the CCP). Likewise, there is also absolutely zero indication of ethnic affiliation, other than obvious geographic ties t o other Zhizo settlements. The pottery only indicates a stylistic affiliation, however, not an ethn ic affiliation. Pottery - making is a learnable skill, and arguably one that can be transferred without special equipment or generations of specialist training . Raw clay is fairly easily acquired - given the demonstrated mobility of historical San communities (Kelly 1995; Thacker 2006) , it can be assumed that even clay sources prized for their purity could be attained by any community 220 with sufficient mobility - and no special tools are necessary for pottery manufacture. This begs the question, why is it that archaeologists continue to assume that finding pottery of a certain style on a site is a definite indicator of actual genetic/ ethnic affiliation? It would bolster the Southern African research paradigm to question the assumption that, for the Earl y Iron Age at least, pottery styles were restricted to language/family groups (and therefore not transferred to other socioeconomic groups). In fact, when it comes to the question of group identity, we should ask ourselves why we care so much about establi shing socioeconomic identities for material assemblages in the EIA when there are much more interesting questions to pursue (such as what other regional subsistence patterns are there, and what other types of information can we get from something as common as pottery) that in time could also help us answer the question of identity in much more detail. In the end, there is no hard positive evidence that rigid social boundaries, or discrete groups of actors, existed during the Early Iron Age. By the same ligh t, there is also no hard positive evidence for a fluid, free - for - all melting pot. What we do appear to have evidence for, in fact, is a range of variation in combinations of material culture and economic practice. We ought to, however, let go of the presum ption that we know which particular, clearly - defined, ancestral ethnic group was responsible for these practices. We should start at the point of understanding the full range of variation of material culture and subsistence practices and trade. As has been said in previous chapters, a much more thorough understanding of settlement processes in South Sowa area is needed. Information about the diversity of sites present as well as secure dates for them needs to be collected before further social in ferences can be made with confidence. Based on the evidence at Thabadimasego and the other contemporary sites discussed above, it is also apparent for the need to frame Early Iron Age socioeconomic processes in terms of flexibility and variation. Denbow (1 982, 1984) laid out a foundation for this framework, but continued fieldwork and multivariate analysis is needed to fill in the specifics for sites as well as regions of Southern Africa. Additionally, ongoing 221 synthesis of case studies (for example, Mitchel l 2004; Mitchell and Whitelaw 2005) will be necessary to provide feedback as to how well any model fits the data. Taking a Southern African Iron Age archaeology, by building comparative datasets and bridging arguments, will help to a ddress many of the issues currently facing the discipline. Building a comparative, synthesized body of evidence for Early Iron Age (not to mention Later Iron Age) socioeconomic practices across multiple countries and for several centuries is a monumental t ask that will almost certainly never end, but is also needed. It will be increasingly important in the future to have collaborative research partnerships in order to accomplish this goal (especially as Batswana are being trained as archaeologists in growin g numbers). We need to work together, and we need to be open to multiple interpretations. The past may now be static, but our understanding of it is not. It is important to recognize that while the past can no longer be changed, our understanding of it doe s continue to evolve. Archaeologists should not be too attached to any one explanatory model, and instead adopt what best fits the data in question. Just as Early Iron Age communities did for their economic practices, adapting our interpretive practices w ill shape our work to best fit the lan dscape of scholarly knowledge. Further considerations This final section addresses a number of issues encountered during the research and analysis for this project. This section also raises potential future avenues of inquiry. It is my hope that this work will pave the way for continued research on the past settlement history of the South Sowa area. Issues encountered during research included sampling bias and analytical limitations. Coverage of the site by survey and excavation was fairly extensive, so it stands to reason that the majority of sub - surface features were identified. Because the initial survey covered all areas of the site equally, the overall distributional density for subsurface material should be an acc urate representation of the site remains possible that for the spatial analysis, both the regularity of the test pit coverage as well as the interval used (10 meters) could have introduced sampling bias that presents challenges to some statistical 222 techniques. Hotspot analysis, as discussed in Chapter 5, was discontinued for a number of reasons. One of these reasons was that 10 meters is the minimum distance between locations in some parts of the site and many artifact clusters wo uld have been much smaller than that. Furthermore, the gridded coverage sure without further testing. As is briefly mentioned in Chapter 4, some mater ials recovered from Thabadimasego were inventoried but not subject to additional analysis. This for the most part included any lithic material. So few flaked lithics were found - perhaps five in all - and they were not identifiable as products of intention al flaking, at least not without seeking out another comparable assemblage from an Early Iron Age site (if such an assemblage even exists). There were also a number of water - rolled pebbles collected during excavations which definitely were not born of the parent rock (calcrete) on the site. Some ideas about what they might be (transports from nearby sand beds, or ostrich gullet stones) were raised in discussion with other archaeologists working at the Botswana National Museum at the time (Staurset 2013, per sonal communication). Without having anything solid to go on, this too was left for future consideration. Finally, the botanical remains, while discussed in the analysis in Chapter 4, still deserve a much more comprehensive evaluation by a specialist at so me point. The research reported on here raises several future lines of inquiry worth pursuing in the future. Unsurprisingly, at the top of the list is continued survey and excavation at other sites in the South Sowa area. Building an understanding of the a rea archaeological record of Botswana. In particular, evaluation of other small, low - complexity sites which may parallel Thabadimasego will be of interest, as these sites may also be of inte rest for other regions where hierarchical settlement patterns exist. It would also be informative to conduct fine - grained spatial evaluations, similar to what has been done for this project, for other Early Iron Age sites in eastern Botswana. The methods u sed here ought to be replicated not only to build comparable datasets, but to 223 evaluate the resilience of the methodology itself. It would be of particular interest to collect spatial data for a site whose chronology is already well - documented so that I can see how accurately what I did in this project represents the depositional history of a site. There is also more research potential for Thabadimasego itself. Numerous charcoal samples were collected from the site but not submitted for AMS dating. Submittin g these for analysis would flesh out the radiocarbon record of the site. Continuing to work out the wrinkles of the higher - order spatial analysis is likely a task that will be pursued in the near future. Eventually, the spatial analysis will take into acco unt vertical depositional data as well - that is, material distributions by soil horizons and depth. Although the rationale for aggregating the materials from each unit and pit was presented in Chapter 5, it would be really interesting to go back and evalu ate those observations quantitatively. That could leave open the possibility of focusing solely on what might be construed the and the site layout might look different. Also, developing a user - friendly way to vis ually model quantified depositional distributions could be very useful to archaeology as a whole. Finally, it needs to be said that the most important future work of all is continuing collaboration with other scholars, both specialists and field re searchers alike. The field of Southern African Iron Age archaeology is, thankfully, ripe with people who are interested in testing models and building bodies of evidence, many of whom are just beginning their careers. Putting together our collective lines of inquiry will result in a robust and exciting research framework for decades to come. 224 AP PENDICES 225 APPENDIX A MISCELLANEOUS TABLES 226 Table 11 Pit and Unit Summary Unit Northing Easting # Levels Total depth (m) Length (m) Width (m) Total area (m^2) Positive 1 48.00 103.00 1.00 0.23 1.00 1.00 1.00 Y 2 65.00 95.00 3.00 0.25 1.00 1.00 1.00 Y 3 102.00 161.00 5.00 0.29 1.00 1.00 1.00 Y 4 81.00 92.00 4.00 0.57 1.00 1.00 1.00 Y 5 101.00 179.00 4.00 0.35 1.00 1.00 1.00 Y 6 111.00 180.00 7.00 0.37 1.00 1.00 1.00 Y 7 80.00 140.00 5.00 0.26 1.00 1.00 1.00 Y 8 80.00 139.00 5.00 0.24 1.00 1.00 1.00 Y 9 111.00 181.00 7.00 0.34 1.00 1.00 1.00 Y 10 100.00 199.00 8.00 0.33 1.00 1.00 1.00 Y 11 111.00 188.00 4.00 0.32 1.00 1.00 1.00 Y 12 71.00 227.00 4.00 0.21 1.00 1.00 1.00 Y 13 100.00 101.00 3.00 0.20 1.00 1.00 1.00 Y 14 99.00 101.00 4.00 0.25 1.00 1.00 1.00 Y 15 108.00 126.00 6.00 0.30 1.00 1.00 1.00 Y 16 88.00 152.00 5.00 0.29 1.00 1.00 1.00 Y 17 78.00 170.00 4.00 0.25 1.00 1.00 1.00 Y 18 90.00 208.00 4.00 0.18 1.00 1.00 1.00 Y 19 103.00 161.00 6.00 0.32 1.00 1.00 1.00 Y 20 97.00 176.00 5.00 0.26 1.00 1.00 1.00 Y 21 - 3.00 69.00 3.00 0.19 1.00 1.00 1.00 Y PTTP 1 10.00 71.00 1.00 0.20 0.40 0.40 0.16 Y PTTP 2 18.00 82.00 1.00 0.18 0.38 0.30 0.11 Y PTTP 3 18.00 72.00 1.00 0.15 0.39 0.37 0.14 Y PTTP 4 13.00 82.00 1.00 0.17 0.35 0.35 0.12 Y PTTP 5 23.00 82.00 1.00 0.21 0.34 0.39 0.13 Y PTTP 6 16.00 85.00 1.00 0.11 0.38 0.42 0.16 Y PTTP 7 23.00 92.00 1.00 0.11 0.39 0.34 0.13 Y PTTP 8 8.00 82.00 1.00 0.13 0.38 0.30 0.11 Y PTTP 9 - 2.00 92.00 1.00 0.15 0.30 0.28 0.08 Y PTTP 10 8.00 92.00 1.00 0.21 0.38 0.42 0.16 Y PTTP 11 23.00 102.00 1.00 0.14 0.42 0.34 0.14 Y PTTP 12 0.00 82.00 1.00 0.11 0.40 0.36 0.14 Y PTTP 13 0.00 85.00 1.00 0.15 0.32 0.27 0.09 Y PTTP 14 20.00 104.00 1.00 0.19 0.40 0.40 0.16 Y PTTP 15 33.00 102.00 1.00 0.12 0.30 0.37 0.11 Y PTTP 16 17.00 102.00 1.00 0.15 0.37 0.40 0.15 Y PTTP 17 17.00 109.00 1.00 0.12 0.37 0.33 0.12 N PTTP 18 30.00 109.00 1.00 0.12 0.24 0.28 0.07 N PTTP 19 43.00 102.00 1.00 0.11 0.30 0.40 0.12 N PTTP 20 33.00 92.00 1.00 0.16 0.32 0.30 0.10 Y PTTP 21 28.00 87.00 1.00 0.19 0.29 0.32 0.09 Y 227 Table 11 (co n PTTP 22 33.00 77.00 1.00 0.21 0.33 0.42 0.14 Y PTTP 23 23.00 72.00 1.00 0.21 0.45 0.33 0.15 Y PTTP 24 18.00 78.00 1.00 0.15 0.29 0.37 0.11 Y PTTP 25 18.00 72.00 1.00 0.26 0.25 0.30 0.08 Y PTTP 26 23.00 67.00 1.00 0.22 0.30 0.42 0.13 N PTTP 27 18.00 67.00 1.00 0.19 0.30 0.40 0.12 N PTTP 28 13.00 67.00 1.00 0.10 0.32 0.33 0.11 N PTTP 29 8.00 67.00 1.00 0.14 0.27 0.32 0.09 N PTTP 30 3.00 67.00 1.00 0.16 0.36 0.36 0.13 Y STTP 1 105.00 110.00 2.00 0.12 0.20 0.25 0.05 Y STTP 2 110.00 113.00 2.00 0.11 0.33 0.38 0.13 Y STTP 3 103.00 114.00 2.00 0.13 0.33 0.37 0.12 Y STTP 4 102.00 117.00 3.00 0.16 0.33 0.35 0.12 Y STTP 5 108.00 118.00 2.00 0.13 0.38 0.41 0.16 Y STTP 6 106.00 120.00 3.00 0.18 0.38 0.34 0.13 Y STTP 7 87.00 134.00 3.00 0.18 0.40 0.37 0.15 Y STTP 8 85.00 131.00 3.00 0.17 0.30 0.31 0.09 Y STTP 9 81.00 134.00 3.00 0.16 0.31 0.40 0.12 Y STTP 10 77.00 133.00 2.00 0.16 0.30 0.33 0.10 Y STTP 11 75.00 134.00 3.00 0.20 0.34 0.40 0.14 Y STTP 12 75.00 135.00 3.00 0.17 0.36 0.39 0.14 Y STTP 13 70.00 132.00 3.00 0.15 0.33 0.37 0.12 Y STTP 14 72.00 155.00 2.00 0.20 0.30 0.36 0.11 Y STTP 15 75.00 155.00 2.00 0.25 0.32 0.43 0.14 Y STTP 16 76.00 153.00 2.00 0.20 0.29 0.37 0.11 Y STTP 17 79.00 151.00 3.00 0.23 0.38 0.30 0.11 Y STTP 18 78.00 158.00 3.00 0.25 0.22 0.26 0.06 Y STTP 19 78.00 160.00 3.00 0.22 0.33 0.35 0.12 Y STTP 20 82.00 171.00 2.00 0.13 0.30 0.35 0.11 Y STTP 21 82.00 174.00 3.00 0.23 0.34 0.39 0.13 Y STTP 22 85.00 178.00 2.00 0.15 0.37 0.27 0.10 Y STTP 23 86.00 179.00 2.00 0.25 0.32 0.36 0.12 Y STTP 24 87.00 173.00 2.00 0.16 0.36 0.25 0.09 Y STTP 25 88.00 178.00 3.00 0.25 0.35 0.24 0.08 Y STTP 26 91.00 155.00 3.00 0.21 0.33 0.37 0.12 Y STTP 27 95.00 153.00 3.00 0.26 0.36 0.28 0.10 Y STTP 28 96.00 155.00 3.00 0.22 0.28 0.38 0.11 Y STTP 29 97.00 158.00 3.00 0.26 0.28 0.33 0.09 Y STTP 30 91.00 151.00 2.00 0.23 0.39 0.30 0.12 Y STTP 31 100.00 154.00 3.00 0.23 0.35 0.27 0.09 Y STTP 32 101.00 149.00 3.00 0.15 0.37 0.27 0.10 Y STTP 33 102.00 141.00 3.00 0.26 0.26 0.35 0.09 Y STTP 34 103.00 144.00 2.00 0.23 0.29 0.40 0.12 Y STTP 35 106.00 146.00 2.00 0.16 0.34 0.36 0.12 Y 228 Table 11 (co n STTP 36 108.00 142.00 2.00 0.17 0.34 0.28 0.10 Y STTP 37 107.00 150.00 3.00 0.11 0.28 0.38 0.11 Y STTP 38 110.00 175.00 3.00 0.23 0.36 0.30 0.11 Y STTP 39 99.00 196.00 2.00 0.22 0.22 0.33 0.07 Y TTP 1 55.00 95.00 1.00 0.20 0.50 0.50 0.25 Y TTP 2 57.00 104.00 1.00 0.19 0.45 0.45 0.20 Y TTP 3 59.00 114.00 1.00 0.26 0.45 0.45 0.20 Y TTP 4 62.00 124.00 1.00 0.16 0.45 0.45 0.20 Y TTP 5 46.00 100.00 1.00 0.20 0.50 0.50 0.25 N TTP 6 49.00 110.00 1.00 0.26 0.45 0.45 0.20 Y TTP 7 54.00 120.00 1.00 0.70 0.45 0.45 0.20 N TTP 8 60.00 130.00 1.00 0.20 0.45 0.45 0.20 N TTP 9 39.00 108.00 1.00 0.16 0.50 0.50 0.25 Y TTP 10 47.00 115.00 1.00 0.13 0.30 0.30 0.09 N TTP 11 51.00 124.00 1.00 0.09 0.50 0.50 0.25 N TTP 12 56.00 132.00 1.00 0.20 0.50 0.50 0.25 Y TTP 13 40.00 117.00 1.00 0.30 0.45 0.45 0.20 Y TTP 14 47.00 124.00 1.00 0.24 0.45 0.45 0.20 Y TTP 15 51.00 132.00 1.00 0.12 0.45 0.45 0.20 Y TTP 16 60.00 142.00 1.00 0.12 0.45 0.45 0.20 Y TTP 17 90.00 200.00 1.00 0.21 0.46 0.38 0.17 Y TTP 18 70.00 180.00 1.00 0.22 0.35 0.40 0.14 Y TTP 19 70.00 160.00 1.00 0.29 0.33 0.37 0.12 Y TTP 20 50.00 160.00 1.00 0.29 0.48 0.36 0.17 Y TTP 21 70.00 170.00 1.00 0.29 0.37 0.40 0.15 Y TTP 22 90.00 170.00 1.00 0.20 0.50 0.40 0.20 Y TTP 23 90.00 180.00 1.00 0.23 0.40 0.37 0.15 Y TTP 24 75.00 165.00 1.00 0.38 0.45 0.30 0.14 Y TTP 25 80.00 170.00 1.00 0.29 0.40 0.42 0.17 Y TTP 26 80.00 180.00 1.00 0.31 0.34 0.31 0.11 Y TTP 27 60.00 160.00 1.00 0.24 0.44 0.40 0.18 Y TTP 28 60.00 170.00 1.00 0.23 0.37 0.42 0.16 N TTP 29 60.00 180.00 1.00 0.12 0.44 0.42 0.18 Y TTP 30 100.00 190.00 1.00 0.26 0.32 0.40 0.13 Y TTP 31 100.00 200.00 1.00 0.24 0.43 0.34 0.15 Y TTP 32 100.00 210.00 1.00 0.19 0.38 0.36 0.14 N TTP 33 90.00 190.00 1.00 0.22 0.36 0.46 0.17 Y TTP 34 90.00 210.00 1.00 0.12 0.43 0.40 0.17 Y TTP 35 80.00 190.00 1.00 0.21 0.41 0.47 0.19 Y TTP 36 80.00 200.00 1.00 0.17 0.40 0.40 0.16 Y TTP 37 80.00 210.00 1.00 0.29 0.35 0.40 0.14 Y TTP 38 70.00 190.00 1.00 0.23 0.41 0.41 0.17 Y TTP 39 70.00 200.00 1.00 0.23 0.39 0.37 0.14 Y TTP 40 70.00 210.00 1.00 0.17 0.35 0.32 0.11 Y 229 Table 11 (co n TTP 41 60.00 190.00 1.00 0.16 0.39 0.38 0.15 Y TTP 42 60.00 200.00 1.00 0.19 0.46 0.40 0.18 Y TTP 43 60.00 210.00 1.00 0.15 0.38 0.40 0.15 Y TTP 44 90.00 160.00 1.00 0.28 0.32 0.27 0.09 N TTP 45 100.00 230.00 1.00 0.19 0.48 0.34 0.16 Y TTP 46 90.00 220.00 1.00 0.21 0.50 0.41 0.21 Y TTP 47 90.00 230.00 1.00 0.08 0.48 0.43 0.21 Y TTP 48 80.00 220.00 1.00 0.11 0.49 0.41 0.20 Y TTP 49 70.00 220.00 1.00 0.15 0.34 0.36 0.12 Y TTP 50 60.00 220.00 1.00 0.21 0.33 0.50 0.17 Y TTP 51 80.00 230.00 1.00 0.10 0.43 0.39 0.17 Y TTP 52 50.00 220.00 1.00 0.30 0.61 0.41 0.25 Y TTP 53 70.00 230.00 1.00 0.15 0.43 0.35 0.15 Y TTP 54 60.00 230.00 1.00 0.20 0.41 0.40 0.16 N TTP 55 110.00 229.00 1.00 0.18 0.50 0.40 0.20 Y TTP 56 120.00 230.00 1.00 0.21 0.37 0.40 0.15 Y TTP 57 110.00 220.00 1.00 0.16 0.35 0.38 0.13 Y TTP 58 120.00 220.00 1.00 0.20 0.34 0.38 0.13 Y TTP 59 110.00 210.00 1.00 0.29 0.44 0.36 0.16 Y TTP 60 120.00 210.00 1.00 0.15 0.46 0.41 0.19 Y TTP 61 128.00 210.00 1.00 0.19 0.38 0.41 0.16 Y TTP 62 110.00 200.00 1.00 0.20 0.40 0.30 0.12 Y TTP 63 120.00 200.00 1.00 0.22 0.40 0.35 0.14 Y TTP 64 128.00 200.00 1.00 0.20 0.35 0.32 0.11 Y TTP 65 110.00 190.00 1.00 0.19 0.38 0.37 0.14 Y TTP 66 120.00 190.00 1.00 0.21 0.40 0.35 0.14 Y TTP 67 110.00 180.00 1.00 0.22 0.34 0.39 0.13 Y TTP 68 120.00 180.00 1.00 0.19 0.39 0.34 0.13 Y TTP 69 110.00 170.00 1.00 0.15 0.39 0.32 0.12 Y TTP 70 120.00 170.00 1.00 0.18 0.44 0.42 0.18 Y TTP 71 110.00 160.00 1.00 0.23 0.31 0.37 0.11 Y TTP 72 120.00 160.00 1.00 0.29 0.47 0.38 0.18 Y TTP 73 100.00 150.00 1.00 0.28 0.52 0.41 0.21 Y TTP 74 110.00 150.00 1.00 0.21 0.35 0.37 0.13 Y TTP 75 120.00 150.00 1.00 0.22 0.43 0.45 0.19 Y TTP 76 100.00 140.00 1.00 0.23 0.39 0.45 0.18 Y TTP 77 110.00 140.00 1.00 0.24 0.34 0.44 0.15 Y TTP 78 120.00 140.00 1.00 0.12 0.43 0.36 0.15 Y TTP 79 100.00 130.00 1.00 0.25 0.46 0.40 0.18 Y TTP 80 110.00 130.00 1.00 0.24 0.40 0.50 0.20 Y TTP 81 122.00 130.00 1.00 0.16 0.43 0.40 0.17 Y TTP 82 100.00 120.00 1.00 0.31 0.39 0.40 0.16 Y TTP 83 110.00 120.00 1.00 0.24 0.45 0.40 0.18 Y TTP 84 120.00 120.00 1.00 0.14 0.40 0.39 0.16 Y 230 Table 11 (co n TTP 85 120.00 111.00 1.00 0.27 0.34 0.35 0.12 Y TTP 86 95.00 110.00 1.00 0.36 0.40 0.38 0.15 Y TTP 87 110.00 110.00 1.00 0.26 0.48 0.41 0.20 Y TTP 88 100.00 110.00 1.00 0.27 0.37 0.33 0.12 Y TTP 89 110.00 100.00 1.00 0.18 0.46 0.44 0.20 Y TTP 90 80.00 100.00 1.00 0.23 0.40 0.51 0.20 Y TTP 91 70.00 100.00 1.00 0.18 0.45 0.36 0.16 Y TTP 92 60.00 100.00 1.00 0.28 0.37 0.45 0.17 Y TTP 93 50.00 100.00 1.00 0.22 0.45 0.35 0.16 Y TTP 94 90.00 150.00 1.00 0.27 0.41 0.36 0.15 Y TTP 95 80.00 150.00 1.00 0.21 0.39 0.39 0.15 Y TTP 96 70.00 150.00 1.00 0.24 0.46 0.33 0.15 Y TTP 97 60.00 150.00 1.00 0.18 0.55 0.44 0.24 Y TTP 98 50.00 150.00 1.00 0.22 0.35 0.44 0.15 Y TTP 99 50.00 140.00 1.00 0.10 0.33 0.43 0.14 Y TTP 100 60.00 140.00 1.00 0.17 0.38 0.40 0.15 Y TTP 101 70.00 140.00 1.00 0.17 0.36 0.34 0.12 Y TTP 102 80.00 140.00 1.00 0.13 0.39 0.34 0.13 Y TTP 103 90.00 140.00 1.00 0.19 0.46 0.45 0.21 Y TTP 104 90.00 130.00 1.00 0.23 0.39 0.35 0.14 Y TTP 105 80.00 130.00 1.00 0.25 0.32 0.46 0.15 Y TTP 106 70.00 130.00 1.00 0.30 0.45 0.45 0.20 Y TTP 107 90.00 120.00 1.00 0.19 0.40 0.36 0.14 Y TTP 108 80.00 120.00 1.00 0.21 0.38 0.32 0.12 Y TTP 109 70.00 120.00 1.00 0.13 0.36 0.38 0.14 Y TTP 110 90.00 110.00 1.00 0.20 0.40 0.40 0.16 Y TTP 111 80.00 110.00 1.00 0.15 0.33 0.32 0.11 Y TTP 112 70.00 110.00 1.00 0.22 0.34 0.40 0.14 Y TTP 113 98.50 101.50 1.00 0.16 0.30 0.30 0.09 Y TTP 114 99.50 101.50 1.00 0.18 0.30 0.30 0.09 Y TTP 115 35.00 80.00 1.00 0.16 0.30 0.28 0.08 N TTP 116 35.00 90.00 1.00 0.11 0.28 0.29 0.08 N TTP 117 40.00 100.00 1.00 0.08 0.42 0.44 0.18 N TTP 118 50.00 230.00 1.00 0.11 0.39 0.36 0.14 N TTP 119 50.00 240.00 1.00 0.22 0.37 0.32 0.12 N TTP 120 60.00 91.00 1.00 0.18 0.40 0.34 0.14 N TTP 121 61.00 240.00 1.00 0.14 0.46 0.40 0.18 N TTP 122 90.00 100.00 1.00 0.14 0.41 0.35 0.14 N TTP 123 100.00 220.00 1.00 0.17 0.53 0.47 0.25 N TTP 124 100.00 240.00 1.00 0.09 0.39 0.33 0.13 N TTP 125 130.00 220.00 1.00 0.22 0.41 0.40 0.16 N TTP 126 100.00 175.00 1.00 0.37 0.40 0.30 0.12 Y TTP 127 100.00 170.00 1.00 0.33 0.38 0.34 0.13 Y TTP 128 100.00 165.00 1.00 0.24 0.35 0.25 0.09 Y 231 Table 12 Flotation inventory Lot Unit Level Location/ matrix Orig. vol. (nearest 100 ml) Flot. sample vol. (to 100 ml) Mass (g) Notes 30 3 2 Feature 1 5000 1000 1300 145 5 4 NW ¼ 1100 1000 1269 150 6 3 SE ¼ 1100 1000 1302 151 6 4 1000 1000 1395 153 6 14 - 28 cm 2300 1000 1354 155 6 28 - 34 cm Feature 2 500 500 858 578 g ceramics withheld from sample and not inc. in flot volume; bagged separate 157 6 24 - 34 cm Feature 2 600 600 964 164 7 4 SW ¼ 700 700 1117 325 g ceramics withheld from sample and not inc. in flot volume; bagged separate 169 8 3 Feature 3 700 700 1091 171 8 4 S ½ 900 900 1298 178 9 2 Along W wall 1100 1000 1298 186 9 6 SW ¼ 600 600 918 197 10 5 450 450 518 Small 1L pitcher is in 50 ml increments 198 10 5 Semi - packed soil 600 600 663 200 10 6 Center of unit, cluster of ash/ dhaka 500 500 566 208 11 2 SE ¼ 600 600 888 218 12 3 W ½ 200 200 305 221 13 1 N ½ 700 700 1090 225 13 2 W ½ 1000 1000 1474 225 13 3 W ½ 900 900 1358 228 14 1 N ½ 2000 1000 1274 230 14 2 NW ¼ 500 500 701 1st sample to use 1 mm sieve 232 14 3 NW ¼ 2400 1000 1332 233 14 4 SW ¼ only 800 800 1016 The outer bag also contained the general level bag which has same lot # 234 15 3 NW ¼, hard - packed soil 900 900 1174 232 Table 1 2 (co n 242 15 4 NE ¼, Semi - packed soil 600 600 945 243 15 4 SW ¼, 350 350 440 251 16 2 1000 1000 1183 253 16 3 NE ¼ 900 900 1106 263 17 3 SE ¼, Hard - packed soil 800 800 937 264 17 3 SE ¼, Semi - packed soil 700 700 903 269 18 2 NW ¼ 600 600 804 271 18 3 NE ¼ 1500 1000 1479 280 19 3 SE ¼, Hard - packed soil 2000 1000 1335 285 19 4 E ½ 600 600 894 287 19 5 NW ¼ 1000 1000 1331 288 19 4 Burnt soil/ stone surface 2500 1000 1498 292 20 1 NW ¼, Hard - packed soil 700 700 957 294 20 2 NW ¼, Hard - packed soil 500 500 736 296 20 3 NW ¼, Semi - packed soil 1000 1000 1333 297 20 2&3 N ½ gravel concentrate 800 800 1031 S ½ gravel in sep. bag inside outer bag, not processed in flot sample 299 20 4 SE ¼ 800 800 1100 347 21 2 NE ¼ 400 400 641 349 N40 E99 - 100 0 - 8 cmbs opening in wall, towards N portion of wall 1500 1000 1583 374 1 - 12 2 SW ¼, Hard - packed soil 2400 2000 2765 376 1 - 12 3 SW ¼ 1500 1000 1404 378 1 - 12 4 NW ¼ 1800 1000 1376 392 1 - 33 2 SE ¼ 2300 1000 1488 394 1 - 33 3 SE ¼ 1500 1000 1454 397 1 - 33 4 SE ¼ 1600 1000 1508 398 1 - 33 5 SE ¼, Semi - packed soil 900 900 1295 399 1 - 33 5 SW ¼, Loose - packed soil 1500 1000 1267 402 1 - 33 6 SE ¼ 1300 1000 1351 233 Table 12 (co n 403 1 - 33 6 E ½, Compact ashy matrix 1200 1000 1308 200 ml of compact ashy chunks withheld from flot 405 1 - 33 7 SE ¼ 1700 1000 1335 162 - 1 7 11 - 18 cm Feature 3, NW ¼ 2000 1000 1486 162 - 3 7 13 - 20 Feature 3, SE ¼ 2900 1000 1672 234 Table 13 Ceramic facies determinations Lot Unit # Level # Sherd type Impression type Motif Paint/ burnish Layout Facies Notes 22 2 1 DB CS single line N/A Unknown Probable Zhizo 22 2 1 DB LI 3 - 4 banded LI N/A Unknown Probable Zhizo 22 2 1 DB LI 3 - 4 banded LI N/A Unknown Probable Zhizo one incision is possibly punctat e 22 2 1 DB CS & LI diagonal banded LI between two rows of CS N/A Unknown Probable Zhizo 22 2 1 DB LI single line red paint Unknown Probable Zhizo 23 2 2 DB CS & LI two rows of angled LI bounded by CS N/A Unknown Probable Zhizo 23 2 2 DB CS & LI two rows of angled LI bounded by CS N/A Unknown Probable Zhizo 23 2 2 DB CS & LI single row angled LI bounded by CS N/A Unknown Probable Zhizo 23 2 2 DB CS angled lines of CS bounded by horizontal CS N/A Unknown Probable Zhizo 25 2 3 DB LI multiple bands N/A Unknown Probable Zhizo 25 2 3 DB CS & LI two rows of angled LI bounded by CS N/A Unknown Probable Zhizo 235 Table 13 (co n 25 2 3 DB wide CS single row wide CS ; angled LI N/A Unknown Probable Zhizo 26 4 1 DB CS single row CS N/A Unknown Probable Zhizo 26 4 1 DB CS & LI single row angled LI bounded by CS N/A Unknown Probable Zhizo 26 4 1 DB CS & LI single row angled LI bounded by CS N/A Unknown Probable Zhizo 26 4 1 DB CS & LI two rows angled LI separated by row of CS N/A Unknown Probable Zhizo 26 4 1 DB CS & LI 2 rows of: horiz CS row above wide - spaced angled LI N/A Unknown Probable Zhizo 26 4 1 DB CS & LI 1 row of: horiz CS row above wid e - spaced angled LI N/A Unknown Probable Zhizo 27 3 2 DB CS one line on neck, on e lin e on shoulder N/A neck and shoulder zhizo 29 4 2 DB triangular punctate alternate large + small triangular punctate, 1 row faint red paint Unknown ? Ziwa 29 4 2 DB CS 2 rows of CS N/A Unknown Probable Zhizo 236 Table 13 (co n 29 4 2 DB CS & LI 2 rows of : ho riz CS row above angled LI band N/A Unknown Probable Zhizo 29 4 2 DB CS & LI 1 row of : horiz CS row above angled LI band N/A Unknown Probable Zhizo 29 4 2 DB CS & LI angled band LI + single row CS N/A Unknown Probable Zhizo 33 4 3 DB CS 2 widely spaced obliquely angled CS rows N/A Unknown Probable Zhizo 33 4 3 DB CS & LI three rows angled LI separated by row of CS N/A Unknown Zhizo 33 4 3 DB CS & LI three rows angled LI separated by row of CS N/A Unknown Zhizo 34 3 3 DB LI multiple banded LI N/A Unknown Probable Zhizo 36 3 3 DB vertical short curved incisions single row N/A Unknown Probable Zhizo 36 3 3 DB CS & LI multiple diagonal LI bordered by linear CS N/A Unknown Probable Zhizo 37 5 1 DB CS Single row of CS N/A neck Probable Zhizo 37 5 1 DB CS Single row of CS N/A Unknown Probable Zhizo 237 Table 13 (co n 37 5 1 DB CS Single row of CS N/A Unknown Probable Zhizo 37 5 1 DB CS two rows CS N/A Unknown Probable Zhizo 37 5 1 DB CS v - shaped CS line N/A Unknown Probable Zhizo 37 5 1 DB CS two CS lines at oblique angles N/A Unknown Probable Zhizo 37 5 1 DB CS & LI three rows angled LI separated by row of CS N/A Unknown Probable Zhizo 37 5 1 DB CS & LI angled LI band bounded by CS rows N/A Unknown Probable Zhizo 37 5 1 DB CS & LI two rows of angled LI bounded by CS N/A Unknown Probable Zhizo 37 5 1 DB LI single band of fine LI N/A Unknown Probable Zhizo 43 5 2 DB CS & LI two rows of angled LI bounded by CS N/A Unknown Probable Zhizo 45 5 3 DB CS & LI Single CS row above LI band N/A Unknown Probable Zhizo 45 5 3 DB LI single band of LI N/A Unknown Probable Zhizo 146 5 4 DB CS & LI three rows angled LI separated by row of CS N/A Unknown Probable Zhizo 238 Table 13 (co n 147 6 1 DB LI single band LI N/A Unknown Probable Zhizo 148 6 2 DB LI single band LI N/A Unknown Probable Zhizo 148 6 2 DB LI single band LI N/A Unknown Probable Zhizo 149 6 3 DB dashed LI single row dashed LI N/A Unknown Probable Zhizo 150 6 3 DB LI single LI N/A Unknown unknown 151 6 4 DB CS Single row of CS N/A Unknown Probable Zhizo 151 6 4 DB CS & LI Single CS row above LI band N/A Unknown Probable Zhizo 159 7 1 DB CS & LI single row CS above angled LI band N/A unknown Probable Zhizo 160 7 2 DB LI single band N/A Unknown Probable Zhizo 160 7 2 DB CS two lines N/A Unknown Probable Zhizo 160 7 2 DB CS & LI one line CS, multiple angled LI N/A Unknown Probable Zhizo 160 7 2 DB CS & LI two lines CS bordering band of angled LI N/A Unknown Probable Zhizo 160 7 2 DB CS single line N/A Unknown Probable Zhizo 160 7 2 DB CS & LI single line CS with band of LI N/A Unknown Probable Zhizo 160 7 2 DB CS three lines of CS N/A Unknown Probable Zhizo 239 Table 13 (co n 160 7 2 DB CS & LI one line CS, multiple angled LI N/A Unknown Probable Zhizo 161 7 3 DB CS & LI three lines of CS with band of LI N/A Unknown Probable Zhizo 161 7 3 DB CS & LI single line of CS with band of LI N/A Unknown Probable Zhizo 161 7 3 DB CS two lines of CS N/A Unknown Probable Zhizo 161 7 3 DB CS single line of CS N/A Unknown Probable Zhizo 161 7 3 DB CS single line of CS N/A Unknown Probable Zhizo 161 7 3 DB CS three lines of CS N/A Unknown Probable Zhizo 161 7 3 DB CS & LI single line of CS with two oblique bands of LI N/A Unknown Probable Zhizo 161 7 3 DB CS & LI two lines of CS wit band of LI N/A Unknown Probable Zhizo 161 7 3 DB LI three lines of LI N/A Unknown Probable Zhizo 163 7 4 DB CS two lines of CS N/A Unknown Probable Zhizo 163 7 4 DB CS & LI one line of CS multiple angled LI N/A Unknown Probable Zhizo 163 7 4 DB CS & LI one line CS, su r rounded by multiple angled LI N/A Unknown Probable Zhizo 240 Table 13 (co n 163 7 4 DB CS single line of CS N/A Unknown Probable Zhizo 163 7 4 DB LI single linear incision N/A Unknown Probable Zhizo 163 7 4 DB CS & LI single line CS with band of LI N/A Unknown Probable Zhizo 163 7 4 DB CS single line of CS N/A Unknown Probable Zhizo 165 7 5 DB CS single line of CS N/A Unknown Probable Zhizo 166 8 1 DB CS single line CS N/A Unknown Probable Zhizo 167 8 2 DB CS single line N/A Unknown Probable Zhizo 167 8 2 DB CS & LI single line CS with band of LI N/A Unknown Probable Zhizo 167 8 2 DB CS three lines at oblique angles N/A Unknown Probable Zhizo 167 8 2 DB LI single line N/A Unknown Probable Zhizo 167 8 2 DB CS single line CS N/A Unknown Probable Zhizo 167 8 2 DB CS & LI two lines CS bordering band of angled LI N/A Unknown Probable Zhizo 167 8 2 DB LI mult i ple LI in angled band N/A Unknown Probable Zhizo 167 8 2 DB CS single line CS N/A Unknown Probable Zhizo 167 8 2 DB CS single line CS N/A Unknown Probable Zhizo 241 Table 13 (co n 170 8 4 DB CS & LI single line of CS and LI N/A Unknown Probable Zhizo 170 8 4 DB LI two lines of LI N/A Unknown Probable Zhizo 170 8 4 DB CS & LI single line of CS bordering band of LI N/A Unknown Probable Zhizo 170 8 4 DB CS & LI single line with multiple bands of LI N/A Unknown Probable Zhizo 170 8 4 DB CS & LI single line of CS with multiple bands of LI N/A Unknown Probable Zhizo 170 8 4 DB CS & LI single line of CS and LI N/A Unknown Probable Zhizo 170 8 4 DB N/A n/a red paint Unknown unknown 170 8 4 DB N/A n/a red paint Unknown unknown 176 9 1 DB CS two lines N/A Unknown Probable Zhizo 176 9 1 DB LI single line N/A Unknown Probable Zhizo 179 9 3 DB li multiple bands of L I N/A Unknown Probable Zhizo 181 9 4 DB CS & LI single line CS with band of LI N/A Unknown Probable Zhizo 181 9 4 DB CS & LI single CS with band of LI N/A Unknown Probable Zhizo 181 9 4 DB CS single line N/A Unknown Probable Zhizo 181 9 4 DB LI band of LI N/A Unknown Probable Zhizo 242 Table 13 (co n 181 9 4 DB CS & LI single line of CS with band of LI N/A Unknown Probable Zhizo 181 9 4 DB CS & LI two lines of CS bordering a band of angled LI N/A Unknown Probable Zhizo 181 9 4 DB LI multiple LI N/A Unknown Probable Zhizo 183 9 5 DB CS & LI single line of CS with multiple LI N/A Unknown Probable Zhizo 187 9 7 DB LI band of multiple LI N/A Unknown Probable Zhizo 190 10 2 DB LI single LI N/A Unknown Probable Zhizo 192 10 3 DB CS two very close lines of CS N/A Unknown Probable Zhizo 211 11 4 DB CS single line of CS N/A Unknown Probable Zhizo 214 12 1 DB LI single LI N/A Unknown Probable Zhizo 215 12 2 DB CS three lines of CS N/A Unknown Probable Zhizo 215 12 2 DB CS two lines of CS N/A Unknown Probable Zhizo 217 12 3 DB CS two lines of CS N/A Unknown Probable Zhizo 219 12 4 DB CS two lines of CS N/A Unknown Probable Zhizo 231 14 3 DB CS three lines of CS N/A Unknown Probable Zhizo 243 Table 13 (co n 231 14 3 DB CS & LI one line of CS with two lines of LI N/A Unknown Probable Zhizo 236 15 1 DB CS two lines of CS N/A Unknown Probable Zhizo 236 15 1 DB LI two LI N/A Unknown Probable Zhizo 238 15 2 DB LI multiple LI N/A Unknown Probable Zhizo 240 15 3 DB CS two rows of CS N/A Unknown Probable Zhizo 241 14 4 DB LI two lines of LI N/A Unknown Probable Zhizo 241 14 4 DB LI f our LI N/A Unknown Probable Zhizo 241 14 4 DB LI single LI N/A Unknown Probable Zhizo 241 14 4 DB CS single line of CS N/A Unknown Probable Zhizo 241 14 4 DB LI double line of LI N/A Unknown Probable Zhizo 241 14 4 DB CS single line of CS N/A Unknown Probable Zhizo 244 15 5 DB CS & LI single line of CS with multiple LI N/A Unknown Probable Zhizo 244 15 5 DB LI multiple bands of angled LI separated by large LI N/A Unknown Probable Zhizo 244 15 5 DB CS numerous lines of CS N/A Unknown Probable Zhizo 244 Table 13 (co n 244 15 5 DB CS & LI single line of CS with multiple LI N/A Unknown Probable Zhizo 244 15 5 DB LI multiple bands of LI N/A Unknown Probable Zhizo 246 15 6 DB CS & LI two lines of CS w i th multiple LI N/A Unknown Probable Zhizo 246 15 6 DB CS two lines of CS N/A Unknown Probable Zhizo 246 15 6 DB LI single LI N/A Unknown Probable Zhizo 246 15 6 DB LI two LI N/A Unknown Probable Zhizo 248 16 1 DB CS & LI three lines of CS bordering a band of LI N/A Unknown Probable Zhizo 250 16 2 DB CS single row of CS N/A unknown Probable Zhizo 250 16 2 DB CS & LI single line of CS with multiple bands of LI N/A unknown Probable Zhizo 250 16 2 DB CS single row of CS N/A unknown Probable Zhizo 250 16 2 DB CS & LI single line of CS with band of LI N/A unknown Probable Zhizo 252 16 3 DB LI multiple LI N/A unknown unknown 254 16 4 DB CS & LI single line of CS with multiple bands of LI N/A Unknown Probable Zhizo 254 16 4 DB CS & LI two lines of CS w i th multiple LI N/A Unknown Probable Zhizo 254 16 4 DB CS single row of CS N/A Unknown Probable Zhizo 245 Table 13 (co n 254 16 4 DB CS & LI angled LI band bounded by CS rows N/A Unknown Probable Zhizo 254 16 4 DB CS & LI two rows of angled incisions bounded by CS N/A Unknown Probable Zhizo 254 16 4 DB CS & LI single line of CS with band of LI N/A Unknown Probable Zhizo 254 16 4 DB LI band of LI N/A Unknown Probable Zhizo 256 16 5 DB CS or triangular punctate single row of punctate N/A Unknown Probable Zhizo 258 16 4 DB CS & LI angled LI band bounded by CS rows N/A Unknown Probable Zhizo 258 16 4 DB CS single row of CS N/A Unknown Probable Zhizo 258 16 4 DB CS & LI angled LI band bounded by CS rows N/A Unknown Probable Zhizo 258 16 4 DB CS & LI angled LI band bounded by CS rows N/A Unknown Probable Zhizo 259 17 1 DB CS & LI single row CS with two LI N/A Unknown Probable Zhizo 259 17 1 DB LI two LI N/A Unknown Probable Zhizo 259 17 1 DB CS & LI two rows CS bordering a band of LI N/A Unknown Probable Zhizo 246 Table 13 (co n 261 17 2 DB LI band of LI bordered by a large LI N/A Unknown Probable Zhizo 265 17 3 DB CS & LI single line of CS with a band of angled LI N/A Unknown Probable Zhizo 265 17 3 DB LI multiple lines of LI N/A Unknown Probable Zhizo 265 17 3 DB CS & LI single line of CS above a band of angled LI N/A Unknown Probable Zhizo 265 17 3 DB CS single line of CS N/A Unknown Probable Zhizo 265 17 3 DB CS & LI single line of CS above a band of angled LI N/A Unknown Probable Zhizo 265 17 3 DB CS & LI two rows of CS bordering a band of LI N/A Unknown Probable Zhizo 265 17 3 DB CS & LI single row of CS with a single LI N/A Unknown Probable Zhizo 265 17 3 DB LI band of LI N/A Unknown Probable Zhizo 265 17 3 DB LI band of LI N/A Unknown Probable Zhizo 265 17 3 DB CS multiple lines of CS N/A Unknown Probable Zhizo 268 18 1 DB CS & LI mul tiple lines of CS bordered by single LI N/A Unknown Probable Zhizo 247 Table 13 (co n 268 18 1 DB CS & LI band of LI bordered by a large CS N/A Unknown Probable Zhizo 268 18 1 DB CS & LI two bands of LI separated by a line of CS N/A Unknown Probable Zhizo 270 18 2 DB CS & LI band of LI above a single line of CS N/A Unknown Probable Zhizo 270 18 2 DB LI two parallel LI N/A Unknown Probable Zhizo 270 18 2 DB CS numerous lines of CS in a parallel band N/A Unknown Probable Zhizo 270 18 2 DB LI single line of LI N/A Unknown Probable Zhizo 270 18 2 DB CS & LI single line of CS forming a 'V' shape with a LI N/A Unknown Probable Zhizo 270 18 2 DB LI two LI on a raised lip N/A Unknown Probable Zhizo 281 19 3 DB CS & LI single line of CS close and parallel to a LI N/A Unknown Probable Zhizo 284 19 4 DB CS two parallel lines of CS N/A Unknown Probable Zhizo 284 19 4 DB CS & LI single line of CS above a band of angled LI N/A Unknown Probable Zhizo 284 19 4 DB CS two lines of CS N/A Unknown Probable Zhizo 248 Table 13 (co n 289 19 6 DB CS & LI two lines of CS bordering two bands of LI N/A Unknown Probable Zhizo 291 20 1 DB CS single line of CS N/A Unknown Probable Zhizo 293 20 2 DB CS single line of CS N/A Unknown Probable Zhizo 293 20 2 DB CS single line of CS N/A Unknown Probable Zhizo 293 20 2 DB LI s ingle LI N/A Unknown Probable Zhizo 293 20 2 DB LI band of multiple LI N/A Unknown Probable Zhizo 293 20 2 DB N/A n/a painted red Unknown unknown 295 20 3 DB CS & LI single line of CS with multiple LI N/A Unknown Probable Zhizo 295 20 3 DB CS & LI two lines of CS bordering a band of angled LI N/A Unknown Probable Zhizo 295 20 3 DB CS & LI two lines of CS bordering a band of angled LI N/A Unknown Probable Zhizo 295 20 3 DB CS single line of CS N/A Unknown Probable Zhizo 295 20 3 DB N/A n/a painted red Unknown unknown 298 20 4 DB LI three lines of dashed LI N/A Unknown Probable Zhizo 298 20 4 DB LI band of LI N/A Unknown Probable Zhizo 249 Table 13 (co n 307 N106 E120 2 DB CS + LI multipl e rows of CS bordering bands of LI N/A unknown Probable Zhizo two fragments that refit 308 N87 E134 3 DB CS single row CS N/A unknown Probable Zhizo 308 N87 E134 3 DB CS + LI row of CS bordering band of LI N/A unknown Probable Zhizo 308 N87 E134 3 DB CS + LI row of CS bordering band of LI N/A unknown Probable Zhizo 310 N81 E134 3 DB CS & LI row of CS bordering band of LI N/A unknown Probable Zhizo 318 N79 E151 3 DB CS & LI two bands LI bordered by rows of CS N/A unknown Probable Zhizo 320 N78 E160 3 DB CS & LI one line of CS and one of LI N/A unknown Probable Zhizo 322 N82 E174 3 DB LI band of LI N/A unknown Probable Zhizo 322 N82 E174 3 DB CS three horiz lines of CS N/A unknown Probable Zhizo 327 N91 E155 3 DB CS & LI multipl e rows of CS bordering bands of LI N/A unknown Probable Zhizo 327 N91 E155 3 DB CS & LI multipl e rows of CS bordering bands of LI N/A unknown Probable Zhizo 327 N91 E155 3 DB CS & LI multipl e rows of CS bordering bands of LI N/A unknown Probable Zhizo 250 Table 13 (co n 328 N95 E153 2 DB CS single row CS N/A unknown Probable Zhizo 328 N95 E153 3 DB CS single row CS N/A unknown Probable Zhizo 328 N95 E153 3 DB CS single row CS N/A unknown Probable Zhizo 329 N96 E 155 2 DB CS & LI multipl e rows of CS bordering bands of LI N/A unknown Probable Zhizo 330 N97 E158 3 DB CS several rows of CS N/A unknown Probable Zhizo 330 N97 E158 3 DB CS several rows of CS N/A unknown Probable Zhizo 330 N97 E158 3 DB CS several rows of CS N/A unknown Probable Zhizo 331 N91 E151 2 DB CS three horiz lines of CS N/A unknown Probable Zhizo 333 N101 E149 1 DB LI band of LI N/A unknown Probable Zhizo 333 N101 E149 2 DB CS & LI multipl e rows of CS bordering bands of LI N/A unknown Probable Zhizo 333 N101 E149 3 DB CS single row CS N/A unknown Probable Zhizo 333 N101 E149 3 DB CS & LI row of CS bordering band of LI N/A unknown Probable Zhizo 333 N101 E149 3 DB LI band of LI N/A unknown Probable Zhizo 333 N101 E149 3 DB LI band of LI N/A unknown Probable Zhizo 251 Table 13 (co n 333 N101 E149 3 DB CS single row CS N/A unknown Probable Zhizo 334 N102 E141 3 DB LI two lines N/A unknown Zhizo 335 N103 E144 2 DB CS single row CS N/A unknown Probable Zhizo 335 N103 E144 2 DB CS & LI row of CS bordering band of LI N/A unknown Probable Zhizo 337 N108 E142 2 DB CS single row CS N/A unknown Probable Zhizo 337 N108 E142 2 DB CS & LI row of CS bordering band of LI N/A unknown Probable Zhizo 339 N110 E175 2 DB CS single row CS N/A unknown Probable Zhizo 339 N110 E175 2 DB CS two rows CS N/A unknown Probable Zhizo 339 N110 E175 3 DB round punctate single curving row round punctate N/A unknown Probable Zhizo 339 N110 E175 3 DB CS rows of CS bordering angled band of CS lines N/A unknown Probable Zhizo 340 N99 E196 2 DB CS single row CS N/A unknown Probable Zhizo 345 21 1 DB LI multiple LI N/A Unknown Probable Zhizo 345 21 1 DB CS & LI two lines of CS bordering a band of LI N/A Unknown Probable Zhizo 252 Table 13 (co n 345 21 1 DB CS & LI single line of CS above a band of angled LI N/A Unknown Probable Zhizo 345 21 1 DB CS & LI single line of CS above a band of angled LI N/A Unknown Probable Zhizo 346 21 2 DB CS & LI two lines of CS bordering a band of LI N/A Unknown Probable Zhizo 346 21 2 DB LI band of LI N/A Unknown Probable Zhizo 346 21 2 DB CS & LI two lines of CS bordering two bands of LI N/A Unknown Probable Zhizo 346 21 2 DB CS & LI two lines of CS bordering two bands of LI N/A Unknown Probable Zhizo 346 21 2 DB CS single line of CS N/A Unknown Probable Zhizo 346 21 2 DB CS two lines of CS N/A Unknown Probable Zhizo 346 21 2 DB CS & LI one line of CS in between two bands of LI N/A Unknown Probable Zhizo 348 21 3 DB CS two lines of CS N/A Unknown Probable Zhizo 348 21 3 DB CS & LI two lines of CS with a line of LI N/A Unknown Probable Zhizo 348 21 3 DB CS two lines of CS with an obliquely angled LI N/A Unknown Probable Zhizo 253 Table 13 (co n 348 21 3 DB LI two parallel lines of LI N/A Unknown Probable Zhizo 168/9 8 3 DB CS & LI two lines of CS with two bands of LI N/A Unknown Probable Zhizo 168/9 8 3 DB LI two LI N/A Unknown Probable Zhizo 168/9 8 3 DB LI two LI N/A Unknown Probable Zhizo 168/9 8 3 DB CS & LI single line of CS with multiple LI N/A Unknown Probable Zhizo 168/9 8 3 DB CS single line N/A Unknown Probable Zhizo 168/9 8 3 DB CS & LI two lines of CS with band of LI N/A Unknown Probable Zhizo 318 STTP N79 E151 2 DR Punctate + LI two rows of short vertical/ oblong punctate bordering three rows of LI; small fragment of diagonal hatching below red paint entire rim+neck Eiland? see p 229, Huffman 2007 37 5 1 DR CS & LI Single CS row above LI band N/A lower rim Probable Zhizo 43 5 2 DR CS two CS lines at oblique angles N/A lower rim Probable Zhizo 250 16 2 DR CS single row of CS N/A lower rim Probable Zhizo 284 19 4 DR CS single line of CS N/A lower rim Probable Zhizo 254 Table 13 (co n 329 STTP N96 E 155 3 DR CS single row wide CS N/A lower rim Probable Zhizo 26 4 1 DR CS & LI angled bands of CS bordered by row of CS (above) and LI (below) N/A lower rim Zhizo 154 6 5 DR CS two rows CS N/A lower rim Zhizo 154 6 5 DR LI two rows angled incisions sep by single LI N/A lower rim Zhizo 330 STTP N97 E158 3 DR LI horiz LI bordering band of angled LI N/A lower rim Zhizo 339 STTP N110 E175 3 DR LI horiz LI bordering band of angled LI N/A lower rim Zhizo 25 2 3 DR CS two rows CS separated approx 1 cm N/A Neck Probable Zhizo 22 2 1 DR CS & LI multiple banded LI beneath row of CS N/A Neck Zhizo 146 5 4 DR LI single LI with possible CS below N/A rim Probable Zhizo 146 5 4 DR LI single band of LI N/A rim Probable Zhizo 43 5 2 DR CS & LI Single CS row above LI band N/A rim Zhizo 2 55 Table 13 (co n 240 15 3 DR CS two rows of CS N/A Unknown Probable Zhizo 241 14 4 DR LI single row of LI N/A Unknown Probable Zhizo 241 14 4 DR LI LI bordering angled band of LI N/A Unknown Probable Zhizo 241 14 4 DR CS & LI single line of CS with one LI N/A Unknown Probable Zhizo 261 17 2 DR CS & LI two lines of CS bordering a band of angled LI N/A Unknown Probable Zhizo 298 20 4 DR CS three parallel lines of CS N/A Unknown Probable Zhizo 272 18 3 DR N/A n/a painted red Unknown unknown 334 STTP N102 E141 3 DR CS three horiz lines of CS N/A upper rim Zhizo 256 APPENDIX B SHELL BEAD DATA 257 Table 14 Finished OES beads Lot Site Unit Level Material Q uanti ty Ext Diam (mm) Int Diam (mm) Burnt? Condition Notes 22 13 2 1 OES 1 4.1 2.2 n finished 22 13 2 1 OES 1 7.5 2.3 n finished 23 13 2 2 OES 1 6.0 2.1 Y finished 24 13 3 1 OES 1 4.0 1.6 n finished 24 13 3 1 OES 1 4.9 2.0 n finished 24 13 3 1 OES 1 4.8 2.1 n finished 24 13 3 1 OES 1 5.1 2.2 n finished 24 13 3 1 OES 1 9.5 2.4 n finished 24 13 3 1 OES 1 4.3 1.3 Y finished 26 13 4 1 OES 1 6.1 2.3 n finished 26 13 4 1 OES 1 10.0 3.1 n finished 26 13 4 1 OES 1 6.4 1.6 Y finished 26 13 4 1 OES 1 5.5 1.9 Y finished 28 13 3 1 OES 1 11.1 2.0 n finished 28 13 3 1 OES 1 6.6 2.1 n finished 28 13 3 1 OES 1 9.3 2.3 n finished 28 13 3 1 OES 1 10.9 2.3 n finished 28 13 3 1 OES 1 11.1 2.3 n finished 28 13 3 1 OES 1 11.0 2.3 n finished 28 13 3 1 OES 1 10.8 2.4 n finished 28 13 3 1 OES 1 11.2 2.4 n finished 28 13 3 1 OES 1 11.3 2.5 n finished 28 13 3 1 OES 1 11.1 2.5 n finished 28 13 3 1 OES 1 9.7 2.6 n finished 28 13 3 1 OES 1 11.3 2.6 n finished 28 13 3 1 OES 1 11.2 2.6 n finished 28 13 3 1 OES 1 11.0 2.7 n finished 28 13 3 1 OES 1 11.0 2.7 n finished 258 Table 14 ( co n 28 13 3 1 OES 1 9.5 2.8 n finished 28 13 3 1 OES 1 11.1 2.9 n finished 28 13 3 1 OES 1 11.0 3.0 n finished 28 13 3 1 OES 1 10.9 2.4 Y finished 28 13 3 1 OES 1 11.2 2.5 Y finished 28 13 3 1 OES 1 11.1 2.6 Y finished 29 13 4 2 OES 1 4.8 2.0 n finished 33 13 4 3 OES 1 5.8 1.5 Y finished 37 13 5 1 OES 1 3.7 1.2 n finished 37 13 5 1 OES 1 4.9 1.5 n finished 37 13 5 1 OES 1 4.9 1.7 n finished 37 13 5 1 OES 1 6.1 1.9 n finished 37 13 5 1 OES 1 4.3 2.3 n finished 37 13 5 1 OES 1 4.8 1.9 Y finished 37 13 5 1 OES 1 5.9 1.9 Y finished 43 13 5 2 OES 1 6.6 2.6 n finished 43 13 5 2 OES 1 4.3 1.7 Y finished 45 13 5 3 OES 1 6.1 1.3 n finished 45 13 5 3 OES 1 5.5 1.6 Y finished 145 13 5 4 OES 1 3.7 1.5 n finished 146 13 5 4 OES 1 6.5 1.7 n finished 148 13 6 2 OES 1 5.7 1.7 n finished 152 13 6 4 OES 1 7.4 2.2 n finished 160 13 7 2 OES 1 3.7 1.7 n finished 160 13 7 2 OES 1 4.5 1.8 n finished 161 13 7 3 OES 1 8.8 3.0 n finished 163 13 7 4 OES 1 8.0 2.7 n finished 166 13 8 1 OES 1 9.8 2.1 n finished 166 13 8 1 OES 1 6.1 2.7 n finished 167 13 8 2 OES 1 5.4 1.8 n finished 167 13 8 2 OES 1 10.8 2.2 n finished 259 Table 14 ( co n 167 13 8 2 OES 1 10.5 2.4 n finished 168 13 8 3 OES 1 4.3 1.6 n finished 168 13 8 3 OES 1 5.3 1.7 Y finished 168 13 8 3 OES 1 6.2 2.2 n finished 170 13 8 4 OES 1 6.0 2.2 n finished 175 13 8 N Wall OES 1 6.1 1.9 n finished 176 13 9 1 OES 1 4.9 2.7 n finished 177 13 9 2 OES 1 6.6 2.9 n finished 181 13 9 4 OES 1 5.1 2.1 n finished 181 13 9 4 OES 1 4.5 2.1 Y finished 187 13 9 7 OES 1 5.5 1.7 n finished 194 13 10 4 OES 1 5.3 2.3 n finished 194 13 10 4 OES 1 5.3 2.4 n finished 207 13 11 2 OES 1 4.8 2.0 n finished 214 13 12 1 OES 1 5.9 1.9 n finished 222 13 13 1 OES 1 6.1 1.8 Y finished 222 13 13 1 OES 1 6.1 1.9 Y finished 222 13 13 1 OES 1 8.3 2.0 Y finished 222 13 13 1 OES 1 6.4 2.5 Y finished 224 13 13 2 OES 1 5.0 2.0 n finished 224 13 13 2 OES 1 6.5 2.2 Y finished 224 13 13 2 OES 1 4.9 2.4 Y finished 226 13 13 3 OES 1 6.2 2.5 Y finished 227 13 14 1 OES 1 7.9 2.0 Y finished 227 13 14 1 OES 1 7.5 2.1 Y finished 227 13 14 1 OES 1 6.7 2.2 Y finished 227 13 14 1 OES 1 8.5 2.3 Y finished 227 13 14 1 OES 1 6.9 2.3 Y finished 227 13 14 1 OES 1 6.5 2.3 Y finished 227 13 14 1 OES 1 6.1 2.9 Y finished 229 13 14 2 OES 1 8.3 2.2 Y finished 260 Table 14 ( co n 229 13 14 2 OES 1 6.8 2.2 Y finished 231 13 14 3 OES 1 6.3 1.8 Y finished 231 13 14 3 OES 1 6.8 2.0 Y finished 231 13 14 3 OES 1 5.3 2.1 Y finished 232 13 14 3 OES 1 6.8 2.0 Y finished 236 13 15 1 OES 1 5.3 2.5 n finished 236 13 15 1 OES 1 3.5 2.5 n finished 236 13 15 1 OES 1 4.1 2.8 n finished 240 13 15 3 OES 1 6.3 2.6 n finished 241 13 15 4 OES 1 7.5 3.1 n finished 244 13 15 5 OES 1 3.2 1.6 n finished 244 13 15 5 OES 1 6.9 2.3 n finished 246 13 15 6 OES 1 4.7 1.9 Y finished 256 13 16 5 OES 1 5.0 1.6 n finished 256 13 16 5 OES 1 7.1 1.7 n finished 259 13 17 1 OES 1 5.6 1.8 n finished 259 13 17 1 OES 1 5.8 2.5 n finished 261 13 17 2 OES 1 4.8 1.7 n finished 261 13 17 2 OES 1 5.7 2.1 Y finished 261 13 17 2 OES 1 4.3 2.2 Y finished 265 13 17 3 OES 1 5.4 1.4 n finished 265 13 17 3 OES 1 6.9 2.1 n finished 267 13 17 4 OES 1 5.9 1.7 n finished 270 13 18 2 OES 1 3.3 1.3 n finished 270 13 18 2 OES 1 6.9 1.5 n finished 270 13 18 2 OES 1 5.6 1.8 n finished 275 13 19 1 OES 1 4.2 1.3 n finished 275 13 19 1 OES 1 8.7 1.7 n finished 276 13 19 2 OES 1 6.5 1.7 n finished 276 13 19 2 OES 1 9.0 2.2 n finished 276 13 19 2 OES 1 7.7 2.4 n finished 261 Table 14 ( co n 276 13 19 2 OES 1 11.2 2.9 n finished 276 13 19 2 OES 1 9.3 2.9 n finished 276 13 19 2 OES 1 9.9 2.9 n finished 281 13 19 3 OES 1 5.4 1.7 n finished 281 13 19 3 OES 1 10.9 2.2 n finished 281 13 19 3 OES 1 11.0 2.4 n finished 281 13 19 3 OES 1 11.1 2.5 n finished 281 13 19 3 OES 1 11.3 2.7 n finished 281 13 19 3 OES 1 8.7 2.7 n finished 281 13 19 3 OES 1 9.4 2.9 n finished 281 13 19 3 OES 1 9.3 3.0 n finished 281 13 19 3 OES 1 8.6 3.2 n finished 281 13 19 3 OES 1 11.2 2.9 Y finished 284 13 19 4 OES 1 8.9 2.2 n finished 284 13 19 4 OES 1 9.7 2.6 n finished 286 13 19 5 OES 1 6.8 2.1 n finished 286 13 19 5 OES 1 11.4 2.7 n finished 286 13 19 5 OES 1 6.4 2.4 Y finished 289 13 19 6 OES 1 10.7 2.9 n finished 289 13 19 6 OES 1 5.6 1.8 Y finished 291 13 20 1 OES 1 5.3 1.7 n finished 295 13 20 3 OES 1 4.6 1.6 n finished 295 13 20 3 OES 1 8.1 2.2 n finished 301 13 20 5 OES 1 5.2 1.8 Y finished 307 13 N106 E120 STR 2 OES 1 6.0 2.2 n finished 308 13 N87 E134 STR 3 OES 1 7.6 2.1 n finished 312 13 N75 E134 STR 2 OES 1 4.5 2.1 n finished 313 13 N75 E135 STR 1 OES 1 6.6 1.7 n finished 315 13 N72 E155 STR 2 OES 1 6.2 2.6 n finished 316 13 N75 E155 STR 2 OES 1 4.2 1.6 n finished 316 13 N75 E155 STR 2 OES 1 6.2 1.7 n finished 262 Table 14 ( co n 316 13 N75 E155 STR 2 OES 1 5.9 2.3 n finished 316 13 N75 E155 STR 2 OES 1 6.2 2.3 n finished 317 13 N76 E153 STR 2 OES 1 6.1 1.8 n finished 319 13 N78 E158 STR 3 OES 1 9.6 2.6 n finished 321 13 N82 E171 STR 2 OES 1 6.6 1.6 Y finished 323 13 N85 E178 STR 2 OES 1 4.6 1.7 n finished 328 13 N95 E153 STR 2 OES 1 6.7 1.9 n finished 329 13 N96 E155 STR 2 OES 1 7.0 2.1 n finished 329 13 N96 E155 STR 3 OES 1 7.2 2.7 n finished 330 13 N97 E158 STR 2 OES 1 5.4 1.9 n finished 332 13 N100 E154 STR 3 OES 1 7.3 1.6 n finished 339 13 N110 E175 STR 3 OES 1 5.9 2.2 Y finished 342 13 TTP2 1 OES 1 6.5 3.0 n finished 348 13 21 3 OES 1 5.3 1.7 n finished 350 13 TTP5 1 OES 1 5.1 2.2 n finished 354 13 TTP9 1 OES 1 4.8 2.2 n finished 355 13 TTP10 1 OES 1 5.6 2.2 n finished V WEATHERED LOOKING 360 13 TTP15 1 OES 1 5.2 1.9 n finished 28/36 13 3 3 OES 1 11.2 2.1 n finished part of a necklace 28/36 13 3 3 OES 1 11.5 2.3 n finished part of a necklace 28/36 13 3 3 OES 1 11.2 2.3 n finished part of a necklace 28/36 13 3 3 OES 1 11.2 2.3 n finished part of a necklace 28/36 13 3 3 OES 1 11.2 2.3 n finished part of a necklace 28/36 13 3 3 OES 1 11.2 2.4 n finished part of a necklace 28/36 13 3 3 OES 1 11.2 2.6 n finished part of a necklace 28/36 13 3 3 OES 1 11.1 2.9 n finished part of a necklace 263 Table 15 Finished Achatina beads Lot Site Unit Level Material Quantity Ext Diam (mm) Int Diam (mm) Burnt? Condition Notes 22 13 2 1 ACH 1 3.13 1.21 n finished 22 13 2 1 ACH 1 3.66 1.6 n finished 23 13 2 2 ACH 1 4.89 2.04 n finished 23 13 2 2 ACH 1 3.27 1.87 n finished 23 13 2 2 ACH 1 4.35 1.87 n finished 23 13 2 2 ACH 1 3.66 2.19 n finished 23 13 2 2 ACH 1 4.95 1.91 Y finished 24 13 3 1 ACH 1 4.02 1.97 n finished 26 13 4 1 ACH 1 5.35 1.53 n finished 28 13 3 1 ACH 1 4.47 1.07 n finished 33 13 4 3 ACH 1 4.72 1.27 n finished 37 13 5 1 ACH 1 4.78 1.28 n finished 37 13 5 1 ACH 1 4.7 1.42 n finished 37 13 5 1 ACH 1 3.76 1.5 n finished 37 13 5 1 ACH 1 4.85 1.58 Y finished 43 13 5 2 ACH 1 3.71 1.39 n finished 148 13 6 2 ACH 1 4.71 1.47 n finished 149 13 6 3 ACH 1 2.98 1.68 n finished 160 13 7 2 ACH 1 5.23 1.18 n finished 160 13 7 2 ACH 1 3.84 1.74 n finished 160 13 7 2 ACH 1 3.13 .95 n finished 163 13 7 4 ACH 1 3.81 1.63 n finished 165 13 7 5 ACH 1 4 1.68 n finished 165 13 7 5 ACH 1 5.25 1.28 n finished 176 13 9 1 ACH 1 4.29 1.55 n finished 177 13 9 2 ACH 1 4 1.4 n finished 177 13 9 2 ACH 1 2.94 1.71 n finished 177 13 9 2 ACH 1 4.61 1.56 Y finished 179 13 9 3 ACH 1 5.49 1.34 n finished 264 Table 15 ( co n ) 181 13 9 4 ACH 1 5.21 1.71 n finished 181 13 9 4 ACH 1 3.8 1.3 n finished 183 13 9 5 ACH 1 3.6 1.32 n finished 185 13 9 6 ACH 1 4.35 1.2 n finished 192 13 10 3 ACH 1 4.15 1.95 n finished 222 13 13 1 ACH 1 4.95 1.72 Y finished 222 13 13 1 ACH 1 5.59 1.73 Y finished 222 13 13 1 ACH 1 5.01 1.34 Y finished 222 13 13 1 ACH 1 5.19 1.81 Y finished 222 13 13 1 ACH 1 4.97 1.9 Y finished 222 13 13 1 ACH 1 4.51 1.69 Y finished 222 13 13 1 ACH 1 5.08 1.68 Y finished 224 13 13 2 ACH 1 4.8 1.54 Y finished 224 13 13 2 ACH 1 4.72 1.98 Y finished 224 13 13 2 ACH 1 5.06 1.54 Y finished 224 13 13 2 ACH 1 4.71 1.76 Y finished 224 13 13 2 ACH 1 4.66 1.59 Y finished 224 13 13 2 ACH 1 4.87 1.71 Y finished 224 13 13 2 ACH 1 4.77 1.44 Y finished 224 13 13 2 ACH 1 5.41 1.19 Y finished 224 13 13 2 ACH 1 4.78 2.08 Y finished 224 13 13 2 ACH 1 4.76 1.59 Y finished 224 13 13 2 ACH 1 4.28 1.56 Y finished 224 13 13 2 ACH 1 5.51 1.62 Y finished 226 13 13 3 ACH 1 5.16 2.14 Y finished 226 13 13 3 ACH 1 4.81 1.86 Y finished 227 13 14 1 ACH 1 4.99 1.97 Y finished 227 13 14 1 ACH 1 4.55 1.52 Y finished 227 13 14 1 ACH 1 5 2.01 Y finished 227 13 14 1 ACH 1 4.92 1.67 Y finished 227 13 14 1 ACH 1 4.58 1.33 Y finished 265 Table 15 ( co n ) 227 13 14 1 ACH 1 4.99 1.4 Y finished 227 13 14 1 ACH 1 5.08 1.66 Y finished 229 13 14 2 ACH 1 3.31 1.98 n finished 229 13 14 2 ACH 1 4.56 1.54 Y finished 229 13 14 2 ACH 1 4.58 1.69 Y finished 229 13 14 2 ACH 1 5.03 1.83 Y finished 229 13 14 2 ACH 1 4.6 1.41 Y finished 229 13 14 2 ACH 1 4.33 1.74 Y finished 231 13 14 3 ACH 1 4.36 1.56 Y finished 231 13 14 3 ACH 1 4.76 1.8 Y finished 231 13 14 3 ACH 1 5.29 1.42 Y finished 231 13 14 3 ACH 1 4.99 1.46 Y finished 231 13 14 3 ACH 1 5.38 1.67 Y finished 233 13 14 4 ACH 1 4.86 2.06 Y finished 233 13 14 4 ACH 1 4.84 1.88 Y finished 236 13 15 1 ACH 1 2.77 2.72 n finished 238 13 15 2 ACH 1 4.94 2.11 n finished 240 13 15 3 ACH 1 3.72 1.76 n finished 240 13 15 3 ACH 1 3.52 1.40 n finished 240 13 15 3 ACH 1 3.42 1.69 n finished 241 13 15 4 ACH 1 4.26 2.08 n finished pale color, no striae or other diagnostic markings 241 13 15 4 ACH 1 3.24 1.77 n finished 241 13 15 4 ACH 1 3.19 2.21 n finished 244 13 15 5 ACH 1 5.12 1.95 n finished 246 13 15 6 ACH 1 4.42 1.41 n finished 246 13 15 6 ACH 1 3.46 1.66 n finished 246 13 15 6 ACH 1 4.34 1.77 n finished 250 13 16 2 ACH 1 3.04 1.2 n finished 250 13 16 2 ACH 1 2.74 1.31 n finished 259 13 17 1 ACH 1 3.39 1.69 n finished 266 Table 15 ( co n ) 259 13 17 1 ACH 1 4.17 2.22 n finished 259 13 17 1 ACH 1 3.81 1.56 n finished 259 13 17 1 ACH 1 4.27 1.45 n finished 259 13 17 1 ACH 1 3.49 1.77 n finished 259 13 17 1 ACH 1 3.59 1.56 n finished 261 13 17 2 ACH 1 4.25 1.53 n finished 265 13 17 3 ACH 1 4.96 1.12 n finished 265 13 17 3 ACH 1 4.61 1.22 n finished 265 13 17 3 ACH 1 4.07 1.48 n finished 265 13 17 3 ACH 1 3.17 1.26 n finished 267 13 17 4 ACH 1 3.64 1.76 n finished 276 13 19 2 ACH 1 4.69 1.45 n finished 284 13 19 4 ACH 1 6.13 1.57 n finished 284 13 19 4 ACH 1 2.93 1.45 n finished 291 13 20 1 ACH 1 4.27 1.74 n finished 295 13 20 3 ACH 1 4.52 1.26 Y finished 303 13 N110 E113 STR 1 ACH 1 3.69 1.43 Y finished 308 13 N87 E134 STR 3 ACH 1 3.95 1.27 n finished 312 13 N75 E134 STR 2 ACH 1 4.55 1.3 n finished 316 13 N75 E155 STR 2 ACH 1 3.63 1.59 n finished 316 13 N75 E155 STR 2 ACH 1 4.05 1.67 n finished 316 13 N75 E155 STR 2 ACH 1 5.03 1.73 n finished 322 13 N82 E174 STR 3 ACH 1 3.83 1.6 n finished 327 13 N91 E155 STR 2 ACH 1 3.53 1.41 n finished 328 13 N95 E153 STR 2 ACH 1 2.86 1.62 n finished 330 13 N97 E158 STR 1 ACH 1 4.31 1.05 n finished 330 13 N97 E158 STR 3 ACH 1 4.72 1.39 n finished 332 13 N100 E154 STR 2 ACH 1 4.23 1.62 n finished 333 13 N101 E149 STR 2 ACH 1 4.39 1.05 n finished 334 13 N102 E141 STR 1 ACH 1 3.94 1.76 n finished 334 13 N102 E149 STR 2 ACH 1 4.18 1.53 n finished 267 Table 15 ( co n ) 336 13 N106 E146 STR 2 ACH 1 3.32 0.88 Y finished 350 13 TTP5 1 ACH 1 2.98 1.59 n finished 355 13 TTP10 1 ACH 1 5.83 2.15 n finished 168/14 13 8 3 ACH 1 4.7 1.55 n finished 168/15 13 8 3 ACH 1 3.57 1.34 n finished 168/16 13 8 3 ACH 1 3.29 1.43 n finished 268 Table 16 Broken OES beads Lot Site Unit Level Material Quantity Ext Diam (mm) Int Diam (mm) Burnt? Condition Notes 24 13 3 1 OES 1 11.4 2.6 n finished 25 13 2 3 OES 1 7.6 2.3 yes finished 26 13 4 1 OES 1 6.4 2.8 yes broken 28 13 3 1 OES 1 11.2 2.9 yes broken 37 13 5 1 OES 1 6.8 3.0 n finished 37 13 5 1 OES 1 5.3 2.4 n finished 37 13 5 1 OES 1 4.8 1.8 n finished 37 13 5 1 OES 1 7.4 2.3 n irregular 149 13 6 3 OES 1 7.3 1.7 yes irregular 160 13 7 2 OES 1 6.5 1.7 yes finished 161 13 7 3 OES 1 6.9 1.4 n JAGGED 163 13 7 4 OES 1 3.8 1.6 yes finished 167 13 8 2 OES 1 8.4 2.5 n finished 167 13 8 2 OES 1 6.9 2.0 yes finished 183 13 9 5 OES 1 5.9 2.2 n finished 185 13 9 6 OES 1 10.3 2.0 yes irregular 199 13 10 5 OES 1 8.4 2.5 n finished 246 13 15 6 OES 1 8.6 1.6 yes JAGGED 259 13 17 1 OES 1 7.6 1.9 n irregular 267 13 17 4 OES 1 7.4 2.6 n finished 267 13 17 4 OES 1 6.4 2.4 n finished 276 13 19 2 OES 1 8.6 2.7 n irregular 281 13 19 3 OES 1 9.4 2.9 n finished 281 13 19 3 OES 1 6.6 2.7 n irregular 281 13 19 3 OES 1 9.9 2.5 yes irregular 284 13 19 4 OES 1 10.9 2.5 n finished 284 13 19 4 OES 1 8.8 2.4 n finished 286 13 19 5 OES 1 7.5 2.5 n finished 289 13 19 6 OES 1 4.6 2.1 n finished 269 Table 16 ( co n ) 298 13 20 4 OES 1 5.2 2.1 n broken 309 13 N85 E131 STR 2 OES 1 8.2 2.5 n broken 316 13 N75 E155 STR 2 OES 1 6.0 1.5 n broken 317 13 N75 E134 STR 3 OES 1 7.6 1.4 yes broken 327 13 N91 E155 STR 2 OES 1 5.9 2.3 n broken 328 13 N95 E153 STR 2 OES 1 8.9 2.6 n broken 328 13 N95 E153 STR 2 OES 1 5.9 1.9 n broken 330 13 N97 E158 STR 2 OES 1 7.8 2.0 n broken 332 13 N100 E154 STR 3 OES 1 7.0 2.2 n broken 338 13 N107 E150 STR 2 OES 1 5.8 2.3 n broken 168/1 2 13 8 3 OES 1 6.2 1.9 yes irregular 168/1 3 13 8 3 OES 1 6.8 2.2 n finished 181 13 9 4 OES 1 4.8 0.0 n finished <50% present 209 13 11 3 OES 1 BROKEN BROKEN yes irregular <50% present 214 13 12 1 OES 1 BROKEN BROKEN yes finished <50% present 214 13 12 1 OES 1 9.9 8.1 no finished Int Diam is probably error; ID'd by EDA as outlier 222 13 13 1 OES 33 BROKEN BROKEN yes irregular <50% present 224 13 13 2 OES 6 BROKEN BROKEN yes finished <50% present 224 13 13 2 OES 29 BROKEN BROKEN yes irregular <50% present 225 13 13 3 OES 4 BROKEN BROKEN no irregular <50% present 270 Table 16 ( co n ) 226 13 13 3 OES 5 BROKEN BROKEN no irregular <50% present 227 13 14 1 OES 7 BROKEN BROKEN yes finished <50% present 227 13 14 1 OES 35 BROKEN BROKEN yes irregular <50% present 229 13 14 2 OES 4 BROKEN BROKEN yes finished <50% present 229 13 14 2 OES 29 BROKEN BROKEN yes irregular <50% present 231 13 14 3 OES 16 BROKEN BROKEN yes irregular <50% present 233 13 14 4 OES 3 BROKEN BROKEN yes irregular <50% present 236 13 15 1 OES 1 BROKEN BROKEN no finished <50% present 240 13 15 3 OES 1 BROKEN BROKEN no finished <50% present 240 13 15 3 OES 1 BROKEN BROKEN yes irregular <50% present 265 13 17 3 OES 1 5.56 2.49 n broken <50% present 267 13 17 4 OES 1 6.01 2.47 n irregular <50% present 276 13 19 2 OES 1 10.59 2.12 n irregular <50% present 281 13 19 3 OES 1 8.25 1.89 n finished <50% present 293 13 20 2 OES 1 5.85 2.31 yes finished <50% present 298 13 20 4 OES 1 5.55 2.68 yes broken <50% present 313 13 N75 E135 STR 3 OES 1 6.7 5.0 n finished Int Diam is probably error; ID'd by EDA as outlier 331 13 N91 E151 STR 2 OES 1 8.34 2.41 n broken <50% present 357 13 TTP12 1 OES 1 BROKEN BROKEN yes finished <50% present 271 Table 17 Broken ACH beads Lot Site Unit Level Material Quantity Ext Diam (mm) Int Diam (mm) Burnt? Condition Notes 23 13 2 2 ACH 1 6.04 1.52 n JAGGED 35 13 4 4 ACH 1 7.42 N/A n finished 167 13 8 2 ACH 1 3.88 1.14 n finished 346 13 21 2 ACH 1 6.6 1.95 n JAGGED 346 13 21 2 ACH 1 5.01 1.84 n JAGGED 348 13 21 3 ACH 1 7.23 1.39 n JAGGED 332 13 N100 E154 STR 3 ACH 1 6.35 1.55 yes finished 322 13 N82 E174 STR 2 ACH 1 4.87 2.15 n finished 326 13 N88 E178 STR 3 ACH 1 4.06 1.52 n finished 33 13 4 3 ACH 1 5.02 1.32 n JAGGED <50% present 35 13 4 4 ACH 1 5.78 1.82 n broken <50% present 222 13 13 1 ACH 9 BROKEN BROKEN yes finished <50% present 224 13 13 2 ACH 8 BROKEN BROKEN yes finished <50% present 226 13 13 3 ACH 4 BROKEN BROKEN yes finished <50% present 227 13 14 1 ACH 9 BROKEN BROKEN yes finished <50% present 229 13 14 2 ACH 3 BROKEN BROKEN yes finished <50% present 231 13 14 3 ACH 6 BROKEN BROKEN yes finished <50% present 232 13 14 3 ACH 1 BROKEN BROKEN yes finished <50% present 272 Table 18 Irregular OES beads Lot Site Unit Level Material Quantity Ext Diam (mm) Int Diam (mm) Burnt? Condition Notes 21 13 1 1 OES 1 6.6 1.9 n irregular 22 13 2 1 OES 1 5.2 1.9 n irregular 22 13 2 1 OES 1 6.9 2.3 n irregular 23 13 2 2 OES 1 5.9 2.2 n irregular 23 13 2 2 OES 1 5.4 1.6 n irregular 24 13 3 1 OES 1 5.3 1.9 n irregular 24 13 3 1 OES 1 10.0 1.9 yes irregular 28 13 3 1 OES 1 10.4 2.2 n irregular 28 13 3 1 OES 1 9.2 1.4 n irregular 37 13 5 1 OES 1 9.5 2.2 n irregular 37 13 5 1 OES 1 7.2 1.9 n irregular 37 13 5 1 OES 1 6.9 1.8 n JAGGED 37 13 5 1 OES 1 8.9 2.0 yes JAGGED 43 13 5 2 OES 1 8.9 1.9 n irregular 149 13 6 3 OES 1 9.0 1.9 yes JAGGED 154 13 6 5 OES 1 9.3 1.7 n JAGGED 155 13 6 FEA 2 OES 1 4.0 1.9 n irregular 177 13 9 2 OES 1 8.5 2.2 n irregular 177 13 9 2 OES 1 6.0 2.5 n irregular 177 13 9 2 OES 1 6.3 2.2 n irregular 177 13 9 2 OES 1 7.6 1.9 yes JAGGED 181 13 9 4 OES 1 7.4 2.0 n irregular 181 13 9 4 OES 1 5.8 2.2 n irregular 181 13 9 4 OES 1 6.9 2.1 yes irregular 187 13 9 7 OES 1 6.4 1.9 yes irregular 194 13 10 4 OES 1 7.2 2.0 n irregular 194 13 10 4 OES 1 7.1 2.0 n irregular 221 13 13 1 OES 1 7.7 2.5 yes irregular 221 13 13 1 OES 1 7.8 2.1 yes irregular 273 Table 18 ( co n ) 222 13 13 1 OES 1 8.2 2.5 yes irregular 222 13 13 1 OES 1 7.7 2.1 yes irregular 222 13 13 1 OES 1 8.2 2.1 yes irregular 222 13 13 1 OES 1 8.2 2.3 yes irregular 222 13 13 1 OES 1 6.1 1.5 yes irregular 222 13 13 1 OES 1 7.9 2.0 yes irregular 222 13 13 1 OES 1 7.9 2.0 yes irregular 222 13 13 1 OES 1 8.4 2.4 yes irregular 222 13 13 1 OES 1 7.3 1.9 yes irregular 222 13 13 1 OES 1 6.4 1.7 yes irregular 222 13 13 1 OES 1 7.4 1.8 yes irregular 222 13 13 1 OES 1 6.3 1.8 yes irregular 222 13 13 1 OES 1 8.3 1.6 yes JAGGED 224 13 13 2 OES 1 6.4 2.1 yes irregular 224 13 13 2 OES 1 7.6 2.3 yes irregular 224 13 13 2 OES 1 8.8 2.2 yes irregular 224 13 13 2 OES 1 7.8 2.2 yes irregular 224 13 13 2 OES 1 6.6 2.0 yes irregular 226 13 13 3 OES 1 8.1 2.4 yes irregular 227 13 14 1 OES 1 7.8 1.9 n irregular 227 13 14 1 OES 1 8.1 2.2 yes irregular 227 13 14 1 OES 1 8.0 2.4 yes irregular 227 13 14 1 OES 1 8.2 2.2 yes irregular 227 13 14 1 OES 1 8.7 2.2 yes irregular 227 13 14 1 OES 1 7.0 2.2 yes irregular 227 13 14 1 OES 1 6.5 1.6 yes irregular 227 13 14 1 OES 1 7.8 2.2 yes irregular 227 13 14 1 OES 1 7.7 2.5 yes irregular 227 13 14 1 OES 1 8.0 1.9 yes irregular 227 13 14 1 OES 1 8.9 2.2 yes irregular 227 13 14 1 OES 1 7.6 2.5 yes irregular 274 Table 18 ( co n ) 227 13 14 1 OES 1 7.7 2.3 yes irregular 227 13 14 1 OES 1 7.6 2.2 yes irregular 227 13 14 1 OES 1 7.6 2.4 yes irregular 228 13 14 1 OES 1 6.9 2.2 yes irregular 229 13 14 2 OES 1 7.8 2.4 n irregular 229 13 14 2 OES 1 7.9 2.3 yes irregular 229 13 14 2 OES 1 8.8 1.9 yes irregular 229 13 14 2 OES 1 8.8 2.2 yes irregular 229 13 14 2 OES 1 7.0 2.1 yes irregular 229 13 14 2 OES 1 9.5 2.5 yes irregular 229 13 14 2 OES 1 6.7 2.2 yes irregular 229 13 14 2 OES 1 7.1 2.2 yes irregular 229 13 14 2 OES 1 7.5 2.1 yes irregular 229 13 14 2 OES 1 7.8 2.2 yes irregular 229 13 14 2 OES 1 6.6 2.0 yes irregular 229 13 14 2 OES 1 7.8 2.0 yes irregular 229 13 14 2 OES 1 8.4 2.1 yes irregular 229 13 14 2 OES 1 7.6 1.9 yes irregular 231 13 14 3 OES 1 7.6 2.2 yes irregular 231 13 14 3 OES 1 6.9 2.0 yes irregular 231 13 14 3 OES 1 7.6 2.0 yes irregular 232 13 14 3 OES 1 7.5 2.4 yes irregular 232 13 14 3 OES 1 6.9 1.7 yes irregular 233 13 14 4 OES 1 7.4 3.0 yes irregular 233 13 14 4 OES 1 6.2 1.9 yes irregular 238 13 15 2 OES 1 5.9 2.6 no irregular 238 13 15 2 OES 1 6.9 2.6 yes irregular 244 13 15 5 OES 1 6.6 2.4 n irregular 248 13 16 1 OES 1 8.7 2.5 n irregular 248 13 16 1 OES 1 7.0 2.1 n irregular 250 13 16 2 OES 1 6.9 1.7 n irregular 275 Table 18 ( co n ) 259 13 17 1 OES 1 6.7 2.0 n irregular 265 13 17 3 OES 1 8.7 1.9 n irregular 267 13 17 4 OES 1 7.1 1.8 n irregular 272 13 18 3 OES 1 7.7 1.8 n irregular 276 13 19 2 OES 1 7.3 1.7 n irregular 276 13 19 2 OES 1 9.3 2.2 n irregular 276 13 19 2 OES 1 9.9 2.0 n irregular 276 13 19 2 OES 1 9.9 2.3 n irregular 276 13 19 2 OES 1 9.1 1.8 n irregular 276 13 19 2 OES 1 3.7 1.8 n irregular 281 13 19 3 OES 1 11.0 2.4 n irregular 281 13 19 3 OES 1 11.4 3.0 n irregular 281 13 19 3 OES 1 10.0 2.2 n irregular 281 13 19 3 OES 1 8.7 1.9 n irregular 281 13 19 3 OES 1 10.5 2.6 n irregular 281 13 19 3 OES 1 9.7 2.8 n irregular 281 13 19 3 OES 1 11.5 2.7 n irregular 281 13 19 3 OES 1 11.1 2.5 n irregular 281 13 19 3 OES 1 9.9 2.9 n irregular 281 13 19 3 OES 1 10.0 2.5 n irregular 281 13 19 3 OES 1 8.4 2.9 n irregular 281 13 19 3 OES 1 9.6 2.0 yes irregular 281 13 19 3 OES 1 8.9 2.1 yes irregular 281 13 19 3 OES 1 9.8 2.1 yes irregular 281 13 19 3 OES 1 9.3 1.9 yes irregular 284 13 19 4 OES 1 9.9 2.0 n irregular 284 13 19 4 OES 1 10.0 2.4 yes irregular 286 13 19 5 OES 1 8.8 2.0 n irregular 286 13 19 5 OES 1 9.9 2.4 n irregular 286 13 19 5 OES 1 7.7 1.9 n irregular 286 13 19 5 OES 1 8.5 1.8 yes irregular 276 Table 18 ( co n ) 291 13 20 1 OES 1 9.3 2.7 n irregular 291 13 20 1 OES 1 5.9 1.8 n irregular 295 13 20 3 OES 1 6.9 2.1 n irregular 295 13 20 3 OES 1 5.7 1.9 n irregular 308 13 N87 E134 STR 3 OES 1 6.9 1.5 n irregular 319 13 N78 E158 STR 2 OES 1 9.4 2.1 n irregular 329 13 N96 E155 STR 2 OES 1 8.5 2.3 n irregular 334 13 N102 E141 STR 3 OES 1 7.0 1.9 n irregular 335 13 N103 E144 STR 2 OES 1 6.3 1.8 n irregular 342 13 TTP2 1 OES 1 6.4 2.2 N irregular 348 13 21 3 OES 1 6.1 1.8 yes irregular 361 13 TTP16 1 OES 1 8.1 2.4 yes irregular 361 13 TTP16 1 OES 1 7.8 2.0 yes irregular 364 13 TTP22 1 OES 1 8.9 2.0 N irregular 277 Table 19 Irregular ACH beads Lot Site Unit Level Material Quantity Ext Diam (mm) Int Diam (mm) Burnt? Condition Notes 23 13 2 2 ACH 1 3.53 1.43 n irregular 23 13 2 2 ACH 1 5.07 1.73 n irregular 23 13 2 2 ACH 1 3.19 1.31 n irregular 23 13 2 2 ACH 1 6.04 1.52 n JAGGED 29 13 4 2 ACH 1 6.09 1.82 n irregular 33 13 4 3 ACH 1 5.02 1.32 n JAGGED <50% present 37 13 5 1 ACH 1 6.48 1.23 n JAGGED 37 13 5 1 ACH 1 8.18 1.5 n JAGGED 152 13 6 4 ACH 1 3.34 1.88 n irregular 170 13 8 4 ACH 1 5.46 1.19 n irregular 181 13 9 4 ACH 1 3.42 1.28 n irregular 190 13 10 2 ACH 1 2.95 1.51 n irregular 209 13 11 3 ACH 1 3.73 1.28 yes irregular 224 13 13 2 ACH 1 2.9 1.53 n irregular 224 13 13 2 ACH 1 4.91 1.82 yes irregular 224 13 13 2 ACH 1 4.71 1.55 yes irregular 226 13 13 3 ACH 1 3.13 2.05 n irregular 226 13 13 3 ACH 1 4.7 1.7 yes irregular 227 13 14 1 ACH 1 4.7 1.44 yes irregular 227 13 14 1 ACH 1 5.35 1.24 yes irregular 227 13 14 1 ACH 1 4.54 1.79 yes irregular 229 13 14 2 ACH 1 4.99 1.8 yes irregular 248 13 16 1 ACH 1 3.34 1.27 n irregular 259 13 17 1 ACH 1 3.41 1.99 n irregular 259 13 17 1 ACH 1 3.68 1.67 n irregular 259 13 17 1 ACH 1 3.14 1.61 yes irregular 265 13 17 3 ACH 1 3.89 1.18 n irregular 265 13 17 3 ACH 1 3.19 1.74 n irregular 267 13 17 4 ACH 1 2.83 1.53 n irregular 278 Table 19 ( co n ) 267 13 17 4 ACH 1 3.34 1.65 n irregular 267 13 17 4 ACH 1 3.79 1.72 yes irregular 281 13 19 3 ACH 1 4.29 1.41 n irregular 293 13 20 2 ACH 1 3.66 1.79 n irregular 325 13 N87 E173 STR 2 ACH 1 3.82 1.2 n irregular 328 13 N95 E153 STR 3 ACH 1 2.94 1.19 n irregular 342 13 TTP2 1 ACH 1 5.23 2.11 N irregular 346 13 21 2 ACH 1 6.6 1.95 n JAGGED 346 13 21 2 ACH 1 5.01 1.84 n JAGGED 346 13 21 2 ACH 1 6.53 1.87 yes JAGGED 346 13 21 2 ACH 1 6.53 1.87 yes JAGGED 348 13 21 3 ACH 1 7.23 1.39 n JAGGED 350 13 TTP5 1 ACH 1 3.28 1.25 N irregular 355 13 TTP10 1 ACH 1 6.28 1.77 N irregular 360 13 TTP15 1 ACH 1 5.68 1.78 N irregular V WEATHERED LOOKING 279 Table 20 OES blanks and fragments Lot Site Unit Level Material Q uanti ty Length (mm) Width (mm) Burnt? Condition Notes 37 13 5 1 OES 1 8.8 8.8 n blank 37 13 5 1 OES 1 8.8 8.8 n blank 332 13 N100 E154 STR 3 OES 1 8.3 8.6 n blank 240 13 15 3 OES 1 6.7 4.9 no blank 163 13 7 4 OES 1 8.4 8.4 yes blank 353 13 TTP8 1 OES 1 8.3 8.3 yes blank 353 13 TTP8 1 OES 1 7.5 7.5 yes blank 373 12 1 1 OES 1 6.4 4.3 n fragment 22 13 2 1 OES 1 8.7 3.8 n fragment 23 13 2 2 OES 1 10.2 6.7 n fragment 24 13 3 1 OES 1 5.9 4.9 n fragment 27 13 3 2(?) OES 1 5.7 3.6 n fragment 27 13 3 2(?) OES 1 4.4 3.5 n fragment 27 13 3 2(?) OES 1 4.8 4.4 n fragment 27 13 3 2(?) OES 1 4.8 3.3 n fragment 27 13 3 2(?) OES 1 5.8 5.4 n fragment 37 13 5 1 OES 1 18.2 22.1 n fragment 37 13 5 1 OES 1 19.9 10.0 n fragment 37 13 5 1 OES 1 12.5 11.7 n fragment 37 13 5 1 OES 1 9.9 7.8 n fragment 37 13 5 1 OES 1 14.7 14.6 yes fragment 37 13 5 1 OES 1 8.8 7.4 yes fragment 43 13 5 2 OES 1 6.9 4.7 n fragment 43 13 5 2 OES 1 12.1 6.3 n fragment 43 13 5 2 OES 1 8.2 5.9 n fragment 145 13 5 4 OES 1 6.4 3.9 n fragment 148 13 6 2 OES 1 10.8 9.2 yes fragment 149 13 6 3 OES 1 17.1 10.8 yes fragment 280 Table 20 ( co n ) 149 13 6 3 OES 1 14.5 9.8 yes fragment 152 13 6 4 OES 1 9.3 6.6 yes fragment 154 13 6 4 OES 1 9.8 8.4 yes fragment 154 13 6 5 OES 1 6.1 6.6 n fragment 163 13 7 4 OES 1 13.2 13.2 yes fragment 163 13 7 4 OES 1 6.3 3.4 yes fragment 168/1 69 13 8 3 OES 1 10.5 7.0 n fragment 168/1 69 13 8 3 OES 1 7.9 5.9 yes fragment 176 13 9 1 OES 1 16.6 13.9 n fragment 176 13 9 1 OES 1 12.8 7.0 n fragment 176 13 9 1 OES 1 8.6 5.9 yes fragment 176 13 9 1 OES 1 8.9 9.5 yes fragment 176 13 9 1 OES 1 5.0 4.5 yes fragment 177 13 9 2 OES 1 19.0 11.4 n fragment 177 13 9 2 OES 1 4.2 4.0 n fragment 177 13 9 2 OES 1 8.0 6.8 yes fragment 177 13 9 2 OES 1 6.5 4.6 yes fragment 179 13 9 3 OES 1 13.8 12.9 n fragment 179 13 9 3 OES 1 13.5 12.3 yes fragment 179 13 9 3 OES 1 12.8 9.4 yes fragment 179 13 9 3 OES 1 9.6 6.8 yes fragment 179 13 9 3 OES 1 5.7 3.4 yes fragment 185 13 9 6 OES 1 15.1 10.0 yes fragment 187 13 9 7 OES 1 14.8 13.5 n fragment 190 13 10 2 OES 1 10.2 10.5 n fragment 201 13 10 6 OES 1 12.0 13.3 n fragment 207 13 11 2 OES 1 10.2 4.6 yes fragment 209 13 11 3 OES 1 7.2 5.1 yes fragment 209 13 11 3 OES 1 3.8 2.6 yes fragment 209 13 11 3 OES 1 6.8 3.7 yes fragment 281 Table 20 ( co n ) 209 13 11 3 OES 1 6.8 3.8 yes fragment 236 13 15 1 OES 1 7.0 4.7 no fragment 238 13 15 2 OES 1 11.9 5.1 no fragment 238 13 15 2 OES 1 8.8 5.6 no fragment 238 13 15 2 OES 1 7.0 6.6 no fragment 240 13 15 3 OES 1 13.8 10.7 no fragment 240 13 15 3 OES 1 7.7 7.6 no fragment 240 13 15 3 OES 1 6.9 6.1 no fragment 240 13 15 3 OES 1 7.8 2.8 yes fragment 246 13 15 6 OES 1 7.7 6.0 no fragment 250 13 16 2 OES 1 12.9 7.1 n fragment 250 13 16 2 OES 1 8.2 5.6 n fragment 250 13 16 2 OES 1 5.6 4.4 n fragment 256 13 16 5 OES 1 8.2 5.0 n fragment 256 13 16 5 OES 1 9.3 6.7 n fragment 259 13 17 1 OES 1 11.7 7.7 n fragment 259 13 17 1 OES 1 6.7 3.1 n fragment 261 13 17 2 OES 1 4.0 8.1 n fragment 261 13 17 2 OES 1 7.7 6.9 n fragment 261 13 17 2 OES 1 5.1 3.4 n fragment 261 13 17 2 OES 1 9.4 5.6 n fragment 265 13 17 3 OES 1 9.6 9.2 n fragment 275 13 19 1 OES 1 6.1 5.4 n fragment 281 13 19 3 OES 1 7.7 8.1 n fragment 286 13 19 5 OES 1 5.8 5.5 n fragment 286 13 19 5 OES 1 6.8 3.9 n fragment 289 13 19 6 OES 1 6.1 7.1 n fragment 295 13 20 3 OES 1 16.0 13.2 n fragment 334 13 N102 E149 STR 2 OES 1 6.3 6.9 n fragment 307 13 N106 E120 STR 3 OES 1 7.2 5.9 n fragment 282 Table 20 ( co n ) 338 13 N107 E150 STR 2 OES 1 8.5 3.1 n fragment 339 13 N110 E175 STR 3 OES 1 9.8 9.2 yes fragment 316 13 N75 E155 STR 2 OES 1 6.0 3.2 n fragment 319 13 N78 E158 STR 2 OES 1 4.0 3.2 n fragment 321 13 N82 E171 STR 2 OES 1 6.4 4.0 n fragment 321 13 N82 E171 STR 2 OES 1 4.3 3.6 n fragment 322 13 N82 E174 STR 2 OES 1 9.7 5.4 n fragment 322 13 N82 E174 STR 2 OES 1 6.0 3.5 n fragment 324 13 N86 E179 STR 2 OES 1 8.1 5.3 n fragment 308 13 N87 E134 STR 2 OES 1 9.6 6.3 n fragment 326 13 N88 E178 STR 2 OES 1 5.0 5.3 yes fragment 331 13 N91 E151 STR 2 OES 1 7.6 3.8 n fragment 327 13 N91 E155 STR 3 OES 1 6.1 6.0 n fragment 330 13 N97 E158 STR 2 OES 1 7.0 4.5 n fragment 357 13 TTP12 1 OES 1 8.0 4.4 yes fragment 342 13 TTP2 1 OES 1 9.8 5.0 N fragment 365 13 TTP23 1 OES 1 10.0 7.1 yes fragment 365 13 TTP23 1 OES 1 4.7 3.8 yes fragment 283 Table 20 ( co n ) 367 13 TTP25 1 OES 18 8.9 6.3 yes fragment Total Mass = 1.0 G; Msmt Is For Largest Fragment 368 13 TTP26 1 OES 31 7.4 6.8 yes fragment Total Mass = 1.1 G; Msmt Is For Largest Fragment 350 13 TTP5 1 OES 1 4.1 4.9 N fragment 353 13 TTP8 1 OES 1 7.3 6.2 yes fragment 353 13 TTP8 1 OES 1 7.6 6.1 yes fragment 284 Table 21 ACH blanks and fragments Lot Site Unit Level Material Quantity Ext Diam (mm) Int Diam (mm) Burnt? Condition Notes 25 13 2 3 ACH 1 7.36 0 n blank 350 13 TTP5 1 ACH 1 7.15 3.89 N fragment 350 13 TTP5 1 ACH 1 6.35 4.86 N fragment 353 13 TTP8 1 ACH 1 20.47 14.51 N fragment 353 13 TTP8 1 ACH 1 13.33 8.55 N fragment 353 13 TTP8 1 ACH 1 9.18 6.69 N fragment 348 13 21 3 ACH 1 6.23 6.01 n fragment 148 13 6 2 ACH 1 11.62 9.72 n fragment 154 13 6 5 ACH 1 7.51 7.65 n fragment 149 13 6 3 ACH 1 12.91 8.76 n fragment 161 13 7 3 ACH 1 16.76 10.08 n fragment 161 13 7 3 ACH 1 9.31 7.12 n fragment 161 13 7 3 ACH 1 4.83 5.63 n fragment 165 13 7 5 ACH 1 18.34 10.97 n fragment 165 13 7 5 ACH 1 6.11 4.68 n fragment 165 13 7 5 ACH 1 3.12 2.08 n fragment 168/16 9 13 8 3 ACH 1 5.1 4.58 n fragment 179 13 9 3 ACH 1 14.45 11.48 n fragment 186 13 9 6 ACH 1 5.99 7.83 n fragment 186 13 9 6 ACH 1 n fragment tiny 186 13 9 6 ACH 1 n fragment tiny 215 13 12 2 ACH 1 19.89 14.17 n fragment 215 13 12 2 ACH 1 13.23 12.53 n fragment 215 13 12 2 ACH 1 21.25 15.98 n fragment 217 13 12 3 ACH 1 27.49 16.22 n fragment 22 13 2 1 ACH 1 12.59 7.16 n fragment 22 13 2 1 ACH 1 12.7 11.03 n fragment 285 Table 21 ( co n ) 224 13 13 2 ACH 1 7.71 5.76 n fragment prob non - cultural (tear - drop - shaped) 24 13 3 1 ACH 1 5.98 3.02 n fragment 248 13 16 1 ACH 1 16.17 13.35 n fragment 248 13 16 1 ACH 1 8.17 6.35 n fragment 248 13 16 1 ACH 1 11.94 8.58 n fragment 25 13 2 3 ACH 1 14.32 9.18 n fragment 250 13 16 2 ACH 1 16.3 6.57 n fragment 250 13 16 2 ACH 1 10.13 8.27 n fragment 250 13 16 2 ACH 1 12.98 10.31 n fragment 250 13 16 2 ACH 1 17.5 10.56 n fragment 265 13 17 3 ACH 1 8.13 5.98 n fragment 280 13 19 3 ACH 1 14.3 12.3 n fragment 281 13 19 3 ACH 1 18.53 14.7 n fragment 307 13 N106 E120 STR 2 ACH 1 7.1 5.57 n fragment 310 13 N81 E134 STR 3 ACH 1 13.69 10.16 n fragment 314 13 N70 E132 STR 3 ACH 1 34.14 20.28 n fragment 314 13 N70 E132 STR 1 ACH 1 6.17 2.63 n fragment 326 13 N88 E178 STR 2 ACH 1 20.66 13.15 n fragment 326 13 N88 E178 STR 2 ACH 1 10.96 6.94 n fragment 330 13 N97 E158 STR 3 ACH 1 11.38 6.73 n fragment 336 13 N106 E146 STR 1 ACH 1 10.04 8.65 n fragment 336 13 N106 E146 STR 2 ACH 1 6.58 6.04 n fragment 286 Table 21 ( co n ) 337 13 N108 E142 STR 2 ACH 1 37.75 21.77 n fragment 337 13 N108 E142 STR 2 ACH 1 17.96 8.22 n fragment 342 13 TTP2 1 ACH 1 8.13 8.85 N fragment 346 13 21 2 ACH 1 22.1 16.34 n fragment 346 13 21 2 ACH 1 15.51 11.11 n fragment part of inner spiral 346 13 21 2 ACH 1 17.71 12.5 n fragment part of inner spiral 346 13 21 2 ACH 1 17.41 8.21 n fragment 346 13 21 2 ACH 1 17.58 7.46 n fragment 346 13 21 2 ACH 1 18.71 9.47 n fragment 346 13 21 2 ACH 1 21.17 16.49 n fragment 346 13 21 2 ACH 1 7.06 6.84 n fragment 346 13 21 2 ACH 1 4.91 4.1 n fragment 348 13 21 3 ACH 1 18.31 8.5 n fragment 348 13 21 3 ACH 1 9.49 4.97 n fragment 348 13 21 3 ACH 1 17.47 9.95 n fragment 348 13 21 3 ACH 1 13.51 11.77 n fragment 348 13 21 3 ACH 1 14.34 9.67 n fragment 348 13 21 3 ACH 1 7.17 5.52 n fragment 349 13 N40 E99 1 ACH 1 6.75 5.31 N fragment 349 13 N40 E99 1 ACH 1 5.04 4.79 N fragment or blank - is squarish 349 13 N40 E99 1 ACH 1 5.88 4.23 N fragment 357 13 TTP12 1 ACH 1 13.54 11.21 N fragment 357 13 TTP12 1 ACH 1 11.97 6.47 N fragment 357 13 TTP12 1 ACH 1 4.25 3.54 N fragment 361 13 TTP16 1 ACH 1 28.13 17.41 N fragment 363 13 TTP21 1 ACH 1 10.59 7.67 N fragment 364 13 TTP22 1 ACH 1 11.55 8.56 N fragment 366 13 TTP24 1 ACH 1 11.93 8.47 N fragment 366 13 TTP24 1 ACH 1 10.2 6 N fragment 214 13 12 1 ACH 1 17.64mm 9.54mm no fragment 287 Table 21 ( co n ) 214 13 12 1 ACH 1 18.42mm 7.43mm no fragment 214 13 12 1 ACH 1 9.75mm 4.85mm no fragment 214 13 12 1 ACH 1 7.05mm 6.17mm no fragment 227 13 14 1 ACH 1 7.15mm 4.73mm no fragment 236 13 15 1 ACH 1 22.40mm 16.61m m no fragment 240 13 15 3 ACH 1 22.61mm 12.98m m no fragment 240 13 15 3 ACH 1 16.44mm 7.42mm no fragment 246 13 15 6 ACH 1 8.33mm 5.67mm no fragment 365 13 TTP23 1 ACH 1 7.94 7.31 no fragment 370 13 TTP28 1 ACH 1 9.42 6.96 no fragment 45 13 5 3 ACH 32 small n/a n/a fragments numerous small intact shells 161 13 7 3 ACH 1 n natural shell 288 Table 22 River mussel beads Lot Site Unit Level Material Quantity Ext Diam (or length) Int Diam (or width) Burnt? Condition Notes 259 13 17 1 Mussel 1 8.65 6.36 n fragment 325 13 N87 E173 STR 2 Mussel 1 5.21 1.6 n irregular 289 APPENDIX C METAL DATA 290 Table 23 Ferrous beads Lot Unit # Level # Material type Mass (g) Magnet Shape Color Length Width Gap Width Strip Width Thickness Notes 24 3 1 ferr metal bead 0.2 strong c rust 6.32 9.04 3.81 2.83 2.54 24 3 1 ferr metal bead 0.5 strong c rust 6.17 10.99 6.34 2.13 4.78 27 3 2 ferr metal bead 0.1 strong c rust 5.72 8.28 4.44 2.8 1.95 27 3 2 ferr metal bead 0.2 strong horseshoe rust 8.41 8.53 1.19 3.06 2.47 36 3 3 ferr metal bead 0.5 strong closed circle rust 8.46 9.31 0 2.96 3.52 36 3 3 ferr metal bead 0.4 strong Cuff rust 7.91 8.37 0 2.48 3.79 heavily corroded; raised bump, center back 36 3 3 ferr metal bead 0.3 strong Cuff rust 7.77 6.8 0 1.88 3.94 raised bump, center back - part of helix? 36 3 3 ferr metal bead 0.2 strong horseshoe rust 8.31 8.16 1.01 3.52 2.42 heavily corroded 29 4 2 ferr metal bead 0.4 strong Cuff rust 7.6 8.41 1.7 2.07 4.44 291 Table 23 ( co n ) 37 5 1 ferr metal bead 0.3 strong butted circle rust 8.05 8.7 0 2.42 4.69 37 5 1 ferr metal bead 0.6 strong c rust 7.46 13.03 5.49 4.41 2.89 37 5 1 ferr metal bead 0.5 strong c rust 8.47 11.63 4.76 4.74 2.97 37 5 1 ferr metal bead 0.2 strong c rust 5.43 8.06 4.67 2.13 2.68 37 5 1 ferr metal bead 0.5 strong Cuff rust 6.46 9.09 5.3 2.04 6.13 37 5 1 ferr metal bead 0.4 strong Cuff rust 4.63 8.05 4.39 2.71 7.63 part of helix? 37 5 1 ferr metal bead 0.4 strong horseshoe rust 7.14 10.37 2.34 3.45 3.44 43 5 2 ferr metal bead 0.6 strong c rust 9.23 12.6 5.31 5.32 3.23 43 5 2 ferr metal bead 0.2 strong c rust 5.3 8.76 4.09 2.65 2.95 43 5 2 ferr metal bead 0.2 strong c rust 5.45 7.47 3.97 2.53 2.6 292 Table 23 43 5 2 ferr metal bead 0.5 strong horseshoe rust 9.13 9.62 0.92 2.73 4.12 43 5 2 ferr metal bead 0.3 strong horseshoe lt rust 8.03 7.62 0.52 1.24 3.54 152 6 4 ferr metal bead 0.4 strong Cuff rust 8.31 9.41 3.48 2.69 4.4 raised bump, center back 154 6 5 ferr metal bead 0.9 strong butted circle rust 9.55 8.98 0 2.97 5.65 157 6 feat 2 ferr metal bead 0.3 moderate Cuff 6.84 9.97 5.58 2.15 5.16 prob fe, c/ cuff 177 9 2 ferr metal bead 0.1 strong c rust 3.82 6.85 3.4 2.37 2.59 179 9 3 ferr metal bead 0.5 strong Cuff rust 7.94 8.28 2.25 2.22 5.23 181 9 4 ferr metal bead 0.6 strong butted circle rust 8.71 9.05 0 2.27 4.46 181 9 4 ferr metal bead 0.4 strong butted circle rust 8.14 8.16 0 2.72 3.91 heavily corroded 185 9 6 ferr metal bead 0.2 strong c rust 5.13 8.83 3.7 3.31 3.66 293 Table 23 187 9 7 ferr metal bead 0.1 strong c rust 4.73 6.91 3.15 2.28 2.14 236 15 1 ferr metal bead 0.2 strong c rust 9.3 6.69 3.34 3.54 1.64 240 15 3 ferr metal bead 0.2 strong butted circle rust 6.68 6.78 0 2.7 2.02 250 16 2 ferr metal bead 0.4 strong c rust 8.27 9.39 4.88 3.98 5.03 250 16 2 ferr metal bead 0.3 strong c rust 5.87 10.33 4.97 3.91 3.98 272 18 3 ferr metal bead 0.4 strong closed circle rust 7.36 7.55 0 2.47 3.59 281 19 3 ferr metal bead 0.3 strong horseshoe rust 6.63 10.25 2.65 3.96 1.96 289 19 6 ferr metal bead 0.2 strong horseshoe rust 8.71 8.23 1.06 2.78 2.13 293 20 2 ferr metal bead 0.3 strong c rust 4.81 8.41 6.69 2.91 3.67 295 20 3 ferr metal bead 0.7 strong Cuff rust 9.78 9.64 2.09 3.25 6.69 294 Table 23 298 20 4 ferr metal bead 0.4 strong c rust 8.37 12.05 4.88 2.93 2.33 359 PTTP 14 1 ferr metal bead 0.4 moderate butted circle 7.3 7.65 NA 1.82 3.58 prob fe, butted circle 359 PTTP 14 1 ferr metal bead 0.3 moderate closed circle 7.07 7.51 NA 1.62 3.52 prob fe, closed circle 362 PTTP 20 1 ferr metal bead 0.2 moderate c 5.52 8.81 4.05 2.6 1.28 prob fe, c - shape, very low surface corrosion 313.3 STTP N75 E135 3 ferr metal bead 0.4 moderate closed circle 8.38 7.62 NA 2.78 2.41 prob fe, closed circle 295 Table 24 Non - ferrous beads Lot Unit # Level # Material type Mass (g) Magnet Shape Color Length Width Gap Width Strip W Thickness Notes 24 3 1 non ferr metal bead 0.5 non flat link br - blk 11.57 5.04 0 2.34 1.88 prob cu 24 3 1 non ferr metal bead 0.4 non horseshoe br - blk 9.44 9.45 2.16 3.11 1.39 prob cu 28 3 2 non ferr metal bead 0.4 non c 6.97 8.37 3.83 2.97 1.94 prob cu, c - shape 36 3 3 non ferr metal bead 0.2 non butted circle br - blk 6.05 6.9 0 1.09 2.63 prob cu 34 3 3 non ferr metal bead 0.2 non flat horseshoe br - blk 8.79 6.27 4.43 3.05 1.43 prob cu 36 3 3 non ferr metal bead 0.2 non flat horseshoe br - blk 8.07 6.4 3.83 2.84 1.38 prob cu 37 5 1 non ferr metal bead 0.4 non closed circle gray - blk 9.22 8.27 0 1.2 2.6 prob cu 45 5 3 non ferr metal bead 0.3 non horseshoe gray - blk 8.92 8.22 0.86 2.4 1.6 prob cu 146 5 4 non ferr metal bead 0.2 non c br - blk 6.84 12.23 5.6 3.95 1.68 prob cu 163 7 4 non ferr metal bead 0.3 non horseshoe gray - blk 8.55 7.03 1.48 3.05 1.47 prob cu 168/9 8 3 non ferr metal bead 0.2 non c ylw - brwn 7.48 7.39 4.15 3.52 2.18 prob cu 181 9 4 non ferr metal bead 0.4 non horseshoe br - blk 6.92 9.77 3.2 1.72 2.28 prob cu 226 13 3 non ferr metal bead 0.1 non c gray - blk 4.99 6.66 4.45 1.44 2.12 prob cu 238 15 2 non ferr metal bead 0.2 non flat horseshoe ylw - brwn 8.44 6.27 4.34 1.86 3.42 prob cu 240 15 3 non ferr metal bead 0.3 non horseshoe gray - blk 7.01 7.49 1.36 1.24 2.7 prob cu 296 Table 24 241 15 4 non ferr metal bead 0.5 non horseshoe br - blk 10.55 8.72 1.98 2.11 3.53 prob cu 244 15 5 non ferr metal bead 0.4 non butted circle br - blk 8.62 7.77 0 1.63 3.43 prob cu 248 16 1 non ferr metal bead 0.2 non c br - blk 4.29 8.19 5.73 2.3 3.5 prob cu 248 16 1 non ferr metal bead 0.8 non horseshoe green - blk 10.64 12.03 2.04 5.16 3.32 prob cu 259 17 1 non ferr metal bead 0.2 non horseshoe gray - blk 6.03 6.71 2.88 1.37 3.07 prob cu 265 17 2 non ferr metal bead 0.5 non c ylw - brwn 6.43 10.07 3.4 4.36 3.22 prob cu 276 19 2 non ferr metal bead 0.2 non flat horseshoe gray - blk 8.23 6.3 4.03 3.15 1.52 prob cu 281 19 3 non ferr metal bead 0.4 non flat horseshoe gray - blk 8.84 6.51 4.37 3.11 2.92 prob cu 281 19 3 non ferr metal bead 0.7 non flat horseshoe gray - blk 9.6 7.08 3.12 3.19 3.5 prob cu 281 19 3 non ferr metal bead 0.2 non flat horseshoe gray - blk 7.78 6.23 4.35 2.76 1.16 prob cu 281 19 3 non ferr metal bead 0.2 non flat horseshoe gray - blk 7.07 8.21 3.97 2.91 1.38 prob cu 298 20 4 non ferr metal bead 0.3 non butted circle gray - blk 8.76 6.74 0 1.06 3.49 prob cu 346 21 2 non ferr metal bead 0.4 non horseshoe gray - blk 8.51 9.77 1.33 2.51 1.28 prob cu 352 PTTP 7 1 non ferr metal bead 0.3 non horseshoe 9.45 13.25 7.02 2.68 1.09 prob cu, horsehoe 315.1 STTP N72 E155 1 non ferr metal bead 1.1 non horseshoe 11.09 12.54 1.71 4.14 1.54 prob cu, horsehoe 297 Table 24 315.2 STTP N72 E155 2 non ferr metal bead 0.5 non c 3.74 6.21 4.78 2.31 3.11 prob cu, c - shape 315.2 STTP N72 E155 2 non ferr metal bead 0.6 non flat horseshoe 7.08 6.16 4.98 2.22/0.83 1.45 prob cu, flat horse 298 Table 25 Metal fragments, bar and wire Lot Unit Level Mat'l type Mass Magnet Shape Color LENGTH WIDTH HEIGHT Notes 22 2 1 ferr metal rod/bar 0.6 strong rectang ular , with lump/ protrusion on one side 13.63 10.72/7.2 2.99 width/width is w/ w/o protrusion 28 3 2 ferr metal rod/bar 2.9 strong evenly rectang. 30.12 6.85 3.77 36 3 3 ferr 0.2 strong wire/cylinder rust 15.39 3.43 2.18 43 5 2 ferr 1.6 strong wire/cylinder rust 32.6 6.24 4.5 43 5 2 ferr 1.2 strong wire/cylinder rust 25.07 6.68 3.97 43 5 2 ferr 0.8 strong wire/cylinder rust 20.24 4.22 5.74 43 5 2 ferr 0.7 strong wire/cylinder blk 19.37 5.64 2.17 43 5 2 ferr 0.5 strong rectangular rust 9.87 3.92 6.91 240 15 3 non - ferr 11.4 non coil of round wire gry - blk, red flecks 39.08 40.81 3.51 height is wire thickness; prob cu 244 15 5 ferr 0.3 strong end - flattened cylinder rust 15.21 2.97 1.67 250 16 2 non - ferr 0.5 non rect blk 20.15 6.42 2.4 252 16 3 non - ferr 0.1 non wire/cylinder blk 11.25 1.48 252 16 3 ferr 0.6 strong rect rust 12.34 7.14 6.25 259 17 1 ferr 0.5 strong rect rust 17.06 5.53 1.71 261 17 2 ferr 0.5 strong rect rust 13.31 6.31 3.78 293 19 6 ferr 0.3 strong rect rust 8.74 4.16 4.2 322 N 82 E 174 ferr metal frag/ flake 0.3 strong approx rectang 11.89 6.42 2.89 322 N 82 E 174 ferr metal frag/ flake 0.6 strong approx rectang 13.06 13.41 2.58 322 N 82 E 174 ferr metal frag/ flake 0.1 strong rectang 9.04 3.45 1.88 299 Table 25 308 N 87 E 134 2 ferr metal wire/rod 0.4 strong sl ighttly assym (prob from corrosion) rectang strip 19.31 3.12 2.21 300 Table 26 Slag Lot Site # Unit # Level # Material type Q uanti ty Mass (g) Magnetism Texture Color Width (mm) Length (mm) Height (mm) N otes 24 13 3 1 ferr 1 0.1 strong semi - spng gray - blk, red - brown flecks 6.31 4.58 27 13 3 2 ferr 1 0.4 weak spongy ylw - brwn, blk - gray 8.75 8.45 6.03 34 13 3 3 non ferr 1 3 non spongy blue - blk, ylw 23.21 19.09 11.45 34 13 3 3 non ferr 1 0.2 non spongy blk/green, ylw 8.09 7.4 4.67 238 13 15 2 non ferr 1 1.4 non spongy ylw - brwn, blk - gray 20.78 15.41 10.72 241 13 15 4 ferr 1 0.2 weak spongy gry - ylw 7.08 5.99 4.16 248 13 16 1 ferr 1 0.6 strong semi - spng red - brown 13.35 7.2 5.33 248 13 16 1 ferr 1 0.2 strong smooth red - brown 10.48 6.34 3.84 248 13 16 1 non ferr 1 0.8 non spongy gry - ylw 10.85 9.49 7.26 248 13 16 1 non ferr 1 0.1 non spongy black 6.24 5 3.89 250 13 16 2 ferr 2 9.4 strong smooth non - spng; red - brwn; surf corrosion 250 13 16 2 ferr 4 0.9 strong smooth red - brown 250 13 16 2 ferr 1 1.2 weak spongy red - brown 16.38 12.89 8.47 250 13 16 2 ferr 1 3.6 weak spongy blk - brown 18.64 17.75 13.91 250 13 16 2 non ferr 2 0.5 non spongy black 13.05 9.47 6.35 250 13 16 2 non ferr 3 0.5 non spongy gry - ylw 11.53 7.9 4.13 252 13 16 3 ferr 1 0.2 strong smooth red - brown 9.71 5.52 2.18 252 13 16 3 ferr 3 0.8 strong spongy ylw - brwn, blk - gray 61.3 51.39 27.63 252 13 16 3 non ferr 2 69.7 non spongy gry - ylw 10 11.54 7.84 301 Table 26 254 13 16 4 ferr 2 8.1 strong smooth non - spng; red - brwn; one piece may be slag attached to a piece of calcrete 254 13 16 4 non ferr 1 2.2 non spongy spng; ylw - gry; could be vit dung 18.63 20.6 8.39 254 13 16 4 non ferr 3 5.1 non spongy spng; blk - gry - ylw - grn could be vd? 254 13 16 4 non ferr 1 2.3 non spongy gry - ylw 20.24 18.7 7.98 256 13 16 5 ferr 1 1.6 strong spongy red - brown 14 12.35 8.98 256 13 16 5 ferr 1 3.2 strong granular red - brown 20.56 14.4 9.92 256 13 16 5 non ferr 1 2.5 strong semi - spng blk - gry - red 18.98 13.49 9.31 256 13 16 5 non ferr 1 5.6 non smooth green - black 24.74 16.78 9.79 256 13 16 5 non ferr 1 0.3 non smooth gry - ylw 8.69 6.29 4.41 265 13 17 2 ferr 1 0.5 weak granular red - brown 10.87 8.32 6.95 267 13 17 4 ferr 1 0.1 strong smooth non - spng; red - brwn 276 13 19 2 non ferr 1 0.3 non spongy wht - gry - blk 8.51 6.05 4.95 276 13 19 2 non ferr 1 0.2 non spongy ylw - brwn, blk - gray 4.88 8.73 3.61 276 13 19 2 non ferr 1 0.1 non semi - spng ylw - brwn, blk - gray 5.67 4.22 3.46 302 Table 26 276 13 19 2 non ferr 1 0.3 non semi - spng ylw - brwn, blk - gray 10.04 8.25 6.13 281 13 19 3 ferr 2 0.9 strong spongy red - brown 9.07 7.85 5.88 281 13 19 3 ferr 1 0.7 strong spongy black 12.66 8.33 7.23 281 13 19 3 non ferr 3 2.4 non spongy black 18.18 11.09 8.24 281 13 19 3 non ferr 2 5.2 non spongy ylw - brwn, blk - gray 24.67 20.27 11.65 284 13 19 4 non ferr 1 1.2 non semi - spng black 16.9 12.66 5.88 284 13 19 4 non ferr 1 0.3 non semi - spng black 11.29 7.38 7.2 295 13 20 3 non ferr 1 0.7 non granular black 14.1 11.86 10.09 295 13 20 3 non ferr 1 0.6 non granular black 13.15 9.32 7.48 305. 3 13 STTP N102 E117 3 non ferr 1 4.8 non smooth non - spng; blk - gry 12.02 19.87 10.51 308. 3 13 STTP N87 E134 3 non ferr 1 0.3 non spongy spng; ylw - gry; could be vit dung 8.14 7.42 6.22 311. 2 13 STTP N77 E133 2 ferr 1 0.5 strong spongy spng; blk - gry - ylw 313. 2 13 STTP N75 E135 2 ferr 1 0.1 strong smooth non - spng; blk - gry 319. 3 13 STTP N78 E158 3 non ferr 1 0.6 non spongy spng; ylw - gry could be vd? 320. 2 13 STTP N78 E160 2 non ferr 1 1.7 non smooth non - spng; blk - gry 8.53 12.96 7.3 303 Table 26 320. 3 13 STTP N78 E160 3 ferr 1 0.2 strong spongy spng; red - brwn 320. 3 13 STTP N78 E160 3 ferr 1 0.3 strong spongy spng; blk - gry - ylw 322. 1 13 STTP N82 E174 1 ferr 3 2.6 strong spongy spng; red - brwn 322. 1 13 STTP N82 E174 1 ferr 3 1.4 strong spongy spng, blk - gry 322. 2 13 STTP N82 E174 2 ferr 5 6.7 strong spongy spng, blk - gry 322. 2 13 STTP N82 E174 2 ferr 2 0.6 strong spongy spng; red - brwn 322. 2 13 STTP N82 E174 2 non ferr 5 1.4 non smooth non - spng; blk - gry 322. 2 13 STTP N82 E174 2 non ferr 8 8 non spongy spng; blk - gry - ylw could be vd? 322. 3 13 STTP N82 E174 3 ferr 1 0.8 strong smooth non - spng; red - brwn 322. 3 13 STTP N82 E174 3 ferr 3 1.6 strong spongy spng; blk - gry - ylw 304 Table 26 327. 2 13 STTP N91 E155 2 non ferr 1 0.2 non spongy spng; blk - gry - ylw could be vd? 328. 2 13 STTP N95 E153 2 ferr 1 0.2 strong spongy spng; blk - gry - ylw 8.45 8.46 5.33 328. 3 13 STTP N95 E153 3 non ferr 7 14.9 non spongy spng; blk - gry - ylw - grn, sml spots of red could be vd? 346 13 21 2 non ferr 1 5.4 non spongy spng; ylw - gry; could be vit dung; high qtz content 17.24 29.03 17.6 this is remarkably dense in compariso n with other non - spng pieces of similar size 346 13 21 2 non ferr 1 5.4 non spongy ylw - brwn, blk - gray 28.8 17.1 17.99 grains of sand attached? 356 13 PTTP 11 1 non ferr 1 0.8 non spongy spng; blk - gry. attached to ceramic fragment 9.8 1026 9.12 msmts and mass inc ceramic frag 375 12 1 2 non ferr 2 0.1 non spongy blk - red - brwn 4.44 5.44 393 33 1 2 ferr 1 1.6 strong semi - spng red - brown 11.1 9.2 305 Table 26 395 33 1 3 ferr 1 0.2 strong smooth gray - blk, red - brown flecks 26.47 20.81 168 /9 13 8 3 ferr 1 5.6 strong spongy ylw - brwn, blk - gray 24.61 20.78 14.39 306 Graphs 0 20 40 60 80 100 120 Beads/ bangle pieces Wire/ bar Total Worked metal object proportions (by count) Copper/ non-ferrous Iron Figure 50 Worked metal object proportions (by count) 307 0 10 20 30 40 50 60 Beads/ bangle pieces Wire/ bar Total Worked metal object proportions, by mass (g) Copper/ non-ferrous Iron 15.7 15.1 12.3 55.1 98.2 11.8 0 12 149.1 172.9 0 20 40 60 80 100 120 140 160 180 200 Beads/ bangle pieces Fragments - amorphous wire/ bar slag Total Metal Items (by mass, g) Iron (ferrous) Copper (or non-ferrous) Figure 51 Metal items by mass 308 2.8 1.6 5.3 3.9 2.1 0 0 15.7 0.9 0.4 2.1 0 5 2.9 0.5 11.8 0 2 4 6 8 10 12 14 16 18 Metal Jewelry (by mass, g) Iron (Ferrous) Copper (or non-ferrous) Figure 52 Metal jewelry by mass 309 44 49 18 51 162 32 0 3 63 98 0 20 40 60 80 100 120 140 160 180 Beads/ bangle pieces Fragments - amorphous wire/ bar slag Total Metal Items (by count) Iron (Ferrous) Copper (or non-ferrous) F igure 53 Metal items by count 310 6 4 18 9 7 0 0 44 3 1 7 0 11 9 1 32 0 5 10 15 20 25 30 35 40 45 50 Metal Jewelry (by count) Iron (ferrous) Copper (or non-ferrous) Figure 54 Metal jewelry by count 311 APPENDIX D OSL REPORT 312 LUMINESCENCE DATING IN KALAHARI DESERT, BOTSWANA 27 October 2014 James Feathers Luminescence Dating Laboratory University of Washington Seattle, WA 98195 - 3412 jimf@u.washington.edu Two sediment samples from an archaeological site in Botswana were submitted for luminescence dating by Adrianne Daggett, Michigan State University. The site is located on a hilltop in the Kalahari Desert near the Makgadikgadi Pans. The sediments are sandy, m uch of it aeolian in origin, but contain gravel and small stones as well. Table 1 lists the samples. Laboratory procedures are given in the appendix. Table 27 OSL Samples Lab # Site lot Burial Depth (cm) UW2850 16: A1.13 278 5 - 10 UW2851 16: A1.13 279 15 - 20 Dose rate was determined by alpha counting, beta counting and flame photometry as described in the appendix. Table 2 8 gives the relevant concentrations as determined from alpha counting (U and Th) and flame photometry (K). The beta dose rate calculated from these concentrations is compared with that measured directly by beta counting, and this is also given in Table 2 8 . There are no significant differences that might be caused, for example, by U - Th disequilibrium in the U decay chain. Moisture content was estimated at 6 ± 3 %, which is more than the measured values of about 1%, which, however, do not reflect seasonal change. Table 29 gives the estimated dose rates, which are similar for both samples. Table 28 Radiation Sample 238 U (ppm) 233 Th (ppm) K (%) Beta dose rate (Gy/ka) ß - counting - counting/ flame photometry UW2850 1.30±0.12 4.84±0.86 1.38±0.04 1.49±0.14 1.46±0.04 UW2851 0.77±0.08 3.00±0.60 1.49±0.03 1.43±0.13 1.42±0.04 Table 29 Dose rates (Gy/ka)* Sample alpha beta gamma cosmic total UW2850 0.01±0.01 1.22±0.05 0.56±0.05 0.27±0.05 2.06±0.09 UW2851 0.01±0.01 1.21±0.05 0.53±0.03 0.25±0.05 1.99±0.08 Luminescence was measured on single - grain quartz using a green laser. Equivalent dose was determined as described in the appendix. Table 30 gives the number of grains measured, the number rejected using the criteria given in the appendix, and the number accepted. The samples showed 313 relatively high luminescence sensitivity for quartz. The acceptance rate for signals from which an equivalent dose could be measured averaged 13%. A high number of zero - aged grains in UW2850 suggests some contamination from the surface. Table 30 Acceptance rates* Sample N No signal Recycle Too high Recuperation Feldspar Zero dose Accepted Rate (%) UW2850 878 679 47 7 0 4 41 100 11.4 UW2851 876 672 36 19 2 4 15 128 14.6 total 1754 1351 83 26 2 8 56 228 13.0 * N refers to the number of grains measured, no signal refers to grains that had no measurable signal, recycle refers to number of grains rejected for failing the recycle test and for no other reason, too high refers to natural signals higher than the sign al from the highest regeneration point, recuperation refers to significant signal after zero dose and preheat, feldspar refers to feldspar contaminated as detected by sensitivity to IRSL, zero dose refers to grains rejected because the equivalent dose was not significantly different from zero. Parameters of the criteria are given in the appendix. A dose recovery test was conducted on 400 grains from both samples, of which 57 passed all the acceptance criteria. The central tendency of the derived/admi nistered (~16 Gy) ratio, from the central - dispersion of the ratio distribution is 7.5%, which is a measure of intrinsic variation due to machine and sample factors and which can be taken as the minimum over - dispersion expected for a single - aged sample. A value of 10% was taken as typical for a single - age sample when evaluating age distributions. Table 31 gives the equivalent dose from the central age model (Galbraith and Roberts 2012) and the over - dispersion for each sample. The over - dispersion is quite high for both samples. A finite mixture model was applied to divide the sample into single - v alue components (see appendix). These are shown in Table 32 . The samples appear quite mixed. Given the mode of deposition, it is unlikely the samples are partially bleached. The younger components probably represent downward movement of grains exposed at the surface in these shallow deposits. Also unlikely to be correct are the old components in UW2851, particularly the fifth one, which would give a late Pleistocene age. Radial graphs are shown in Figure 1 . For UW2850 most values seem to cluster aro und the third component. This probably best represents the depositional age. For UW2851, even discounting the larger and smaller values, there is still substantial scatter. Short of any better information, the central age model, represented by the red line with a value similar to the third component, may present the best estimate, although arguments could be made for a younger age. Table 31 Equivalent dose (central age model) Sample N Corrected Age (ka) Over - dispersion (%) UW28 50 100 0.80±0.08 82.6±8.3 UW2851 128 2.53±0.34 144±10 314 Table 32 Finite Mixture Model components Sample Component Equivalent dose (Gy) Proportion (%) UW2850 1 0.20±0.02 18.6 2 0.63±0.03 29.9 3 1.33±0.06 43.1 4 2.29±0.31 8.4 UW2851 1 0.41±0.03 12.9 2 1.21±0.06 32.7 3 2.36±0.09 29.2 4 6.16±0.36 5.7 5 29.4±1.57 19.5 Table 33 gives the ages. Using the largest component on UW2851 makes that sample about the same age as UW2850, which would suggest rapid deposition of the deposit. The older age, from the central age model, separates the layers in time and also agrees with a rad iocarbon date (Daggett, personal communication) Table 33 Ages Sample Age (ka) Error (%) basis Calendar (years AD) UW2850 0.65 ± 0.04 6.9 Largest component 1370 ± 40 UW2851 1.27 ± 0.18 14.3 Central age model 740 ± 180 UW2851 0.61 ± 0.04 6.8 Largest component 1410 ± 40 315 Radial graphs of each sample. A radial graph plots precision against the equivalent dose, standardized to the number of standard errors the value is from a reference point. The middle reference point in both graphs, shown as the red line is from the central age model. Reference points for UW2850 are the first (pink) and third (blue) components. Shaded areas encompass all points within two standard errors of the reference. Reference p oints for UW2851 are also the first and third components. UW2850 UW2851 Figure 55 Radial graphs of each OSL sample 316 Procedures for Luminescence Analysis of Coarse - grained Quartz from Sediment Samples Sample preparation Sample material is removed from the collection container, leaving aside any portions (such as the ends of tubes) that may have been exposed to light. The latter may be used for dose rate information. From the unexposed portions, about ¼ is set aside as a voucher (material that can be used at a latter date if necessary). If separate samples for measuring moisture content have not been collected, the voucher can be used for this. For moisture the sample is simply weighed wet, and then dried for severa l days at 50°C before weighing again. The wet minus the dry weight divided by the dry weight gives the percent moisture by weight. Remaining unexposed material is separated into size fractions by sieving. If the material contains abundant silt or clay, t he sample is wet sieved through a 90µm screen. Otherwise it is sieved dry. The greater than 90µm fraction is treated with HCl and H 2 O 2 , rinsed three times with water and dried. It is then dry sieved to retrieve the 180 - 212µm fraction (or any other fract ion deemed appropriate). This fraction is etched for 40 minutes in HF and then rinsed with water, HCl and water again. After drying, it is passed through the 180µm screen to remove any degraded feldspar. The material caught in the screen is density sepa rated using a lithium metatungstate solution of 2.67 specific gravity. Equivalent dose Grains are placed in specially - manufactured disks for single - grain measurement. Luminescence is measured on either a Risø TL - DA - 15 reader or a Risø TL/OSL DA - 20 re ader with single - grain attachment. Stimulation is by a 532nm laser delivering 45 W/cm 2 . Detection is through 7.5 mm U340 (ultraviolet) filters. Exposure is for 0.8s on each grain at 125°C. The first 0.06s is used for analysis and the last 0.15s for bac 240°C for 10 seconds follows each dose, except for the cali brating test doses after which a 200°C for 1 second preheat is employed. The test dose usually employed is about 5 Gy. Doses are delivered by a 90 Sr beta source which provides about 0.1 Gy per second to coarse - grained quartz. Equivalent dose (D e ), whic h is a measure of the total absorbed dose through time, is determined using the single - aliquot regenerative dose (SAR) protocol (Murray and Wintle 2000, Wintle and Murray 2006). The SAR method measures the natural signal and the signal from a series of re generation doses on a single aliquot. The method uses a small test dose to monitor and correct for sensitivity changes brought about by preheating, irradiation or light stimulation. SAR consists of the following steps: 1) preheat, 2) measurement of natur al signal (OSL or IRSL), L(1), 3) test dose, 4) cut heat, 5) measurement of test dose signal, T(1), 6) regeneration dose, 7) preheat, 8) measurement of signal from regeneration, L(2), 9) test dose, 10) cut heat, 11) measurement of test dose signal, T(2), 1 2) repeat of steps 6 through 11 for various regeneration doses. A growth curve is constructed from the L(i)/T(i) ratios and the equivalent dose is found by interpolation of L(1)/T(1). A zero regeneration dose and a repeated regeneration dose are employed to insure the procedure is working properly. An advantage of single - grain dating is the opportunity to remove from analysis grains with unsuitable characteristics by establishing a set of criteria grains must meet. Grains are eliminated from analysis if they (1) had poor signals (as judged from errors on the test dose greater than 30 percent or from net natural signals not at least three times above the background standard deviation), (2) did not produce, within 20 percent, the same signal ratio (often c alled recycle ratio) from identical regeneration doses given at the beginning and end of the SAR sequence, suggesting inaccurate sensitivity correction, (3) yielded natural signals that did not intersect saturating growth curves, (4) after a zero dose had a signal 317 larger than 10 percent of the natural signal or a signal not distinguishable from background, (5) produced a zero D e (within 1 - sigma of zero), or (6) contained feldspar contaminates (judged visually on growth curves by a reduced signal from infrared stimulation before the OSL measurement; done on two doses to lend confidence the reduction in signal is due to feldspar co ntamination). At the end of each SAR sequence, linear - modulated OSL (where the laser power is ramped from 0 to 90% power in 30 seconds) is measured for each grain. Grains clearly dominated by a component other than the fast component (judged visually) a re marked, and if the D e from these differ significantly from those of fast - component grains, they are also removed from analysis. A dose recovery test is performed on some grains. The luminescence of the grains is first removed by exposure to the laser ( using the same parameters mentioned earlier). A dose of known magnitude is then administered. The SAR procedure is then applied to see if the known dose can be obtained. Successful recovery is an indication that the procedures are appropriate. A D e val ue is obtained for each suitable grain. Because of varying precision from grain to grain, the same value is not obtained for each grain even if all are of the same age. Instead a distribution is produced. The common age model and central age model of G albraith (Galbraith et al. 1999, 2005, Galbraith and Roberts 2012) are statistical tools used in evaluation of D e distributions. These models are used in reference to D e e values by the bulk dose rate provides model controls for differential precision by computing a weighted average using log D e values. The central age model is similar except rather than assuming a single true value it assumes a normally - distributed natural distribution of D e values, even for single - aged samples, because of non - statistical sources of variation. The central age is the mean of that distribution and the standard deviation is the over - dispersion ( b ), which represents that deviation beyond what can be accounted for by measurement error. Empirical evidence suggests that b of between 10 to 20 percent are typical for single - aged samples (Olley et al. 2004, Jacobs et al. 2006). For samples of mixed ages, either a minimum age model or a finite mixture model is employed for evaluation. The minimum age (Galbraith et al. 1999) calculates a statistical minimum using a truncated normal distribution and is suitable for partially ble ached samples. Finite mixture model (Roberts et al. 2000) uses maximum likelihood to separate the grains into single - aged components based on the input of a given b value and the assumption of a log normal distribution of each component. The model estim ates the number of components, the weighted average of each component, and the proportion of grains assigned to each component. The model provides two statistics for estimating the most likely number of components, maximum log likelihood (llik) and Bayes Information Criterion (BIC). The finite mixture model is appropriate for samples that have been post - depositionally mixed (although with limitations). Dose Rate Radioactivity is measured by alpha counting in conjunction with atomic emission for 40 K. Samples for alpha counting are crushed in a mill to flour consistency, packed into plexiglass containers with ZnS:Ag screens, and sealed for one month before counting. The pairs technique is used to separate the U and Th decay series. For atomic emission measurements, samples are dissolved in HF and other acids and analyzed by a Jenway flame photometer. K concentrations for each sample are determined by bracketing between standards of known concentration. Conversion to 40 K is by natural atomic abundance. Radioactivity is also measured, as a check, by beta counting, using a Risø low level beta GM multicounter system. About 0.5 g of crushed sample is placed on each of four plastic sample holders. All are counted for 24 hours. The average is converted t o dose rate following Bøtter - Jensen and Mejdahl (1988) and compared with the beta dose rate calculated from the alpha counting and flame photometer results. 318 Additional soil samples are analyzed where the environment is complex, and gamma contributions dete rmined by gradients (after Aitken 1985: appendix H). For some samples, in situ measurements are done using a CaSO 4 :Dy dosimeter. About 0.1 g of the powder is sealed in a copper capsule. After heating to remove any latent luminescence, the capsule is pla ced in the ground as near to the location of the sample as possible and left for one year. The dosimeter is returned to the laboratory accompanied by a travel dosimeter which measures any radiation absorbed enroute. In the laboratory, the luminescence fr om the CaSO 4 is measured by thermoluminescence and the signal calibrated against a beta source. The source is used with the shutter closed to provide a low calibrating dose, about 0.001 Gy/1000s. Cosmic radiation is determined after Prescott and Hutton (1 988). Radioactivity concentrations are translated into dose rates following Guérin et al. (2011). Age is calculated using a laboratory constructed spreadsheet based on Aitken (1985). All given error terms are computed at one - sigma. 319 BIBLIOGRAPHY 320 BIBLIOGRAPHY Adamiec, G., and Aitken, M. J., 1998, Dose rate conversion factors: update. Ancient TL 16:37 - 50. Aitken, M. J., 1985, Thermoluminescence Dating , Academic Press, London. Bøtter - Jensen, L, and Mejdahl, V., 1988, Assessment of beta dose - rate using a GM multi - counter system. Nuclear Tracks and Radiation Measurements 14:187 - 191. Galbraith, R. F., Roberts, R.G., Laslett, G. M., Yoshida, H., & Olley, J. M., 1999, Optical datin g of single and multiple grains of quartz from Jinmium rock shelter, northern Australia: Part I, experimental design and statistical models. Archaeometry 41:339 - 364. Galbraith, R. F., Robert s, R. G., & Yoshida, H., 2005, Error variation in OSL palaeodose estimates from single aliquots of quartz: a factorial experiment. Radiation Measurements 39:289 - 307. Galbraith, R. F., and Roberts, R. G., 2012, Statistical aspects of equivalent dose and error calculation and display in OSL dating: an overview and some recommentdations. Quaternary Geochronology 11:1 - 27. Guérin, G., Mercier, N., and Adamiec, G., 2011, Dose - rate converstion factors: update. Ancient TL 29:5 - 8. Jacobs, Z., Duller, G. A. T., & Wintle, A. G., 2006, Interpretation of single grain D e distr ibutions and calculation of D e . Radiation Measurements 41:264 - 277. Murray, A. S., and Wintle, A. G., 2000, Luminescence dating of quartz using an improved single - aliquot regenerative - dose protocol. Radiation Measurements 32:57 - 73. Olley, J. M., Pietsch , T., & Roberts, R. G., 2004, Optical dating of Holocene sediments from a variety of geomorphic settings using single grains of quartz. Geomorphology 60:337 - 358. Prescott, J. R., and Hutton, J. T., 1988, Cosmic ray and gamma ray dose dosimetry for TL and ESR. Nuclear Tracks and Radiation Measurements 14:223 - 235. Roberts, R. G., Galbraith, R. F., Yoshida, H., Laslett, G. M., & Olley, J. M., 2000, Distinguishing dose populations in sediment mixtures: a test of single - grain optical dating procedures using mixtures of laboratory - dosed quartz. Radiation Measurements 32:459 - 465. Wintle, A. G., & Murray, A. S., 2006, A review of quartz optically stimulated luminescence characteristics and their relevance in single - aliquot regeneration dating protocols. Radiation Measurements 41:369 - 391. 321 APPENDIX E GLASS BEAD ANALYSIS REPORT 322 Report on Thaba di Masego glass beads Prepared by Marilee Wood, Dec. 1 2013 Description The two sites at Thaba di Masego (16 - A1 - 13 and 16 - A1 - 12) combined produced 48 glass beads and one fragment of a substance that might be glass. Only the beads will be discussed. Forty - two of th e beads belong to the Zhizo series of which 3 are too corroded to allow determination of color, 36 are light cobalt blue (the most common color found in this series), one is yellow and two are rather vivid bluish - green (a very unusual color but glass quali ty and bead shape indicate they do belong to the Zhizo group). Six beads may belong to the Chibuene series of which 2 are light grayish cobalt blue and 4 are grayish blue - green. Most of the beads are corroded to varying degrees, a condition that is often found with the Zhizo series due to the composition of the glass from which the beads are made. I cleaned some of them with an ultrasonic cleaner but determined the color of others by scraping away a small section of patina. All of the beads were made fro m drawn tubes that were cut into bead lengths and left in that condition (rather than being reheated to round the sharp ends). This lack of secondary heating is typical of the Zhizo series. The glass used to make both the Zhizo and Chibuene series beads was probably made in the Near East, east of the Euphrates ( Robertshaw et al. 2010:1903; Wood et al. 2012:71) possibly in Persia or nearby. The drawing method used to make the beads, however, is a south Asian technology so they were probably made by sout h Asian artisans - possibly by ones who worked in the Persian Gulf region. This proposition - that the beads were made in that region - is based on the probability that the glass was manufactured there and the likelihood that the beads were brought to Sou thern Africa by traders from Oman and the Persian Gulf, whose ships were the most active in the western Indian Ocean from the 7 th or 8 th century CE to the mid - 10 th century (Wood 2011: Introduction:28 - 29). Interpretation This assemblage of beads adds important new evidence of Indian Ocean trade reaching the interior of Southern Africa before the end of the first millennium CE. Zhizo and Chibuene series beads are the only types that have been found in the interior prior to the mid - 10 th century. The Chibuene series has up to now only been recognized at the port site of Chibuene, in Southern Mozambique, and at Nqoma ( Wood et al. 2012:67) , in the Tsodilo Hills well west of Thaba di Masego. Archaeologists working at Nqoma hav e proposed that the trade bringing foreign goods may have travelled via a Zambezi route, rather than a more southerly one. They have not, however, been able to identify any Zhizo beads in the Zambezi region. The presence of such a large number of Zhizo b eads, along with potential Chibuene series beads, at the edge of the Sowa Pan suggests that a trade route originating at the port of Chibuene and reaching as far as Nqoma may have passed through the Sowa Pan region. At the site of Chibuene most of the Chibuene series beads were found near the base of the occupation layers so it is believed they slightly predate the Zhizo series (Wood et al. 2012:66). If this is the case, Thaba ther back than the mid - 9 th c. radiocarbon date for the site. The assemblages of both 16 - A1 - 13 and 16 - A1 - 12 are very similar and the latter, even though few beads are present, contains one possible Chibuene series bead. This suggests that the two sites wer e occupied more or less contemporaneously. Finally, because no beads of any later series are present, it suggests that the sites were abandoned sometime before the mid - 10 th century. 323 BIBLIOGRAPH Y 324 BIBLIOGRAPHY 2010. Robertshaw, P., Wood, M., Melchiorre, E., Popelka - Filcoff, R.S., & Glascock, M.D. Southern African glass beads: chemistry, glass sources and patterns of trade . Journal of Archaeological Science 37 : 1898 - 1912. 2011. Wood, M. Interconnections: Glass beads and trade in So uthern and eastern Africa and the Indian Ocean 7th to 16th centuries AD. Studies in Global Archaeology 17 . Department of Archaeology and Ancient History, Uppsala University, Uppsala, Sweden. 2012. Wood, M., Dussubieux, L. & Robertshaw, P. Glass finds from Chibuene, a 6th to 17th century AD port in Southern Mozambique. South African Archaeological Bulletin 67 (195): 59 - 74. 325 Table 34 Glass bead compositions (major and minor oxides only) Sample ID Lot # Bead ID Diaphan eity Muns ell # Color group Series SiO 2 Na 2 O Mg O Al 2 O 3 P 2 O 3 Cl K 2 O CaO Mn O Fe 2 O 3 Cu O Sn O 2 Pb O 2 13Lot2 1 21 T DM 07 tsp - tsl 2.5PB 5/4 blue Zhizo 66.7 3% 12.6 9% 3.2 3% 3.6 6% 0.2 1% 0.7 3% 4.4 2% 5.23 % 0.9 4% 1.3 9% 0.0 8% 0.0 7% 0.54 % 1 3Lot2 3R 23 TDM 8 tsp - tsl 2.5PB 5/4 blue Indet. 71.4 2% 0.15 % 0.6 1% 8.1 4% 0.1 8% 0.2 1% 1.2 9% 10.7 6% 4.0 0% 2.4 8% 0.0 4% 0.0 0% 0.06 % 1 3Lot2 6 26 TDM 09 tsp 2.5PB 5/4 blue Zhizo 64.1 4% 13.5 7% 4.0 0% 3.2 5% 0.1 7% 1.3 4% 5.1 2% 5.61 % 0.9 4% 1.3 2% 0.1 1% 0.0 4% 0.35 % 1 3Lot2 7 27 TDM 02 tsp - tsl 5PB 4/6 blue Zhizo 64.6 3% 15.8 2% 3.9 6% 2.9 4% 0.1 0% 1.8 1% 3.7 4% 4.89 % 0.7 8% 0.7 7% 0.0 8% 0.0 3% 0.38 % 13Lot3 6R 36 TDM 1 opaque NA black Indet. 63.2 6% 0.94 % 2.2 4% 5.5 0% 0.4 2% 0.0 9% 4.5 9% 22.6 8% 0.1 7% 0.1 1% 0.0 0% 0.0 0% 0.00 % 1 3Lot3 7A 37A TDM 03 tsp - tsl 2.5PB 5/4 blue Zhizo 63.0 7% 15.2 5% 3.6 4% 3.3 6% 0.2 1% 1.2 1% 4.2 6% 5.79 % 1.3 6% 1.1 1% 0.0 8% 0.0 5% 0.53 % 1 3Lot3 7B 37B TDM 04 tsp - tsl 2.5PB 5/4 blue Zhizo 65.3 5% 13.3 5% 3.8 7% 3.2 4% 0.2 2% 1.3 7% 3.9 5% 5.38 % 1.4 9% 1.0 2% 0.0 7% 0.0 7% 0.58 % 1 3Lot3 7C1 37C - 1 TDM 05 tsl 5BG 5/1 gray - b - g CHB 66.7 9% 14.6 7% 2.4 0% 3.3 7% 0.4 9% 1.8 7% 4.4 7% 3.37 % 0.5 3% 1.6 8% 0.1 9% 0.0 1% 0.07 % 1 3Lot3 7C2 37C - 2 TDM 05 tsl 5BG 5/1 gray - b - g CHB 65.6 6% 14.6 0% 2.3 2% 3.5 8% 0.5 4% 1.9 7% 5.3 0% 3.43 % 0.5 7% 1.6 3% 0.2 0% 0.0 1% 0.08 % 1 3Lot3 7C3 37C - 3 TDM 05 tsl 5BG 5/1 gray - b - g CHB 65.5 7% 14.8 7% 2.4 3% 3.5 1% 0.5 0% 1.9 5% 5.0 2% 3.53 % 0.5 8% 1.6 8% 0.2 1% 0.0 1% 0.07 % 13Lot3 7D 37D TDM 06 tsl - tsp 5PB 4/6 blue Zhizo 64.4 8% 14.2 6% 4.5 7% 2.9 3% 0.1 3% 1.1 7% 4.2 7% 5.08 % 0.9 0% 0.8 7% 0.0 8% 0.1 4% 1.08 % 13Lot3 7F 37F TDM 48 tsl 2.5PB 5/4 blue CHB 65.5 3% 16.0 3% 2.2 9% 3.1 2% 0.4 1% 1.8 7% 4.8 0% 3.52 % 0.5 9% 1.4 0% 0.3 2% 0.0 0% 0.03 % 1 3Lot3 7G 37G TDM 47 tsp - tsl 2.5PB 5/4 blue Zhizo 65.2 8% 13.8 4% 4.6 2% 2.7 2% 0.1 2% 1.2 2% 4.4 7% 5.62 % 1.0 2% 0.9 2% 0.0 8% 0.0 0% 0.03 % 1 3Lot3 7H 37H TDM 46 tsl 5Y 6/8 yellow Zhizo 64.8 6% 14.8 6% 4.7 9% 2.5 3% 0.1 9% 0.6 0% 3.2 4% 5.77 % 0.8 2% 0.5 5% 0.0 0% 0.2 7% 1.51 % 1 3 Lot4 5 45 TDM 10 tsp - tsl 2.5PB 5/4 blue CHB 68.5 6% 15.5 9% 2.1 0% 3.0 9% 0.3 9% 0.7 2% 3.7 6% 2.91 % 0.4 0% 1.8 3% 0.2 2% 0.0 3% 0.28 % 326 Table 34 13Lot1 46R 146 TDM 11 tsp - tsl 2.5PB 5/4 blue Indet. 75.8 0% 0.02 % 0.5 9% 7.2 4% 0.0 8% 0.1 5% 2.0 2% 4.53 % 0.0 4% 9.4 9% 0.0 3% 0.0 0% 0.00 % 13Lot1 59 159 TDM 12 tsp 2.5PB 5/4 blue Zhizo 64.0 8% 13.9 5% 4.4 3% 3.2 7% 0.1 4% 1.1 2% 4.1 8% 5.48 % 0.7 9% 1.1 8% 0.0 9% 0.1 2% 1.12 % 13Lot1 62 - 3 162 - 3 TDM 14 tsp - tsl 2.5PB 5/4 blue Zhizo 65.0 7% 14.0 4% 3.4 8% 2.5 2% 0.1 3% 0.6 6% 3.6 6% 4.24 % 1.1 9% 0.4 7% 0.0 0% 0.8 0% 3.68 % 13Lot1 63 163 TDM 13 tsp - tsl 2.5PB 5/4 blue Zhizo 66.4 1% 15.1 2% 3.8 8% 3.4 5% 0.1 4% 0.5 1% 3.4 1% 4.94 % 0.5 5% 1.1 1% 0.0 6% 0.0 3% 0.35 % 13Lot1 66 166 TDM 15 tsl - tsp 5PB 4/6 blue Zhizo 65.8 4% 14.2 5% 4.2 9% 2.9 7% 0.1 9% 1.5 5% 3.9 2% 4.73 % 0.8 0% 0.8 2% 0.0 4% 0.0 5% 0.51 % 13Lot1 72R 172 TDM 16 tsl - tsp 2.5PB 5/4 blue Zhizo 65.0 3% 14.2 4% 3.6 2% 3.3 9% 0.2 7% 0.5 3% 4.2 1% 6.49 % 0.0 3% 2.1 8% 0.0 2% 0.0 0% 0.00 % 13Lot1 81 181 TDM 17 tsl - tsp 2.5PB 5/4 blue CHB 67.3 6% 14.0 8% 2.4 9% 3.6 3% 0.5 4% 0.6 2% 3.8 9% 3.94 % 0.4 6% 1.7 9% 0.5 9% 0.0 9% 0.44 % 13Lot2 01 201 TDM 18 tsl 10B 3/2 blue CHB 65.1 9% 15.1 8% 2.8 6% 3.9 0% 0.6 0% 0.5 4% 5.0 2% 3.66 % 0.5 7% 1.6 3% 0.5 3% 0.0 2% 0.21 % 13Lot2 0 5 205 TDM 19 tsp - tsl 5PB 4/6 blue Zhizo 66.0 2% 14.8 3% 5.0 5% 2.8 5% 0.1 8% 0.4 4% 3.8 8% 4.61 % 0.7 4% 0.7 0% 0.0 6% 0.0 8% 0.51 % 13Lot2 22A 222 A TDM 21 tsp - tsl 5PB 4/6 blue Zhizo 66.1 8% 14.4 2% 3.5 8% 3.1 4% 0.2 4% 0.3 9% 4.6 3% 4.68 % 0.9 3% 1.1 1% 0.0 9% 0.0 8% 0.47 % 13Lot2 22B 222 B TDM 20 tsp - tsl 2.5PB 5/4 blue Zhizo 64.4 4% 14.3 8% 2.8 7% 4.0 3% 0.3 0% 0.7 1% 4.2 5% 6.51 % 0.8 9% 1.3 5% 0.0 5% 0.0 2% 0.10 % 13Lot2 23 223 TDM 22 tsl - tsp 5PB 4/6 blue Zhizo 66.7 3% 14.0 7% 4.1 7% 3.0 0% 0.2 2% 0.4 9% 3.3 8% 4.71 % 1.1 1% 0.9 7% 0.0 8% 0.1 6% 0.84 % 13Lot2 24A 224 A TDM 24 tsp - tsl 2.5PB 5/4 blue Zhizo 68.9 2% 14.6 4% 4.1 6% 2.7 1% 0.1 4% 0.3 4% 3.8 4% 4.44 % 0.1 1% 0.6 3% 0.0 4% 0.0 0% 0.01 % 13Lot2 24B 224 B TDM 23 tsp - tsl 7.5BG 5/6 bluish - green Indet. 32.7 7% 6.67 % 2.2 5% 1.8 4% 0.1 5% 0.2 3% 1.7 9% 2.93 % 0.0 3% 0.5 1% 0.6 3% 0.1 3% 49.9 3% 327 Table 34 13Lot2 27AR 227 A TDM 25/2 6 Indet. Indet. Indet. Indet. 68.6 7% 15.2 0% 4.2 2% 2.8 1% 0.1 7% 0.5 0% 3.3 9% 5.00 % 0.0 1% 0.0 2% 0.0 0% 0.0 0% 0.00 % 13Lot2 27B 227 B TDM 27 tsp - tsl 2.5PB 5/4 blue Zhizo 65.7 6% 17.0 4% 4.3 6% 2.6 7% 0.1 4% 0.3 7% 3.9 0% 4.25 % 0.2 6% 0.7 2% 0.0 6% 0.0 5% 0.36 % 13Lot2 27C 227 C NA Indet. Indet. Indet. Indet. 63.2 7% 16.5 1% 3.4 3% 3.4 5% 0.2 3% 0.4 1% 3.7 3% 5.45 % 1.1 7% 1.4 6% 0.0 7% 0.0 9% 0.62 % 13Lot2 27D 227 D TDM 28 tsp - tsl 2.5PB 5/4 blue Zhizo 65.9 4% 16.7 3% 4.1 1% 2.9 2% 0.1 3% 0.3 6% 3.5 2% 4.24 % 0.4 9% 0.8 1% 0.0 6% 0.0 7% 0.55 % 13Lot2 29 229 TDM 29 tsp - tsl 2.5PB 5/4 blue Zhizo 70.1 6% 12.9 9% 3.2 3% 3.3 3% 0.2 8% 0.5 0% 4.0 9% 3.97 % 0.4 4% 0.9 1% 0.0 5% 0.0 0% 0.02 % 13Lot2 32 232 A TDM 31 tsl - tsp 5BG 5/1 gray - b - g CHB 65.8 5% 16.5 0% 2.3 0% 3.9 0% 0.4 9% 0.6 4% 4.0 9% 3.48 % 0.4 7% 1.8 9% 0.2 4% 0.0 1% 0.06 % 13lot2 32BR 232 B TDM 30 Indet. Indet. blue Indet. 67.6 1% 0.00 % 0.5 6% 8.2 4% 0.1 9% 0.2 6% 1.1 9% 16.1 1% 0.0 5% 5.7 7% 0.0 1% 0.0 0% 0.00 % 13Lot2 59 259 A TDM 33 tsl 2.5PB 5/4 blue CHB 65.1 4% 16.6 5% 2.4 5% 3.7 9% 0.6 0% 0.6 2% 4.7 1% 3.37 % 0.3 7% 1.6 3% 0.3 8% 0.0 3% 0.20 % 13lot2 59BR 259 B TDM 32 Indet. Indet. blue Zhizo 64.9 1% 14.8 8% 3.8 9% 3.0 6% 0.2 5% 0.6 0% 3.9 9% 5.27 % 0.0 2% 3.1 1% 0.0 2% 0.0 0% 0.00 % 13Lot2 61 261 TDM 34 tsl - tsp 5PB 4/6 blue Zhizo 66.3 1% 14.0 3% 3.8 8% 3.2 0% 0.2 4% 0.5 5% 3.9 1% 5.49 % 1.0 7% 0.8 6% 0.0 7% 0.0 5% 0.32 % 13Lot2 81 281 TDM 35 tsl 2.5PB 5/4 blue CHB 66.1 3% 15.8 8% 2.4 4% 3.7 1% 0.6 1% 0.6 0% 4.4 4% 3.67 % 0.4 0% 1.6 9% 0.2 4% 0.0 2% 0.10 % 13Lot2 84 284 TDM 36 tsl - tsp 2.5PB 5/4 blue Zhizo 63.8 4% 15.1 1% 3.6 3% 3.4 2% 0.2 7% 0.7 0% 4.4 2% 6.41 % 0.9 7% 0.9 6% 0.0 4% 0.0 3% 0.15 % 1 3Lot2 95 295 TDM 37 tsl 5BG 5/1 gray - b - g CHB 66.4 0% 16.0 3% 2.3 3% 3.7 2% 0.4 8% 0.6 2% 4.3 3% 3.65 % 0.4 6% 1.5 8% 0.3 0% 0.0 0% 0.02 % 13Lot3 28 328 TDM 38 Indet. Indet. blue CHB 64.9 9% 15.6 2% 2.7 3% 3.7 8% 0.6 5% 0.6 3% 4.2 1% 3.91 % 0.6 0% 1.8 3% 0.7 6% 0.0 3% 0.17 % 13Lot3 30 330 TDM 39 tsl - tsp 2.5PB 5/4 blue Zhizo 67.3 2% 12.5 4% 3.3 6% 3.3 9% 0.2 2% 0.5 3% 4.3 5% 6.08 % 0.7 0% 1.2 8% 0.0 6% 0.0 2% 0.11 % 328 Table 34 13lot3 46R 346 TDM 40 tsl - tsp 2.5PB 5/4 blue Zhizo 65.4 1% 15.2 5% 4.8 0% 2.7 1% 0.1 6% 0.5 1% 4.1 5% 5.80 % 0.0 1% 1.1 8% 0.0 1% 0.0 0% 0.00 % S12Lot 373 373 TDM 41 tsp - tsl 7.5BG 5/6 bluish - green Zhizo 63.0 8% 11.8 9% 3.8 7% 2.4 9% 0.1 9% 1.1 8% 3.5 1% 5.58 % 0.2 3% 0.9 5% 1.7 9% 0.2 4% 4.56 % S12Lot 375A 375 A TDM 42 Indet. Indet. Indet. Indet. 66.3 3% 12.7 8% 3.6 6% 2.4 9% 0.1 8% 0.9 8% 5.0 2% 5.12 % 1.2 2% 0.5 9% 0.0 0% 0.1 1% 1.47 % S12Lot 375B 375 B TDM 43 tsl - tsp 5BG 5/1 gray - b - g CHB 64.5 2% 14.3 5% 2.6 7% 4.3 3% 0.5 7% 1.7 5% 4.6 7% 3.70 % 0.5 0% 1.9 7% 0.6 9% 0.0 2% 0.20 % S12Lot 377 377 TDM 44 tsl - tsp 2.5PB 5/4 blue CHB 66.6 1% 14.2 9% 2.2 4% 3.8 1% 0.4 2% 1.1 6% 5.1 2% 3.49 % 0.4 9% 1.9 1% 0.2 7% 0.0 1% 0.08 % 12Lot3 79R 379 TDM 45 Indet. Indet. blue Zhizo 65.6 5% 13.8 7% 4.3 2% 2.8 7% 0.1 9% 0.4 3% 3.7 7% 5.25 % 0.9 8% 1.5 1% 0.1 4% 0.1 7% 0.77 % 329 Table 35 Mean values (%) of major and minor oxides for S ite 12 and 13 beads (As compared to published values for established bead series ; source for bead series values: Robertshaw et al. 2010 ) Bead Series SiO 2 Na 2 O MgO Al 2 O 3 K 2 O CaO Fe 2 O 3 Zhizo 69.62 13.15 4.31 3.26 3.23 5.5 0.94 K2 64.51 16.22 0.43 11.85 3.34 2.34 1.3 K2 GR 61.05 14.36 0.37 16.63 3.39 2.85 1.35 Indo - Pacific 63.08 14.75 0.59 13 3.46 2.85 2.27 Islamic 63.21 13.71 4.83 6.05 3.91 6.63 1.66 Map ungubwe Oblate 61.88 13.47 5.8 7.67 3.47 6.66 1.04 Zimbabwe 60.98 15.81 4.33 6.71 3.74 6.94 1.48 Khami 61.4 18.66 1.21 9.81 2.82 3.39 2.7 Thabadimasego (Site 13) & Site 12 65.69% 14.67% 3.46% 3.26% 4.21% 4.68% 1.25% 330 Table 36 Elements and oxides included in principal components analysis and their PCA values Variable PC1 PC2 ZrO 2 0.087 - 0.495 Cr - 0.182 - 0.423 HfO 2 0.134 - 0.351 Al 2 O 3 0.182 - 0.325 NbO 2 0.271 - 0.249 Ti 0.185 - 0.239 CaO - 0.207 - 0.222 V 0.269 - 0.144 Y 2 O 3 0.305 - 0.044 ThO 2 0.23 - 0.034 U 3 O 8 0.257 0.024 Rb 2 O 0.279 0.034 MgO - 0.223 0.053 BaO 0.271 0.061 Cs 2 O 0.307 0.091 SrO 0.2 0.142 Li 0.268 0.182 K 2 O 0.217 0.187 Na 2 O 0.14 0.224 331 Table 37 Series determinations for glass beads from Site 12 and Thabadimasego Site # Zhizo Chibuene Undetermined 12 Lot 373 Lot 375B Lot 375A Lot 377 Lot 379 Thabadimasego Lot 159 Lot 181 Lot 23 Lot 162 - 3 Lot 201 Lot 36 Lot 163 Lot 232 Lot 146 Lot166 Lot259A Lot 232B Lot 172 Lot 281 Lot 224B Lot 205 Lot 295 Lot 21 Lot 328 Lot 222A Lot 37C - 1 Lot 222B Lot 37C - 2 Lot 223 Lot 37C - 3 Lot 224A Lot 37F Lot 227A Lot 45 Lot 227B Lot 227C Lot 227D Lot 229 Lot 259B Lot 26 Lot 261 Lot 27 Lot 284 Lot 330 Lot 346 Lot 37A Lot 37B Lot 37D Lot 37G Lot 37H 332 APPENDIX F MATERIAL DISTRIBUTION HISTOGRAMS FOR THE ORIGINAL DATA 333 Figure 56 Achatina total, count, per pit 334 Figure 57 Achatina >50%, count, per pit 335 Figure 58 Finished Achatina beads , count, per pit 336 Figure 59 Irregular Achatina beads , count, per pit 337 Figure 60 Achatina <50% beads , count, per pit 338 Figure 61 Bone , mass , per pit 339 Figure 62 Burnt seed , mass , per pit 34 0 Figure 63 Charcoal , mass , per pit 341 Figure 64 Dhaka , mass , per pit 342 Figure 65 Ferrous material , mass , per pit 343 Figure 66 Ferrous beads, count, per pit 344 Figure 67 Ferrous beads, mass, per pit 345 Figure 68 Ferrous fragments, mass, per pit 346 Figure 69 Ferrous wire, count, per pit 347 Figure 70 Ferrous wire, mass, per pit 348 Figure 71 Glass beads, count, per pit 349 Figure 72 Metal, mass, per pit 350 Figure 73 N on - ferrous metal, mass, per pit 351 Figure 74 Non - ferrous beads, count, per pit 352 Figure 75 Non - ferrous beads, mass, per pit 353 Figure 76 Non - ferrous wire, count, per pit 354 Figure 77 Non - ferrous wire, mass, per pit 355 Figure 78 OES, count, per pit 356 Figure 79 OES <50%, count, per pit 357 Figure 80 OES beads finished, count, per pit 358 Fig ure 81 OES beads irregular, count, per pit 359 Figure 82 OES beads >50%, count, per pit 360 Figure 83 OES fragments, count, per pit 361 Figure 84 Pottery, mass, per pit 362 Figure 85 Decorated body sherds, mass, per pit 363 Figure 86 Decorated rim sherds, mass, per pit 364 Figure 87 Undecorated body sherds, mass, per pit 365 Figure 88 Undecorated rim sherds, mass, per pit 366 Figure 89 Shell items, count, per pit 367 Figure 90 Slag, mass, per pit 368 APPENDIX G MATERIAL DISTRIBUTION FOR STANDARDIZED DATA 369 Figure 91 Achatina per pit, count, standardize d 370 Figure 92 Metal per pit, mass, standardize d 371 Figure 93 Shell per pit, count, standardize d 372 Figure 94 Non - ferrous items per pit, mass, standardize d 373 Figure 95 Ferrous items per pit, mass, standardize d 374 Figure 96 OES per pit, count, standardize d 375 Figure 97 Achatina >50% per pit, count, standardize d 376 Figure 98 Achatina finished per pit, count, standardize d 377 Figure 99 Achatina irregular per pit, count, standardize d 378 Figure 100 Ac hatina <50% per pit, count, standardize d 379 Figure 101 Bone per pit, mass, standardize d 380 Figure 102 Burnt seed per pit, mass, standardize d 381 Figure 103 Charcoal per pit, mass, standardize d 382 Figure 104 Ferrous beads per pit, count, standardize d 383 Figure 105 Ferrous beads per pit, mass, standardize d 384 Figure 106 Ferrous fragments per pit, mass, standardize d 385 Figure 107 Ferrous wire per pit, count, standardize d 386 Figure 108 Ferrous wire per pit, mass, standardize d 387 Figure 109 Glass per pit, count, standardize d 388 Figure 110 Non - ferrous beads per pit, count, standardize d 389 Figure 111 Non - ferrous beads per pit, mass, standardize d 390 Figure 112 Non - ferrous wire per pit, count, standardize d 391 Figure 113 Non - ferrous wire per pit, mass, standardize d 392 Figure 114 OES beads >50% per pit, count, standardize d 393 Figure 115 OES beads finished per pit, count, standardize d 394 Figure 116 OES beads irregular per pit, count, standardize d 395 Figure 117 OES beads <50% per pit, standardize d 396 Figure 118 OES fragments per pit, count, standardize d 397 Figure 119 Pottery per pit, mass, standardize d 398 Figure 120 Decorated body sherds per pit, mass, standardize d 399 Figure 121 Decorated rim sherds per pit, mass, standardize d 400 Figure 122 Undecorated body sherds per pit, mass, standardize d 401 Figure 123 Undecorated rim sherds per pit, mass, standardize d 402 Figure 124 Slag per pit, mass, standardize d 403 BIBLIOGRAPHY 404 BIBLIOGRAPHY Archaeological Record: A Multi - American Antiquity 73 (3): 464 90. - Azania: Archaeological Research in Africa 49 (3): 411 28. The Iron Age Of Southern Africa: A Critique South African Archaeological Bulletin 64 (190): 148 55. South African Archaeological B ulletin 66 (194): 167 72. South African Archaeological Bulletin 67 (196): 264 67. Barker, Graeme. 2006. The Agricultural Revolution in Preh istory: Why Did Foragers Become Farmers? Oxford: Oxford University Press. Behar, Doron M, Richard Villems, Himla Soodyall, Jason Blue - Smith, Luisa Pereira, ene Metspalu, Rosaria American Jo urnal of Human Genetics 82 (2): 1 11. Confronting Scale in Archaeology: Issues of Theory and Practice , edited by G Lock and B Molyneaux, 217 34. New York: Springer. Biesele, Megan. 1993. . Bloomington: Indiana University Press. Hu American Antiquity 43 (3): 330 61. - Gatherer Settlement Systems and Archaeological American Antiquity 45 (1): 4 20. r - Herder Interactions Reflected in the South African Journal of Science 95: 171 80. Internation al Handbook of Historical Archaeology , edited by T Majewski and D Gaimster, 51 65. New York: Springer. 405 Ethnicity, Hunter - Assimilation in Africa , edited by Susan Kent, 206 29. Washington D.C.: Smithsonian Institution Press. Journal of Arid Environments 82: 156 64. Relations in Northern South Africa, Southwestern Zimbabwe, and Eastern Botswana, AD 1000 to African Archaeological Review 17 (4): 183 210. Relations in Northern South Africa, Southwestern Zimb abwe, and Eastern Botswana, AD 1000 to The African Archaeological Review 17 (4): 183 210. . 2007. The Emergence of Social and Political Complexity in the Shashi - Limpopo Valley of Southern Africa, AD 900 to 1300: Ethnicity, Class, and Polity . Ca mbridge Monographs in African Archaeology BAR International Series; 1617 . Oxford: Archaeopress. Campbell, Alec C, and Mike Main. 1991. North - West District Remote Area Dwellers Socio - Economic Survey, Remote Area Development Region South of Lake Ngami . Repo rt Submitted to the Ministry of Local Government and Lands and Norwegian Agency for International Development Co - Operation . Gaborone. Department of Anthropolog y . Albuquerque: University of New Mexico. Man 20 (3): 454 74. Journal of Anthropological Research 43 (2): 121 38. Central District Land Use Planning Unit. 2000. Central District Integrated Land Use Plan . Gaborone, Botswana. - The Unity of African Ancient History 3000 BC to AD 500 , 129 46. Dar Es Salaam: E & D Limited. African Archaeology , edited by Ann Brower Stahl, 276 300. Malden, MA: Blackwell Publishing. The African Archaeology Network: Reports and a Review , edited by Felix A Chami, Gilbert Pwiti, and Chantal Radimilahy, 161 79. Dar es Salaam: Dar es Salaam University Press Ltd. 406 . 200 Journal of Social Archaeology 7 (1): 72 100. Antiquity 82 (318): 976 94. c Residue Evidence for the Processing of Marine Animal Products in Pottery Vessels from the Pre - Colonial South African Journal of Science 100: 279 83. Marian Vanhaeren, and Karen van Niekerk. 2005. Journal of Human Evolution 48 (1): 3 24. tifacts: Functions, Operating Sequences, Ethnoarchaeology in Action , 138 67. Cambridge: Cambridge University Press. Deacon, H J, and Janette Deacon. 1999. Human Beginnings in South Africa: Uncovering the Secrets of the Stone Age . Walnut Cree k: Altamira Press. Deacon, Janette. 1984. The Later Stone Age of Southernmost Africa . Cambridge Monographs in African Archaeology 12 . Oxford: British Archaeological Reports. South African Journal of Science 75: 405 8. - Sett lement in Botswana , edited by R Hitchcock and M Smith. Johannesburg: Heinemann. Frontiers: Southern African Archaeology Today , edited by M Hall, M L Wilson, and A J B Humphreys, 24 39. Oxford: British Archaeological Reports. . 1985. Report on Archaeological Investigations at Sowa Pan . British Petroleum. Journal of African History 27 (1): 3 28. The African Archaeological Review 8: 139 76. tics of Identity in the Kalahari: AD 700 - Beyond Chiefdoms: Pathways to Complexity in Africa , edited by Susan Keech McIntosh, 110 23. Cambridge: Cambridge University Press. Africanizing Knowledge: African Studies across the Disciplines , edited by Toyin Falola 407 and Christian Jennings, 345 74. New Brunswick: Transacti on Publishers. Botswana Notes and Records 43: 76 94. . 2014. The Archaeology and Ethnography of Central Africa . Cambridge: Cambridge University Press. Denbow, James R., Carla Klehm, and Laure Dussubi on Pre - Antiquity . Journal of African Archaeology 5 (2): 271 31 3. Denbow, James R., Jeannette Smith, Nonofho Mathibidi Ndobochani, Kirsten Atwood, and Duncan - Journal of Archaeological Science 35: 459 80. Science 234 (4783): 1509 15. Department of Anthropology . University of Texas at Austin. - ICP - MS Analysis of African Glass Beads: Laboratory Inter - Comparison with an Emphasis on the Impact of Corrosion on International Journal of Mass Spect rometry 284 (1 - 3): 152 61. Journal of Anthropological Archaeology 24 (4): 316 34. Eggert, Manfr African Archaeology , edited by Ann Brower Stahl, 301 26. Malden, MA: Blackwell Publishing. The Archaeologi cal and Linguistic Reconstruction of African History , edited by Christopher Ehret and Merrick Posnansky, 158 81. Berkeley: University of California Press. Elphick, Richard. 1985. Khoikhoi and the Founding of White South Africa . New Southern African Histor y Series . 2nd ed. Johannesburg: Ravan Press. Measurement Science and Technology 14 (9): 1493 1509. . 2014. Luminescence Dating in Kalahari Desert, Botswana . Seattle. Feely, J M, and S M Bell - South African Archaeological Bulletin 66 (194): 105 12. 408 Agriculture Division of Land Utilisation. South African Archaeological Bulletin 68 (197): 63 71. Journal of World Prehistory 20 (1): 1 86. Geograph ical Analysis 40 (3): 297 309. Gifford - - Saharan African Archaeological Review 17 (3): 95 139. Gifford - African Archaeology , edited by Ann Brower Stahl, 187 224. Malden, MA: Blackwell Publishing. ss Modern Methods for Analysing Archaeological and Historical Glass , edited by Koen Janssens, 201 34. West Sussex: John Wiley & Sons, Ltd. Modern Methods for Analysing Arch aeological and Historical Glass , edited by Koen Janssens, 312 43. West Sussex: John Wiley & Sons, Ltd. Gratuze, B, M Blet - Lemarquand, and J. - Journal of Radioanalytical and Nuclear Chemistry 247 (3): 645 56. Ndondondwane, South Africa: The Question of Cultural Continuity between the Early an d Late Iron Journal of Archaeological Science 31: 1511 32. Pa Space and Spatial Analysis in Archaeology , Robertson, 61 68. Calgary: University of Calgary Press. South African Archaeological Soci ety Goodwin Series 5: 83 87. - Gathering and Farming Modes of Hunters and Gatherers: History, Evolution and Social Change , edited by Tim Ingold, D Riches , and I Woodburn, 137 47. Oxford: Berg. South African Archaeological Bulletin 42 (146): 140 52. 409 - Goodwin Series 8: 30 46. Hall, Thomas D, and Christopher Chase - - Systems Perspective and Archaeology: Forward into the Past." Journal of Archaeological Research 1 (2): 121 43. Hall, Thomas D, P Nick Kardulias, and Christopher Chase - - Systems Analysis and Journal of Archaeological Resea rch 19: 233 79. Hammond - South African Archaeological Bulletin 53 (1): 9 15. Hammond - The South African Archaeological Bulletin 54 (170). Hammond - - Tooke 2000 - Ethnicity and Ethnic Group in Iron Age Southern SOUTH AFRICAN JOURNAL OF SCIENCE 96 (8). Biodiversity and Conservation 4 (3): 220 32. Annals of the Association of American Geographers 74: 298 307. Hellenthal, Garrett, George B J Busby, Gavin Band, James F Wilson, Cristian Capelli, Daniel Falush, and Science (New York, N.Y.) 343 (6172): 747 51. sm in Southernmost Africa: New Antiquity 70: 945 49. Science 304 (5669): 404. Herbert, Eugenia W. 1984. Red Gold of Africa: Copper in Precolonial History and Culture . Madison: University of Wisconsin Press. Environment in Southern Hitchcock, Robert K. 1978. - Gatherers, Pastoralists and Agriculturalists in the Western Sandveld Region, Central District, Botswana . Gaborone, Botswana: Ministry of Local Government and Lands. 410 Hodder, Ian. 1982. Symbols in Action: Ethnoarchaeological Studies of Material Culture . New Studies in Archaeology . Cambridge: Cambridge University Pre ss. Hogg, Alan G, Quan Hua, Paul G Blackwell, Mu Niu, Caitlin E Buck, Thomas P Guilderson, Timothy J - Saharan Africa: A Maximum - Proceedings: Biological Sciences 269 (1493): 793 99. ArcGIS Help 10.1 . Accessed February 4. http://resources.arcgis.com/en/help/main/10.1/index.html#//005p00000010000000. Howell, Nancy. 2000. . 2nd ed. New York: Aldine de Gruyter. Wo rld Archaeology 18 (1): 84 95. - African Studies 49 (2): 1 12. Goodwin Series 8: 14 29. Southern African Humanities 13 (1): 19 35. The South African Archaeological Bulletin 59 (180): 66 69. Southern African Humanities 17: 57 79. . 2007. Handbook to the Iron Age: The Archaeology of Pre - Colonial Farming Societies in Southern Africa . Scottsville: University of KwaZulu - Natal Press. . - Journal of Archaeological Science 35: 2032 47. South African Archaeological Bulletin 67 (196): 231 43. - South African Archaeological Bulletin 67 (195): 1 4. South African Archaeological Bulletin 66 (194): 161 66. Kronos 3: 3 13. 411 nd Hunter - South African Archaeological Bulletin 62 (186): 98 103. Continuity: Dental Affinities among American Journal of Physical Anthropology 155 (1). Ironbridge Gorge Museums Trust. 2009. The National Slag Collection: A Simple Catalogue . Association of Southern African Professional Archaeologists Biennial Conference . Gaborone, Botswana: Session: The Archaeology of Farming Communities. s Relevance The South African Archaeological Bulletin 42 (145): 55 58. - Eastern San: Implications for Southern African Rock Art Studies, Ethno graphic Analogy, and Hunter - Gatherer Cultural Current Anthropology 37 (2): 277 305. - Temporal Changes at Two Early Farming Communit y Sites in the Southern Association of Southern African Professional Archaeologists Biennial Conference . Gaborone, Botswana: Session: The Archaeology of Farming Communities. on Sequences of Ostrich Eggshell Beads and Journal of Archaeological Science 32 (12): 1711 21. The Foraging Spectrum: Diversit y in Hunter - Gatherer Lifeways , 111 60. Washington: Smithsonian Institution Press. Journal of African Archaeology 4 (1): 1 2. Journal of World Prehistory 22: 399 414. Kinahan, Jill. 2000. Cattle for Beads: The Archaeology of Historical Contact and Trade on the Namib Coast . Studies in African Archaeology . Uppsala: Department of Archaeology and Ancient History. - Azania: Archaeological Research in Africa . ive Views on the Acquisition of Livestock by Hunter - Gatherers in Southern Africa: The South African Archaeological Bulletin 51 (164): 106 8. 412 . 2001. got . Windhoek: Namibia Archaeological Trust. South African Archaeological Bulletin 68 (197): 1 2. Kiyaga - - Central Bot The Archaeology of Africa: Foods, Metals and Towns. , edited by T Shaw, P J J Sinclair, B W Andah, and A Okpoko, 386 90. London: Routledge. - Th e South African Archaeological Bulletin 29 (113/114): 19 23. Africa 50: 8 23. - Gatherer? Variation in the Archaeological Record of Easter Journal of Archaeological Research 13 (4): 337 66. Azania 29/30: 51 64. Past, Present Ditswa Mmung: The Archaeology of Botswana , edited by Paul Lane, Andrew Reid, and Alinah Segobye, 177 205. Gaborone: Pula Press and The Botswana Society. of Ethnographic African Archaeology , edited by Ann Brower Stahl, 24 54. Malden, MA: Blackwell Publishing. eTopoi: Journal for Ancient Studies 3 (Proceedings of the International Con ference Held in Berlin, 6th 8th June 2012): 353 59. Environmental and Ecological Statistics 17 (2): 231 45. hnography and the Tolerance of History in Africa 20: 185 235. - Gatherers: Survival, History, and the African Study Monographs Suppl.26: 257 80. Lee, Richard B. 1979. . Cambridge: Cambridge University Press. 413 Lewis - Williams, J D. 2002. A Cosmos in Stone: Interpreting Religion and Society through Rock Art . Walnut Creek: Altamira Press. Southern African Prehistory and Palaeoenvironments , edited by R G Klein, 329 60. Rotterdam: Balkema. Maggs, Tim, a - The Journal of African History 32 (1): 3 24. A Botswana Notes and Records 40: 46 59. Journal of World Prehistory 16 (2): 99 143. Masundire, H M, S Ringrose, F T K Sefe, and C Van der Post. 1998. Inventory of Wetlands in Botswana . Gaborone: University of Botswana (compiled for the National Conservation Strategy (Co - ordinating) Agency Ministry of Local Government, Lands and Housing). Pula: Botswana Journal of African Studies 15 (1): 75 90. - South African Archaeol ogical Bulletin Goodwin Series , no. 6: 33 41. Annual Review of Anthropology 31: 279 301. Goodwin Series 8, African: 4 13. Miller South African Archaeological Bulletin 56 (173): 83 103. Journal of Archaeological Science 29 (10): 1083 1131. Journ al of African Archaeology 2 (1): 23 47. Journal of Archaeological Science 21: 101 15. Mitchell, Peter. 2002. The Archaeology of S outhern Africa . The Archaeology of Southern Africa . Cambridge: Cambridge University Press. - Before Farming 1 (3): 1 18. 414 African Archaeology , edited by Ann Brower Stahl, 150 73. Malden, MA: Blackwell Publishing. Mountains: Late First Millenni um AD Hunter - Before Farming 2: 1 22. Journal of Afric an History 46: 209 41. Pula: Botswana Journal of African Studies 15 (1): 91 107. Modern Methods for Analysing Archaeological and Historical Glass , edited by Koen Janssens. Morris, A G. 1992. The Skeletons of Contact: A Study of Protohistoric Burials from the Lower Orange River Valley, South Africa . Johannesburg: Witwatersrand University Press. Mitochondrial Human Genome from a Pre - Genome Biology and Evolution 6 (10): 2647 53. South African Archaeological Bulletin 68 (197): 52 62. Processing of Wild Cereal Grains at Ohalo II, a 23 000 - Year Old Campsite on the Shore of the Sea of Antiquity 86: 990 1003. African Archaeology , edited by Ann Brower Stahl, 249 75. Malden, MA: Blackwell Publishing. American Antiquity 66 (4): 679 85. Residues from Ceramics Found in the Southwestern Cape, South Afri Archaeometry 27 (2): 231 36. Pearsall, Deborah M. 2000. Paleoethnobotany : A Handbook of Procedures . San Diego: Academic Press. of Its Formation i Pula: Botswana Journal of African Studies 15 (1): 125 43. - Hunter - Gatherers: A Lithic Department of Anthropo logy . East Lansing: 415 Michigan State University. - African Archaeology , 249 73. Cambridge: Cambridge University Press. Pickrell, Joseph K, Nick Patterson, Po - Ru Loh, Mark Lipson, Bonnie Berger, Mark Stoneking, Brigitte Proceedings of the National Academy of Science . Pikirayi Journal of Social Archaeology 7 (3): 286 301. Wild Cereal Grains in Nature 430: 670 73. Pleurdeau, David, Emma Imalwa, Florent Détroit, Joséphine Lesur, Anzel Veldman, Jean - Jacques Bahain, t Direct Evidence of Caprine Domestication PloS One 7 (7): e 40340. Goodwin Series 8: 117 26. Plug, I, and S Badenhorst. 2001. The Distribution of Macromammals in Southern Africa Over the Past 30000 Years as a Reflected in Animal Remains from Archaological Sites . Transvaal Museum Monograph . Vol. 12. Pretoria: Transvaal Museum. Studies of Iron Age Communities in Southern Advances in World Archaeology 4 , edited by F Wendorf and A Close, 189 238. London: Academic Press. - IC P - Laser Ablation - ICP - MS in Archaeological Research , edited by R J Speakman and H Neff, 84 93. Albuquerque: University of New Mexico Press. Zambezia 18 (2): 119 29. Zambezia 18 (2): 119 29. - Zambezi Valley, No The South African Archaeological Bulletin 51 (163): 3 6. African Archaeology , edited by Ann Brower Stahl, 378 91. Malden, MA: Blackwell Publishing. African Archaeology , edited by Ann Brower Stahl, 353 77. Ma lden, MA: Blackwell Publishing. 416 Reid, Andrew, Karim Sadr, and Nick Hanson - Ditswa Mmung: The Archaeology of Botswana , edited by Paul Lane, Andrew Reid, and Alinah Segobye, 81 100. Gaborone: Pula Press and the Botswana Society. Goodwin Se ries 8, African: 58 68. Antiquity 74 (284): 326 31. Research, Department of Environmental Affairs and Centre for Applied. 2010. Makgadikgadi Framework Management Plan . Gaborone: Government of Botswana. Confronting Scale in Archaeology: Issues of Theory and Practice , edited by G Lock and B Molyneaux, 145 61. New York: Springer. Riede l, Frank, Sebastian Erhardt, Chrispen Chauke, Annette Kossler, Elisha Shemang, and Pavel Tarasov. the Last Millennium Indicated by Distribution of Baobab Tre Quaternary International 253 (March). Elsevier Ltd: 67 73. Ringrose, S, C Harris, P Huntsman - Mapila, B W Vink, S Diskins, C Vanderpost, and W Matheson. 2009. gadi Pans (Botswana Kalahari) as Deduced Sedimentary Geology 219: 262 79. Ringrose, Susan, Philippa Huntsman - Mapila, Ali Basira Kampunzu, William Downey, Stephan Coetzee, Bernard Vink, Wilma Matheson, and Corn Geochemical Evidence for Palaeo - Environmental Change in the Makgadikgadi Subbasin, in Relation Palaeogeography, Palaeoclimatology, Palaeoecology 217 (3 - 4): 265 87. Robbin s, Lawrence H, Alec C Campbell, Michael L Murphy, George A Brook, Pradeep Srivastava, and Current Anthropology 46 (4): 671 77 . Robbins, Lawrence H, Alec C Campbell, Michael L Murphy, A George, Abel A Mabuse, Robert K Hitchcock, and Grace Babutsi. Traditions at Lake Ngami , Botswana South African Archaeological Bullet in 64 (189): 13 - 32. of Specular Hematite in the Kalahari Ca. A.D. 800 - Current Anthropology 39 (1): 144 50. Robertshaw, Peter, Marilee Wood, Erik Me lchiorre, Rachel S Popelka - Filcoff, and Michael D Glascock. Journal of Archaeological Science 37: 1898 1912. - Referenced R adiocarbon Database for Early Iron Age Sites in Sub - Southern African Humanities 21: 327 44. 417 Current Anthropology 38 (1). The University of Chicago Press: 104 12. African Archaeological Review 15 (2): 101 32. Ethnicity, Hunter - Gatherers, and , edited by Susan Kent, 28 47. Washington D.C.: Smithsonian Institution Press. Journal of African History 44: 195 209. - Gatherers and Herd Desert Peoples: Archaeological Perspectives , edited by Peter Veth, Mike Smith, and Peter Hiscock, 206 21. Malden: Blackwell Publishing. Sadr, Karim, and Francois - Xavier Fauvelle - nding Hollows on the West Coast of Southern African Humanities 18 (2): 29 50. The South African Archaeological Bulletin 56 (173/174): 76 82. - Walled Structures in Southern Journal of Archaeological Science 39: 1034 42. Journal of African Archaeology 4 (2): 235 52. Sa dr, Karim, Andrew Smith, Ina and Foragers on Kasteelberg: Interim Report of Excavations 1999 - Th e South African Archaeological Bulletin 58 (177): 27 32. Makgadikgadi from AD 800 - Schapera, Isaac. 1930. The Khoisan Peopl es of South Africa . London: Routledge. Schiffer, M B. 1995. Behavioral Archaeology: First Principles . Salt Lake City: University of Utah Press. American Antiquity 66 (4): 729 37. ers in the Modern World: Conflict, Resistance, and Self - Ditswa Mmung: The Archaeology of Botswana , edited by Paul Lane, Andrew Reid, and Alinah Segobye , 101 14. Gaborone: Pula Press and the 418 Botswana Society. The South African Archaeological Bulletin 60 (182): 79 83. Setshogo, Moffat P., and Fanie Venter. 2003. Trees of Botswana: Names and Distribution . Pretoria. Archaeology, 1870 - Journal of Southern African Studies 29 (4): 823 44. Silitshena, R M K, and G McLeod. 1998. Botswana: A Physical, Social and Economic Geography . 2nd ed. Gaborone: Longman Botswana. de and Society on the South - East African Antiquity 86: 723 37. Combining the Past and the Present: Ar chaeological Perspectives on Society. Proceedings from the Conference , edited by Terje Oestigaard, Nils Anfinset, a nd Tore Saetersdal, 171 79. Oxford: Archaeopress. African Study Monographs Suppl. 26: 15 25. storical Evidence for Recent Hunter - South African Archaeological Bulletin 52: 52 58. - WESTERN C APE , SOUTH AFRICA , AND THE ARCHAEOLOGICAL IDENTITY OF PRELUSTORIC HUNTER - 91. African Studies 49 (2): 51 - 73. Journal of Anthropological Archaeology 17: 201 15. African Herders: Emergence of Pastoral Traditions , 151 92. Walnut Creek: Altamira Press. History Compass 4 (1): 1 7. for Recent Hunter - South African Archaeological Bulletin 52 (165): 52 58. 419 - Western Cape , South Africa, and the Archaeological Identity of Prehistoric Hunter - Gatherers within the Last The South African Archaeological Bulletin 46 (154): 71 91. Cultural Identity and Site Integrity on Open Sites: Evidence from Bloeddrift 23, a Pre - Colonial Herder Camp in the The South African Archaeological Bulletin 56 (173/174): 23 33. Smith, Jeannette, Julia Lee - Settlement in the Shashe - South African Archaeological Bulletin 62 (186): 115 25. Solway, Jacqueline S., and Richard B. Lee. 1990. Current Anthropology . ArcGIS Help 10.1 . Accessed February 4. http://resources.arcgis.com/en/help/main/10.1/index.html#//005p0000000n000000. Spielm Annual Review of Anthropology 23: 303 23. Blackwell Studies in Global Archaeology . Malden: Blackwell. Studies in Culture Contact: Interaction, Culture Change, and Archaeology , edited by James G. Cusick. Carbondale: So uthern Illinois University Press. South African Archaeological Bulletin 59 (180): 45 51. Journal of Archaeological Science 36: 798 806. American Anthropologist 116 (2): 251 64. Tanaka, Jiro. 1980. The San, Hunter - Gatherers of the Kalahari . Tokyo: University of Tokyo Press. Pula: Botswana Journal of African Studies 15 (1): 60 74. - A rchaeology and Ethnoarchaeology of Mobility , edited by Frederic Sellet, Russell Greaves, and Pei - Lin Yu, 240 61. Gainesville: University Press of Florida. - Gatherer Archaeology in Arid 420 Southe Desert Peoples: Archaeological Perspectives , edited by Peter Veth, Mike Smith, and Peter Hiscock, 161 76. Malden: Blackwell Publishing. Development C Botswana Notes and Records 43: 151 65. Thomas, D , and P A Shaw. 1991. The Kalahari Environment . Cambridge: Cambridge University Press. ata, Synthesis, Quaternary Science Reviews 21 (7): 783 97. Thorp, Carolyn R. 2000. Hunter - Gatherers and Farmers: An Enduring Frontier in the Caledon Valley, South Africa . Cambridge Monographs in African Archaeology . Vol. 50. Oxford: Archaeopress. Tobias, Phillip V., and Megan Biesele. 1978. The Bushmen: San Hunters and Herders of Southern Africa . Edited by Phillip Tobias. Cape Town: Human & Rousseau. Cupules from South Africa Archaeological Bulletin 59 (179). - Limpopo F aculty of Science . Johannesburg: University of the Witwatersrand. van Waarden, Catrien. n.d. The Origin of Zimbabwe Tradition Walling . Citizens; an Department of Anthropology. Binghamton: State University of New York. Soufh African Archaeological Bulletin 68 (198): 173 87. van Zyl, Wynand J, Shaw Badenhorst, Elene Taljaard, James R. Denbow, Edwin N. Wilmsen, and Wynand Annals of the Ditsong Nat ional Museum of Natural History 3: 49 58. The Journal of African History1 25 (2): 129 45. Journal of African History 36 (2): 173 95. Vogel, J, RADIOCARBON 43 (1): 133 37. South African Archaeolo gical Bulletin 47 (155): 8 12. Gatherers after Agropastoralist 421 Journal of Anthropological Archaeology 15: 205 17. Walker, N. 1 Botswana Notes and Records 26: 1 36. Speaking for the Bushmen , edited by A G M Sanders, 54 87. Gaborone: Botswana S ociety. Ditswa Mmung: The Archaeology of Botswana , edited by Paul Lane, Andrew Reid, and Alinah Segobye, 65 80. Gaborone: Pula Press and the Botswana Society. Watson, E, K Bauer, R Aman, G Weiss, A vonHaeseler, and S P AMERICAN JOURNAL OF HUMAN GENETICS 59 (2): 437 - 444. - Foragers and Agro - Pastoralists: Lithic Use in Botswana from the Austin: University of Texas. American Antiquity 48 (2): 253 76. Spec ular Haematite in Botswana, Ca. 200 - World of Iron , 35 45. History in Africa 30. African Studies Association: 327 420. Journal of Southern African Studies 20 (3, Special Issue: Ethnicity and Identity in Southern Africa): 347 53. Wilmsen, Edwin N. 200 Journal of Southern African Studies 28 (4): 825 41. Southern African Humaniti es 21: 263 74. - Speaking Peoples and Current Anthropology 31 (5). The University of Chicago Press: 489 524. Wilmsen, Edwin N., David Killick, Dana D . Rosenstein, Phenyo C. Thebe, and James R. Denbow. 2009. Journal of African Archaeology 7 (1): 3 39. - European Trade in the S hashe - Faculty of Humanities . Johannesburg: University of the Witwatersrand. 422 Journal of African Archaeology 9 (1): 67 84. Wood, Marilee, Laure Du South African Archaeological Bulletin 67 (195): 59 74. Woodborne, Stephan, Marc Pienaar, and Sian Tiley - Journal of African Archaeology 7 (1): 99 105.