WM" '1"? I I. In ._ .vll H‘ '5“ 9:! “5. a! gjr‘vl’." 0:! .i'aqm‘ :.l “le1‘ ~ :‘Zfifwé I430 . I‘I' . ‘ . b.’ ' I LV .1 m «‘4 I . . I‘: .‘ . , VG turd. I...» h ' 9 .' ‘1 II ' y.,,,_.:x... 3*, gm yum-a \N“ 1‘ .1‘ .. A Is“.- I ' -' I: - . . '2..a~'.~'I, .__..‘I‘.. ‘I“I':)1‘§:-‘0w:w; I; III [ca I .:‘ ;:‘- £I:§'H‘tq 52"“_ i” 1‘ s... _ .... ‘. 2:32;...‘2 . fihfL‘v‘uri‘fi: b3 ”.1“ :4 ' “WWI? 'W 3. "'..‘<:¢‘-?o!€',l".t7‘.’$$ I'I- \ . ':'.. la‘l,..,“'lw‘ (I ".- . a. b ‘:?""‘l“‘":" ‘W' 1'" .0»; 2-358» ‘ vat-2 my mailing V] 35:31 . has}... .. ‘. ‘ n ' "I ' ‘ - ’5 ‘V‘ '- .4 v .I‘ . I‘ I] V‘ .’ ' $- 5 JETS] )- “k. . ' In: .22" ‘1."3'” MIN “I ~ .. ‘1 {:ij I.'"”.'.§'I}\‘_.‘.a ”(I I .5 I . ' ~A- '; ‘ \:‘l-‘| .. ’1, -. '.L"( .- .- I. u‘ ‘. ‘ .v%‘ I,“ “"p . I} II, . A!" y! n - I I ‘ , . I l "' ‘ . , ”J .— .‘ II V '. I'.‘ I v M I “a . u. :1}... ' A‘I \z" I \ '~‘ L- , .m.. I .2. . ‘ §£q7-¢P':%W’”bfi '33“ .. .k ‘ N 3‘33. w Min.“ .'.o 'JI\'-‘J.'/n H. . ~ _I snub". 5 LIX: M. «W E“. :‘chfiq‘. 330.1‘ 115‘ %’2 I 2:4} kw! J . 34'.) . 1: I H ~ :IFVI‘ \_ I ”FIN" " I. “$24., ‘ U", JIGK may x- - \ gm.“ "9133 Q_"- 'i'vu' ' 3:51;: 3‘. -:=:c .\ V3453": ‘ .;".I‘:II‘J.A.IIII\V2 '. (ll 3 :4 9.1: - .\'~——'. .: .- t 4 . a I I. ' I . ‘ :5“, " ‘|'I I ‘. , VI' ‘ l." V”. ”H 72 I: .1 I l I ‘ . 3mm"! I ‘ " ‘ ‘ rt: KI. Jlu; .o‘flfiu 5': I; : " 6; II I." I}, 'I‘ i; In . I {”1“ t ;: H I; ? IN.“ . I I .1! ‘ I I 4 I a.“ 'I V. H" “I“ :.l|, ’: “"1: "j "‘ :' h I ' ' HI: '1 MI, ”ML”: ‘L J‘l‘vifo '. ‘ “ J. . , I’Ijul "H'? I 1]“ , ‘ .' l ' HII- H .H LIBRARY Michigan State University ”flow-val \|v~ THE NAMORATUNGA CEMETERY AND ROCK ART SITES OF NW KENYA: A STUDY OF EARLY PASTORALIST SOCIAL ORGANIZATION By B. Mark Lynch A DISSERTATION Submitted to Michigan State University in partiaT fquiTTment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Anthropoiogy 1978 Gard/g) - ABSTRACT THE NAMORATUNGA CEMETERY AND ROCK ART SITES OF NW KENYA: A STUDY OF EARLY PASTORALIST SOCIAL ORGANIZATION By B. Mark Lynch This study analyzes the three Namoratunga Cemetery and rock art sites west of Lake Turkana in Nw Kenya. The three sites were utilized by pastoralists during the first millennium B.C., a time when early pastoralists were first entering East Africa. The goals of this study are: (l) to reconstruct the social organization of these early herders through the use of burial information, (2) to examine the relevance of assumptions made by archaeologists when analyzing mortuary data and finally, (3) to place these sites into the broader framework of East African prehistory. While the Namoratunga sites display a wide geographical distribution west of Lake Turkana, the analysis concentrated on the two largest sites which were located near the Kerio River. These two sites are located on eroded volcanic outcrops separated by less than one Km. and contained 162 and ll graves respectively. In addition both sites contained over l,000 examples of geometric rock art which occurred in the form of engravings. B. Mark Lynch From the analysis of the burials and rock art at these two sites it appears that the social organization of these early pastoralists was largely egalitarian. In addition individual kin groups had their own areas within each of these cemeteries to bury their dead. The two sites in turn were occupied by different social units, perhaps clans. The Namoratunga burial program was then compared with a series of ethnographic models drawn from Nilotic and Cushitic speaking peoples to test for best fit. It was found that the Namoratunga burial routine most closely resembles present day Eastern Cushitic burial practices. This apparent connection between the sites and Cushites was supported by the third Namoratunga site located llO miles further to the north. Here rows of stone columns were found to align with a series of stars and constellations used by present day Eastern Cushites to compute a sophisticated calendar system. The Namoratunga sites thus provide one of the first clear links between Cushitic speaking peoples and early pastoralists in East Africa. Finally, certain assumptions made by archaeologists when analyzing mortuary data were examined in light of the special problems posed by pastoralists in terms of their disposal of the dead. It was found that due to a number of circumstances any thorough analysis of pastoralists‘burial practices would be almost impossible. ACKNOWLEDGMENTS A special debt of gratitude is owed to my Doctoral Committee: Lawrence H. Robbins, Chairman, Charles Cleland and William Lovis. I would also like to thank Joseph Chartkoff and Moreau Maxwell who helped with this manuscript in many informal discussions. I am especially grateful to Dr. Robbins. Dwaddition to making this research opportunity possible, he was also a constant source of both ideas and encouragement. Aid on much of the statistics and computer programing was provided by William Lovis, Randy Donahue and Anne Nieberding. The help of these individuals was instrumental in the completion of the analysis. Without the help of Robert Victor of the Department of Astrophysics it would not have been possible to determine the astronomical significance of Namoratunga 3. Mrs. Margaret Robbins helped greatly with ethnographic data on the Pokot and Turkana. The excellent drawings in this manuscript are the work of Margaret Uland. Additionally the encouragement of my friends and colleagues Peter Andrews and Mark and Dorothy Van Benschoten must be acknowl- edged. This field work was generously supported by the National Science Foundation. I an also grateful to the Government of Kenya for permission to do this research. In addition I would like to ii thank R. E. F. Leakey and A. J. C. Onyango-Abuje of the National museums of Kenya and N. Chittick and D. W. Phillipson of the British Institute in Eastern Africa for facilitating this research. R. Soper then of the British Institute was most helpful in giving me access to the results of his initial work at the Namoratunga sites. Finally I would like to thank my crew of J. Salenyi, S. Tekal, H. Kiyionga, A. Loumua, E. Lokuasito, E. Lokoonua, P. Nyangaita, K. Engatelepit, S. Kaleba, L. Lotela, E. Nyagaluk, and Engatuny. This manuscript would not have been possible without their hard work under very difficult field conditions. Any errors of analysis or interpretation that might occur are the sole responsibility of the author. TABLE OF CONTENTS ACKNOWLEDGMENTS LIST OF TABLES LIST OF FIGURES LIST OF APPENDICES Chapter I. INTRODUCTION . Brief History of Pastoralism in East Africa . II. SETTING AND ARCHAEOLOGY Location and Discovery . The Setting Excavation Strategy. . Artifacts in Grave Fill Skeletal Remains . Site Survey III. MORTUARY ANALYSIS Cluster Analysis . Burial Orientation Flexure . . Decorated Graves . . Spatial Analysis (Placement of the Graves) Petroglyphs as Cattlebrands: The Ethnographic . Evidence . . Summary and Interpretation IV. ART ANALYSIS . . . . . . . . . . . Step l: Independent Binomial Probability Distribution . Step II. Pearson Product Moment Correlation iv Page ii vi viii xiii Chapter Page Step III: Paired Differences Technique . . . . . 149 Step IV: Comparison of Means . . . . . . . . 152 Intra Site Variability . . . . . . . . . . . 154 Summary and Conclusions . . . . . . . . . . 160 V. OTHER NAMORATUNGA SITES . . . . . . . . . . . 166 Astronomical Alignments . . . . . . . . . . 173 VI. ETHNOGRAPHIC ANALYSIS . . . . . . . . . . . . 181 Namoratunga . . . . . . . . . . . . . . 183 Turkana . . . . . . . . . . . . . . . . 187 Pokot . . . . . . . . . . . . . . . . 188 Nuer . . . . . . . . . . . . . . . . . 190 Masai . . . . . . . . . . . . . . . . 193 (onso . . . 197 East African Pastoralists and Mortuary Archaeology . 205 VII. CONCLUSION: NAMORATUNGA IN HISTORICAL PERSPECTIVE . . 221 APPENDICES . . . . . . . . . . . . . . . . . . 227 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . 364 Table w .b LOUDVOSU'I 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. LIST OF TABLES Initial 15% Random Sample at Site I Second 10% Sample Breakdown of Orientation, Decoration, Sex, Age and Grave Goods by Sample . . . . Five Burial Groups Derived from ISODATA Routine Breakdown of Clusters by Weight of Stone in Grave . Breakdown of Burial Clusters by Age Breakdown of Burial Clusters by Orientation . Orientation vs. Age and Sex Flexure vs. Sex Flexure vs. Orientation Flexure vs. Age Flexure vs. Burial Cluster Decorated Graves (Excluding Those Excavated) Breakdown of Decorated Graves by Cluster . Excavated Decorated Graves Grave Diameter vs. Number of Petroglyphs . Designs Used by Turkana, Pokot, Masai, and Samburu Grave Diameter vs. Average Number of Different Designs Grave Size vs. Symbol for Excavated Graves Key to Design Motifs vi Page 41 42 43 57 57 6O 63 63 66 67 69 71 75 79 80 82 113 118 119 134 Table Page 21. Distribution of Designs by Weathering Category . . . . 137 22. Frequency of Design by Weathering Category . . . . . 151 23. Ratio of Number of Different Designs to Total Number of Designs . . . . . . . . . . . . . . . 153 24. Time Trends of Most Common Designs . . . . . . . . 155 25. Distribution of Art at Site 1 . . . . . . . . . 157 26. Distribution of Designs by Rock Face . . . . . . . 161 27. Rock Art Designs Found at Kalokol Namoratunga . . . . 172 28. Azimuths of Stars in 300 BC Compared with Present Day Azimuths . . . . . . . . . . . . . . 176 29. Summary of Mortuary Distinctions Symbolized . . . . . 203 30, Key to Appendix A . . . . . . . . . . . . . 229 31. Cranial Measurements from Namoratunga Skeletons . . . 312 32. Soil Ph Values . . . . . . . . . . . . . . 314 33. Soil Samples Collected from East-West Transect from Namoratunga l to the Kerio River . . . . . . . . 315 34. Distribution of Designs and Weathering Categories at Site 1 . . . . . . . . . . . . . . . 321 35. Distribution of Designs and Weathering Categories at Site 2 . . . . . . . . . . . . . . . 324 vii LIST OF FIGURES Figure 43-0) 12. 13. 14. 15. 16. 17. 18. 19. 20. OOCDNO‘U‘I The Highlands and Rift Valley of Kenya Modern Locations of Cushites; Plains, River-Lake, and Highland Nilotes . Location of Namoratunga Sites View of Volcanic Outcrops Containing Namoratunga Sites View of Southernmost Lokori Site Graves Initially Excavated by R. Soper Grave 41 at Namoratunga 1 . Initial 15% Excavation Sample Namoratunga 2 Excavated Graves Grave Construction Grave 82-Bund1e Reburial Second 10% Excavation Sample . North-South Transect at Namoratunga 1 . Survey Area and Site Locations Distribution of Decorated Graves Contour Distribution of Decorated Graves at Site 1 Distribution of 0 Design Distribution of U Design Distribution of 0 Design Distribution aid” Design viii Page 10 14 18 18 23 24 25 26 28 35 39 44 46 83 85 86 87 88 89 Figure Page 21. Distribution of 9 Design . . . . . . . . . . 9o 22. Distribution of 9 Design . . . . . . . . . . 91 23. Distribution ofl" Design . . . . . . . . . . 92 24. Distribution on Design . . . . . . . . . . 93 25. Distribution of ‘63 Design . . . . . . . . . . 94 26. Distribution of on) Design . . . . . . . . . . 95 27. Distribution of II Design . . . . . . . . . . 96 28. Distribution of © Design . . . . . . . . . . 97 29. Distribution of @ Design . . . . . . . . . '. 98 30. Distribution of Graves over 2.0 Meters in Diameter . . 102 31. Contour Distribution of Graves at Site 1 . . . . . . 103 32. Distribution of Graves 2.0 Meters or Less in Diameter . 104 33. Example of Light Weathering . . . . . . . . . . 131 34. Example of Moderate Weathering . . . . . . . . . 131 35. Example of Heavy Weathering . . . . . . . . . . 132 36. Example of More Than one Type of Weathering on Same Rock Surface . . . . . . . . . . . . . . 132 37. Distribution of Designs for Site 1 . . . . . . . . 143 38. Distribution of Designs for Site 2 . . . . . . . . 144 39. Comparison of Frequency of Designs at Sites 1 and 2 . . 148 40. Distributions of Paired Differences: (H],i-H2,i) . . . 150 41. Distribution of Art at Site 1 . . . . . . . . . 156 42. Distribution of Art at Site 2 . . . . . . . . . 159 43. Location of Namoratunga 3 . . . . . . . . . . 167 44. Plan of Site 3 . . . . . . . . . . . . . . 169 ix Figure Page 45. Stone Pillars at Kalokol Namoratunga 3 . . . . . . 171 46. Alignments of Stones at Namoratunga 3 . . . . . . . 175 47. Mortality Profile of Namoratunga Sites Compared with Four Prehistoric American Indian Populations . . . . 185 48. Summary of Namoratunga Mortuary Archaeology . . . . . 186 49. Turkana Burial Schedule . . . . . . . . . . . 189 50. Pokot Burial Schedule . . . . . . . . . . . . 191 51. Nuer Burial Schedule . . . . . . . . . . . . 194 52. Masai Burial Schedule . . . . . . . . . . . . 198 53. Konso Burial Schedule . . . . . . . . . . . . 201 54. Diagram of a Perfect Paradigm From Saxe 1970 . . . . 213 55. Diagram of a Perfect Tree From Saxe 1970 . . . . . . 214 56. Burial 7 and 9 . . . . . . . . . . . . . . 231 57. Grave 9 . . . . . . . . . . . . . . . . . 233 58. Burial 16 and 17 . . . . . . . . . . . . . . 235 59. Grave 17 . . . . . . . . . . . . . . . . 237 60. Burial 30 and 31 . . . . . . . . . . . . . . 241 61. Grave 32 . . . . . . . . . . . . . . . . 243 62. Burial 32 and 37 . . . . . . . . . . . . . . 246 63. Burial 41 and 45 . . . . . . . . . . . . . . 250 64. Grave 45 . . . . . . . . . . . . . . . . 251 65. Burial 46 and 53 . . . . . . . . . . . . . . 253 66. Burial 54 and 56 . . . . . . . . . . . . . . 257 67. Grave 60 . . . . . . . . . . . . . . . . 260 68. Burial 65 and 76 . . . . . . . . . . . . . . 263 Figure 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. Grave 77 . Burial 77 and 78 Burial 79 and 83 Grave 89 . Burial 89 and 97 Burial 98 and 104 . Burial 105 and 122 Burial 138 and 143 Burial 159 and 160 Grave 160 Buria1 161 and 162 Burial 168 and 170 Grave 1A . Turkana Burial 1 and 2 Turkana Burial 3 and 4 Artifacts Artifacts Artifacts Artifacts Artifacts Artifacts Artifacts in Grave Fill . in Grave Fill . in Grave Fill . in Grave Fill . in Grave Fill . in Grave Fill . Found in Burial Pits Apo Hill Site Lithics Apo Hill Site Lithics Kangatet Site Lithics xi Page 266 267 271 275 277 280 283 289 292 294 296 300 305 306 308 330 332 334 336 338 340 342 345 347 349 Figure Page 94. Kangatet Site Lithics . . . . . . . . . . . . 351 95. Kangatet Site Lithics . . . . . . . . . . . . 353 96. Kangatet Site Lithics . . . . . . . . . . . . 355 97. Kangatet Site Lithics . . . . . . . . . . . . 357 98. Morulim Hill Site Lithics . . . . . . . . . . . 359 99. Namoratunga 1 Surface Artifacts . . . . . . . . . 361 100. Namoratunga 2 Surface Artifacts . . . . . . . . . 363 xii Appendix A. Burial Descriptions . B. Turkana Graves Excavated Near the Namoratunga Sites C. Skeletal Measurements D. Soil Samples E. Site Survey F. Distribution of Designs and Weathering Categories at Site 1 and 2 . . . . . . . . . G. Artifacts Associated with Graves at Namoratunga 1 and 2 . H. Artifacts Collected on Site Survey . LIST OF APPENDICES xiii Page 228 302 311 313 316 320 328 343 Who knows how the standing stones were forged? Who knows how to make the fire of Namoratunga? The Hell fire will return worse than ever The fires of Hell will come more terrible yet Who can count the people in heaven? Who can count the stars in heaven? The Hell fire will come worse than ever The Hell fire will be worse than the last time. There is no one who can obstruct fate - no one. There is no one not even a diviner. The Hell fire will return worse than ever The Hell fires to come will be more horrible than the first.* * Turkana song about Namoratunga sites. xiv CHAPTER I INTRODUCTION The origin and development of pastoralism, particularly in East Africa has a long and complicated history. Unfortunately due to the absence of written records or reliable oral traditions it has been difficult if not impossible to link archaeological assemblage bearing on this problem with known peoples. Most of the available archaeological evidence is very general. For example, it has been demonstrated that pastoralists were established in parts of Kenya and Tanzania by the first millennium B.C. At Prospect Farm (960 :_110 and 740 :_80 B.C.), Narosura (310 1 110 B.C.), Tunnel Rock Shelter (100 :_60 and 780 :_60 B.C.) and Muringa (undated) remains of domestic cattle have been found. At Narosura sheep/goat were also present. On the basis of overall technological similarities and present day distributions of language groups in Southern Ethiopia, Kenya and northern Tanzania some investigators have associated these sites with known pastoralists, especially with ancestral Cushitic and in some cases, Nilotic speaking peoples. However, these linkages are based almost entirely upon indirect evidence. This approach is valid, of course, particularly since it has been impossible thus far to make direct links between the archaeological data and linguistic and ethnographic information. But, specific fits between known peoples and the archaeological record would lend considerable strength to these arguments. The Lake Turkana region in NW Kenya figures prominently in the problem of pastoral development in East Africa. It is an area which played a major role in the southward spread out of the Sahara and into East Africa of these early pastoralists since it offered a natural corridor into the Rift Valley and as such is ideally suited to the study of pastoralism. This study attempts to apply burial data from three related cemetery and rock art sites west of Lake Turkana to this historical problem of pastoral movements in East Africa. In particular the study attempts to directly link an excavated archaeological assemblage with the prehistory of a known people. The three mortuary sites display a wide geographical distribution west of Lake Turkana as well as providing a large sample of burials which fall within the time period when pas- toralists were first spreading into East Africa, around the first and second millennium B.C. The second goal of this work is to apply burial data to a study of the social organization of these early pastoralists; a subject of which almost nothing is known. Although burial evidence has been used in East Africa, it has been utilized primarily to show population affinities and racial types (Fagan 1964; Leakey 1931; Galloway 1959; Rightmire 1970). Thus far no one has attempted to apply burial data to the study of social organization (Lynch N.D.). However, burial data can yield infor- mation concerning prehistoric social systems which is not readily extracted from other aspects of the archaeological record (Binford 1971; Tainter 1975). Using the archaeological data from the three cemetery sites a model will be formulated of past social organization. This model will then be compared with present day East African pastoral societies. Particular emphasis will be placed upon the Cushitic and Nilotic speaking groups. These two linguistic groups are thought to figure prominently in the spread of pastoralism in East Africa. Formal analysis as employed by both Brown (1971) and Saxe (1970) will be used to implement the comparisons between the archaeological and ethnographic models. The models can then be compared in terms of relative complexity and variables which are significant in partitioning each society. In such a way it should be possible to assign the archaeological data to a known group. This will provide one of the firstinstances where an excavated archaeological assemblage will be linked to a particular people in East Africa. As such, this work will yield the first firm historical framework for the study of pastoral social organization and population movements in East Africa. Brief History of Pastoralism in East Africa Although pastoralism is now practiced in much of East Africa its origins are almost completely unknown beyond the admittedly limited time depth provided by oral history. Linguistic data has been used to extend this time depth, but it too is severely limited in the type of information it can yield. What follows is a brief description of what is known about the development of pastoralism in East Africa. It makes use of archaeological and linguistic data as well as oral traditions of modern African people. The earliest evidence for animal domestication in Africa occurs in what is now the Sahara desert region on north Africa. By the fourth millennium B.C. at the site of Es Shaheinab near Khartoum (about 3200 B.C.) dwarf domestic goat occur (Clark 1970:191). At Haua Fteah, near the Mediterranean coast by about 5000 B.C. at least 80% of the meat eaten was from domestic sheep or goat. At yet another site, the Uan Muhriggiag rock shelter, at the eastern end of the Tassili Range, a skull of what has been described as a domestic Bos brachyceros has been found. It is dated to the early fourth or fifth millennium B.C. (4000 i 120 B.C.), (Clark 1970). Domestic cattle may also be present in the lowest layers of the site which date to about 5500 B.C. An additional date of 3810 B.C. was obtained from a 805 skeleton at the site of Adrar Bous also in the Sahara (Carter and Clark 1976). In the Tassili Range rock art also occurs which may depict herds of domestic cattle and sheep. Although the early occurrances of domesticated animals are scarce, it is apparent that by at least the fifth millennium B.C., and likely earlier, pastoralists are present in the Sahara. After 2500 B.C. the Sahara became increasingly dessicated (Clark 1976). Many of these pastoralists were forced to move out of the Sahara region. These movements may have taken several directions. Some moved into the Nile Valley and Ethiopia while others likely turned southward into southern Sudan and northern Uganda and into the high grasslands of the Rift Valley in Kenya and northern Tanzania (Clark 1970:202). The Lake Turkana region in northwest Kenya figured prominently in this spread of pastoral peoples into the Rift Valley since it offered a natural geographical corridor into Central Kenya and Tanzania (Figure l). The area around Lake Turkana constitutes one of the largest low lying gaps in the rugged Rift Valley. To the north of the lake is the Ethiopian Plateau which rises sharply some 1500 meters above the Lake Turkana region. To the south and west of Lake Turkana is the western edge of the Rift Valley and the Elgego escarpment which rises abruptly to some 1500+ m. above the surrounding valley floor. By the end of the first millennium B.C., pastoral peoples occupied the Eastern Rift Valley of Kenya and Tanzania (Clark 1970). These people are for the most part associated with the "neolithic" stone bowl cultures, an assemblage typified by the occurrence of stone bowls and grindstones. The Prolongs Drift site in the Nakuru Basin provides perhaps the earliest data for domestic cattle in the Rift Valley and is dated at 1500 B.C. (Sutton 1972). However, recent work by Bartholme (1977) has uncovered evidence of domestic stock (sheep/goats) in early lake beds east of Lake Turkana which may prove earlier. Ethiopian Pla . -- —-—'\ ETHIOPIA '\._._ ....._. Ww Lake Victoria . over 1500 mg \l \- ‘. r0 KM Figure 1 THE HIGHLANDS AND RIFT VALLEY OF KENYA It is believed that this first southward migration of pastoralists into Eastern Africa can be associated with the south- ward movement of peoples speaking a Cushitic language (Clark 1970; Sutton 1977). "There seems no reason not to identify the early Southern Cushites with those earliest East African food producing societies the material remains of which have been discovered by archaeologists in parts of central Kenya and northern Tanzania" (Ehret l97l:8). Cushitic language groups belong to the larger Afro-asiatic family which includes Arabic and other Semitic language groups. They are presently centered in the Ethiopian highlands and are divided into four main groups of which only two are of concern here. Only the Eastern and Southern Cushitic peoples seem from present evidence to have played a significant role in the history of East Africa (Sutton 1973). The archaeology of this period is almost completely unknown and so the reconstruction of this period relies almost entirely upon linguistic data. Based upon linguistic evidence it would seem that the first area of Cushitic establishment was southern Ethiopia near the Kenya border (Ehret 1974). Beginning in the second millennium B.C. and lasting well into the first millennium there was a proliferation of Cushitic communities and spread of Cushitic influence southward. It was during this period that Southern and Eastern Cushitic speaking groups became differentiated with the Southern Cushites spreading over large parts of East Africa. By the first millennium B.C. the original Southern Cushitic community was spread in a number of scattered groups as far as Lake Victoria and the Indian Ocean (Ehret 1974). The original community by this time had evolved into at least three successor Southern Cushitic speaking groups--one probably in southeastern Kenya, another in northeastern Tanzania, and still another further into the East African interior, perhaps in southwestern Kenya (Ehret 1974). At the same time the Eastern Cushites differentiated into several communities, one attained particular pre-eminence in the country along the southern fringes of the Ethiopian highlands. They strongly influenced other Afro-asiatic speakers as well as several Eastern Sudanic peoples, most notably the early Nilotes. Like the Cushitic-speaking peoples, though many centuries later, the Nilotes, also pastoralists spread southward from the Ethiopian, East African borders (Ehret 1974). The ancestral Nilotic peoples probably inhabited an area along the southern fringe of the Ethiopian highlands near the Lake Turkana region. "The overall distribution of Nilotic languages in earlier times thus suggests the proto-Nilotic homeland as falling in a broad belt of territory running from northern Kenya and the Lake Turkana region west and northwest toward the Nile River" (Ehret 1974:164). The original Nilotic people later differentiated into several successor communities. These were three in number, speaking dialects ancestral to the modern, (1) River Lake (Western) Nilotes, (2) Plains (Eastern) Nilotes, and (3) the Highland (Southern) Nilotes (Ehret 1971). The Highland branch of the Nilotes consists of the Kalenjin cluster in the western Kenyan highlands. This group comprises the Nandi, the Kipsigi, the Tugen, the Elgego, the Marakwet, the Pdkot, and the Elgon Kalenjin. The Plains Nilotes fall into three main sub-groups: the Masai in Kenya and Tanzania; the Karimojong-Teso cluster of northwestern Kenya and northeast Uganda with extensions into southeastern Sudan; and the Bari cluster, mostly in southern Sudan. The River-Lake Nilotes consist of the Dinka and the Nuer of the Sudan as well as the Luo-speaking peoples, who are divided into a number of tribes extending from the southern Sudan to northern Uganda and the lakeshore of western Kenya (Sutton 1968). (Figure 2). Both the Highland and Plains Nilotes were influenced by Cushitic-speaking peoples already in these areas; the River-Lake Nilotes to a much lesser degree. The Highland Nilotes were par- ticularly influenced with strong evidence of contact in both cultural practices and vocabulary. The Highland Nilotes probably had contact with Eastern Cushitic speaking groups around the first millennium B.C. probably near the southern fringe of the Ethiopian highlands (Ehret 1974). This contact took place before the spread of the Highland Nilotes southward into Kenya. When they did spread southward sometime in the first millennium B.C., they carried with them a culture incorporating many practices of Cushitic origin and a Nilotic language incorporating many Eastern Cushitic loanwords (Ehret 1974). 10 Figure 2 MODERN LOCATIONS OF CUSHITES; PLAINS, RIVER-LAKE, AND HIGHLAND NILOTES 0 150 Miles go‘ e‘” GALLA ° .._ .r ’ “'1' L Turkana ME "I .1" .I 9 I I SOMALI I i szra : TANZANIA 0‘ . \ o‘.‘ Southern Cushites RIVOr-lake Nilotes (Luo peoples) @ Highland Nilotes G) "a“ Q) Gm“ g Plains Nilotes @ Burungi and Alagwa J 11 The borrowed features indicate that the particular Eastern Cushites who influenced the Highland Nilotes were closer both culturally and linguistically to the Galla and related peoples such as the Konso than to other Eastern Cushites such as the Somali, Afar and Sidamo (Ehret 1974). Once they moved south into the Rift Valley and the western Highlands of Kenya the Highland Nilotes encountered the Southern Cushites, but this group had much less of an effect on the Highland Nilotes. By 1000 A.D. the River-Lake Nilotes were in the Nile Region. The Highland and Plains Nilotes had spread southward and covered a wide area around the modern borders of Sudan, Ethiopia, Kenya and Uganda. In more recent times there have been a number of other major population movements in Eastern Africa, also in the Lake Turkana region. By the 12th century or slightly earlier the Somali, Eastern Cushites, moved out of the southern edges of the Ethiopian highlands into the Kenyan, Somalia borderlands. The Galla, also Eastern Cushitic-speaking peoples were already established in this area. The subsequent Somali expansion to the south and west forced the Galla to migrate. This migration took place in the 16th century A.D. when they traveled toward Lake Turkana (Were 1972). The Galla entered Ethiopia from the Lake Turkana region about A.D. 1522-30, an event which can be dated from the evidence of the Ethiopic document of 1593, confirmed by the Ethiopian Chronicles (Huntingford 1955:76). 12 It is likely that at this time the Galla had contact with the ancestors of the Masai (Huntingford 1955). Masai oral tradi- tions suggest that they originated in the area west of Lake Turkana (Jacobs 1975:411). Their traditions refer to the place of origin as endigirr ee Kerio. This place may be identified with the Turkana escarpment where the Kerio River flows into Lake Turkana (Were 1972). They slowly moved southward into the central Rift Valley beginning perhaps as early as the 14th century A.D. This movement was caused by the increased dessication of the Lake Turkana region (Jacobs 1975). By the 17th century the Masai had reached the Uasin Gishu plateau. According to oral traditions (Gulliver 1955) the Turkana who were formerly part of the Jie tribe in Uganda moved into the Rift Valley of Kenya in the early 1800s. The Turkana moved into the area west of Lake Turkana pushing back the Samburu to the south and east and other groups to the north. Throughout this brief discussion of the origin and spread of pastoralism in East Africa, the importance of the Lake Turkana region was apparent. It figured prominently in all of the major migrations of pastoral peoples into East Africa and as such it is an area ideally suited to a study of the prehistory pastoral societies. CHAPTER II SETTING AND ARCHAEOLOGY During parts of 1975-1976 fieldwork was conducted by the author as part of a project from Michigan State University led by Lawrence H. Robbins. This project was a continuation of earlier work by Robbins into the prehistory of the area west of Lake Turkana. The author's role in this project was to excavate and analyze what was thought to be an extensive burial area at the Lothagam Late Stone Age site (Lothagam is situated approxi- mately 16 km. south of the Turkwell delta--see Figure 3). Initial work at the site by Robbins in 1965-1966 identified a large burial area at the site on the basis of human bone scatter. Excavation of these burials was particularly important given the paucity of Late Stone Age skeletal remains in East Africa. However, upon initial sampling of this area it soon became apparent that due to extremely poor skeletal preservation only severely limited kinds of information could be obtained from these burials. As such, it was decided that Lothagam did not warrent extensive excavations. Because of this, the author initiated work at two large rock art and cemetery sites reported to Robbins by R. Soper of the British Institute in Eastern Africa. This work was 13 MILES I ‘ ETHIOPIA .'— O c—a—I-a F ERGUSON'S GULF Namoratunga Site 3 —. Lodwar Namoratunga Cemetery a. Rock Art Sites Turkwe Lokori Figure 3 LOCATION OF NAMORATUNGA SITES y’ TANZANIA\. 15 generously supported by the National Science Foundation grant #SOC 75-15569. Location and Discovery The two sites are situated 11 km. northeast of Lokori, a small and relatively isolated settlement in northwest Kenya (see Figure 3). The sites are located on basalt cones two km. west of the Kerio River and are separated by one kilometer. The southernmost basalt cone is located at 2° 0' 5" N 36° 6' 50" E. These two sites were initially reported to Robert Soper by Mr. Ben Aldridge of the African Inland Mission at Lokori in 1966. In 1968 Soper made a brief visit to the site and recorded a number of petroglyphs found there (Soper 1968), (see also Gwynne 1969). He later returned and conducted limited excavations at the southernmost cemetery in 1970 (Soper 1977). The present author visited the sites in 1975-1976 and conducted the first large scale excavations at these important sites. While these two sites were being excavated a third site was reported to me by one of my workers. This site was 110 miles further to the north just south of the town of Kalokol near Ferguson's Gulf. It is two miles west of the present shore of Lake Turkana. Unfortu- nately excavations could not be conducted at this third site (this site is discussed separately on pg. 166 to 180). Interestingly, all three sites share the same Turkana name of Namoratunga. This means literally "people of stone." Turkana legend, as reported to the author by a number of elders in the 16 region, has it that a number of people were dancing in circles when an evil spirit dressed as a beggar appeared. When the dancing people laughed at his appearance, he turned them into stone. The Turkana also have a song about these sites. The Setting Since both the geography and geology of the area where the Lokori sites occur is discussed in detail by Morgan (1971) and Rhemtulla (1970), only a brief summary will be given here. The area southwest of Lake Turkana lies in the Eastern Rift Valley in an area characterized by, "Divergent faults and marginal monoclinical downwarps" (Rhemtulla 1970:71). This area ranges from desert to semidesert with an average rainfall of less then 10 inches per year at Lokori and slightly less than six inches at Lodwar. This usually falls in a rainy season from March to May, although in more desert regions these seasons are less well defined. The Kerio River is the main river in the area. It is also one of the two major rivers draining the highlands flowing to Lake Turkana. It flows for the greater part of the year only drying during August to November. This is because most of its water comes from the Elgeyo escarpment to the southwest where rainfall is much higher. The drainage of the river which flows into Lake Turkana is dictated by the overall northward slope of the rift floor which falls from 1830m at Nakuru to about 366m at Lake Turkana. 17 At Lokori the Kerio River turns eastward before resuming its northerly alignment. North of Lokori the plain of the Kerio is divided by a lava hill called Morulim and by the gravel terraces beyond it. Because of this obstruction the river takes its easterly course for about 10 km. before returning to its north- east course. Formerly the Kerio flowed to the west of Morulim, the change having occurred within the past 30 to 40 years. Just north of Morulim, the river skirts the Loriu Plateau to the east and a broad alluvial plain to the west before draining towards Lake Turkana. This broad alluvial plain is bounded by tertiary volcanics further to the west. The alluvial plain is of low relief and consists of hori— zontal sediments ranging in age from Pliocene to recent. Elevation of the plain is about 580m above sea level. Ten soil samples were taken across the alluvial plain at 200m intervals. Samples ranged from 98.7% sand to 83.5% sand; 9% silt to 2% silt and 7.1% clay to 1% clay. A complete listing of soil composition as well as grain size is given in Appendix D. The Namoratunga sites are situated on weathered basalt cones on these alluvial plains (Figure 4). Both sites are located on what were major tertiary volcanoes and stand at 640m above sea level, 60m above the floor of the plain. The eastern side of the southernmost hill has been weathered to a broad raised platform about 150m x 150m. It is on this raised platform that the mortuary site is located (Figure 5). The west portion of the hill consists Figure 4 View of the volcanic outcrops near Lokoricontaining Namoratunga sites Figure 5 View of southernmost Lokori site. Cemetery is on raised lava platform to the east 19 of a rocky peak formed by the surviving rim of the cone. Extending from this hill in a north-north-westerly direction is an irregular lava ridge extending for about 400m roughly level with the eastern platform. The northern hill is marked by a crater like depression at the north and the peak left by the surviving rim to the south. On both hills the exposed rocks are broken into irregular slabs with smooth surfaces. The vegetation in the area, although scarce can be divided into a number of distinct units based on the distance from the Kerio River and the nature of the surface (Morgan 1971). The area bordering the river itself, two km. east of both hills, contains the most luxuriant vegetation in the area. It consists of a gallery forest with, Acacia Tortiles, Cordia gharaf, and Salvadora persica being the most common trees. Further from the river on the alluvial plain beyond the gallery forest is a zone dominated by Salvadora persica which forms discrete clumps and sometimes including Acacia tortilis. Furthest from the river (about three kms) is a zone of scattered mixed bush varying from moderately dense to very widely scattered. This area is characterized by Acacia nubica but includes Acacia tortiles, Salvadoragpersica, Boscia coriocia and bushes of Moerva subcordata, M. Crossifolia and M. Oblongifolia. Along the former course of the Kerio the forest is degenerating with much dead wood. This is due to the Kerio changing course and leaving its former gallery forest without a source of water. 20 Along the Kerio today the Turkana cultivate small fields where they grow millet. These fields are located adjacent to the river on its floodplain. Each year soil is deposited on this floodplain as the river overflows its banks during the height of the rainy season. As such this area contains soils that are more fertile than the surrounding areas. It is doubtful that these surrounding areas could support millet. On the Namoratunga hills themselves there is generally a sparce vegetation cover. Despite this sparce cover, however, the hills contain a surprisingly diverse number of plants. These include: Salvadora persica L.; Aristida mutabilis Trin. and Rupr; Indigofera spinosa (Forsk.); Acacia nubica Benth; Acacia tortilis (Forsk); Adenuim obesum (Forsk); Maerica crossifolia (Forsk.); Caralluma russelliana (Brogn.); Boswellia sp.; Barleria acanthoides Vahl; and Ruellia patula Jacq, (personal communication J. Kahurananga, East African Herbaruim) (Samples were collected by L. H. Robbins from Hill 1. All of these plants are Used by the local Turkana inhabitants.) Although herds of large game are common in Kenya, the region west of Lake Turkana contains a relatively sparce animal population. Although 25 species have been recorded in the Turkana district (Stewart and Stewart 1963), almost none were actually observed in the Lokori area by the author. However, elephant tracks were observed in the vicinity. It should be noted that local elders claim that in the past the area around Lokori 21 contained much more game, including zebra, giraffe, and gazelle. (Lokori itself means giraffe crossing in the Turkana language). In all likelihood the depletion of game in this region is fairly recent due both to hunting by the local Turkana and overgrazing of the area by the Turkana herds (Allan 1965). These same elders claimed that the Lokori area was much greener during their boyhood (30 to 40 years ago). (This corresponds to data reported for the Kerio Delta area collected by Robbins 1974). In the past it appears that the Kerio flowed year round. It would appear then that the semidesert conditions now found in the area are fairly recent in origin. It seems reasonable to assume that in the past the area could have supported a much higher population density. Today the population density of the area is one of the lowest in East Africa, 3.3 persons per square mile (Allan 1965). Given the sparce vegetation and the food requirements of the Turkana herds, the carrying capacity of the area is very low. Naturally this carrying capacity would increase greatly if more plant food was available for the Turkana herds. Excavation Strategy The only previous work conducted at the Lokori sites was that done by R. Soper (1968; 1977). His initial report was on a number of the petroglyphys found at the sites (1968). Soper subsequently returned in 1970 (Soper 1977) and excavated four of the cairns near the center of the southernmost of the two sites 22 (see Figure 6). He reported 110 cairns on the southern outcrop and seven on the other. These figures were based on aerial photographs. However, many graves could only be observed by a closer "on the ground" inspection. This is because many of the graves, particularly in the western portion of the site, were partially obscured by wind blown sand. As a result my research revealed that the southern site contained 162 graves and the other 11; an increase of 52 and four graves respectively. The individual graves consisted of an outer circle of upright irregularly shaped basalt slabs (Figure 7). The interior of the circles were covered with horizontal slabs and small stones, which sometimes rose to a low mound about 40 cm above the sur- rounding ground. All of these stone slabs were easily acquired from the outcrops on the hills where the graves were located. Here the basalt has exfoliated into regular slabs 10 to 15 cm thick. The external diameter of the graves ranged from 1.2m to 4.7m with a mean diameter of 2.3. The outer standing slabs ranged in size from 5 cm above the ground surface to as much as 175 cm (grave #49). Due to the large number of cairns as well as the effort required to excavate them, a sampling strategy was employed. The initial stage of the excavation, consisted of selecting a 15% random sample of all the graves at the two sites (Figure 8, 9) (24 at one site and two at the other). The usefulness of a random sampling design is more than adequately discussed elsewhere (Clarke SITE 1 Figure 6 GRAVES lNlTlALLY EXCAVATED BY R. SOPER [INDICATED 3v HATCHINGI 24 Figure 7 Grave 41 at Namoratunga I 25 O 2] s R m1 V. E M 5|] 0L SITE 1 INITIAL 15% EXCAVATION SAMPLE [INDICATED BY HATCHING] 26 O I 3 - I__I_I__l Excavated METERS @208 Figure 9 NAMORATUNGA 2 EXCAVATED GRAVES 27 1968) and so will not be dealt with further here. This 15% sample was found to be significant at the .05 level. All graves were profiled whenever possible removing only the inner horizontally layered stone slabs (e.g. Figure 10). The outer circle of vertical slabs was left in situ if possible. In doing so it was hoped that the site, upon completion of the exca- vations, could be returned as closely as possible to its original configuration. The cairns were excavated as a unit, however,any artifacts found were mapped with depths recorded. At the onset of the excavations two graves were excavated with all soil being screened through 1/8" mesh (7 and 9). At the same time two other graves (143 and 16) were excavated but not screened. Since all the graves had to be painstakingly excavated by trowels, it was found that the artifact recovery rates for screened VS non-screened graves were nearly identical. Since screening greatly increased the amount of time necessary to excavate a single grave, it was discontinued. The graves were initially cleared of a layer of wind blown sand which covered the horizontal slab interior. This deposit of wind blown sand often reached a depth of as much as 30 cm. The graves were then mapped (see Appendix A for drawings of graves excavated). The graves were excavated in two halves to maintain a vertical profile. In profiling these graves it became apparent that the inside of the circle consisted of horizontally layered slabs to a depth of 50 cm or more. These slabs ranged in size 28 On 305 _au.te> .0 tonnam .230 3...» 2.2“ 13:3) .0 miu .atan I """.‘.I.‘ . I ; m><¢0 :m .atan r,...».~.}.4».y.1.t.vhrtr.._\.;..n .A .1 4...... new ZOZUDF—mZOU m>z P P ' IXIM Test Squar .Tl e LCantaur Interval-10 meters 0 50 Llell \. Figure 13 1 METERS NORTH-SOUTH TRANSECT AT NAMORATUNGA 1 45 levels and terminated when sterile soil was encountered. All squares were terminated in the second level (at a depth of about 30 cm). These seven squares only yielded two chert flakes and lacked any traces of additional burials. A second set of 23, 1 X 1 meter squares was excavated just to the west of the cemetery area but still on the same basalt outcrop. Squares were dug every 10 meters in a line running north-south across the outcrop (see Figure 13). These squares were also dug in arbitrary 15 cm levels and were all terminated by 30 cm. No artifacts were recovered in these test squares. From these 30 squares it appears that there were no non-cairn burials at the site and that a large segment of the burial population was interred elsewhere or was not buried. Site Survey In addition to the excavations carried out on the mortuary sites, an attempt was made to locate any associated habitation sites. It was thought that habitation sites would reveal aspects of the material culture and subsistence, which could not be deter- mined from the burial data alone. To facilitate this a survey was conducted of 190 square kilometers around the two sites with the two sites located in the approximate middle of this area (see Figure 14). This area was defined largely by geographical fea- tures: The Loriu Plateau to the east, the Kerio River to the south, one mile west of the Kangatet River to the west and the Kangatet River to the north where it joins the Kerio. This survey area was 46 . Figure I4 SURVEY AREA A SITE LOCATIONS 47 in turn stratified for sampling purposes. The 12 square kilometers immediately surrounding the two sites were intensively surveyed, while the remaining 178 square km were surveyed using standard survey procedures. This intensive survey entailed in addition to standard survey on foot, the digging of 50 cm test holes every 200 meters across this area (see Lovis 1976 for rational of this tech- nique) from the Kerio River in the east to the Kangatet River in the west. The depth of 50 cm was chosen arbitrarily as was the spacing of test holes every 200 m. Here it was assumed that much like the present day settlement pattern in the area the presence of water would have dictated site location. In all 325 test holes were dug. They recovered no artifacts or evidence of habitation sites. Appendix 0 contains the results of a series of soil samples taken across the area. What these indicate is that the Kerio and Kangatet Rivers have in the past meandered extensively in the area around the Namoratunga l and 2 sites. In fact, at one time the rivers may have been as close as 100 meters to the two sites. The entire area surveyed was found to be flood plain. To try to get some idea of the depth of these flood plain deposits three additional 2 X 2 meter squares were excavated running due east from Namoratunga 1 extending towards the Kerio River. The first square was dug approximately 200 meters east of the site, the second at 1200 meters and the third, near the present bank of the Kerio, approximately 1800 meters to the east. The first two squares were taken to a depth of 2 m. the third to a depth of 3 meters. All three squares were terminated at those depths. No 48 artifacts were found, but it does appear that the flood plain deposits extend to a depth of at least 3 meters. The remainder of the 190 square kilometers in the survey area was surveyed either by foot or by motor vehicle. Because the area lacked any appreciable vegetation, except near the Kerio River survey by motor vehicle was efficient in most areas. Where this was not feasible, survey on foot was conducted. In addition, erosion gulleys near the Kangatet River as much as 4 meters deep revealing extensive profiles were inspected. No sites were found in this enlarged survey area between the Kerio and Kangatet rivers. However, just to the west of the Kangatet River, where low lava ridges begin to replace the horizontal flood plain deposits several sites were noted. Three Lower Paleolithic as well as one Late Stone Age site were recorded (see Figure 14). (Only small representative samples of artifacts were collected from these sites.) (See Appendix E for site descriptions.) The survey produced only four sites and none of these could be associated with the cemetery sites. There are a number of possible explanations for this. (1) There are no habitation sites of the proper antiquity in the survey area. Individuals lived outside the survey area and traveled some distance to intern the dead at these sites, in this case at least over 4 kilometers. Some support for this may be provided by burial 83, a bundle reburial. (2) Habitation sites were buried under flood plain deposits from the Kerio and Kangatet Rivers at a depth greater than 50 cm. Fifty 49 centimeters was the arbitrary depth chosen for the test holes in the first survey strata. The three subsequent 2 X 2 meter test squares dug from Namoratunga 1 east to the Kerio River indicated that the flood plain deposits reached a depth of at least 3 meters in the areas tested. (3) The third explanation may have to do with the nature of the habitation site itself. In the area today among the Turkana,settlements are very small and temporary often consisting of little more than 3-5 small thatched huts and an animal kraal made of brush. Given the demands of the local environment, a certain degree of mobility is required and thus a limited material culture. Just west of Namoratunga 1 an abandoned Turkana camp was found. According to the local Turkana the camp had been abandoned for less than two years. In that time all that remained of the camp were small circles of stone marking what used to be the outline of several huts. This is because in addition to normal processes such as wind action, the Turkana often reuse the wood from abandoned settlements. Given several more years, this particular settlement would be completely unrecognizable archaeologically. Robbins (1973) came to similar conclusions and noted that at least 63% of Turkana material culture consists of perishable items. If the assumption of analogous types of settle- ments for the people responsible for Namoratunga l and 2 is valid, then habitation sites would be extremely difficult to locate. CHAPTER III MORTUARY ANALYSIS Mortuary customs and mortuary sites have been the subject of study by archaeologists and social anthropologists throughout most of the history of professional research. However;most of the earlier studies that utilized mortuary data (and many as recently as the 1960's) tended to focus on only one dimension of mortuary practices. Burials were described in great detail (Webb 1946, Lewis and Lewis 1946 to name but a few) but analyzed only in terms of the dimension of time. Similarities or differences between burial populations were explained in terms of different cultures mixing through time and influences from other areas (Kroeber 1927, Dragoo 1963). These early studies display an unsophisticated view of cultural variability in assuming uniformity in burial customs within a society and that sites could be compared on the basis of formal similarities in burial practices. Since the manner in which different social positions are symbolized at death is often quite arbitrary, any comparisons of this nature are bound to be highly misleading. Different cultures may symbolize the same idea in very different ways (Binford 1971). More recently, studies have viewed (Binford 1971, Goldstein 1976, Saxe 1970, Peebles 1974, Mainfort 1976) mortuary practices as 50 51 a critical facet to the understanding of cultures (Brown 1971). These studies have attempted to derive elements of social organi- zation from burial data. However, in doing so many of these studies have concentrated on a single dimension in extracting inferences about extant social organization. Many, for example, have relied almost entirely upon artifactual associations and while examining other dimensions, have treated them as only secondary in importance (Larson 1971, Peebles 1971, 1974, to cite just a few cases). Other researchers have relied largely upon studies of social and political organization (Fried 1967, Service 1962). Several of these studies have achieved a certain degree of success (Stickel 1968, King 1969, Renfrew 1973). Unfortunately such approaches have tended to conceptualize social variables as essentially dichotomous entities--egalitarian or ranked. There is no room for varying degrees of rank differentiation in such an approach. The recent ethnographic surveys of Saxe (1970) and Binford (1971) have demonstrated the multidimensional nature of mortuary data. They provide empirical support for the proposition that a wide range of social inferences can be drawn from mortuary data. Both the organization and structure of societies as well as the status positions occupied by these members of the societies is encapsulated in the mortuary routine. Hence, these studies suggest that of all data categories available to the archaeologist, mortuary data may offer the most complete picture of the organization of the social system. 52 Both Saxe and Binford (see references above) rely heavily for their theoretical framework upon the work of Ward Goodenough and his discussion of role theory (Goodenough 1965). Status is viewed as a combination of rights and duties termed social identities. Status involves "(1) what legal theorists call rights, duties, privileges, powers, liabilities and immunities . . . and (2) the ordered ways in which these are distributed in what I shall call identity relationships" (Goodenough 1965:2). The ordered patterns of rights and duties in the interaction between distinct social identities are identity relationships. However, in any given interaction an individual must select from a range of social identities at once. These must combine to make a grammatically possible composite identity, that is, the social identities must not be conflicting with the occasion on which they are used or with other identities. "The composite of several identities selected which is appropriate to an interaction constitutes the selectors social persona in the interaction" (Goodenough l965:7). Saxe (1970) has observed that at death the social persona of the deceased involves the possible interaction of the deceased's entire range of social identities and the alternate identities recognizing relationships to the deceased. Thus at death, a composite social persona is symbolized through the mortuary ritual. Expanding upon this, Binford (1971:17, 21) suggested that two components of social significance participate in structuring the form of the mortuary ritual. The first is the social persona of 53 the deceased. The second is the size and composition of the social group recognizing status responsibility to the deceased. A direct correlation between these two can be inferred. The more important the social persona, the more people recognizing duty-status relationships to him. Hence the number of identity relationships that an individual might have varies with the organizational form of the society in question (Saxe 1970:7). In egalitarian societies one might expect infants to have few identity relationships while elders would have many by virtue of their having participated more fully in the society. Such a society should exhibit graded increases in the number of identities from infants to elders since statuses are achieved. If on the other hand an infant is found to have a larger social persona than certain adults, some form of hereditary ranking is likely indicated. In this case the death of an infant may have been a matter of concern for the entire community. The ethnographic sections of the works of Saxe (1970) and Binford (1971) bear directly upon this point. Although Saxe proposed a series of eight hypotheses, only the first two are significant here (see Chapter V1 for complete discussion of all 8 hypotheses). Hypothesis 1: The Components of a Given Disposal Domain Cooperate in a Partitioning of the Universe, the Resultant Combinations Representing Different Social Personae (Saxe 1970:65). Hypothesis 2: In a Given Domain, the Principles Organizing the Set of Social Personae (Produced by the 54 Cooperative Partitioning of the Universe of Disposal Components) are Congruent with those Organizing Social Relations in the Society at Large (Saxe 1970:66). Saxe was able to confirm these two propositions using a small sample of three societies. Binford (1971) proposed three further generalizations utilizing a much more extensive ethnographic sample. "1. The specific dimensions of the social persona commonly given recognition in the differentiated mortuary ritual vary significantly with the organizational complexity of the society as measured by different forms of subsistence practices. 2. The number of dimensions of the social persona commonly given recognition in mortuary ritual varies significantly with the organizational complexity of the society, as measured by different forms of subsistence practices. 3. The forms, which differentiations in mortuary ritual take, vary significantly with the dimensions of the social persona symbolized" (Binford 1970:23). Given a set of interments, then, it becomes possible to make inferences not only about the social persona of the deceased, but also the level of complexity of the society itself. Two other recent works that have made significant contri- butions to the study of mortuary data are those of Tainter (1975) and Goldstein (1976). Both rely heavily upon the works of both Saxe and Binford for their basic framework. Tainter, borrowing extensively from Brown (1971) proposed energy expenditure as a measure of social structure and organization. Tainter uses entrophy measures to quantify this energy expenditure (see Goldstein 1976 for discussion of difficulties inherent in the use of such entrophy 55 measures). Goldstein (1976) while utilizing a range of distinct burial dimensions relies heavily upon locational variables. She deals specifically with Saxe's hypothesis 8 revising it with a much larger ethnographic sample. While the works of Saxe and Binford will provide much of the theoretical underpinnings of this work, the author also borrows heavily from the works of both Tainter and Goldstein since their particular approaches are well suited to the author's own data base. Cluster Analysis The first stage of the Namoratunga mortuary analysis entailed the grouping of the burials into meaningful culsters. As mentioned previously, very little in the way of grave furniture was recovered. As such, only characteristics of the grave itself and body disposition could be utilized in classifying the burial data. To implement this classification a clustering routine was utilized. This program was drawn from the Statistical Package for the Social Sciences (SPSS 1975) on file at Michigan State University. All analysis was conducted on a CDC 6500 computer. The variables used in this study included: (1) external grave diameter; (2) total grave depth; (3) burial pit depth; (4) range and average size of vertical slabs standing above the surface; (5) age, and (6) sex of the individual; (7) compass orientation as measured in degrees; (8) presence of petroglyphs on the vertical slabs. These represent all of the variables that were recorded 56 for each grave. All the excavated graves were utilized in the cluster analysis. The clustering routine used was ISODATA. It is a polythetic agglomerative program (Nie et a1. 1975). ISODATA is an average link method which fuses subordinate clusters based upon the average similarities between all cases previously existing in a cluster and the potential new member. From this routine five burial groups were derived (see Table 4). Although these burial clusters are treated as distinct groups in Table 4 for ease of study, they should be viewed more accurately as forming a continuum with a certain amount of overlap between categories. In examining these groups it is apparent that one burial, 89, was isolated. However, this is understandable when the characteristics of this grave are examined more closely. In addition to being one of the largest graves in diameter, 3.6 m, it was also the deepest (2.3 m--next closest was 1.8 m). As such, the isolation of 89 is both understandable and apparently warranted. When these individual burial groups are examined it is found that they closely parallel the weight of stone, as expressed in tons, utilized in grave construction (see Table 5). This is true despite the fact that this variable was not utilized in forming the original five groups. This is particularly interesting since Tainter (1975) and to a certain extent Brown (1971) have proposed using energy expenditure as a means of measuring social complexity. Using a sample of 103 ethnographic groups Tainter found that the 57 TABLE 4.--Five Burial Groups Derived from ISODATA Routine. Cluster I II III IV V VI #89 41 120 98 138 65 46 132 122 53 39 78 160 82 37 83 45 159 32 76 97 16 7 54 74 27 105 79 97 17 60 77 9 30 143 56 31 TABLE 5.--Breakdown of Clusters by Weight of Stone in Grave. Cluster I II III IV 10.56 6.59 4.75 4.69 1.57 3.5 5.78 3.86 3.75 2.11 2.36 6.38 4.69 3.94 2.11 1.25 4.69 4.69 3.94 1.88 2.37 7.33 2.90 2.60 2.65 2.12 3.20 2.11 2.60 2.64 2.36 1.88 2.90 1.28 .62 1.96 l 28 10.56 6.15 4.01 3.34 2.06 ave. tons 58 status an individual held in life correlated directly with the amount of energy expended on that individual's interment. To quantify this energy expenditure Tainter employed entrophy measures. Goldstein (1976) while finding no difficulty with Tainter's basic premise that energy expenditure correlates with status, did note a number of problems inherent in the use of entrophy measures. Excluding his use of entrophy, Tainter's hypothesis is useful in terms of the Namoratunga burial data. If we assume the weight of each grave reflects the energy expended in their construction, then at the Namoratunga sites not everyone had similar amounts of energy, as measured by grave weight, expended on their burial. For example, grave 89 contained 10.5 tons of stone whereas grave 143 only contained .62. This variability can be viewed in several ways. As Tainter has suggested this variability may reflect differences in the status of these two individuals, the individual whose grave contained 10.5 tons of rock being of higher status than the individual's who grave only contained .62. However, this difference may also be due to other con- siderations. For example, large grave size may reflect the status of that individual's kin group with kin groups differing in their relative statuses. Among the Masai, an East African pastoral group, members of the E1 Kiboran lineage warrant a distinctive form of burial (see page 196). In addition, grave size might also reflect the size of an individual's kin group. Larger kin groups would have more manpower available for grave construction. In 59 East Africa it is quite common for relatives to bury the deceased (see Chapter VI). Finally, grave size may simply correlate with who constructs the grave. If, for example, women construct the graves of other women, then one might expect the graves of women to be smaller in size due to differences in physical strength between the sexes. These alternate hypotheses indicate that differential grave size may reflect other variables besides an individual's social status. These alternate hypotheses will be dealt with later in this chapter. Returning to the five burial groups, no females are in- cluded in clusters 1, II, III and IV. Females only occur in cluster V, along with other males. It is apparent that access to the larger graves as reflected in grave weight is limited to males. A view of the sex distribution for the entire excavated burial sample reveals a skewed distribution. It was observed in the previous section (see Chapter III) that less than 25% of the individuals interred at Namoratunga l were female. Given the excavation procedures employed, the possibility of sampling error can be ruled out. Assuming a normal ratio of males to females in the parent population, at least half of the females warranted an alternative form of interment, or they were not interred. It would appear that burial at the Namoratunga sites in itself entailed a certain degree of status that was largely confined to males. Given the proposition that females were interred elsewhere, it might be hypothesized that the few who were interred at 60 ARV czocxcz “my upeeu Amy ezoexea Aev o~=o< me=o> “my o_=e< mesa» Amy e_=e< wheewz Amy o_=o< o_eeez Apv czocxcz Aev csocxca Amv opee< opeeaz A_v o_=e< opooez APV epeo< o_eoaz .mm< An mememzpu megzm Co czouxomcmii.m m4m asaxmpmia.op m4m N NeeNNNN N N N N ooeeoaxm NNo_ N ANNNV _ N N eoNN N “eoeeoem ANNNN N ANNNN N N ANNNN P oNooN ANNNN N omoeo N “Noee N ANNNN N ANNNN _ ANNNV N ENNNo: Noee ANNN_N N “NNNN Nee, ANNNN N ANNNN _ ANNNV N ANNNN N NNNNN Aev opus ”Npau< m::o> new: we cowumooN compwmoa atom zonpm gmupzocm mmcx am: I it'lT . .mm< .m> mgsxm—mii.p_ m4m

mgaxmpmii.mp m4m Lmumzpu o umucmuxu o :mno ANNNN N omooN N ENNNoz NNNNN N NNNNN Nev >N qumspu ace: No :oNquoN NNNNNNNN NNNN aN: III..- 51-...» Ill. NNNNNNNNNN--.NN NNNNN 73 being that females entering from outside the area would display greater heterogeneity in their burial positioning (Saxe 1971). Using a similar set of assumptions the Namoratunga 1 population is matrilocal, since male burials exhibit a greater degree of variation. However, the author feels that given the sample sizes involved here conclusions of this nature are entirely unwarranted. Furthermore, Saxe's contention has not been supported by ethno- graphic data. It may well be that the degree of flexure is totally unrelated to social distinctions. For example, the degree of flexure may be related to the time elapsed between death and interment and the amount of rigor mortis which has set in. (See Allan and Richardson, 1971 for discussion of difficulties in the assumptions made by approaches similar to that of Saxe [1971].) The differences found between males and females likely reflect the greater number of males interred at the site, almost three to one. Similar comparisons were made with flexure and age, flexure and orientation, and flexure and grave size. The only pattern found was that middle adults exhibited greater variability in flexure and body position than any other age category. However, it should be noted that the middle adult age category is twice as large as any other age group. Once again greater variability would be expected solely by virtue of the number of individuals involved. No regularities whatsoever were observed between flexure and orientation and flexure and grave size. 74 Decorated Graves As mentioned earlier a large number of the graves at the Namoratunga 1 site were decorated with petroglyphs that had been pecked into the standing vertical slabs of the grave (see Chapter IV for complete discussion of art). There is at least one com- pelling reason for assuming that this art is directly associated with the graves, despite the fact that the art theoretically could have been placed on the grave any time after its construction. Despite the presence of both males and females, children and adults at the sites, only males are interred in decorated graves. Thirty-seven of the graves at Namoratunga l (23%) were decorated, 27% at Namoratunga 2 (3 graves). The number of petroglyphs these graves contained ranges from 1 to 10 with an average of 3.13 petroglyphs per decorated grave. In most cases a single grave contains a number of different designs (see Table 13). Eight of these decorated graves fall into the excavation sample and repre- sent 23% of all the graves excavated. Thus the excavation sample ratio of decorated to non-decorated graves is identical to that found in the total population. The excavated decorated graves contained only males. In addition only burial clusters 11, III and IV contain decorated graves. None occur in Cluster V despite the fact that this cluster contains 45% of all the graves excavated. Cluster I contains no decorated graves, but since this cluster contains only one grave it is impossible to determine if this is indeed a meaningful pattern. 75 TABLE 13.--Decorated Graves (Excluding Those Excavated). .” Grave Location on Grave Direction Facing 5 0 SE SE 11 0* (2) N N ‘/3427* N N 0* N N (pt N s (D N N 12 /<£D sw SN 19 O N N 0* W W 11* w w 29 0 SN NE 36 0 N N O (2)“... E w 0 E E 950 N s (a N s 42 CD NNE ssw é * N s a * N s O * N s (7* N s 0...... E w 9 *** E E 76 TABLE l3.--Continued. Grave Location on Grave Direction Facing 44 O (2)* E NEW (9 * E NEW 0 (4)* E E <:) *** E E Q9 N... E E 47 0 SW NE 49 C? * N s w * N s 50 O (2)* N s W* N s 51 o E SE 0 SE SE 52 0 SE SE 61 O (2) S SSE 86 09 (2) NE NE 88 G) (4)* E E 22* E E SE SE of” C9 0 ** N NE (D ‘7 © SE S TABLE 13.-—Continued. . 77 Grave Location on Grave Direction Facing * 0 NE SW e; NE SW 0 SN NE 94 C) N SE 117 @ * N S 1‘ N S 127 @ s N * O (2) NE SN (1) * NE SN 128 O * E N 'A' @ E N 134 (n) N N 136 C? N SE 137 Q! (2) SE SE 141 f? E N <17 (2) SE E 144 ll * SN SN * 0 SN SN * 9 SN SN 3"” SN SN 145 Q9 N E 154 0 SE SE TABLE l3.--Continued. 78 Grave Location on Grave Direction Facing Decorated Graves at Namoratunga 2 164 95’ * 0 * U1) ® A. m 169 © (3) 170 9 N W SE SE NE SW NE SSE * 0n same rock surface. * *** 0n same rock but different surface. * On same rock surface. 79 TABLE l4.--Breakdown of Decorated Graves by Cluster. I II III IV V 0 2 2 4 0 0% 20% 33% 57% 0% When the graves were compared on the basis of the designs they contain, no patterns were noted (see Table 15). Graves within the same burial group did not necessarily share the same design motifs. Of the eight different designs found among the Cluster II graves (41 and 46) only one is shared, 0 . For the Cluster III graves (105 and 136) one of the total of two design elements is shared,U . There are 12 different design elements within the Cluster IV group (4) graves. These four graves have no designs in common. Similarly, no regularities are found between design motif and burial age or orientation. However, if the average number of designs is compared with grave diameter there is a general trend for larger graves to have more occurrences of art. Table 16 compares grave diameter with the number of occurrences of art. With one exception there is a tendency for larger graves to contain more petroglyphs than smaller graves. The exception is the grave diameter category of 2.5 m to 2.9 m. Here the average number of occurrences of art is 4.25, the highest average found. However, this is due to the fact that this 80 TABLE l5.--Excavated Decorated Graves. Location Direction Burial No. Decoration 0n Grave Facing Age Orientation Cluster 32 09 SE NW YA 272° IV VW" N SE * 48 <2) . a ll * N E I (2)* N E CG * N E 41 (9) SE SE 2 190° II (‘1) N S * O N S Q * N N W ** SE NW *‘A’ CD SE NW 46 (D N N 1 175° 11 * C9 SN NE C.) * SN NE 54 W N s YA 185° IV ‘60 N s 'k 9 (2) SE NN 0 * SE NW 60 O * NNE SN '2 East? IV © * NNE SN * GP NNE SN 81 TABLE 15.--Continued. Location Direction Burial No. Decoration On Grave Facing Age Orientation Cluster © *** NE SW @ *** NE NE 79 0 SE SE MA Nest IV 105 Q9 SN SN MA 180° III 132 © N SE ? East III Q9 SSE SE * On same rock surface. ** On same rock surface. *** 0n same rock but different surface. 82 TABLE l6.--Grave Diameter vs. Number of Petroglyphs. Diameter 1.5-1.9 m. 2.0-2.4 m. 2.5-2.9 m. 3.0-3.4 m. 3.5+ N=5 N=14 N=4 N=6 N 7 Ave. No. of Petroglyphs 2.2 3.0 4.25 3.1 4.1 category only contains four decorated graves and one of them, #44, contains 9 occurrences of art, the second highest total at the site. If this one idiosyncratic grave is excluded, the average number of petroglyphs per grave would closely parallel grave size. Spatial Analysis (Placement of the Graves) The Namoratunga l cemetery was also examined to determine if there were any spatial regularities, the assumption here being that if a discernible spatial structure does exist within the organization of the Namoratunga 1 cemetery this structure is a reflection of the organizational complexity of the society it represents. Goldstein (1976) in her examination of Mississippian social organization found a strong correlation between the spatial organization of Mississippian cemeteries and social structure. The first variable considered in the Namoratunga 1 spatial analysis was the location of decorated graves. Figure 15 indicates the location of decorated graves at the Namoratunga 1 cemetery. To perform the spatial analysis a grid was imposed on the entire cemetery. The roughly square configuration of the cemetery readily 83 AAETERS SITE 1 Figure ‘15 DISTRIBUTION OF DECORATED GRAVES [INDICATED BY HATCHINGI 84 lended itself to such a grid. The grid employed consisted of 12 cells of comparable size. A 12 cell grid was the smallest size that could be utilized, given the necessity of having at least one decorated grave in each cell (a requirement of chi-square statistic is no empty cells). Once the grid was imposed each decorated grave was given an X and Y coordinate which located it within a particular cell. The program CROSSTABS drawn from the Statistical Package for the Social Sciences (Nie, et a1. 1975) was then used to compute the chi-square statistic for the distribution of decorated graves within the cemetery. This distribution was found to be non-random and significant at the .03 level. Figure 16 displays this distri- bution in the form of a contour interval based upon the percent of decorated graves in each cell. A simply visual inspection shows that decorated graves are most frequent along in the eastern periphery of the site, particularly in the northeast and decrease in frequency rather sharply moving to the west and southwest. The distribution of certain design motifs found on graves within the cemetery was also tested despite the fact that it does not lend itself to a similar statistical analysis due to the small numbers of graves involved. However, Figures 17 through 29 Show the distribution of designs with the cemetery. Certain designs appear to have extensive distributions. Fifty percent contain design,O 33% have v , 25% have 0 . But even these designs concentrate in certain areas of the site. The (3 design for example, is concentrated in a broad belt extending from the southeast 85 Figure 16 CONTOUR DISTRIBUTION OF DECORATED GRAVES AT SITE 1 1 ' graves 2 meters or less in diameter 30 10 4° 50 ' ( HIGH LOW \ graves over 2 meters in diameter 86 s R maul F. E M 5'. 0L SITE 1 Figure 17 DISTRIBUTION OF 0 DESIGN [INDICATED BY HATCHiNG} 87 J S R m1 n E M 51 OL 63 «=8 69 Figure 18 SITE 1 DISTRIBUTION OF C9 DESIGN (INDICATED BY HATCHINGI SITE 1 Figure 19 DISTRIBUTION or (1) DESIGN [INDICATED BY HATCHINGI 89 0 5 10 20 ® I_ I l I I SITE 1 @ METERS Q @C’D @ Figure 20 DISTRIBUTION or c/" DESIGN [INDICATED BY HATCHING} 90 0 5 10 20 L I I I 1 (9 SITE 1®® METERS Figure 21 DISTRIBUTION or 9 DESIGN [INDICATED BY HATCHINGI 91 SITE 1 6, N. 1‘... ”I Q I V N. NK (.”’01'1 INQI / V Figure 22 DISTRIBUTION or 99 DESIGN [INDICATED BY HATCHINGI 92 O 5 10 20 @ ‘_‘—'—'—' SITE 1 @@ ths 61°99 69: @@ @® 639., GD 1 “03:0 £969.: 0 :%8®q€9:@® @®@®@@ .. s 1W “’96:? Figure 23 DISTRIBUTION OF A DESIGN [INDICATED BY HATCHINGI 93 SITE 1 ($66569 Figure 24 DISTRIBUTION OF L9 DESIGN [INDICATED BY HATCHINGI "I 94 AAETERS Figure 25 DISTRIBUTION or :65 DESIGN [INDICATED BY HATCHING} 95 ® , °. 69 Figure 26 .. E T N111 .. so 0 DISTRIBUTION OF (A) DESIGN BY HATCHINGI (INDICATED 96 2J S R m1 N... E M 51 CL SITE 1 Figure 27 DISTRIBUTION OF II DESIGN [INDICATED BY HATCHING] L__ 97 SITE 1®@ METERS ' 1D Figure 28 DISTRIBUTION OF @ DESIGN [INDICATED BY HATCHING] 98 SITE 1 @619 METERS Figure 29 DISTRIBUTION or ((4) DESIGN [INDICATED BY HATCHING] 99 portion of the site to the east. The ('9 design is concentrated in the northeast, whereas the OD is almost exclusively in the southeast. Although there is extensive overlap between these three designs, the area of concentration of each is clearly different. Designs which occur infrequently appear to be even more strongly centralized. The graves which contain the 9’ design are separated by less than five meters. Four of the five graves with then) design occur within a 10 meter area. The A motif occurs on two graves which are separated by only three meters. Similar concentrations are found for 9 ,C.) , J’ , and 9 designs. Only the @ and the (5) appear not to be strongly associated with any one portion of the site. Although these distributions cannot be tested statistically, due to the small sample size of most of the designs, the fact that 10 of the 12 design distributions tested tend to cluster geographically cannot be overlooked. Rather than being randomly distributed throughout the site, these designs appear to delineate areas of the site which share similar design motifs. However, it should be noted that these areas are in no way clearly defined. Conceptually, these areas sharing similar designs can best be viewed as clines with certain design elements overlapping where their areas meet. The CROSSTABS procedure was also employed to test the distribution of different size graves at the site. In the initial 15% excavation sample at the site it was observed that only males occurred in graves that were decorated and graves that were over 100 2.0 meters in diameter. Hence using only external grave diameter and decoration it appeared to be possible to predict grave content. The second 10% sample of graves that was excavated provided empirical support for the proposition that grave content could be predicted. The 2.0 m. figure for external diameter appears to be an important dividing point, above which only males were found and below which either males, females or children might be found. This observation was verified to some extent by the clusters derived from the ISODATA routine. Although more variables were used in this stage of analysis (see page 55 for complete list), the clusters which were derived verified the initial field observations. Clusters I through IV only contained males, were the only decorated graves and were also all over 2.0 meters in diameter. The Cluster V burials contained either males, females or children and with two exceptions were all less than 2.0 meters in diameter (138 at 2.2 m. and 39 at 2.2 m. were the only exceptions). As a final independent check of the validity of utilizing diameter to divide the burial population into two broad groups, a multiple regression analysis was performed of the excavated graves. All of the same variables used in the clustering routine were employed (see page 55). All variables were run in turn as the independent variable. Of all the variables grave diameter was found to best predict the remaining variables (R = .8) (John Davis, personal communication). 101 Grave diameter then appears to be a valid criterion for dividing the entire burial population into two broad groups: graves that are greater than 2.0 meters in diameter, and graves that are less than or equal to 2.0 meters in diameter. Figure 30 shows the distribution of graves over 2.0 meters in diameter. Once again a 12 cell grid was imposed upon the entire cemetery and graves over 2.0 m. were assigned X and Y coordinates. The program CROSSTABS revealed a non-random distribution signifi- cant at the .045 level. Figure 31 expresses this distribution in the form of contour intervals. The greatest concentration of graves over 2.0 m. occurs in the northeast portion of the cemetery, the lowest in the southwest with a gradual decrease in grave size as one proceeds from the northeast to the southwest portion of the site. The same method was used to check the distribution of graves 2.0 m. or less in diameter (Figure 32). Here the distribution was also found to be non-random and significant at the .023 level (see Figure 32 for distribution of these graves). The contour intervals of graves equal to or less than 2.0 m. based on the percentage of these graves in each cell reveals a pattern opposite of that found for the grave over 2.0 m. Here the greatest concentration was found in the southwest portion of the site gradually decreasing toward the northeast (see Figure 31). The possibility existed that the spatial patterns discussed above were Simply a reflection of the availability of the raw 102 @ H SITE 1 ER’R,@ METERS Figure 30 DISTRIBUTION OF GRAVES OVER 2.0 METERS IN DIAMETER [INDICATED BY HATCHING] 103 Figure 31 CONTOUR DISTRIBUTION OF GRAVES AT SITE 1 Q 215 /. graves 2 meters or less in diameter / 6° 50 80 HIGH LOW / /40 /6O graves over 2 meters in diameter 104 METERS Figure 32 DISTRIBUTION OF GRAVES 2.0 METERS OR LESS IN DIAMETER [INDICATED BY HATCHINGI 105 material for grave construction. The outcrop on which the cemetery was located contained an abundant supply of lava exfoliating in regular slabs about 10 to 15 cm. thick. These slabs were used in the grave construction both for the vertical perimeter and the horizontal layers. However, not every grave in the cemetery was equidistant from available slabs. For example, graves on the eastern periphery of the cemetery were the closest to the raw material. Those in the west and in the center of the cemetery were further away. If grave size was determined by proximity to available lava slabs there should be a direct correlation between the distance from the raw material and grave size. To test this proposition a simple linear regression was run with the distance from lava slab outcroppings as the independent variable and the weight of rock used in the construction of a particular grave as the dependent variable. The weight of rock used was felt to best represent grave size since it included both diameter and depth variables. All of the excavated graves were used in this linear regression. This analysis showed almost no correlation between grave tonnage and distance from raw material (R = .4). A simple linear regression was also run using the distance from the raw material as the independent variable and average size of slab utilized as the dependent variable. The only data on the Size of slab used in the construction of particular graves came from the outer circle of standing slabs. The length of these was measured for each excavated grave. These lengths were then averaged 106 for each excavated grave and compared with the distance of the grave from the lava outcroppings. Once again the analysis showed no correlation with distance from raw material being of little predictive value in determining the weight of stone in a particular grave or the size of the stone employed (for stone Size R = .36). The spatial patterns observed at the site then appear to be independent of resource availability and as such would seem to be culturally significant. The distribution of graves within the Namoratunga l cemetery appears to be markedly non-random (the Namoratunga 2 site was not large enough to apply a similar spatial analysis). Graves over 2.0 m. are strongly associated with the eastern portion of the site and conversely, 2.0 m. or less are more common in the western portion of the site. Decorated graves of all Size diameters are significantly associated with the eastern half of the site. The cemetery seems to be divided into at least two broad areas, one in the east and the other in the west. There is no sharp boundary between the two. The differences appear to be more of a gradational nature (see Figure 31). Simple visual inspection of the location of the cemetery areas on their respective outcrops tends to reinforce this apparent preference for the east. The outcrop which contains the Namoratunga 1 cemetery as discussed earlier is oriented in an east-west direction. The very westernmost portion of the outcrop is unsuitable for the location of graves, with its raised rocky peak. However, the 107 remaining eastern half of the outcrop consists of a broad raised platform and provides a suitable area for grave construction. It is noteworthy that only the very easternmost portion of this raised platform was used. At Namoratunga 2 the outcrop is oriented north-south with the raised rocky peak in the southern half of the outcrop. Here the north half of the outcrop provides a broad raised platform. At Namoratunga 2 much like Namoratunga l the graves are located on the easternmost portion of the raised plat- form area. In neither case did the physical constraints imposed by the outcrops themselves dictate the placement. The final visual observation which can be made concerns the Namoratunga I cemetery and its overall configuration. The cemetery is roughly diamond shaped and is oriented with the four cardinal directions. This is particularly interesting when it is recalled that all of the burials were also oriented in one of the four cardinal directions. Petroglyphs as Cattlebrands: The Ethnographic Evidence As mentioned earlier, 34 of the Namoratunga l graves were decorated. Interestingly, when work was first initiated at the Lokori Namoratunga sites, the author observed a number of the same geometric designs found on Namoratunga graves branded on the live- stock of the local Turkana herders. When this apparent connection between the Namoratunga designs and Turkana brands was followed up it was found that the Turkana were able to recognize a large number 108 of these designs despite the fact that the sites and the art are not related to them (Lynch and Robbins 1977). In addition to claiming that they were not responsible for the art, there is no evidence of the Turkana having a rock art tradition. As such it seems unlikely that the art is the work of the Turkana. Since the art at the two Namoratunga sites was only associated with the graves of males, the art and cemeteries are clearly related and the work of a group other than the Turkana. A total of twenty Turkana informants from the Lokori area were questioned about the brands through an interpreter. The individuals questioned were all male Turkana elders who owned livestock and lived in isolated homesteads up to 11 kilometers from Lokori itself. The Turkana were interviewed in groups of from three to six individuals at a time. Each of the Namoratunga designs was drawn on the ground and the elders were asked if they recognized the designs as being Turkana. In no case was there a discrepancy in the information obtained from different groups. Of the 142 different kinds of geometric designs at the site, 99 (70%) were recognized by the Turkana as brand symbols which had individual names in their language. These signs were branded on cattle, camels, donkeys and other small livestock as marks of ownership. However, not all the animals owned by an individual receive the same design. Animals were sometimes differentiated on the basis of sex and species. For example, goats usually received a different design than cattle. Thus an individual can use more than one brand symbol. l09 Only males possess brand symbols. Among the present day Turkana, one of a male‘s brand symbols is inherited through the male line. As such, these symbols serve to delineate Turkana lineages. Lineages of the same clan while having different brands would identify them in the same way in terms of clan names. For example, men possessing quite different brands but belonging to the same clan, would still identify their brands by the common clan name of Ngikaleso. However, this ideal can be obscured by other Turkana practices. For a variety of reasons an animal may bear a number of different brands. One way this can occur is if an animal passes in ownership from one lineage to another, particularly if it is a head of breeding cattle. Animals may be sold, traded, or even given from one person to the next. Some designs may only belong to one individual who is wealthy. These are applied in addition to the lineage brand. Finally, besides ownership marks some animals, especially cattle, may have other designs which are burned in simply for adornment. A similar pattern was found among other East African tribes such as the Samburu, Pokot and Masai, all pastoralists. The Masai recognized 4% of the Namoratunga designs, the Samburu 9% and the Pokot ll%. However, it should be noted that for these groups it is likely that these percentages may be extremely low in comparison to the number of designs these groups actually have in common with the Namoratunga sites. Only two Masai informants were questioned llO while only one individual from the Samburu and Pokot was interviewed. The same interviewing procedures employed with the Turkana were used with these three groups. "Cattle, donkeys and small livestock bear ownership marks. These consist of cuts or brand lines on one or both ears and of lines and curves burned on the left side of the body. The marks if put on the body of cattle and donkeys show what lineage or sub-lineage they belong. Those on small livestock the owner chooses for himself. Probably every lineage has its own distinct marks" (Merker l9lO:l67, on the Masai). Among the Pokot (L. H. Robbins, personal communication) it was reported that a new brand could be obtained as a result of a cattle raid when the opponent was killed and his cattle were taken. This is one way in which similar brands become incorporated into different groups. This information serves to further support the use of these designs as brand symbols and their association with pastoralists. To the author's knowledge, none of these pastoralists alter the design already on an animal when they acquire it. They simply apply their own brand in addition to the one already on the animal. Because of this, one might expect there to be a certain amount of difficulty in readily differentiating livestock. However, this does not appear to be the case. Despite the presence on some animals of several brands these herders have no difficulty deter- mining the exact ownership of an animal. It is likely that much like the Nuer, (Evans-Pritchard l940) pastoralists in general become quite familiar with each individual animal in their herd. In addition, the decorations found on certain animals may also aid in distinguishing livestock with multiple brands. lll Some support for the assumption that the art may have been used in a similar fashion at the Namoratunga sites was provided by the fact that much like the ethnographic groups the designs are only associated with males at the Namoratunga sites. These graves also appear to indicate that the people responsible for Namoratunga were also at least in part pastoralists. As indicated in Appendix A, teeth of domesticated cattle and sheep/goat were common occurrances in the grave fill. It appears that the association of brand symbols with pastoralists and males in particular is quite ancient and that at Namoratunga the use of the designs at least in part parallels present day practices. Because of the similarities in the use of the art between the Namoratunga sites and the ethnographic groups, it is possible to make further inferences about the social organization of Namoratunga l and 2. As noted above, ethnographically the brands serve to delineate kin groups, in the case of the Turkana, patrilineages. In the previous section it was observed that graves sharing similar designs tended to be clustered within the site. On the basis of the ethnographic evidence it can be suggested that these represent individual kin groups within the larger cemetery. Such an explanation may in part also explain the geographical separation of Namoratunga l and 2. As mentioned earlier, these two sites are separated by less than 2 km. Namoratunga l contains l62 graves while Namoratunga 2 contains ll. There are no geographical variables to account for this distribution. The Namoratunga l site could have easily accommodated another eleven graves. This physical llZ separation seems to be due to social considerations. Perhaps the two sites represent separate corporate groups. This is verified in part by the designs which are found on the decorated graves at the two sites. Three graves were decorated at Namoratunga 2, l64, 169, and l70. Two of these graves, l64 and 170 contained three symbols which were not found on any of the graves at Namoratunga T. If the assumption that these symbols represent particular kin groups is a valid one, then obviously there is some social distinction between the two sites. However, as the ethnographic examples pointed out, the brand system has the potential for a great deal of ambiguity. An individual only inherits one design from his father. Additional designs can be acquired in a number of different ways as alluded to earlier. This may in part explain why some of the Namoratunga l graves contained as many as seven different symbols. However, despite the presence of large numbers of different designs on a single grave, underlying regularities still exist at Namoratunga 1. If the number of different designs per grave is examined in terms of grave size it becomes apparent that certain regularities can be found. The average number of designs on a grave correlates directly with grave size. This again fits the expected ethnographic pattern, since wealthy individuals or persons of high status often have more brand symbols (Merker lQlO:l67). The particular groups represented by these designs do not seem to be ranked in any way. All of the symbols found on the 113 TABLE l7.--Designs Used by Turkana, Pokot, Masai, and Samburu. Decorations Design Turkana Pokot Masai Samburu on Graves @ Eonai t X Q9 Ngi kaleso X X C/ Ngimeturona X )\ ngikatekok X(m) p ngiduya 9 ngi raraka 9 ngikaruok X C) ngikamosorok X X CD ngitaparakolong Q? ngipucho X X {)2 ngiduya £2: ngikaleso X m ngimacharikota X m ngikinom (/\J ngikaleso X 55%;; ngitogor ngigolereto fig} eonait :1; ngikinom X W ngidocha X p ngikaleso e ngigona TABLE l7.--Continued. ll4 Design Turkana Pokot Masai Samburu Decorations on Graves éfib ngikamosoroko ngikaleso ngimeturona ngiraraka ngitogor ngiduya ngimasula ngiraraka ngimacharakwat ngisiger ngikamosoroko ngikaleso ngikaleso ngikatek ngitogor ngikaruwok ngimasula Isa‘afi t—‘afiseaefi’DHQWfiéiC 9. ngikatap V n ngirarak ngisiger ekinomit ¥®Bxl epongait 115 TABLE l7.--Continued. Design Turkana Pokot Masai Samburu Decorations on Graves ngfififgbaifibb®.Z}X>-) @§%Bmz:i ekurerut ngimacharikota X ngimeturona ngitogor ngithinger ngimacharikota X ngikaleso ngirarak ngithinger ngimeturona ngikaleso ngimeturona ekurerut edongait ngikuruk ekar ngikureru ngithinger ngirap ngikamosoroko X ngitaparakolong X ngikaleso TABLE l7.--Continued. 116 Turkana Pokot Masai Samburu Decorations on Graves =0 waelbtellsé ea «Ami-Bi“ ngirarak ngikatap ngikamasorok ngichilla ngimasula ngipunga ngigorereto ngidogiro ngiponga ngimeturuna ngigolereto ngikureru ngikaleso emeturonait ngikamosoroko emedoit ekalesoit ngikatek ngitacha ngikureru ngithinger ngigona X(m) ><><><><>< 117 TABLE l7.--Continued. Decorations Design Turkana Pokot Masai Samburu on Graves ;§;§i ngikatap f/ ngiduya X 1‘0)” ngi thinger "]_‘ ngikaruwoko an ngigona X 3:5? ngikaleso Q ngiduya J ngi thinger X X A ngi rarak X gig; ngikaleso Total 99 (77%) l6 (ll%) 6 (4%) l3 (9%) * For Pokot, Masai and Samburu lineage names were not collected, (M) indicates from Merker, l9lO. The remaining designs were col- lected by the author and L. H. Robbins. 118 TABLE lB.--Grave Diameter vs. Average Number of Different Designs. Grave diameter 3.5+ 3.4-3.0 2.9-2.5 2.4-2.0 l.9-l.5 Average number of different designs 2.9 2.5 2.25 2.1 1.8 larger graves also appear on much smaller graves. If the kin groups represented at the Namoratunga sites were ranked in any way with certain groups as symbolized by these designs being of higher status than others, we would have expected graves of similar size to share the same designs. Thech design for example, occurs on a grave over 3.5 m. in diameter but also on one less than 2 m. in diameter. The O , U , and ® , designs occur on all size categories. The symbols W and 9 occur in four of the five categories and w , @ occur in three. Twelve designs occur in at least two size categories. Nine are found in only one size category and all but two of these are idiosyncratic designs (V0 ). The same pattern was found when all of the decorated graves at Namoratunga l were examined. Much like what was found for the excavated decorated graves, graves of similar size did not necessarily contain the same designs. The very fact that cemetery areas exist at all may further support this hypothesis of kin groups. Saxe (l970) had originally proposed this in his hypothesis 8. 119 TABLE l9.--Grave Size vs. Symbol for Excavated Graves. 1.5 - 1.9 2.0 - 2.4 2.5 - 2.9 3.0 - 3.4 3.5+ 9 9 9 9 C) (3 C) C) C) w W W cm '11) Q? (9 © @ @ @ ‘59 139 (37 EU’ ‘19 L) i? u U 7? W .x" x‘“ 3E if ef" @ ® @ O o o ‘D 4 £2 AL "To the degree that corporate group rights to use and/or control crucial but restricted resources are attained and/or legitimized by means of lineal descent from the dead (i.e. lineal ties to ancestors) such groups will maintain formal disposal areas for the exclusive disposal of their dead, and conversely (Saxe l970:l19). Saxe found this hypothesis to be verified by his sample of three ethnographic groups. Goldstein (1976) elaborated upon this hypothesis 120 with an ethnographic sample of 30 groups. Based on this, Goldstein restated Saxe's hypothesis 8 A. To the degree that corporate group rights to use and/ or control crucial but restricted resource(s) are attained and/or legitimized by lineal descent from the dead (i.e. lineal ties to ancestors), such groups will, by the popular religion and its ritualization, regularly reaffirm the lineal corporate group and its rights. 9ge_means of ritualization is by the maintenance of a permanent, specialized, bounded disposal area for the exclusive disposal of their dead. B. If a permanent, specialized bounded disposal area for the exclusive disposal of a group's dead exists, then it is likely that this represents a corporate group who has rights over the use and/or control of crucial but restricted resource(s). This corporate control is most likely attained and/or legitimized by means of lineal descent from the dead, either in terms of an actual lineage or in the form of a strong, established tradition of the crucial resource passing from parent to offspring. C. The more structured and formal the disposal area, the less number of alternative explanations of social organization apply, and conversely (Goldstein l975:61). What she suggests is that if there is a formal bounded disposal area, used exclusively for the dead, then the culture is probably one which has a corporate structure in the form of a lineal descent system. The more organized and formal the disposal area is, the more conclusive this interpretation. 121 The use of a cemetery is certainly compatible with the hypothesis that kin groups, whether they represent either lineages or sub-lineages are buried within the larger lineage or clan cemetery. The distinction between Namoratunga l and 2 would be of the same order, although it is impossible to determine if this distinction is on the moiety, clan, or lineage level. However, in any case, the most important corporate group would appear to be that represented by the cemetery areas as a whole. The cemetery itself shows a number of organizational principles which have been discussed earlier. The cemetery displays an overall pattern with regard to burial orientations. A significant number of the Namoratunga l burials are oriented away from the center of the site. An individual's orientation apparently was determined by his or her location within the cemetery. It does not seem to reflect sub group affiliation. Burials with the same design on their graves do not necessarily share the same orientation. The same appears to be the case with respect to an individual's location within the cemetery. Graves over 2.0 m. tend to occur in the east and graves 2.0 m. or less in the west. The entire cemetery then appears to be roughly divided into two broad areas. On the basis of these spatial considerations it seems reasonable to suggest that this social unit took precedence over the smaller groups represented by graves sharing similar symbols. It seems that despite the potential for a great deal of ambiguity in the use of these symbols, it does not appear to obscure 122 the underlying patterning at the Namoratunga l site. All of the regularities observed can be explained in terms of present day uses of the designs. As such the ethnographic model appears to account for the variability encountered at the sites. The hypothesis that kin groups are buried within the larger cemetery, and that the two sites represent different corporate groups, cannot be rejected using the available evidence. Unfortunately at the present time it is impossible to fully test this hypothesis. Ideally burials within graves sharing similar designs could have been compared using non metric traits to try and determine genetic affinities as could burials from the two cemeteries (Buikstra l974, Lane and Sublett l972). Due to poor skeletal preservation this could not be done. However, as will be seen in the next chapter, the practices of particular subgroups being buried within the larger cemetery is not without ethnographic analogy. Among certain Eastern Cushitic speaking peoples now inhabiting parts of southern Ethiopia such as the Konso, clans have their own distinct burial area. These areas are further subdivided on the basis of lineage membership (Haberland l963). (See Chapter V1 for further discussion of mortuary practices among Eastern Cushitic speaking peoples.) Summary and Interpretation As discussed at the beginning of this chapter, the variability in grave size at the Namoratunga sites can be viewed in a number of alternate ways. Having discussed the Namoratunga burial routine it 123 is now possible to return to these hypotheses to discern if any fit the available data more closely than Tainter's (l975) suggestion that energy expenditure correlates with status. The first two suggestions can be treated singly since they are to a certain degree linked and would be reflected in a similar fashion archae- ologically. One hypothesis was that grave size reflected the status of the deceased kin group. The second was that it reflected the size of that kin group. In either case individuals would warrant larger graves by virtue of their membership in a particular kin group. Given either of these suggestions, one would expect graves of comparable size to share similar designs. It is assumed of course that these designs symbolize particular kin groups. An examination of the decorated graves at Namoratunga l revealed that graves of similar size did not contain the same designs. As such there is no support for either of these two related propositions. The final hypothesis proposed was that differences in grave size reflected who constructed the grave. If women constructed the graves of other women, for example, simply due to differences between the sexes in physical strength we would expect the graves of women to be smaller than those of men. This hypothesis is not supported by present day burial practices among East African pastor- alists. Among these groups graves are always constructed by male kindred regardless of the sex of the decreased (see Chapter VI) although women may sometimes aid in the work (Evans-Pritchard l956). secondly in the case of graves 9, 17, 37, women were interred in 124 graves considerably larger than those of certain males. There also appears to be no support for this hypothesis archaeologically. At least for the Namoratunga l data there appears to be no reason not to suggest that, as Tainter has shown elsewhere, variation in energy expended, as expressed in grave weight, can best be interpreted as differences in status (see Tainter l975 for complete derivation of this argument). The five burial groups derived from the cluster analysis should be viewed as forming a continuum in relative status, from group V which would have the lowest relative status to group I with the highest. As observed in the previous section, burial at the Namoratunga l and 2 cemeteries was confined to a segment of the total population. Females (less than 25% of burial population) and subadults (8%) appear to be interred elsewhere or were not buried. Access to the site than appears to entail a certain status; this status being limited for the most part to males. Females and subadults that were buried at the sites occupied the lowest status positions rela- tive to males given energy expenditure as a measure of status. Age also appears to be a determining factor in burial at the Namoratunga sites. The population profile from Namoratunga l is heavily weighed with adults. Only two subadults were found in the 38 graves excavated. In addition, the top status positions or largest graves were only occupied by middle adults. Young adults occupied smaller graves and lower status positions. The two children found belonged to the lowest status group at the site (Group V). While 125 the presence of these children may indicate that at least some status is ascriptive it might also simply reflect idiosyncratic behavior. Burial in decorated graves, while confined to males, contained both middle and young adults. Almost all of the dis- tinctions in the mortuary routine at the Namoratunga l site can be explained in terms of age or sex distinctions and so status is apparently achieved. This burial routine, then, would appear to fit an egalitarian social pattern (Service l962). One might suggest that the women and children who were interred at Namoratunga l warranted this disposal due to their special relationship with one of the males buried at the site. These females appear to accrue the social status of the male partner either through marriage or kinship. All of the female and subadult burials at the site were buried next to much larger male graves. Furthermore, the excavated female graves at the site are in close proximity to a decorated male grave. Graves 9, 76, l43 and l62, all containing females, are actually in direct contact with decorated graves. Graves l7 and 37, also containing females, are less than 2 meters from a decorated grave. Grave 3l, that of a child is also in contact with a decorated grave. The grave of 65, a child, is not near a decorated grave but is contiguous to a larger male grave. This may lend some support to the proposition that the status of females and subadults at the site was dependent upon certain males. 126 The spatial organization of the cemetery also revealed several patterns. It was found that there were significant associations between the east half of the cemetery and graves over 2.0 m. and the west half and graves 2.0 m. or less in diameter. Hence, not only was burial at the site indicative of a certain status, but status was further reflected by an individual's location within the cemetery. From the available evidence it seems that the Namoratunga cemeteries can best be visualized as being composed of a number of family units with high status males surrounded by their relatives. This is the pattern we see with the graves of women and children being found in association with larger adult male graves. Such a pattern is not incompatible with the spatial patterns found with grave size and decoration since even in egalitarian societies individuals of high status tend to some extent to have high status relatives (Service 1962). Finally, it also agrees with the clustering of graves sharing similar designs, if these do indeed symbolize kin groups. Unfortunately, since skeletal preservation was extremely poor, it was impossible to determine, using either metric or non-metric information, if graves which were in close proximity were more closely related physically. CHAPTER IV ART ANALYSIS Rock art, particularly geometric designs, have a wide distribution in East Africa. In Tanzania, paintings as well as petroglyphs depicting geometric designs have been found (Shorter 1967, Odner l97l, Tanner l953, Soper and Golden 1969, Chaplin l974). These Tanzanian sites contained circles, sun figures, concentric circles, spiral designs and beetle like figures (Collinson l970). Similar designs are also widespread in Uganda, particularly in the Lake Victoria region (Lawrence l953, l954, Lanning l960, Morton l964). Here, too, both paintings and petroglyphs were recorded. Until recently it was thought that Kenya unlike its neighbors, Uganda and Tanzania. had little rock art. Soper in his research at Namoratunga (l968) noted only a few other sites in Kenya which had art similar to that found at the Namoratunga sites. However, more recent research has uncovered a great deal of rock art in Kenya (Odak l976) "What can be said at this stage is that Kenya, though little known with regard to prehistoric art, is comparable with and perhaps surpasses the neighboring countries about which much has been done in this line of research" (Odak 1976, p. l0). Art is found in the Eastern Rift Valley, the west side of Lake Turkana, the Uasin Gishu plateau and the plains stretching southward 127 128 from Nairobi to the Tanzania border (Gramly l975). Finally, similar geometric designs have been found as far north as Ethiopia near the Red Sea (Graziosi l964). However, at most of these sites the number and range of these designs is limited. In addition, precise interpretations of this geometric art is extremely difficult and usually highly sub- jective. The work of Gramly (l975) at Lukenya Hill in southern Kenya is the sole exception. Here the geometric symbols that were painted on the rock shelter walls were found to closely resemble Masai cattle brands. The designs were apparently associated with Masai "meat feasting" sites where the designs painted on the walls represent the brands of the particular animals eaten. Chapman (l974) also implies that the art is connected with pastoralists, but does not link the geometric symbols with animal brands. For the most part the art at all of the sites in East Africa has simply been described with no attempt at any form of systematic analysis. In terms of rock art the Namoratunga sites are relatively unique in East Africa. In addition to providing a very large sample of rock art, over l,OOO different engravings, the sites mark an instance where the meaning of these geometric designs can be inferred. These sites also represent one of the only instances where the art can be linked to an excavated archaeological assem- blage, the accompanying cemetery areas. Since the art on the graves was discussed earlier, this section will only deal with the art on the outcrops which contained 129 the two cemeteries near Lokori. This constitutes the vast majority of the art (93%). As discussed earlier, these two eroded volcanic cones, l km apart, are the only elevated areas on the broad flood plain between the Kerio and Kangatet Rivers. No other occurrences of rock art were found elsewhere in the Lokori area despite the fact that the l9O sq km around the sites were intensively surveyed. As such, the two outcrops appear to be important rock art centers for at least the 190 sq km surrounding the two sites and may prove to be the only art centers in the entire region. One thousand seventy different engravings were found at these two sites representing l42 distinct geometric designs. The designs consisted largely of circle, spiral and line motifs, however, four examples of naturalistic art did occur; two giraffe figures and two figures which resembled single hump camels.* The petroglyphs were formed by taking a hard pointed lava hammerstone and pecking through the relatively soft exterior "skin" of the basalt slabs to the lighter colored interior. This art ranged from examples that were well patinated and with desert polish to instances where the art looked very "fresh." Because of this wide range in the relative amounts of weathering, the art was broken into three broad categories. These groupings provide a rough relative chronology for the art. The first group consists of all examples which appeared to be * If these are indeed camel representations, they mark knowledge of camels by 330 BC. These may well represent the only "dated" evidence of them in East Africa. 130 "fresh.“ These designs were much lighter in color than the weathered rock surface they were picked through and were not patinated (see Figure 33). The second category contained rock art that unlike the first category displayed some patination. The second category was darker in color than category one but was still lighter than the parent rock surface. This group exhibited no desert polish (see Figure 34). The third category represents the other extreme and contained art which waspatinathmiand had desert polish. These examples, for all intents and purposes, displayed the same color, patinated and desert polish as the original rock surfaces through which they were engraved (see Figure 35). The second group was by far the most difficult to assign of the three, with the first and third categories being readily differentiated. However, since one individual categorized the art, it is felt that judgments about the groups are at least internally consistent and as such are still of value. Viewing these groups chronologically, category one would be the youngest with category three the oldest. Unlike other areas where marked stylistic changes can be traced through time (Newcomb l976), the Namoratunga art displayed no stylistic variability through these three time periods. The possibility that these three different categories reflect the fact that certain engraved rock surfaces were simply more exposed to the elements than other surfaces is not likely. In many cases all three stages of weathering were found on the same rock surface (see Figure 36). In addition, well patinated mt .._ .- Figure 3 Example of Light Weathering Example of Moderate Weathering Figure 35 Example of heavy weathering Figure 36 Example of more than one type of weathering on same rock surface 133 engravings were often found in "protected" areas whereas "fresher" engravings could be found on more exposed surfaces. Despite the fact that the Turkana were not responsible for any of the art (see page 108), they were still able to recognize 99 out of the l42 different designs found at Namoratunga (70%) as brand symbols (see Chapter III, pages lO7 to llO for complete discussion of the use of brands). Each of the 99 designs were given lineage names in their own language (see Table l7). The Pokot utilized ll% of these designs, the Masai 4% and the Samburu 9%. As mentioned previously all three are also East African herding peoples. Once again, assuming that this art was used in much the same way at the Namoratunga sites, as lineage markers, it becomes possible to use the art on the two outcrops as a key to further interpreting the past social organization reflected at the Namoratunga sites. As mentioned earlier, the two Lokori sites are separated by approximately l km. The two sites contained l62 and ll graves respectively. Geographical constraints do not appear to account for this distribution, since either of the two volcanic outcrops could have easily accommodated all l73 graves. As such this distri- bution appears to be determined by cultural practices. If different subgroups within the larger social unit used the two cemetery areas, such as distinct clans, one might hypothesize that the two sites would differ in the designs they contained. The designs could thus be used as an index or rough measure of social distance. The greater the difference in the designs at the two sites, the greater the TABLE 20.--Key to Design Motifs. 134 OGDVO‘Ul-DWN-d eugmolkwcaES‘bfi-OGOGGG‘orcee ‘9 N .b 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 4°) .wewaewow 0‘" u a \\ 0 3.3 bmfie/égh 49 U“ Q ]@%%£p§e$om§£@owJ+J\©j‘ Ex: §@ 51 52 53 54 55 56 57 UT CD 05 0 \I 0) “12.3191: 9‘!»ng g gags N “gamma 135 TABLE 20.--Continued. 97 0—9 113 f? 129 V62) 98 film 114 0 130 || 99 :0: 115 911' 131 A 100 a], H6 100 132 "170 101 W 117 3’2 133 /ID 102 @ ll8 -O- 134 if 103 , 119 ~29 135 W 104 9 120 flay l36 44> 105 fi? 121 p 137 i? 106 122 z 138 A 107 l’ 123 G 139 017:) lOB M 124 (fig 140 €13, 109 6’ 125 2%: 141 2 110 5 l26 CI) 142 f3 111 Q 127 (a 112 0% l28 3 *Designs l30-l42 did not occur on the hills and were unique to graves. 136 social distance between the people responsible for the sites. The following analytical procedures were employed to examine this problem (see Table 20). Step 1: Independent Binomial Probability Distribution The first stage in the analysis was to determine if the two sites differed significantly in the designs they contained. To examine this problem a model was formulated to calculate the probability of all of the occurrances of a particular design occurring at one site only. The Independent Binomial probability distribution was used. It can most simply be expressed by the formula _ _ n-l P(Y — N) — 5 Where P = probability of success N = number of trials or sample size Y = is a random variable describing the number of successes in N trials or in the sample. So for example, for design l which occurred 3 times at site l and 3 times at site 2 in the heavy weathered category we would have or .56“ = .0125. This procedure is followed for each design .sn-l motif in each of the three weathering categories. These probabili- ties are in turn summed for each category of weathering and divided by the number of designs. This gives the mean probability or 137 TABLE 2l.--Distribution of Designs by Weathering Category. Hill 2 Hill 1 L W M W M W L W H W H W Design # 27 19 21 56 14 22 29 l0 50 61 10 10 16 10 11 11 12 13 10 14 15 16 ll 13 17 18 l9 20 21 22 138 TABLE 21.--Continued. Hill 2 Hill 1 L W M W M N L W H W H H Design # 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 139 TABLE 2l.--Continued. Hill 2 Hill 1 M W L W H W M W L W H W Design # 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 14 63 64 65 66 140 TABLE 2l.--Continued. Hill 2 Hill 1 M W L W H W M W L W H W Design # 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 141 TABLE 21.--Continued. Hill 2 Hill 1 M W L W H W M W L W H W Design # 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 142 TABLE 21.--Continued. Hill 2 Hill 1 L W M W M W L W H W H W Design # 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 143 Figure 37 Distribution of Designs for Site 1 of Designs Number Number 01 Designs Nu'mber oi Designs 20 i - e 163 12: Md = 2.0 ,_ : 3t 3: “i I e e 4 . - e - C O a e e e O 4 - 6ffj; 1 1 1 éj 1 51121 5 Iéj ‘20 Frequency of occurance 16 : 12 3 Md = 2.0 m i '2 8 : g —l o 8 4 Z . 2 - e e - e o .. 0 4 8 12 16 Frequency of occurance 16] . 12j ' - Md =1.0 : 5:" 8 .. 1 .9. 4 .1 1 e O C O '1 €1j1111é' I 1‘12j1 1116 Frequency of occurance 144 Number of Designs Number oi Designs Number of Designs Figure 38 Distribution oi Designs for Site 2 12 hick-9.0 8 e 4 e O C .0” C. C O C . 615551 r1116! 1 Imii T T§2T1 T4tor1 118' I [5'65 '62 Frequency of occuronce 34! O 1 241 Md=l.0 163 83 . O: . e'e eeee ee e e 0 8 16 24 32 4O 48 Frequency of occuronce 12 . 3 . Md=l.0 4 e O C 0 8 14 20 Frequency of occuronce HEAVY MODERATE lIGHT 145 expected probability of finding all of each design at only one site. The find formula is Where I = the total number of designs in each category. Probabilities for finding all of each design at one site. Heavily weathered 42.45034814/72 = .5895881686 .590 Moderatelygweathered 39. 40930224/62 = .635633907 .636 Lightly weathered 21.l9543457/32 = .6623573303 .662 Total 74. 7684491/113 = .6616676912 .662 This method was first employed for the heavily weathered designs. The chi square technique was used to test if the expected frequencies of designs differed significantly from the actual frequencies of designs at the two sites. Heavily weathered (45 - 42.48)2 = = .3545 42.48 df 1 at .05 C = .3841 146 One Site Only Both Sites N Observed count 45 27 72 Expected count N(.590) N(l-.590) 42.48 29.52 For heavily weathered category the difference between the two hills was not found to be significant at the .05 level. For the moderately weathered category the null hypothesis of no difference was rejected. Therefore, it is concluded that the two hills are significantly different in the designs they contain. Moderatelngeathered One Site Only Both Sites N Observed count 49 13 62 Expected count N(636) N(l-.636) 39.432 22.568 x2 = 6.378 df = 1 at .05 = 3.841 Lightly Weathered One Site Only Both Sites N Observed count 27 5 32 Expected count N(.662) N(l—.662) 21.184 10.816 x2 = 4.72 at .05 c = 3.841 As was the case for the moderately weathered examples, the null hypothesis of no difference between the two sites is rejected. The two sites differ significantly (.05 level) for the lightly weathered category. 147 However,when all the designs are examined, regardless of weathering category, the two sites are not found to differ signifi- cantly. Total One Site Only Both Sites N Observed count 81 32 113 Expected count N(.662) N(l-.662) 74.806 38.194 x2 = 1.517 C(at .05) = 3.841 The conclusions drawn from the analysis are interesting when viewed in terms of differences in time. Through time the two sites become increasingly differentiated in terms of their designs. Given the initial assumption that differ- ences in rock art indicates social differences, it would appear that initially the two sites were quite similar in the social groups they represented. Through time, however, this social distance between the sites increased. Hence, the initial hypothesis that the difference between the two sites is culturally determined appears to be validated with different kin groups within the larger social unit utilizing the two cemeteries. Step II: Pearson Product Moment Correlation The two sites were also compared in terms of the relative frequencies of designs they had in common. To facilitate this the relative chronology for the art described previously was utilized. 148 Hill 1 (Frequency oi Desigml Hill 1 Hill 1 16 .15 10 Comparison of Frequency of Designs at Sites 1 and 2 Figure 39 . C 0833' IFTII'11erIIIIIFIIIIIIITrIIIIrW 243240485662 0 8 16 Hill 2 (Frequency of Designs) M 0.. O 0 '3. 8 16 24 32 40 Hil|2 0 10 Hill 2 ’IIYUTVrrIF1 20 ’ItFIITTIlTITIIUFIITTIIIYII 50 HEAVY MODERATE lIGHT 149 For each of the three categories of weathering, termed heavy, moderate and light, a Pearson product moment correlation was performed between the two sites (Roscoe 1975). So for example, weathered examples at site one were compared with weathered examples from site two. Because of the close proximity between the two sites, 1 km, it was hypothesized that there would be a high correlation between the two sites through time in terms of the relative frequencies of shared designs at both sites. The resulting correlation coefficients were .800, .605 and .052 for the heavy, moderate and light categories respectively. For the heavy and moderate categories the most frequent designs at site one are also the most frequent at site two. This correlation is not evident for the light designs. This indicates that in terms of frequency of occurrence of designs there is a progressive weakening in the relationships between the two sites through time. Step III: Paired Differences Technigue To test the correlation derived from the Pearson product moment correlation for the heavy and moderate categories the paired differences technique for testing the difference in the two means was performed. Both heavy and moderate categories showed signifi- cant differences (99th percentile level) with site two having con- sistently larger frequencies of individual designs. However, it should be noted that the Pearson product moment correlations for sites one and two disappears for the heavy and moderate categories if the five most common designs at the two sites 150 Figure 40 Distributions oi Paired Differences: (H1) ' H2,” 4 - 1- _ 24: 2.53 815 . $1 18} 25" 123 ' W 2 6: . . 0: e e e e eee.ee° .e e -46T I I I .3I4I T I .22I I I I IqTOI I I I0 8 (X1'X2) A -353 £5.55 on 8 X .. ’1‘— i O O O O O . ”.0 0. v o w fifi l I I IT r I I I I I I I I I ITI I I I I I Ii 1 ~45 -34 -22 -1o 0 8 (XI-X21 A -.088 i 2.52 N 3 ' ’5 . >2" - . ,. w o T I I I 1 I I I I I I I I I I T I I T I I Iul VT I I 1 -46 -34 '22 -1O 0 8 (X1 - X21 HEAVY MODERATE llGHT 151 m.mm Ne m.mo emN o.¢c Fem Nme Pmpop Fmauu< o.o o F.e NP m.o p mp um ¢.N _ w.¢ 8p m.o m mp mo o.o o e.N N m.p 6 mp om m.¢ N “.8 PF w.m mp mN up P.“ m N.p m N.P 8 NF mp 0.0 o ¢.m op —.N m up 8? m.m v 8.? e N.m F_ mp Np m.m e o.m_ om m.N. pm mp, op m.a e o.N o m.N m mp N e.N F m.u NN m.w mN Nm 8 m.e N _.N _N 8.0, mm mm m m.m e N.m RN m.m mP om N N prz m.m¢ NN o.m¢ we m.NN Rep mm Page» 111. .111 111. 111. Fmauo< m.op c _.m o o.o o o_ om m.m_ m m.~ _ ~.N 8 PF up o.o o F.o v p.¢ o op NF m.oN oN N.m_ oF ¢.op up an op o.o o N.m_ op N.o P FF 8 m.m N m.~ P m.m «P N— m _ ppm: N szuu< acme; N _m:pu< mpmcmuoz N pmzpu< umgmnpmmz Pouch * commmo $6 .cmcu $0 .awcm mo .cmcu .xgommumu mcwcogumoz an common 4o aucmzcmcu11.NN u4m0; .0 do Nb 5 S K G .V ”3 .V we ow mu 156 Figure 41 _ DISTRIBUTION OF ART AT SITE 1 L 1 . ‘51" 19° comoun lNTERVAI. - IO METERS METERS csmercar ART AREAS 157 The four areas with art were compared to see if there were significant differences between the art in each area. TABLE 25.-~Distribution of Art at Site 1. Design No. Greater Than 9 West East North South Total 10 9 22 4 O 35 14 0 20 O 0 20 48 ll 2 0 O 13 4 10 l 0 0 ll 17 2 7 1 O 10 36 10 O O 0 10 12 O 10 O O 10 34 0 9 0 O 9 Total 42 71 5 O 118 2 I J f2 ij X = N [ Z Z -1] i=1 j=1 fi fij 22 = 1.666 - 1.0 = .666 X 118 = 78.588 df (3-1) X (8-1) = 14 2 = 29.141 99th percentile X Reject Hypothesis of Independence Of the four areas, the southern quarter had almost no art. A chi-square test was first performed on all of the designs having total frequencies of 9 occurrences or greater (see Table 24). This indicated that the location of designs was not independent of spatial areas. The four areas differed from each other significantly 158 (99th percentile). In addition to being spatially distinct the four areas also differed in the designs they contained. Table 25 shows that designs l4, l2, and 34 occurred only in the east while designs 48, 4, and 36 occurred only in the west. Of these six, 12, 36, and 48 displayed this pattern in two time periods while the other three designs occur in high frequency in only one time period (heavy weathered). The designs which occurred less than 9 times also displayed a similar pattern. They, too, are spatially segregated. Tracing these patterns temporally is more difficult. In many cases the designs remain spatially segregated through time. However, in other instances a design appears in only one area and for only a single time period. Once again, time is measured in relative terms using the broad weathering categories discussed earlier. At site two the spatial patterns were not as distinct. Here the art on the outcrop occurs in three distinct geographical areas, to the south of the cemetery, to the west and to the northeast (see Figure 42). The west area contained much less art than either the northeast or south. These areas were separated by areas which contained no art, even though suitable rock surfaces were present. Unlike site one, the designs were not found to be spatially segre- gated. All but five of the designs were found in both the south and northeast areas. This lack of patterning may reflect the lack 159 Figure 42 DISTRIBUTION OF ART AT SITE 2 Cemetery 0 ART AREAS Contour 0 METERS Interval = 10 meters 160 of spatial organization of the site two cemetery. Here unlike the cemetery at site one, the site two cemetery was not organized in any particular manner. However, the cemetery at site two was extremely small (11 graves). The burials at site two were oriented the same as site one, in one of the four cardinal directions indicating a shared mortuary routine with site one. The art was also examined to determine if any patterns could be found in which designs occurred on individual rock surfaces. A single large rock surface, for example, could contain as many as 17 different designs. To facilitate this 50% of all the engraved rock surfaces at site one were examined with only heavily weathered engravings being utilized. A simple pattern matrix (see Table 26) revealed that the art was apparently randomly placed on particular rock surfaces. It appears that spatial con- sideration for the designs are important only in terms of what area of the site the art is located in and not in terms of what particular surface the art is engraved upon. Summary and Conclusions In summary, there was both inter and intra site variability in the art. First, there was a significant difference in the art found at the two sites for the moderate and light weathering categories. It appears that different subgroups within the larger social unit as represented by these symbols are to be found at the two sites. Through time the art at both hills becomes more divergent 161 N RN _. FF mmumm F 0’) 050 P x55“ P r- P F r— '— Nr—N MQPr—r—Q F F F F ,.... '1.- F P r—NMVLOKDNQ) «FF mm mm Nw Fm cm as ms co mo om mm mm om me me me me Ne mm mm em om NN mF NF NF 9F mF eF mF NF FF OF m m s m e m N F .“.”“.Huufia"l} 11‘1““ “H""Uld..""lls“u n 11111”... .5555 seem Ne mememea to eoFeseFeemFo--.eN uem~ §':‘\ \ 1 .3“ iffy-:1.) I ___—-x\\ '1'. s ' V .‘1 7S "” “00-0-5“ “.. I .\ ~ ”.T‘h“ ‘l‘s ' '\ “ . ‘s‘l, \L~ I -II[ N Y II III Age Categories Figure 47 Mortality Proiile oi Namoratunga Sites Compared with Four Prehistoric American indian Populations 186 >uo.oe0..u.< {03.0.2 09.30.2202 .0 >.0EE:m 3 2...... 0.. oc/\ne> m..//\ue> o../\ue> oc/\ue> 0: =0... =0... =0: :0: =0... .33 .33 .3? 30m 30m .23... .e.eE0... .230... .e.eE0._u .2200... 33.—.0... m. w :. EON“ :. Eodw 5 EON“ 0. EDNA 120.930 5 EOdA 30.0 20.0 30.0 30.0 30.0 30.0 EESE. 285.2... .ceE.e.... 3E3...- >.0E.... EOE... >22.— {01:35 3.31.0.5... . 2.32.013 V 2.22.003 V < c2220 3.09.5“. 3.220 3.003.. 3.02 veton 32 3.30 < <0ZDF<¢Oi530 I c. 30... 187 Turkana According to Gulliver (195l), burial among the Turkana is extended only to certain classes of people, namely, heads of families and all married men, and all married women. However, based on information from local elders in the Lokori area collected by the author, regional differences would seem to exist. At Lokori, the Turkana claimed that all individuals warranted interment. When the head of a nuclear family dies (a fully independent adult man) he is buried in the center of the central goat kraal of his homestead by his sons or brothers. "He is laid on his side, head pointing east" (Gulliver 1951, p. 227). "The body is deposited in a shallow trench together with blood, meat, milk, water, tobacco and grain. According to the importance of the dead person in life, a sheep, goat or other domestic animal is killed and eaten by those present sitting in a circle around the grave" (Emley l927, p. l86). Very old women who are grandmothers or senior wives are likewise buried in the central kraal by their nearest male relative. A full wife, a married woman who is a mother, is buried under her day hut which is pulled down over her (Gulliver l951). "Young and middle aged married women are left in the hut, which is broken down over the corpse" (Emley 1927, p. 186). All married women are buried barren of any ornaments which are distributed to female relatives. Although not stated in the ethnographic literature, the author was informed that in addition stone cairns were constructed over these burials. 188 All other dead people are put outside the homestead. Gulliver (l951) was informed that they were either just left in the bush, or a grave was dug and large stones and brush piled on top (Gulliver l95l). Here too some regional variation may be in evidence. In the Lokori area in all cases cairns were constructed over interments for the expressed purpose of keeping wild animals from the bodies. However, further to the north where suitable stones are not available for cairn construction individuals are often placed in erosional gullies or simply placed in a sand blowout area. Exceptions to the above rules occur where the head of a homestead dies away from the homestead and is buried out in the bush and when males are killed while raiding and are left unburied at the scene (Gulliver and Gulliver l955). In summary, the components which in various combinations define the disposal type reflect the following social identity categories: (1) Married or not, (2) age in a broad way (i.e., child vs. adult), (3) sex, (4) degree of importance, in terms of respect and wealth. These distinctions are based largely on burial location; burial within the homestead at various locations vs. burial in the bush (Figure 49). m Unfortunately very little ethnographic information is available on the burial practices of Highland Nilotes. The infor- mation on the Pokot was obtained from M. Robbins (personal communi- cation). At the death of a man (married?), the body is first washed 189 0.2.2.; .0100 0000.5... 3 .52. 302.000.. .0 .020 00.002000 A 2.30 0. 2.00 32.6 __< o 5... >00 .005. .0..00u .00... .0200...» .00.: .0..00U .30 2.00 0. 5.00 0. 5.00 0. 5.00 02003 «3.3 3.00... 002 002 t0...02 .0 22.30.0020 00.202 22.00.... «0.200.. 00.02 /\\ nu—D‘v< 053.02 < 000... d .250 >000... a .250 >000 00 0.6.0 02 22.5 ..< 22.30.0020 u.2..0 ..< 22.0.0090 >000... d .0200 .220“. .0 000000... .0 0000/\ 00.50 >.0.00.00 00000... >.0.00.00 00000.. >.0.2000 00000... >.0.00.00 00000... .02 .000..0.0. >.00..... .000..0.0. >.00..... .000..0.0. >.00..... .000..0.0. >50.me 00.2050: 00...02 200.0 ..< 2%.... 00.05....— /£\ 020...... 50.3 05.02 < 00 20.. 202 organization (Binford l97l). However, beyond this very general level, there are clear distinctions between the various models. When the Namoratunga burial patterns are compared with the ethnographic sample in terms of mortuary distinctions symbolized, only the Masai and Konso upon initial inspection provide good fits. Both symbolize the age, social position and social affiliation of the deceased. However, upon closer inspection, the Masai clearly have to be excluded. Although the Masai do symbolize social affiliation, it is only in the case of one lineage. Secondly, unlike Namoratunga the Masai do not bury the vast majority of their dead nor do they have clearly defined disposal areas. The Konso provide a much tighter fit in terms of symbolizing social affiliation. The Konso have disposal areas which are based entirely upon social affiliation with an individual being buried in his lineage burial ground. The only possible disparity between the Konso and the Namoratunga burial patterns concerns the distinctions in burial based upon the sex of the deceased. At the Namoratunga sites burial distinctions are clearly made on the basis of the sex of the deceased. This distinction is not as clear among the Konso. However,among the Konso, a distinction is made between where males and females are interred. If married, a female is buried with her husband, if not, she is buried with her father's lineage. Un- fortunately ethnographic data on Konso mortuary distinctions is not detailed enough and as such it is impossible to ascertain if the 203 mm:=.m.osmz .mm..;m:u 0.0.mmm. omcog .mm.o._z mucm.;m.=. .oxo. .mmpo..z mzmmPQV .mmmz .mmuo..z m:.m.a. m:m..=» Ammuo..z 0.0. a 00>.m. .mzz co..m.....< .m.uom co.p.mo. .m.uom ummmmumo .o xwm nmmmmumo .o mm< 50mm. .o :o..muo. 50mm. .0 mmamu wsmz .‘l' lll‘:|‘l.'l. ll .cmN..oas>m mco..uc..m.o xgmzu.oz .o >0msazmuu.m~ 000

emmmca coon op mac :mxmu ma cpzou mpcmemcammme Pmpcmcu umoa ozra .mpmaup>wn=w me we mpasmm a Eogua -- -- m.mm m.mm -- -- em mm -- -- mmcmpnam .auwawuuo om -- mop mop -- -- No, mop -- -- ugogu Papwawouo Km -- m.¢op FFP mm -- -- PF, -- mm UL< ,mpmawuuo NN. -- m.mop m.~FP ¢o_ -- PP, eF. -- cop ecogu Paoapeaa mm, -- mp. m.omp mpp -- cm. FNP -- m_P ue< Papawcma Sm -- m~._m Pm em em Pm mm -- em mmcmpnzm Pmpcoca mpp -- m.mp. m.mp. -- -- EFF _FP -- -- ecocu Peacoca mm, -- m._mp m.om_ mm? -- Pm_ omF -- mm_ uc< Fmocoea mm -- mm mm -- -- mm -- -- -- .2m “Page 8cm; mm -- mm mm -- -- mm -- -- -- Seawaz “Page “Lad NSF -- _Np FNF -- -- _m_ -- -- -- mammcm-=owmam -- -- Km Km -- -- Km -- -- -- .2m PEHEOEL Ezswcwz -- -- NF_ NF, -- -- NF. -- -- -- .2m Paococa s=swxmz -- Ne, c.5mp m.F¢P mNF we. Amp -- -- mN_ .Em E=s_xmz mm_ om, N.om_ ow_ MAP mm_ mm, mmF Nm_ MAP spasm; szspxmz Aacaxcahv Aeemp xswccm_: _ESOL cam: op _op om mm m <~ asmcu , Ammowgmao ccmummuv mppmo III. II'III II‘III II 5 . E. II ‘ 5 .mcopmpmxm mmcaumcosmz 502$ ‘55: -55 5551.5- mucmEmgammwz mecmLUnn.Pm MAm N.mm m.nm F.mm o.mm ¢.om m._P o.o~ w.em ~.Np ucmm mcwu m.mm m.~m ¢.~m w.Pm e.o¢ ¢.~e m.mm ~.e~ N.om ccmm Eamum: m.m m.¢ m.m m.m m.m— m.~m c.m~ m.o P.wm ucmm mmgmou N._ —.N o.~ o.~ m.m o.m p.m m.m o.—_ ucmm mmcmou Em> Nu.wm &N.wm &N.mm xo.mm xe.~m fim.mm &w.mm xm.mw &m.mm ucmm EN 5? Eoom EF Euom Euom zoom zoom Euom :pqmo soomp Eoomp soomp soomF Eoomp soow Eooc Amuwm we muwm mo ummmv Eoom mmcm :gmummm .I'I' 55-5-i:‘--7551’-!91‘ -.-L'|‘l.‘l-“lt‘ I'IIIIIII' .Lm>wm owcmx mg» op _ consumcoamz Eogm pummcmch ummzuummm cm :6 umuumppou mmpasmm Pmomna.mm m4m