Holocene Foragers, Fishers and Herders of Western Kenya 9781841714172, 9781407324227

Citation preview

BAR S1037 2002

Cambridge Monographs in African Archaeology 54 Series Editor: John Alexander Assistant Editor: Lawrence Smith

KAREGA-MŨNENE: HOLOCENE FORAGERS, FISHERS AND HERDERS OF WESTERN KENYA

Holocene Foragers, Fishers and Herders of Western Kenya Karega-Mũnene

BAR International Series 1037 2002 B A R

Cambridge Monographs in African Archaeology 54 Series Editor: John Alexander Assistant Editor: Lawrence Smith

Holocene Foragers, Fishers and Herders of Western Kenya Karega-Mũnene

BAR International Series 1037 2002

Published in 2016 by BAR Publishing, Oxford BAR International Series 1037 Cambridge Monographs in African Archaeology 54 Holocene Foragers, Fishers and Herders of Western Kenya © Karega-M nene and the Publisher 2002 The author's moral rights under the 1988 UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher.

ISBN 9781841714172 paperback ISBN 9781407324227 e-format DOI https://doi.org/10.30861/9781841714172 A catalogue record for this book is available from the British Library BAR Publishing is the trading name of British Archaeological Reports (Oxford) Ltd. British Archaeological Reports was first incorporated in 1974 to publish the BAR Series, International and British. In 1992 Hadrian Books Ltd became part of the BAR group. This volume was originally published by Archaeopress in conjunction with British Archaeological Reports (Oxford) Ltd / Hadrian Books Ltd, the Series principal publisher, in 2002. This present volume is published by BAR Publishing, 2016.

BAR

PUBLISHING BAR titles are available from:

E MAIL P HONE F AX

BAR Publishing 122 Banbury Rd, Oxford, OX2 7BP, UK [email protected] +44 (0)1865 310431 +44 (0)1865 316916 www.barpublishing.com

PREFACE The problem of subsistence has received little attention in East African archaeology. Various models of human subsistence strategies have been constructed and a linear chronology from a hunting-gathering economy to pastoralism and agriculture has been the dominant conceptual framework for the research in the last few decades. It is the contention of this monograph that this overarch model masks the subtle and perhaps overlapping true nature of a mosaic of adaptation to the local resource base. A broad approach, involving examination of the transition from food collecting to food production as a process rather than as a single event is adopted. The approach also involves the examination of several causes of culture change in the region. It is anticipated that this approach will enable us to better understand the subsistence strategies of the human groups who occupied the Gogo Falls site in the Lake Victoria basin during the Neolithic and Iron Age periods. The opening chapter introduces the research problem and discusses theoretical issues relating to the advent of food production in East Africa and the theoretical framework that is employed in interpreting the artefactual and faunal data explicated in the study. Chapter two reviews previous research on the Neolithic and on the earliest manifestations of the Iron Age in the region. The chapter also addresses the shifts in definition and paradigm of especially Neolithic research through time. Chapter three presents information about the environmental and climatic setting of East Africa during the Holocene, that is, the last 10,000 years. This information provides a general ecological framework within which the economic behaviour of the site’s inhabitants is examined. The geographical setting of the Gogo Falls site is described in chapter four. The chapter also discusses, albeit briefly, the history of human settlement in the Kenya Lake Victoria basin as well as modern land-use patterns in the region. Chapter five describes the history and nature of the archaeological investigations that have been undertaken at the site. Chapter six describes the artefactual and faunal data recovered from the site and also discusses the approaches used in analysing those finds. Chapter seven presents the faunal data – taxonomic representation, relative abundance and age spectra of the taxa identified in the faunal assemblage, representation of various anatomical elements, and incidence of bone modification by both humans and animals. These data are employed in Chapter eight to reconstruct the taphonomic history of the assemblage and the subsistence economies of the human groups who occupied the site during the Neolithic and Iron Age periods. The chapter also discusses the butchery and culinary practices that may have been adopted by the site’s occupants, and offers a brief assessment of the environmental conditions that may have prevailed at the site during its occupation, as suggested by the faunal data. The concluding chapter evaluates the validity or otherwise of making inferences about the subsistence activities and the linguistic and/or ethnic identity of the people who occupied Holocene East Africa from pottery ‘wares’. It also assesses whether pottery can be used to build a reliable chronological framework for the region. This is done using the artefactual, faunal and chronometric data currently available from the site and data from other sites in East Africa. In addition, the chapter offers conclusions and a perspective on future research and suggests there is need to seek archaeological explanations for archaeological phenomena, rather than heavily relying on ‘other lines of evidence’. The field research, which is the subject of this monograph, was made possible by generous grants from L.S.B. Leakey Foundation; Smuts Memorial Fund, Anthony Wilkin Fund and Bartle Frere Exhibition of the University of Cambridge; Boise Fund of the University of Oxford; and the Deans’ Committee of the University of Nairobi to whom I am most grateful. I am also grateful to the British Council for awarding me the scholarship, which enabled me to embark on the doctoral programme at Cambridge and to the National Museums of Kenya for logistical support. Sincere gratitude also goes to the late Gideon S. Were then Director of the Institute of African Studies, University of Nairobi, for providing me with a vehicle and to Richard E. Leakey then Chairman of the National Museums of Kenya Board of Governors for allowing me to use his personal Land Rover during field work. Thanks are due to the late Denis Owuor and to Nyapala Onyuor for permission to excavate on their farms. Peter Robertshaw is thanked for introducing me to the Gogo Falls site and for supplying me with his excavation notes and other relevant information. Samuel M. Kahĩnju is thanked for his assistance during field research and subsequent analyses. Paul W. Kyenze, John Kimengich, Jonathan Rotich and the late Elijah Ogutu are also thanked for their assistance during the cataloguing and analyses of the finds. I am also very grateful to my teacher, friend and colleague Simiyu Wandibba and to my friend and colleague Herman O. Kiriama for their help with the analysis of the ceramics. My gratitude also go to Gilbert Oteyo, Zachary Otieno and Dennis Milewa for their help with some of the illustrations. I am also grateful to Geoff N. Bailey and David W. Phillipson for their invaluable comments on the University of Cambridge dissertation, whose revision has resulted in this monograph. I am also grateful to Simiyu Wandibba for his comments on the monograph. I salute David W. Phillipson and Rupert Housley for their kind help with the dating of the charcoal samples at the University of Oxford’s Radiocarbon Accelerator Unit. Thanks are also due to Charles French for

his comments on the site’s stratigraphy. Marsha Levine is thanked for supplying me with her computer programme for inputting the faunal data. Her friendship, help and advice and the wonderful get-togethers she organised for her ‘old’ students and friends made my stay at Cambridge enjoyable. John M. Lonsdale is thanked for his friendship and advice throughout my stay at Cambridge and for the many stimulating discussions we had on a wide range of topics, including archaeology and history and the politics of the disciplines in East Africa. I am grateful to Stanley H. Ambrose, John W. Barthelme, Diane Gifford-Gonzalez, D. Bruce Dickson, Frederick Kang’ethe, Wanjĩra wa Kariũki, David Kuehn, Herman O. Kiriama, Mũchũgũ Kĩirũ (D.H.), Meave G. Leakey, Fiona B. Marshall, Bernard N. Mbae, J. Mworia-Maitima, Godfrey Mũriũki, Theresa N. Ng’ang’a, Kathleen Ryan, Sultan H. Somjee, Mũthoni wa Thang’wa, Bernard Wailes, Simiyu Wandibba and Catherine Wanjikũ for their friendship and encouragement. I am thankful to John W. K. Harris, Purity W. Kĩũra, Sibel and Chap Kusimba, Onesphor Kyara, Godwin Mollel, Janet Monge, Jackson Njau, Lisa and Michael Rogers, Kathleen Ryan, Marisa Kai Trenkle and Bernard Wailes for their friendship, encouragement and for brightening my long days at the University of Pennsylvania and Rutgers University when revising the manuscript. Kagũnda wa Kairũ, Mbũgua wa Kariũki, Nyingĩ wa Kĩmani and their families are thanked for their friendship and encouragement and for taking care of my family during my long absences from Kenya. Lastly, but not least, I am grateful to my son Irũngũ for teaching me how to smile and laugh when I was bed-ridden with severe back pains. It is to him that I dedicate this monograph.

ii

For Irũngũ

TABLE OF CONTENTS

Preface Dedication

i iii

Table of contents

iv

List of Figures

vi

List of Tables

vii

Chapter 1: Introduction

1 1 4 5 5 6 8

Origins of Food Production Migrations The Evidence The Problem Current Sequence Evolutionary Ecological Approach

Chapter 2: Previous Research on the Neolithic and Iron Age Pioneer Neolithic Research The ‘Stone Bowl Culture’ The Neolithic ‘Debate’ The ‘Pastoral Neolithic’ Attempts at Regional Syntheses Neolithic research Iron Age research Urewe ware Early food production

10 10 13 14 15 16 25 27 27 29

Chapter 3: Holocene Environmental and Climatic Setting Modern environmental and climatic setting The Lake Victoria basin Palaeoenvironmental and palaeoclimatic setting Localities of studies of Holocene conditions Terminal Pleistocene Early Holocene (ca. 10,000 BP - 6,000 BP) Middle Holocene (ca. 6,000 BP - 3,000 BP) Later Holocene (ca. 3,000 BP - present) Faunal distribution and human settlement

33 33 37 40 41 41 44 45 46 47

Chapter 4: Geographical Setting and Human Settlement Geographical setting of the site Human settlement Soils and land-use patterns

49 49 51 53

Chapter 5: Archaeological Investigations Recent investigations Sampling and excavation Site formation and stratigraphy

56 58 59 66 iv

Incorporation of archaeological material Dating Sorting and cataloguing

69 69 71

Chapter 6: The Finds and Approaches to Analysis Iron artefacts Ceramics Composition of the ceramic assemblage Spatial representation Chronological representation Faunal remains Identification Age estimation Quantification

73 73 73 79 79 83 84 91 92 92

Chapter 7: Patterning of the Faunal Assemblage Taxonomic representation Spatial representation Chronological representation Taxonomic abundance NISP counts MNI counts Age categories Skeletal element representation Modification patterns

94 94 97 99 101 101 104 107 108 110

Chapter 8: Subsistence Activities and Environment Taphonomic history Subsistence strategies Gathering and cultivation Fowling Fishing Hunting Herding Butchery Palaeoenvironmental implications

115 115 117 118 119 119 121 123 126 126

Chapter 9: Subsistence Strategies and Technology Occupation of Gogo Falls Pottery, chronology and subsistence Interpretive model Summary and conclusions

128 128 128 134 137

Appendix I Appendix II Appendix III Appendix IV Appendix V References Cited

140 145 146 151 153 158

v

List of Figures

Figure 1.1 Figure 2.1 Figure 2.2 Figure 2.3 Figure 2.4 Figure 2.5 Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 3.5 Figure 3.6 Figure 4.1 Figure 4.2 Figure 4.3 Figure 5.1 Figure 5.2 Figure 5.3 Figure 5.4 Figure 5.5 Figure 5.6 Figure 6.1 Figure 6.2 Figure 6.3 Figure 6.4 Figure 6.5 Figure 6.6 Figure 6.7 Figure 6.8 Figure 6.9 Figure 6.10 Figure 6.11 Figure 6.12 Figure 7.1 Figure 7.2 Figure 7.3 Figure 7.4 Figure 7.5 Figure 7.6 Figure 7.7 Figure 8.1

East Africa: countries and major towns Location of sites named in the text Maringishu ware Remnant (Elmenteitan) ware Nderit ware Narosura ware East Africa: natural vegetation Vegetation zones on East African mountains The Great Rift Valley system East Africa: mean annual rainfall Lake Victoria drainage basin Localities of studies of Holocene conditions Location of Gogo Falls Kenya: ethnic groups South Nyanza District: agro-ecological zones Gogo Falls site location Gogo Falls: excavations at Area A Gogo Falls: stratigraphy of square 20 Gogo Falls: stratigraphy of square 23 Gogo Falls: stratigraphy of square 25 Gogo Falls: stratigraphy of square 28 Gogo Falls: Elmenteitan vessel Gogo Falls: spatial distribution of the ceramic assemblage Gogo Falls: spatial distribution of ceramics in Area A Frequency of the ceramic assemblage Gogo Falls: Kansyore pottery Gogo Falls: Elmenteitan pottery Gogo Falls: Urewe pottery Composition of the faunal assemblage Spatial distribution of the 1989 faunal collection Spatial distribution of the 1989 faunal collection in Area A Frequency of the 1989 faunal collection Frequency of the 1983 faunal collection Composition of the faunal assemblage Original composition of the 1983 faunal collection Stratigraphic location of dated sample, cow elements and pottery in Square 26 Stratigraphic location of dated sample, caprini elements and pottery in Square 29 Incidence of modified specimens in the faunal assemblage Incidence of specimen modification in the faunal assemblage Incidence of weathered specimens in the faunal assemblage Relative importance of taxonomic groups represented in the faunal assemblage

vi

2 11 17 18 19 20 34 36 38 39 42 43 50 52 54 57 60 61 63 64 65 75 76 77 78 80 81 82 85 86 88 89 90 95 96 100 102 111 112 113 120

List of Tables Table 2.1 Table 2.2 Table 5.1 Table 5.2 Table 5.3 Table 5.4 Table 6.1 Table 6.2 Table 6.3 Table 6.4 Table 6.5 Table 6.6 Table 6.7 Table 6.8 Table 6.9 Table 6.10 Table 6.11 Table 7.1 Table 7.2 Table 7.3 Table 7.4 Table 7.5 Table 7.6 Table 7.7 Table 7.8 Table 7.9 Table 7.10 Table 7.11 Table 7.12 Table 7.13 Table 8.1 Table 9.1

Chronometric dates on East African Neolithic sites Chronometric dates on East African Iron Age Urewe sites Chronometric dates from the 1983 excavations at Gogo Falls Chronometric dates on charcoal from the 1989 excavations at Gogo Falls Frequency of artefacts and faunal remains by excavated areas Frequency of artefacts and faunal remains by excavated squares Frequency of iron artefacts by squares Frequency of iron artefacts by levels Excavated areas, sample size and density of the ceramics Frequency of the ceramic assemblage by squares Composition of the ceramic assemblage Frequency of the wares by areas Frequency of the wares by squares and type of deposit Frequency of the 1989 faunal collection by squares Excavated areas, sample size and density of the 1989 faunal collection Frequency of the 1983 faunal collection by trenches Frequency of the 1989 faunal collection by squares and type of deposit Taxonomic representation in the faunal assemblage Taxonomic representation in the squares excavated in 1989 Taxonomic representation in the trenches excavated in 1983 Stratigraphic representation of taxa and artefacts in selected squares Taxonomic quantification of the 1989 faunal collection by NISP Taxonomic quantification of the 1983 faunal collection by NISP Overall taxonomic quantification by NISP Overall taxonomic quantification by MNI Comparison of overall NISP and MNI values Age categories represented by postcranial elements Age categories represented by teeth Overall skeletal element representation Incidence of modification on anatomical elements Environmental characterisations of the Gogo Falls fauna pottery wares and subsistence strategies by squares and type of deposit

vii

22 29 58 71 71 71 73 73 79 79 79 83 83 87 87 87 91 97 98 99 103 105 106 106 107 107 108 109 109 110 127 130

viii

CHAPTER ONE

INTRODUCTION

By international standards, archaeologists have paid little attention to the study of the subsistence economies of the peoples who occupied East Africa, that is, the countries of Kenya, Tanzania and Uganda (Figure 1.1) during the Holocene. In consequence, our knowledge about this subject is rather poor. Equally poor is our knowledge about the process of subsistence change from food collecting activities like hunting and gathering to food production – the keeping of domestic animals and crop cultivation – which occurred during that period.

Origins of Food Production Several theories have been proposed to explain the origins of food production in East Africa. The salient points in these theories are fairly similar; therefore, instead of reviewing them separately, they will all be reviewed together. The review will begin outside East Africa because food production is generally viewed as alien to the region. In so doing, it is hoped that the resulting summary will be a fair presentation of the theories. There seems to be a general agreement among archaeologists that food production in Africa was partly exotic and partly autochthonous (e.g., Harlan et al. 1976; Phillipson 1977, 1984a; Shaw 1976; Stahl 1984; Zohary and Hopf 1988). Among the exotic plants and animals are emmer wheat (Triticum turgidum), pigs (Sus scrofa), sheep (Ovis aries) and goats (Capra hircus), all of which were first domesticated in the Near East. It is thought that these spread to Africa from the Near East via the Egyptian Nile valley, where remains of both the animals and plants have been dated to between 8,000 BP and 6,000 BP (Stemler 1980; Wendorf and Hassan 1980).

The research reported here was conducted at the Gogo Falls archaeological site on the Kenyan side of Lake Victoria. Excavations conducted at the site indicate a multi-component occupation of long duration extending from the Neolithic to the Iron Age. Compared with the region’s other known sites dating to the same period, the site is unique because of both its size − which has been estimated to be 25,000 m² (Robertshaw 1985) − and its composite nature and long period of occupation. As such, it offers an ideal situation where the subsistence economies of the Neolithic and Iron Age populations and the change from food collection to food production can be investigated. In order to give a regional perspective, data from the site are compared with findings from researches that have been conducted in other parts of East Africa.

From Egypt the domestic animals spread to northern Sahara, where they were readily adopted by the pre-Neolithic populations. These latter are thought to have led a sedentary lifestyle, involving specialised hunting and gathering (Clark 1962, 1967a, 1976, 1980). It has been suggested that this lifestyle could have enabled them to domesticate indigenous wild cattle (Bos sp.) prior to, or after the introduction of the Near Eastern domesticates (Clutton-Brock 1989; Ehret 1979, 1984; Gautier 1984, 1987). If the domestication of cattle preceded the introduction of the Near Eastern animals, this could explain why the exotic animals were readily accepted.

The present study attempts to do three things. One, to document the exploitation of animals for purposes of subsistence by the human populations who occupied the Kenyan side of the Lake Victoria basin during the Neolithic and Iron Age periods, with specific reference to the Gogo Falls site. Two, to examine the correlation, if any, between the subsistence strategies, which were adopted at the site and the pottery wares represented there. Three, to seek an explanation for the apparent coexistence of Kansyore pottery with Akira and Elmenteitan Neolithic pottery wares and the Iron Age Urewe ware at the site.

The sedentary lifestyle of the Saharan preNeolithic populations had been made possible by the favourable climatic conditions in the region. From about 12,000 BP there was a considerable

1

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 1.1 East Africa: countries and major towns

2

Karega-Mũnene

increase in rainfall, as a result of which, water became abundant. This, in turn, caused the appearance of shallow lakes and swamps in the depressed areas, as well as open grassland habitats and cooler Mediterranean flora. These conditions enabled various wild terrestrial and aquatic fauna and plants to flourish in the region. They also induced migrations of the northern Sahara populations who had acquired the Near Eastern domestic fauna (in addition to their domestic cattle) to central Sahara by about 7,000 BP. Other human groups who lived by intensive plant collecting and hunting also migrated to central Sahara from the Upper Nile, Ethiopia and the forest fringes of West Africa. While in central Sahara, the latter migrants learnt about the keeping of domestic animals from the north Sahara migrants (Clark 1962, 1967a, 1980).

where water was available. The symbiotic relationship that developed between the humans and wild plants and animals at the watering points eventually led to the domestication of some of the plants and animals. Besides central Sahara, it has been suggested that West Africa, Ethiopia and East Africa were also ‘centres’ of domestication. Unlike domestication in the Sahara, domestication in the latter areas has not been attributed to climatic stimulus. The crops that are believed to have been domesticated in West Africa include several species of yam (Dioscorea sp.), African rice (Oryza glaberrima) and oil palm (Elaeis guineensis); the only animal domesticated there being the guinea fowl (Numida meleagris). The Ethiopian domesticates include such crops as teff (Eragrostis tef), noog (Guizotia abyssinica), ensete (Ensete ventricosa), khat (Catha edulis) and arabica coffee (Coffea arabica). As in West Africa, only one animal − the donkey (Equus asinus) − is believed to have been domesticated in Ethiopia. The East African domesticates are thought to have been finger millet (Eleusine coracana) and robusta coffee (Coffea canephora); no animal is believed to have been domesticated there (Ehret 1979, 1984; Harlan 1977; Harlan et al. 1976; Murdock 1959; Purseglove 1972a, 1972b, 1976; Sauer 1969). Like the central Sahara domesticates, some of the West African and Ethiopian domesticates are believed to have reached East Africa through migrations.

The abundance of food resources in the Sahara encouraged the migrants to live in larger groups. Therefore, they intensified their food collecting activities further. This, in turn, led to experimentation with some of the plants and eventually to the domestication of such plants as bulrush millet (Pennisetum americanum), finger millet (Eleusine coracana) and gourds (Lagenaria siceraria) (Ehret 1974; Harlan 1977; Stemler et al. 1975). The domestication process may also have been necessitated by rapid population growth, which together with increased desiccation of the Sahara from about 4,500 BP, may have exerted pressure on the diminishing wild food resources. The desiccation could have been hastened by over-grazing and indiscriminate cutting of bush for cultivation (Clark 1976, 1980).

Proponents of the theories under review are, however, divided about the antiquity of domestication in Africa compared to that of the Near East. Some of them suggest that African domestication is more recent and that its roots lie in the introduction of the Near Eastern domesticates. To take one example, sorghum (Sorghum bicolor) is thought to have grown as a weed on wheat/barley farms in Ethiopia before its potential as a cultivar was realised (Purseglove 1972a). It has also been suggested that domestication of plants and animals in Africa was an autochthonous development, independent of influence from the Near East (Clark 1976; Ehret 1974, 1979, 1984; Sauer 1969; Simoons 1965). Proponents of this hypothesis claim greater antiquity for Saharan and Ethiopian domestication, placing the date as early as 7,000 BP (Ehret 1979, 1984).

All this forced the populations who had become herders and grain cultivators out of the region. Some of them migrated to the Mediterranean basin, others southwards to the forest fringes of West Africa and others eastwards to the presentday Sudan and Ethiopia (Clark 1962, 1964, 1967a, 1976, 1980). It is these population movements which are believed to have led to the introduction of food production into East Africa, an issue to which we will be returning below. There is little doubt that the above invocation of climatic stimulus for domestication was borrowed from Childe’s (1951) riverine-oasis theory on the origins of domestication. The theory suggests that the onset of aridity during the early Holocene caused wild plants and animals as well as humans to move to areas 3

Holocene Foragers, Fishers and Herders of Western Kenya

now referred to as Cushites persists to date (e.g., Ambrose 1982, 1984; Barthelme 1985; Ehret 1974; Sutton 1966; Marshall 1986; Odner 1972; Phillipson 1977). This view is not supported by direct physical anthropological evidence. On the contrary, the latter now indicates that the attribution of the concerned human skeletal remains to ‘Caucasoids’ was wrong. Indeed, according to Rightmire (1984: 196) the identification was based on “evidence [that] has never been very solid”. The identifiable specimens are ‘Negroid’, meaning they are of individuals who were related to the Bantu (Rightmire 1975a, 1975b, 1984). As we shall presently learn, the identity of the Cushites is solely based on historical linguistic evidence.

Migrations As mentioned above, the introduction of food production in East Africa has been attributed to the migration of human populations from the Sahara. However, it is thought that the populations concerned did not migrate to East Africa directly, but to the Sudan and Ethiopia and thence to East Africa. Initially, the populations who eventually reached East Africa were thought to have been ‘Caucasoid’, a term that was meant to imply that they were not related to the region’s Bantu-speaking peoples. Use of the term ‘Caucasoid’ for the populations concerned, was based on anthropological evidence, viz., the identification of human skeletal remains from some East African Neolithic sites by Leakey (1931, 1935, 1936, 1942). Although Leakey never claimed the identification was unequivocal, Cole (1963: 48) asserted:

The theories under discussion suggest that the Cushites entered northern Kenya with their domestic animals and cereal crops about the sixth to fifth millennium BP, thence to the rest of East Africa (Ambrose 1982, 1984; Barthelme 1985; Ehret 1974; Marshall 1986; Phillipson 1977). The theories are, however, silent about why the traditional Ethiopian crops of teff, noog and ensete were not introduced in East Africa.

...we know very well what these people looked like. They were tall and longheaded, almost indistinguishable skeletally from Mediterranean Caucasoids and from the Hamitic peoples of the Horn [emphasis added].

Although other writers did not make such claims, the ‘Caucasoid’ myth characterised most of what was published about the East African Neolithic (e.g., Clark 1962, 1964; Leakey M.D. 1945; Leakey and Leakey 1950). This myth was perpetuated regardless of O’Brien’s (1939: 295) warning that:

Some authors (e.g., Marshall 1986; Murdock 1959; Sauer 1969) have, however, argued that the ‘spread’ of food production over East Africa cannot be adequately explained in terms of population movement alone. Rather, they have sought to explain the phenomenon by a combination of migration and diffusion. Murdock (1959: 89), for instance, has argued that food production ‘spread’ not through

...it is no longer possible or desirable for African [sic] archaeologists to turn their eyes northwards to Europe for guidance and enlightenment in their many problems. We can no longer regard Africa, at least south of the Mediterranean area of colonisation, as having been continuously influenced from the north and east, but must recognise that, apart from any occasional foreign impetus, African Stone Age Man achieved his own, essentially African culture, built out of African materials, in an African environment to which that culture was especially and purposely suited. In that realisation lies, I believe, a promise of tremendous significance for the future of African archaeology and for the solution of racial and kindred anthropological problems in that continent.

...a single migrating people but [through]... diffusion to successive peoples of different linguistic stocks − a slow and laborious process since each in turn had painfully to learn the complex new techniques of food production and to substitute these for the old familiar modes of acquisition through hunting and gathering.

The southward spread of food production into East Africa is thought to have taken much longer than the spread over North Africa. Three reasons have been advanced to explain this difference. One, that East Africa was very rich in wild animals and plants; therefore, there was no immediate need to engage in food production. Two, the Neolithic technology of the migrant Cushitic-speaking groups was inefficient in clearing the forests and wooded grassland of East Africa for crop cultivation. Three, the small size of the indigenous hunting and gathering groups

Over half a century after this statement was made, it remains germane in East African archaeology. That is because the tendency to attribute food production to what Wrigley (1960: 189) calls ‘quasi European intruders’ who are 4

Karega-Mũnene

made food production a less attractive economic activity (Clark 1962, 1964, 1967a, 1980; Gabel 1974).

1972b, 1976; Stahl 1984; Stemler 1980). The ‘identification’ of Ethiopia as a ‘centre’ of domestication early this century was, for instance, based not on archaeological evidence but on the presence of a wide variety of domestic wheat, a situation which was viewed as suggesting great genetic diversity of the plant (Vavilov 1926, 1951).

Like the Near Eastern, Saharan and Ethiopian domesticates, the West African cultivars, notably tuber and root crops as well as iron working technology, are thought to have been introduced in East Africa by Bantu migrants. The Bantu are said to have originally migrated from their west African ‘cradle land’ between about 5,000 BP and 2,500 BP because of population pressures. This movement spread the cultivation of tuber and root crops to the western Sudan, thence to East Africa. It has been argued that while in the Sudan some of the Bantu groups learned about animal husbandry and metallurgy from indigenous populations before migrating to East Africa (Clark 1962, 1964; Ehret 1968, 1974, 1982a, 1982b, 1984).

To support the suggestion that East Africa was a ‘centre’ of domestication, Lwanga-Lunyiigo (1976) has argued for a more ancient presence of the Bantu − who he presumes were responsible for local plant domestication − than is generally acknowledged. To demonstrate this argument, he cites three forms of evidence. One, physical anthropological evidence from Holocene eastern and southern African sites which suggests an earlier Bantu presence. Two, chronometric evidence from Rugomora Mahe (Katuruka) in Tanzania which places the Iron Age at the site in the third millennium BP. Three, differences in the traditional crops and methods of cultivation used in East and West Africa. Using this evidence, he inverts the hypothesis about Bantu migration from West Africa. He argues that if there was any Bantu migration, it was from East Africa to central and southern Africa − as the movement of the sickle-cell gene from East Africa to central and southern Africa suggests − rather than from West Africa to East Africa. But the evidence cited in support of this argument is not sufficient to qualify the inversion of the Bantu migration hypothesis.

The evidence The obvious direct evidence for food production, which can be used to support the theories outlined above, consists of remains of prehistoric domestic plants and/or animals. However, because these are not always available from archaeological sites, additional evidence is sought from indirect sources like rock art or engravings, artefacts associated with food production and plant impressions on pottery, bricks or daub. The modern geographical distribution of domestic animals and plants and their nearest modern wild relatives is also used as indirect evidence (Zohary and Hopf 1988).

The Problem

In the whole of East Africa and in most of subSaharan Africa, there is as yet no direct evidence for cultivation during the Neolithic. But there is considerable evidence for animal husbandry in the form of skeletal remains. The available evidence for East Africa is reviewed at length in the following chapter. Our concern here is to briefly review the evidence that has been used to construct the theories discussed above.

As indicated above, the question of food production has not been adequately addressed in East Africa. Indeed, as we shall see in the next chapter very little effort has been expended on systematic field research specifically on the problem. This contrasts sharply with the situation in such places as Europe, the Near East and Mesoamerica where knowledge about food production is currently quite intensive and extensive (e.g., Flannery 1986; Zohary and Hopf 1988). This notwithstanding, there exists a fairly impressive corpus of literature on the region’s mid-late Holocene period. However, as the following chapter will demonstrate, this literature tells us very little about the origins and development of food production because it is biased towards technological aspects.

Although specific ‘centres’ of domestication have been identified in Africa, as the above discussion indicates, it is not clear exactly where, when and how the plants and animals concerned were domesticated. The ‘centres’ have been inferred from circumstantial evidence, notably the current distribution of the domestic, wild and weedy races of the plants in question and not from direct archaeological evidence (Harlan 1977; Harlan et al. 1976; Purseglove 1972a, 5

Holocene Foragers, Fishers and Herders of Western Kenya

This condition is greatly influenced by the history of archaeological research in the region and by a marriage of convenience between archaeology and historical linguistics. In fact, the influence of the latter is evident in virtually all archaeological writing (e.g., Ambrose 1982, 1984; Barthelme 1985; Gramly 1975; Phillipson 1977; Marshall 1986; Robertshaw 1991). Historical linguistics, for instance, suggests that East Africa was inhabited by hunting and gathering populations during the pre-Neolithic period. From about mid-Holocene these populations were displaced and/or assimilated by Cushitic-speaking peoples who migrated from Ethiopia. The migration of the Cushites was followed by the migration of Bantu-speaking peoples from somewhere in west-central Africa, thereby causing further displacement and/or absorption of the hunting and gathering groups. The linguistic evidence also suggests that the Cushites were pastoralists and the Bantu predominantly cultivators with a few domestic animals (Ehret 1974; Schoenbrun 1990, 1993). These hypotheses have been uncritically accepted by the majority of archaeologists, who correlate the Cushitic- and Bantu-speaking peoples with the Neolithic and the earliest manifestations of the Iron Age, respectively. These correlations can be summed up as follows:

attempt to explain the change in terms of population movements. The available archaeological evidence for food production during the period under investigation is far from representative of East Africa as a whole. That is largely because the geographical coverage of the research is skewed. For instance, virtually all the Neolithic data come from sites restricted to a narrow corridor in the eastern Rift Valley, stretching from northern Kenya to northern Tanzania (Bower 1984). In consequence, virtually nothing is known about the Neolithic to the east or west of that corridor. The spatial distribution of sites bearing the earliest manifestation of the Iron Age is also skewed since virtually all the sites are located in the Lake Victoria basin. Investigations of the majority of both the Neolithic and Iron Age sites have generally been limited to a few test trenches, representing very small fractions of the sites (Bower 1984). In consequence, only fairly small samples of both cultural and faunal remains have been recovered. Although the representativeness of the cultural and faunal samples recovered in this way may be open to question, most of what we ‘know’ about early food production and the presumed population movements in East Africa is based on those samples. As a result, the ‘knowledge’ is not only incomplete, but also largely speculative.

Cushites = Neolithic = Animal husbandry Bantu = Iron Age = Agriculture.

Current Sequence

Besides treating these correlations as an acceptable way of interpreting the Neolithic and Iron Age phenomena, archaeologists have also attempted to ‘reconstruct’ the routes of migration of both the Cushites and Bantu. The problem with this approach is that it has involved nothing more than speculations about the routes of migration and descriptions of the subsistence activities practised by the linguistic groups concerned (e.g., Ambrose 1982; Soper 1982). Apart from isolated chronometric determinations and associated cultural and/or faunal remains, the approach is wholly dependent on linguistic hypotheses. As such, it adds very little to the current archaeological knowledge because, as Schmidt (1975: 133) aptly observes, “archaeological phenomena cannot be tied to generalised linguistic phenomena, certainly not in terms of dispersal of language groups over the terrain”. The approach also tends to blur such crucial issues as cultural change during the Neolithic and Iron Age periods, as archaeologists

The current sequence of the East African Neolithic and Iron Age has been constructed using chronometric dates and cultural remains like pottery and stone artefacts. The manner in which this has been done and the resulting chronologies are discussed at length in the next chapter. Therefore, it will suffice to make only a few pertinent observations here about that sequence. To begin with, the Iron Age period is traditionally divided into two phases, viz., the ‘Early Iron Age’ and the ‘later/Later Iron Age’ (Phillipson 1977; Soper 1971a, 1982). Each of these phases is known by pottery wares and not from iron artefacts, as the name seems to imply. As we shall see in the succeeding chapter, this division is arbitrary. The East African Neolithic is locally known as ‘Pastoral Neolithic’ (‘PN’). This term was coined following the recovery of Neolithic faunal samples in which domestic animals were

6

Karega-Mũnene

dominant. Corresponding with this was lack of direct evidence for cultivation, not because plant remains were actually unavailable, but because very little efforts were made to recover them. Thus, the emphasis on animal husbandry that the term ‘PN’ implies is largely due to negative evidence. This emphasis could be misleading because experience in East Africa has clearly shown that “absence of evidence is not evidence of absence”, to use Martin Rees’s words (cited in Gribbin and Cherfas 1982: 48). Indeed, as the discussion of these issues in the next chapter will demonstrate, there have been radical changes in knowledge about the region’s Neolithic from early this century. There is hardly any reason to firmly assume that remains of domestic plants may not be recovered from Neolithic sites in the future.

The appearance of both the Neolithic and Iron Age finds by which both periods are known has, as the theories discussed above indicate, been associated with population movements. Attempts have been made to establish the ethnic/linguistic identity of the Neolithic and Iron Age populations concerned. This has involved the treatment of pottery wares and lithic assemblages as discrete cultural entities, which are presumed to represent specific ethnic/linguistic groups. Kansyore ware has been correlated with ancestors of Khoisan hunter-gatherers; Elmenteitan with Southern Nilotes (i.e., ancestors of modern Dadog and Kalenjin); and Akira, Narosura, Nderit and Maringishu with Southern Cushites (i.e., ancestors of the Alagwa, Aramanik, Asa, Burungi, Dahalo and Iraqw). Thus, the creators of Kansyore ware are thought to have been hunter-gatherers, whereas the creators of the other Neolithic wares − Elmenteitan, Akira, Narosura, Nderit and Maringishu − are regarded as the earliest food producers in region. On the other hand, Iron Age wares have been associated with the Bantu, who are credited with the introduction of agriculture (Ambrose 1982, 1984).

Since the term Neolithic is generally understood to imply food production (i.e., the keeping of domestic animals and/or cultivation), the present study proposes that the term ‘PN’ be replaced by Neolithic. This proposition is made with the knowledge that the term Neolithic means different things in the former French colonies in Africa. In the latter areas, the term is used to refer to cultures that are characterised by polished or ground stone artefacts and pottery (De Barros 1990; Shaw 1977).

These correlations are extremely tenuous and can be misleading. For instance, the association between Kansyore pottery and ancestors of modern Khoisan − whose presence in the region has not been documented (Schepartz 1988) − is based on previous lack of secure association between the pottery and domestic animals. As the present study shows, this is no longer a valid observation. Furthermore, the correlations are not significantly different from the population movements hypotheses discussed above. Unfortunately, these hypotheses are untestable because precise correlations between material culture and language cannot be demonstrated archaeologically. In fact, as Stahl (1984: 20) has observed, “the attempt to attribute cultural remains to a particular linguistic group represents an archaeological cul-de-sac”. Thus, there is urgent need to employ better approaches in the interpretation of the Neolithic and Iron Age data from East Africa. It is because of this as well as the need to attempt to understand the process of the development of food production that the present study will use the evolutionary ecological approach discussed below.

The pottery wares by which the Neolithic is known are Elmenteitan, Akira, Maringishu, Narosura and Nderit (Bower et al. 1977; Wandibba 1977, 1980). The first ware is associated with a lithic industry that is known by the same name, whilst the other four wares are associated with the Savanna Pastoral Neolithic lithic industry (Ambrose 1984; see also Nelson 1980). Both the lithic assemblages and the wares are considered Neolithic because they have been found in association with skeletal remains of domestic animals (Bower et al. 1977). The other widely known pre-Iron Age pottery in East Africa is Kansyore. This pottery is thought to be the oldest pre-Iron Age ware in the region, since it has hitherto not been associated with domestic animals. Its place in the region’s chronology has also been problematic because it has hitherto not been reliably dated. Recently, however, the ware was found in secure association with domestic animals at Gogo Falls. This association has been radiocarbon dated to the first half of the fourth millennium BP.

7

Holocene Foragers, Fishers and Herders of Western Kenya

[because] they have evolved by natural selection and to do so have had to confer some benefit of adaptive value” (Mithen 1990: 4, emphasis added). Viewed in this way, cultural change is a selective and adaptive process rather than a transformational process (Dunnell 1980). Although the evolution of both cultural behaviour and organisms is selective and adaptive, culture and organisms are far from similar; therefore, their respective evolution proceeds differently:

Evolutionary Ecological Approach The key words of this approach are evolution and ecology. In biological parlance the former refers to a natural process involving the gradual development of organisms and the latter to the study of organisms in relation to one another and to their physical and biological environments (Pianka 1974). These terms are used in archaeology to mean the same things, the only difference being the inclusion of the term phenomena (e.g., culture) besides the term organisms. Ecological and evolutionary approaches have been derived from these terms and used by archaeologists to explain cultural change. As one might expect, the explanations offered by each of the approaches are unsatisfactory on their own. The ecological approach, for instance, has offered functional explanations and the evolutionary approach historical explanations. This difference is best illustrated by the views of each of the approaches to the question about the origins of food production. The ecological approach explains that it was “an adaptive response to stress arising in specific historical situations” and the evolutionary approach that it was a consequence of the unfolding of a “historical situation... [in which the] people, or their culture, came to recognize the benefits agriculture could bring” (Rindos 1984: 91).

Culture must be viewed as a system of inheritance separate and distinct from the genetic system in all ways save the genetic capacity for culture. The rules governing cultural inheritance differ vastly from those governing genetic evolution. Genes are transmitted equally by both parents to the offspring at the time of fertilization and no further change or modification of the genetic message is possible after this time. Culture is not bound by consanguinity, specific timing of inheritance, or need for unitary transmission. It may be transmitted between individuals having no close genetic relationship. Cultural inheritance may occur at any point in an individual’s life. Culture is not transmitted in toto and individuals may possess only a portion of the total available cultural information (Rindos 1984: 54, emphasis added).

Expounding on these differences, Rindos (1984: 61) adds: ...the fact that human cultural behaviors are not the result of genetic differences does not imply that the traits could not have been the result of selection. Variation in human culture has some meaning − it is not merely the result of totally random and adaptation-neutral processes.

The evolutionary ecological approach combines elements of both the ecological and evolutionary approaches, for a better appreciation of cultural behaviour through time. This, in turn, enables the approach to offer better explanations for cultural change. That is because the approach not only recognises that cultural change is evolutionary, but also that it occurs within given ecological, social and historical contexts. Additionally, the approach views culture and its social and natural environments plus the interrelationships between them as dynamic. This is very important because both culture and its social and natural environment are ever-changing. As used above, the term culture refers not only to technology, but also to capacities to learn and to make rational decisions (Mithen 1989, 1990). These capacities, it has been argued, are “product[s] of evolution” (Mithen 1989: 483), which means that they are universal to humans regardless of their social organisation and subsistence activities.

At the very heart of the evolution of cultural behaviour and evolutionary ecology are the capacities to learn and to make decisions. There can be little doubt that both learning and decision making are conscious and intentional and that they are aimed at solving problems. Because individuals, who together form the corporate society, are different in many ways, there is considerable “behavioural flexibility” amongst them (Mithen 1990: 1). In any given society, decision making is by both the corporate society and the individual. Therefore, it is inappropriate to emphasise the role of one at the expense of the other. No doubt, individuals are different − even in the most egalitarian society (Mithen 1989, 1990) − but they all act within given social contexts. Thus, while individuals may ‘gather’

As products of evolution, decision making capacities are “ultimately of a biological nature 8

Karega-Mũnene

information on their own, they eventually share that information between themselves. This not only enables them to hunt, collect wild plants, and/or tend domestic animals and/or plants as a group, but also tends to reduce potential conflict, as Jochim (1979: 97) aptly observes:

prohibit the killing of warthog − established by actions and attitudes of the corporate society. The preference for wildebeest over warthog could also be influenced by such considerations as differences in the behaviour of those animals and/or the amount of flesh each of them provides.

...decisions or goals themselves may be in conflict: each individual... [making] multiple choices involving numerous aspects of behavior, each demanding time and energy, and each perhaps guided by a different set of preferences or norms. The result may be that the actual behavior of an individual represents a compromise solution to these competing demands.

Employing this approach in the present study will enable us to study food collecting and production in East Africa as an evolutionary process in which the individual and the society played significant and complementary roles. This could explain, for instance, why there is similarity in certain aspects of the archaeological phenomena (e.g., pottery belonging to a specific ware) and why there are significant differences in the phenomena (e.g., between wares). This approach will also enable us to study the apparent coexistence of different pottery wares at Gogo Falls in evolutionary terms within specific ecological, social and historical contexts. Thus, both food collecting and production and use of different pottery at the site will be viewed as attempts by the humans concerned to adapt to their ever-changing social and natural environment, rather than as simple transformations precipitated by presumed human migrations.

Thus, appreciation of the contributions made by the individual and by the society is crucial in the study of cultural change. That is because all aspects of cultural change cannot be explained solely in terms of the individual or the society. This is demonstrated by the fact that some aspects of cultural behaviour may represent compromise solutions, which may be reached in the formalisation of individual behaviour by the corporate society. For example, different individuals may independently decide to hunt wildebeest for meat as opposed to warthog over a long period. This preference may be due to social requirements − e.g., totemism, which may

9

CHAPTER TWO

PREVIOUS RESEARCH ON THE NEOLITHIC AND IRON AGE

The term Neolithic was first used in an archaeological context in East Africa towards the end of the nineteenth century by Gregory (1896, 1921), a geologist who investigated the geology of the Rift Valley system in 1892-93 and 1919. He used the term to describe obsidian artefacts he had found on the Athi Plains, Kikuyu Escarpment and in the areas near Lakes Bogoria and Baringo in Kenya. Those artefacts were “of a Neolithic type” and belonged to what he called “the Newer Stone Age” (Gregory 1921: 219, 220). Early this century isolated finds of bored stones (or stone ‘rings’ which were presumably used as digging stick weights), stone bowls and polished stone axe-heads made in Kenya were also described as Neolithic (Dobbs 1914, 1918; Gregory 1921; Hobley 1913).

be relied on as a major indicator of Neolithic cultures. As a result, it was, together with other artefacts, excluded from the definition of the Neolithic. As a result, European and Near Eastern Neolithic are now defined solely in terms of the presence of direct evidence of food production, that is, remains of domestic plants and/or animals. These changes in definition did not have an immediate impact on the East African Neolithic, as we shall see below.

Pioneer Neolithic Research Research on the Neolithic of East Africa dates back to the late 1920s, when the first systematic archaeological research was conducted in the region, specifically in the Lakes NakuruNaivasha basin (Figure 2.1). The principal aim of that research was not to study the Neolithic per se, but to reconstruct the region’s cultural sequence. However, some of the sites investigated in the course of the research yielded polished stone artefacts, stone bowls, human burials and pottery. Since some of these finds, notably pottery and polished artifacts, had been recognised in Europe and in the Near East as diagnostic Neolithic finds, their presence in East Africa was taken to signify the existence of the Neolithic. Thus, it was thought that the pottery, stone bowls and polished artefacts were used for preparation of domesticated cereals; therefore, their presence in the region indicated the existence of “some sort of agriculture” (Leakey 1931: 198).

This usage of the term Neolithic was in accordance with its application in European and Near Eastern archaeology at the time. Towards the end of the nineteenth century, for example, the term was used to refer to the ...later or polished Stone Age; a period characterized by beautiful weapons and instruments made of flint and other kinds of stone, in which, however, we find no trace of the knowledge of any metal, excepting gold, which seems to have been sometimes used for ornaments. (Lubbock 1872: 2-3)

Early in the twentieth century the definition of Neolithic was expanded to include reference to cultivation, animal husbandry and pottery. Thus, the hallmarks of Neolithic cultures in Europe and in the Near East became evidence of food production, production of pottery and polishing of stone artefacts (Burkitt 1925; Childe 1934).

Consequently, the three East African cultures − ‘Njoroan’, ‘Gumban A’ and ‘Gumban B’ − which were characterised by those artefacts were classified as Neolithic (Leakey 1931). The ‘Njoroan’ culture was named for archaeological material exposed by construction workers near present-day Njoro town. The culture was classified as Neolithic because it had pottery and a polished stone axe-head; these were found together with obsidian artefacts and human remains at the type-site. The ‘Gumban’ cultures

Subsequent research in Europe and in the Near East revealed that polished artefacts and pottery did not necessarily appear at the same time as domestic animals and/or plants. In Greece and western Asia, for example, domestic plants predated pottery (Moore 1995; Renfrew 1972) while in northern Europe pottery pre-dated domestic crops and/or animals (Gebauer 1995; Phillips 1980). This meant that pottery could no longer 10

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 2.1 Location of sites named in the text 11

Karega-Mũnene

were named after the mythical pit-dwelling pigmy peoples known as Gumba to the Embu, Gĩkũyũ and Meru of central Kenya (Fadiman 1976; Leakey 1931; Muriuki 1974; Mwaniki 1973). ‘Gumban A’ was considered to be Neolithic because it had pottery and stone bowls; obsidian artefacts and human remains were also present. The pottery was internally scored and had extensively decorated external surface. The type-site for this culture was Stable’s Drift, but it was also found at Makalia and Willey’s Kopje. ‘Gumban B’ had stone bowls, pestle rubbers, obsidian artefacts and pottery which was characterised by rouletted-decoration (Leakey 1931). The type-site for this culture was Nakuru burial site, the present-day Lion Hill near Lake Nakuru (Figure 2.1).

also identified in the Kenyan Victoria basin, but on shell mounds (Leakey 1936; Leakey and Owen 1945; O’Brien 1939). The latter culture was eventually attributed to the Neolithic and the Tumbian renamed Sangoan-Lupemban, hence its attribution to the Middle Stone Age period (McBrearty 1986). In Uganda three ‘Neolithic’ cultures: ‘Kageran’, ‘Wilton-Neolithic A’ and ‘Wilton-Neolithic B’ were identified. The ‘Kageran’ was named after River Kagera on whose bank it was found. It was characterised by cores, choppers, scrapers and flakes. ‘Wilton-Neolithic A’ was represented at Nsongezi rockshelter where its characteristic features were pottery, backed blades, crescents and thumbnail scrapers. The pottery was decorated by cord rouletting, herringbone and cross-hatching designs (O’Brien 1939; Wayland 1934). However, subsequent excavations at both the Kagera river bank site and Nsongezi did not confirm the presence of the ‘Kageran’ culture (Nelson and Posnansky 1970).

These Neolithic cultures were viewed as derivations of two ‘Mesolithic’ cultures, namely, ‘Kenya Wilton’ and Elmenteitan, which were also identified in the Lakes Nakuru-Naivasha basin. ‘Kenya Wilton’ was found at Long’s Drift in Kenya and at Apis Rock (present-day Nasera Rock or Soit Nasera) in Tanzania. The artefacts that were identified in the culture included stone bowls, crescents, burins, thumbnail scrapers and nondescript pottery. The Elmenteitan was identified at Gamble’s Cave II in the Elmenteita basin. Its hallmarks were two-edged blades, backed blades, crescents, scrapers, outils ecailles, burins and pottery. The pottery forms consisted of small bowls and large vessels with pointed, rounded, or flat bases. Other diagnostic features of the pottery were ‘rivet’ holes drilled through some vessels (Leakey 1931). Elmenteitan materials were also found at Bromhead’s burial site, where the most significant finds were human skeletal remains which were identified as of ‘Caucasoid’ origin (Leakey 1935).

‘Wilton-Neolithic B’ was identified at the Chui (Leopard) Cave near Mount Elgon on the Uganda/Kenya border where it was characterised by crescents, burins and thumbnail scrapers (O’Brien 1939). Although the polished stone artefacts, stone bowls and pottery which had been used to identify the Neolithic in the Lakes Nakuru-Naivasha basin were absent from the Ugandan sites, the cultures represented there were still classified as Neolithic. That was because of the presence of thumbnail scrapers, artefacts that were regarded as fossiles directeurs for ‘Wilton’ lithic industry because of their presence in the Kenyan ‘Wilton’ cultures. Since radiocarbon dating was unknown at the time the researches discussed above were undertaken, all the cultures which were identified in the region were dated using Wayland’s (1924) pluvial/inter-pluvial theory. The theory suggested that prehistoric East Africa experienced climatic changes that could be reconstructed from beach deposits in the region. As a result of the reconstructions, specific periods of wetness (‘pluvials’) which were thought to have been separated by periods of aridity (‘inter-pluvials’) were identified. Beginning with the oldest, the periods were named the ‘Kamasian’, ‘Gamblian’, ‘Makalian’ and ‘Nakuran’, the first two being attributed to the pre-Holocene period and the latter two to the Holocene. More specifically, the ‘Makalian’ was

Subsequent research in East Africa failed to yield more sites with finds similar to those of the ‘Njoroan’ culture; therefore, the term was dropped from the region’s archaeological vocabulary. The research, however, revealed two more cultures, viz., ‘Tumbian’ and ‘Kenya Wilton C’ which were then considered to be Neolithic (Leakey 1936). The former was named after Tumba in the modern Democratic Republic of Congo (formerly Zaire), where it was reported to be characterised by polished stone axe-heads. Somewhat similar artefacts were found in the Uganda and Kenya Lake Victoria basins and attributed to this culture. ‘Kenya Wilton C’ was 12

Holocene Foragers, Fishers and Herders of Western Kenya

presumed to date between about 12,000 BP and 4,500 BP and the ‘Nakuran’ from about 3,000 BP to 2850 BP. The cultures which were attributed to the former period included the Elmenteitan, ‘Kenya Wilton A’ and ‘Kenya Wilton B’. The rest of the cultures, that is, ‘Kenya Wilton C’, ‘Njoroan’, ‘Gumban A’ and ‘Gumban B’ were placed in the Nakuran ‘pluvial’ (Leakey 1931, 1936).

Finds that were somewhat similar to those from Njoro River Cave were made at the nearby Egerton Cave. They included remains of human skeletons, pottery, stone bowls, fragments of a lower grindstone, a pestle rubber, obsidian twoedged blades, scrapers and utilised flakes, plus faunal remains which were attributed to antelope (Faugust and Sutton 1966). At Hyrax Hill a Neolithic culture was identified at two localities or ‘sites’, as the excavator called them. The characteristic finds of the culture were pestle rubbers, stone bowls, obsidian artefacts, human burials, pottery and skeletal remains of domestic cattle and sheep. (The remains were reportedly lost or destroyed in Britain during the Second World War; therefore, the identifications cannot be validated or invalidated.) The pottery was classified as ‘Gumban B’ in spite of its lack of internal scoring and all-over decoration on the external surface which were characteristic of ‘Gumban B’ (Leakey 1945).

The ‘Stone Bowl Culture’ Leakey’s (1931) definition of the East African Neolithic remained in use for several decades. About a decade after it was first applied, it was upheld by Mary Leakey (1943), who suggested that the presence of polished artefacts, pestle rubbers, stone bowls and systematic burials be the criteria on which the Neolithic was to be identified. She argued that it was not necessary that a site be considered Neolithic only if it had evidence of cultivation and/or animal husbandry in the form of remains of domestic plants or animals. That was because knowledge of the region’s Neolithic at the time was extremely scanty. Such negative evidence, she correctly reasoned, could not

Both the ‘evolved Elmenteitan’ culture of Njoro River Cave and the Hyrax Hill culture were considered to have a genetic relationship with the Neolithic cultures discussed earlier. That was because they all had stone bowls in common. Besides the then well known Neolithic sites, stone bowls had also been found at Naivasha Railway rockshelter in Kenya (Leakey 1945) and at Ngorongoro in Tanzania (Leakey 1966). Subsequently, all the cultures that had stone bowls were grouped together and described as variants of the ‘Stone Bowl Culture’, thus elevating stone bowls to Neolithic fossiles directeurs. This development, in turn, made the identification and description of sites bearing stone bowls an important pursuit of Neolithic research in East Africa.

...be regarded as particularly significant owing to the fact that the known characteristics of many [Neolithic] cultures... [in East Africa was] still confined to the one-sided and incomplete evidence supplied by either a habitation or a burial site. (Leakey 1943: 182)

Although this definition was, as we shall learn presently, not accepted by all archaeologists working in East Africa, it played a significant role in the recognition of new Neolithic cultures at Hyrax Hill and Njoro River Cave in Kenya. Njoro River Cave yielded several cremated human burials, stone artefacts, pottery, remains of gourd, basketwork and an elaborately decorated carbonised wooden vessel. The lithic assemblage from the site included grindstones, beads, bowls, pestle rubbers and the characteristic Elmenteitan two-edged blades, segmented blades, crescents, burins and outils ecailles. The pottery was also typically Elmenteitan (Leakey and Leakey 1950). Overall, it was assumed that the culture represented at the site had a genetic relationship with the ‘Mesolithic’ Elmenteitan culture, as a result of which, it was described as ‘evolved Elmenteitan’.

Subsequently, more sites bearing stone bowls were reported. These included Ilkek and Ol Orien in the Lakes Nakuru-Naivasha basin, where the artefacts were associated with poorly preserved human remains, pestle rubbers, beads, obsidian artefacts and remains of a gourd (Brown 1966). At Prospect Farm stone bowls, a polished stone axe-head, fragments of a grindstone, pestle rubbers, obsidian artefacts, pottery and remains of domestic and wild animals were recovered. The lithics and potsherds bore some similarities with the finds from Njoro River Cave, but other potsherds showed some affinities with Hyrax Hill Neolithic pottery. These observations were interpreted as indications of a broad 13

Karega-Mũnene

contemporaneity between the Prospect Farm culture and the Njoro River Cave and Hyrax Hill cultures (Cohen 1970).

cited that evidence in an earlier publication in support of his argument for a Neolithic period in sub-Saharan Africa. In that publication he had argued that the stone bowls and grindstones of East Africa “strongly suggest some form of plant cultivation” (Clark 1962: 217).

More stone bowls and other Neolithic fossiles directeurs were reported from several sites in East Africa. Keringet Cave, for instance, yielded stone bowls, pestle rubbers, a grindstone and largely nondescript pottery (Brown 1966; Cohen 1970). Polished stone axe-heads, pestle rubbers and grindstones were also reported at Randhore and Agoro rockshelters in the Lake Victoria basin (Gabel 1969) and various other locations in Kenya (Brown 1969; Gramly 1975; Leakey 1943; Owen 1941). Similar finds were reported at Ngorongoro (Leakey 1966; Sassoon 1968), Seronera (Bower 1973a) and several other locations in Tanzania (Brown 1969; Mturi 1986; Odner 1971c; Sassoon 1967) and in Uganda (Posnansky and Sikibengo 1959; Soper 1971c). The chronology and associations of most of these finds remain uncertain because they were either isolated finds, surface collections, or finds from ‘mixed’ archaeological deposits.

Sutton (1966) also argued that the East African Neolithic could be subsumed within the Later Stone Age. According to him, the Hyrax Hill Neolithic as well as Kansyore pottery (which is discussed at length below) were creations of the remnants of ‘the aquatic civilization of middle Africa’. The latter ‘civilization’ belonged to the Later Stone Age period and was believed to have originated in the Sudan (Sutton 1974, 1977, 1980). Although by definition the ‘aquatic civilization’ is pre-Neolithic, Sutton’s (1980: 322) assertion that it was meant to be essentially ‘anti-neolithic’ makes it necessary that we briefly examine what it entails. The ‘aquatic civilization’ was characterised by cultures that were geared towards the exploitation of aquatic resources for subsistence and other needs like transport. Thus, the creators of such cultures were “a small but successful and hence assimilating group wedded, psychologically as well as economically, to the waters” (Sutton 1974: 535). As such, their material culture was to be found close to lakes, rivers and swamps, where it is represented by bone harpoons and pottery. It was also assumed that boats, fishing nets and baskets were useful to the people in question in their practice of an aquatic lifestyle. The examples of aquatic sites in East Africa, which were cited, were Gamble’s Cave II and Lion Hill. But the evidence for the existence of that lifestyle at Gamble’s Cave, for instance, was a fragment of a bone harpoon and a single sherd of ‘wavy-line’ pottery. Bearing this sort of evidence in mind, one cannot help considering Sutton’s (1974, 1977, 1980) ‘aquatic civilization’, especially with regard to the region south of Lake Turkana, as overly imaginative.

The Neolithic ‘Debate’ As hinted above, the existence of the East African Neolithic was not accepted by all researchers working there; it was questioned by some and dismissed by others. It was argued, for instance, that the term ‘Neolithic’ was inappropriate for the East African cultures concerned because there was virtual “absence of evidence of widespread stone using settled agricultural societies” (Posnansky 1967b: 644). The only direct evidence of food production known at the time, the domestic faunal remains from Hyrax Hill, was ignored when offering this view. Nevertheless, the view was supported by Clark who argued that: Material culture by itself without the direct proof of cultivated plants and cereals, of domestic stock, or of permanent dwelling and settlement patterns is at best ambiguous evidence on which to establish the existence of fully, or even of incipient, food-producing cultures in subsaharan [sic] Africa before the introduction of metallurgy and the spread of early Iron Age culture in the years immediately preceding and following the beginning of the present era. (Clark 1967b: 621)

The existence of the East African Neolithic was also disputed by Cole, whose views seem to have arisen from the colonial attitude towards the indigenous inhabitants of East Africa and not from a genuine professional interest. According to her, the ancestors of the present sub-Saharan Africans were incapable of accomplishing ‘ingenious’ feats like pot-making, iron-working and food production. That was because they suffered “mental stagnation” which she

Surprisingly, Clark also ignored the Hyrax Hill evidence. That was in spite of the fact that he had 14

Holocene Foragers, Fishers and Herders of Western Kenya

attributed to “climatic and biological causes” (Cole 1963: 39). She reasoned:

the Neolithic status of the sites concerned, to paraphrase Bower (1991). It is hardly surprising, therefore, that use of the description persisted until the mid-1970s, when research conducted in the region confirmed the existence of the Neolithic.

...apart from not absolutely certain evidence of domesticated goats in the Sudan during the fourth millennium B.C., there are few indications of stockkeeping in sub-Saharan Africa, before the Iron Age, which started at various times during the first millennium A.D. in different areas; and the evidence for agriculture is equally nebulous. A Neolithic stage, in fact, is generally absent (or at least cannot be distinguished). (Cole 1963: 49)

To take a few examples, research conducted at Narosura in southwestern Kenya yielded not only all the Neolithic fossiles directeurs, but also skeletal remains of domestic animals (Odner 1972). Research conducted at several sites in the Lakes Nakuru-Naivasha basin also yielded similar artefacts and faunal remains in which domestic animals were predominant (OnyangoAbuje 1977a, 1977b). However, none of the sites investigated then yielded remains of domestic plants mainly because little effort was made to recover any. Indeed, until fairly recently there had been no concerted search for plant remains since Neolithic research started in East Africa. In the absence of direct evidence for cultivation, circumstantial evidence in the form of polished artefacts and grindstones was used to support the case for Neolithic cultivation. It was, for instance, reasoned that grindstones were used for preparing domestic grains for consumption and polished stone axe-heads for preparing land for cultivation (Onyango-Abuje 1976, 1977b).

In addition, she disputed the presence of domestic stock at Hyrax Hill, not because she had re-studied the faunal remains from the site and discovered they had been misidentified, but because early food production in the region contradicted her belief or the general colonial mentality. Her arguments, however, did not demonstrate there was no local Neolithic. In the above quotation, for example, she acknowledges that there were “few indications of stockkeeping... before the Iron Age”, while denying the very existence of the Neolithic. This acknowledgement is very significant when viewed against the little Neolithic research that had been done in East Africa by the early 1960s.

The ‘Pastoral Neolithic’

The existence of the East African Neolithic was also demonstrated by the findings of another research conducted in central Kenya, involving the dating of Later Stone Age and Neolithic sites as well as analyses of museum collections, controlled surface collections and finds made through test excavations of several sites. Analysis of the faunal samples associated with Neolithic pottery – Narosura, Elmenteitan, Akira and Maringishu – revealed a preponderance of domestic cattle and sheep/goat (Bower et al. 1977).

The above arguments reinforced use of the description ‘Stone Bowl Culture’ for the Neolithic cultures. Although this description emphasised the elevation of stone bowls to Neolithic fossiles directeurs, no significant efforts were made to study what the artefacts were used for. So far, the only study devoted to the stone bowls has been Merrick’s (1973) examination of their variation in size and shape. Given the contexts from which many of the stone bowls have been recovered, they could have been used for food preparation (domestic or wild), ochre pulverisation, mortuary ritual, or some other purposes.

All this confirmed the status of the East African Neolithic and demonstrated that domestic animals were known in the region prior to the Iron Age. Thereupon, the Neolithic was renamed the ‘Pastoral Neolithic’ because of the observed dominance of domestic animals in Neolithic faunal samples. The ‘Pastoral Neolithic’ itself was defined as a period that was characterised by a Later Stone Age technology and a pastoral economic base relying heavily on domestic cattle and sheep/goat (Bower et al. 1977: 119, fn. 2). Consequently, the description ‘Stone Bowl Cultures’ was replaced by the term ‘Pastoral

In spite of the failure to investigate what the stone bowls were used for, archaeologists working in East Africa continued to use the description ‘Stone Bowl Culture’, probably because it was less ‘controversial’. For example, although Gramly (1975) argued that the ‘Culture’ was associated with the earliest food production in the region, whenever he used the term Neolithic, he enclosed it in inverted commas, preferring to use the description ‘Stone Bowl Culture’ instead. This practice obscured 15

Karega-Mũnene

Neolithic’, thus effectively ending use of fossiles directeurs and circumstantial evidence for food production in the identification of Neolithic cultures.

Nderit ware encompasses the pottery that had originally been labelled ‘Gumban A’ by Leakey (1931). It is characterised by carinated or narrow-mouthed bowls, which are generally decorated on both surfaces. Decoration on the external surface is extensive and consists of closely set cuneiform impressions. The internal surface is decorated by scores or grooves, line incisions, or jabs, or not decorated at all. This definition fitted pottery from the type-site (Nderit Drift), Stable’s Drift, Makalia burial site, Hyrax Hill, Salasun, Akira and Lukenya Hill in Kenya and at Nasera and Seronera in Tanzania (Wandibba 1977, 1980). Other sites where the ware has been reported include site GaJi 2, Dongodien, Kangatotha and Jarigole in the Lake Turkana basin (Barthelme 1984, 1985; Mehlman 1989; Nelson 1995; Robbins 1972).

Attempts at Regional Syntheses Throughout the first half of this century no attempts were made to undertake regional syntheses of the cultures that had been identified in East Africa. Such an attempt was first made in the early 1960s by Sutton (1964), who reclassified the Neolithic pottery (and some Iron Age pottery) into categories A, B and C. Class A consisted of Elmenteitan pottery as represented at Gamble’s Cave II, Njoro River Cave, Naivasha Railway rockshelter and Long’s Drift. Class B was used for ‘Gumban A’ and Hyrax Hill pottery, whilst Class C consisted of all roulette-decorated pottery, some of which belonged to ‘Gumban B’. Class A was considered to be the oldest because it was presumably created by hunter-gatherer communities, whilst Class C was the youngest because of its association with iron working and, therefore, sedentary communities (Sutton 1964).

Narosura ware was named after the site where it was first identified (Odner 1972). Its typical vessels are open- and narrow-mouthed bowls, bowls with slightly everted rims and beaker-like vessels. Decoration is by comb stamping or line incisions which are sometimes delimited with horizontal lines or divided by zig-zag reserved bands (Wandibba 1977, 1980). This ware is represented at the type-site (Narosura), Prolonged Drift, Naivasha Railway rockshelter, Ndabibi, Lukenya Hill (GvJm 2, GvJm 3, GvJm 22, GvJm 44), Crescent Island, Crescent Causeway site, Maringishu, Prospect Farm, Salasun and Akira in Kenya. In Tanzania it is represented at Mumba-Hohle, Soit Nasera, Jangwani I and Ishimijenga (Bower et al. 1977; Gramly 1975; Mehlman 1989; Wandibba 1977, 1980).

Each of these classes encompassed a lot of variation. For instance, while the majority of the pottery used in the classification was Neolithic, some Iron Age pottery, notably ‘Gumban B’ which was renamed Lanet ware after the Lanet Iron Age site near Nakuru town (Posnansky 1967a), was also included. Secondly, although radiocarbon dates had been obtained on some of the sites which yielded the pottery, beginning with Njoro River Cave in the late 1950s (Barendsen et al. 1957), the dating was ignored in the classification. In consequence, no meaningful comparisons could be made between the resulting classes, hence the persistence of a need for a better classification of all the Neolithic pottery.

Maringishu ware was originally associated with the Hyrax Hill variant of the Neolithic in Kenya. Its typical vessel is the ovoid beaker which was first identified at Hyrax Hill by Leakey (1945). Decoration on this and other less common Maringishu vessel types is of a curvilinear motif formed by broad undulating horizontal ridges. The elliptical spaces between the ridges are filled with small linear impressions, punctations, or very fine cord roulette impressions (Wandibba 1977, 1980). The ware is represented at the eponymous site (Maringishu), Hyrax Hill and Nderit Drift (Bower et al. 1977; Wandibba 1977, 1980).

Wandibba (1977, 1980) made an attempt towards that goal; but his work was limited to the southern part of the Kenya Rift Valley. In that study, specific attributes – firing, decorative techniques and motifs, vessel shapes and such features as lugs, handles, spouts and knobs – were used to compare and group the pottery into wares. Consequently, five Neolithic wares, namely, Akira, Maringishu, Narosura, Nderit and Remnant (Elmenteitan) were defined (Figures 2.2 - 2.5).

16

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 2.2 Maringishu ware (after Wandibba 1980) 17

Karega-Mũnene

Figure 2.3 Elmenteitan (Remnant) ware (after Wandibba 1980)

18

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 2.4 Nderit ware (after Wandibba 1980)

19

Karega-Mũnene

Figure 2.5 Narosura ware (after Wandibba 1980)

20

Holocene Foragers, Fishers and Herders of Western Kenya

Akira ware was originally described as Thin Incised Panelled ware because of the small thickness (5 mm or less) of the walls of the vessels (Bower 1973a). The ware is characterised by highly burnished vessels, most of which are flat-based. Decoration consists of panels of very fine incised lines, but narrow bands of applied decoration occur on some vessels (Wandibba 1977, 1980). The ware is represented at Akira and Lukenya Hill – sites GvJm 2, GvJm 14, GvJm 22 and GvJm 44 – in Kenya; Seronera and Nasera in Tanzania; and in Karamoja in Uganda (Bower 1973a; Bower et al. 1977; Gramly 1975; Mehlman 1989; Robbins 1972; Wandibba 1977, 1980).

were found. The pottery was, however, included in an alternative scheme in which Nderit was the oldest, followed by Kansyore, Narosura, Akira and Maringishu, in that order (Bower et al. 1977). Interestingly, this latter scheme excluded Remnant (Elmenteitan) ware. While Sutton’s (1964) scheme has now been relegated to history, Wandibba’s wares and chronological scheme (Bower et al. 1977; Wandibba 1977, 1980) are open to question. Indeed, neither the wares nor the scheme can be used to clarify the problem of chronology and spatial variation in Neolithic pottery. Besides Maringishu and Narosura, which are thought to have been contemporaneous (Wandibba 1977, 1980), two or more wares have often been found together in the same deposits. At Seronera in Tanzania, for example, Kansyore pottery was associated with Nderit ware (Bower 1973a) and at Gogo Falls Kansyore was found in the same deposits with Elmenteitan, Akira and Urewe wares (Robertshaw 1985, 1991; Personal observation). In addition, there is considerable overlap in the chronometric dates that are currently available for the wares (Table 2.1). As we shall learn especially in the closing chapters of this study, these conditions could be indicative of one of several things. They could be a reflection of differences in the social and ritual functions of the pottery, contemporaneity between the wares, poor definition of the wares, exchange/trade patterns between neighbouring groups, cultural diversity, intermarriages, or stratigraphic mixing (e.g., Bower et al. 1977; Karega-Mũnene 1996).

Remnant (Elmenteitan) ware encompasses what had been described as Elmenteitan and ‘evolved Elmenteitan’ pottery (Leakey 1931; Leakey and Leakey 1950). The ware is generally undecorated, but panels of punctations and irregular rim milling occur on some vessels. The common vessel types are open-mouthed bowls with slightly everted lips, large open-mouthed cauldron-like vessels, platters and carinated cups. Features like lugs, handles and spouts occur on some vessels (Wandibba 1977, 1980). Sites bearing this ware include Remnant site, Gamble’s Cave II, Bromhead’s site, Keringet Cave, Njoro River Cave and Maasai Gorge rockshelter (Bower et al. 1977; Wandibba 1977, 1980). Wandibba (1977, 1980) also made attempts to order the five wares into a provisional chronological sequence. On the basis of the radiometric dates available at the time and the stratigraphic sequences at Salasun and Lukenya Hill, he suggested that Nderit ware was the oldest, followed by Narosura, Remnant (Elmenteitan) and Akira and Maringishu, the latter two being contemporaneous. Preliminary studies of faunal remains from Neolithic sites in the region suggested that Maringishu, Narosura, Remnant (Elmenteitan) and Akira wares were associated with domestic animals (Bower et al. 1977).

More recent attempts to unravel the problem of chronology and spatial variation in Neolithic pottery have been unsuccessful. One of these has involved use of average link cluster of such variables as location of decoration on the vessel, decorative motifs and techniques and vessel forms on the known wares. Consequently, it has been suggested that the wares should be referred to as ‘traditions’ and that their present names should be replaced with “nonsense names with an African flavour”. Thus, Kansyore should be renamed’Oltome’ tradition, Narosura ‘Oldishi’ tradition and Maringishu ‘Olmalenge’ tradition (Collett and Robertshaw 1983a).

It is important to note here that the schemes proposed by Sutton (1964) and Wandibba (1977, 1980) excluded Kansyore, the only pre-Iron Age pottery then known in the Lake Victoria basin. At the time when those schemes were proposed Kansyore pottery was poorly known largely because its distribution lay outside the Rift Valley, where the rest of the Neolithic pottery

It has been argued that these names are preferable to using a letter system, as Sutton (1964) had done, because they do not require “the elimination of an acceptable name, 21

Karega-Mũnene

Table 2.1 Chronometric dates on East African Neolithic sites. Site Name

Site type

Pottery

Date (BP)

Lab/specimen No.

Material

Reference

Ileret

Open habitation

Ileret

4000 ± 140

GX-4643/A

Apatite

Barthelme 1985

Dongodien

Open habitation

Nderit Nderit Nderit Nderit Nderit

4580 ± 170 4100 ± 125 3860 ± 60 3945 ± 135 3405 ± 130

GX-4642-II-A SUA-637-B P-2610 SUA-637 GX-4642-I-A

Apatite Charcoal Charcoal Charcoal Charcoal

Barthelme 1985 Barthelme 1985 Barthelme 1985 Barthelme 1985 Barthelme 1985

GaJi 2

Open habitation

Nderit Nderit

4160 ± 110 3970 ± 60

SUA-634 P-2609

Charcoal Charcoal

Barthelme 1985 Barthelme 1985

Lopoy

Open habitation

Turkwel Turkwel

870 ± 80 950 ± 80

UCLA-2124G UCLA-2124J

Burnt soil Charcoal

Lynch & Robbins 1979 Lynch & Robbins 1979

Apeget I

Open habitation

Turkwel

1800 ± 300

UCLA-2124K

Charcoal

Lynch & Robbins 1979

Kangatotha

Open habitation

Nderit

5020 ± 220

N-814

Burnt soil

Robbins 1972

Namoratunga

Burial

?Turkwel

2285 ± 165

GX-5042-A

Bone

Lynch & Robbins 1979

Ngenyn

Open habitation

?Turkwel ?Turkwel ?Turkwel

2080 ± 130 2020 ± 130 1970 ± 150

UCLA-1322 BIRM-770 BIRM-767

Collagen Charcoal Collagen

Hivernel 1978 Hivernel 1978 Hivernel 1978

Maringishu

Open habitation

Maringishu

1695 ± 105

GX-4466-A

Apatite

Bower et al. 1977

Njoro River Cave

Burial

Elmenteitan Elmenteitan Elmenteitan Elmenteitan

2920 ± 80 2900 ± 75 3090 ± 65 3165 ± 100

Y-91 Y-220 Y-221 Y-222

Charcoal Charcoal Charcoal Charcoal

Barendsen et al. 1957 Merrick & Monaghan 1984 Merrick & Monaghan 1984 Merrick & Monaghan 1984

Keringet Cave

Burial

Elmenteitan Elmenteitan Elmenteitan

2910 ± 115 2430 ± 110 2050 ± 110

N-653 N-654 N-655

Charcoal Charcoal Charcoal

Cohen 1970 Cohen 1970 Cohen 1970

Prolonged Drift

Open habitation

?Narosura ?Narosura

2530 ± 160 2315 ± 150

GX-5753G GX-5753A

Collagen Apatite

Gifford et al. 1980 Gifford et al. 1980

Prospect Farm

Open habitation

Elmenteitan Elmenteitan

2690 ± 80 2910 ± 110

UCLA-1234 N-651

Charcoal Charcoal

Cohen 1970 Cohen 1970

Remnant

Open habitation

Elmenteitan Elmenteitan

2315 ± 150 1355 ± 145

GX-4324 GX-4634

Charcoal Charcoal

Bower et al. 1977 Bower et al. 1977

Enkapune ya Muto

Rockshelter

Nderit

3990 ± 70

ISGS-2308

Charcoal

Marean 1992

Maasai Gorge

Rockshelter

Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan

2865 ± 150 2595 ± 135 2515 ± 140 2495 ± 150 2325 ± 145 1560 ± 135 1545 ± 135

GX-4462-A GX-5346 GX-4471-A GX-4345-A GX-5344 GX-4311-C GX-4312

Charcoal Charcoal Apatite Apatite Charcoal Collagen Charcoal

Bower & Nelson 1978 Ambrose 1985 Onyango-Abuje 1977b Ambrose 1985 Ambrose 1985 Bower & Nelson 1978 Bower & Nelson 1978

Crescent Island

Open habitation

Narosura Narosura Narosura Narosura Narosura

2795 ± 155 2660 ± 120 2660 ± 160 2535 ± 140 2405 ± 150

GX-4587-G GX-4585-A GX-4589-G GX-4586-G GX-4588-A

Gelatin Apatite Gelatin Gelatin Apatite

Onyango-Abuje 1977b Onyango-Abuje 1977b Onyango-Abuje 1977b Onyango-Abuje 1977b Onyango-Abuje 1977b

Naivasha Railway

Open habitation

Narosura

2000 ± 135

GX-4583-G

Gelatin

Onyango-Abuje 1977b

Ndabibi

Open habitation

Narosura Narosura Narosura

2225 ± 155 1665 ± 145 1415 ± 150

GX-4465-G GX-4463-A GX-4464-G

Gelatin Apatite Gelatin

Bower et al. 1977 Bower et al. 1977 Bower et al. 1977

Causeway site

Open habitation

Narosura

2045 ± 125

GX-4319-A

Apatite

Bower et al. 1977

Rangong’

Rockshelter

?Kansyore

2315 ± 185

GX-1100

Charcoal

Gabel 1969

22

Holocene Foragers, Fishers and Herders of Western Kenya

Table 2.1 continued Site Name

Site type

Pottery

Date (BP)

Lab/specimen No.

Material

Reference

Luanda

Shell midden

Kansyore Kansyore

6740 ± 80 8240 ± 245

Pta-3139 GX-8743

Shell Apatite

Robertshaw et al. 1983 Robertshaw et al. 1983

Kanjera West

Shell midden

Kansyore Kansyore

5845 ± 310 5700 ± 100

GX-8744 OxA-828

Apatite Collagen

Robertshaw et al. 1983 Robertshaw 1991

White Rock Point

Shell midden

Kansyore

4015 ± 260

GX-8745

Apatite

Robertshaw et al. 1983

Gogo Falls

Open habitation

Kansyore Kansyore Elmenteitan Elmenteitan Elmenteitan Elmenteitan Kansyore Kansyore Kansyore Kansyore

7300 ± 500 5805 ± 185 1710 ± 70 1900 ± 70 1990 ± 80 1770 ± 80 3020 ± 100 3613 ± 115 2617 ± 194 2474 ± 65

OxA-745 GX-8536 HAR-6261 HAR-6260 HAR-6254 HAR-6252 HAR-6262 334-106 292-8 292-9

Burnt tooth Apatite Charcoal Charcoal Charcoal Charcoal Charcoal Obsidian Obsidian Obsidian

Gowlett et al. 1987 Robertshaw et al. 1983 Robertshaw 1991 Robertshaw 1991 Robertshaw 1991 Robertshaw 1991 Robertshaw 1991 Robertshaw 1991 Robertshaw 1991 Robertshaw 1991

Nsongezi

Rockshelter

?Kansyore

1025 ± 150

M-1113

Hearth

Crane & Griffin 1962

Akira

Open habitation

Akira Akira Akira

1965 ± 140 1775 ± 150 1255 ± 140

GX-4386-G GX-4385-G GX-4384

Gelatin Gelatin Charcoal

Bower et al. 1977 Bower et al. 1977 Bower et al. 1977

Salasun

Open habitation

Nderit Nderit Narosura Narosura

7255 ± 225 6595 ± 235 2990 ± 170 2680 ± 150

GX-4422-A GX-4469-A GX-4468-A GX-4421-A

Apatite Apatite Apatite Apatite

Bower et al. 1977 Bower et al. 1977 Bower et al. 1977 Bower et al. 1977

Ngamuriak

Open habitation

Ngamuriak

Open habitation

Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan Elmenteitan

2135 ± 140 1940 ± 140 2038 ± 78 2076 ± 81 1764 ± 62 1661 ± 141 1540 ± 51 1783 ± 99 1952 ± 72 2927 ± 74

GX-8533 GX-8534 334-1019 334-1032 411-10 411-5 411-6 411-7 411-8 411-9

Charcoal Charcoal Obsidian Obsidian Obsidian Obsidian Obsidian Obsidian Obsidian Obsidian

Marshall 1986 Marshall 1986 Robertshaw 1990b Robertshaw 1990b Robertshaw 1990b Robertshaw 1990b Robertshaw 1990b Robertshaw 1990b Robertshaw 1990b Robertshaw 1990b

Sambo Ngige

Open habitation

Elmenteitan Elmenteitan Elmenteitan

1326 ± 74 1346 ± 78 1419 ± 52

292-1 292-2 292-3

Obsidian Obsidian Obsidian

Robertshaw 1990b Robertshaw 1990b Robertshaw 1990b

Lemek Northeast

Open habitation

Narosura Narosura Narosura

2225 ± 140 2506 ± 70 2566 ± 78

GX-8532 334-25A 334-25B

Charcoal Obsidian Obsidian

Marshall 1986 Robertshaw 1990b Robertshaw 1990b

Narosura

Open habitation

Narosura

Open habitation

Narosura Narosura Narosura Narosura Narosura

2760 ± 115 2660 ± 115 2640 ± 115 2360 ± 115 2260 ± 115

N-702 N-701 N-703 N-700 N-496

Charcoal Charcoal Charcoal Charcoal Charcoal

Odner 1972 Odner 1972 Odner 1972 Odner 1972 Odner 1972

Lukenya 3

Rockshelter

Nderit Nderit

1806 ± 105 1520 ± 155

GX-3539 N-1827

?Charcoal Charcoal

Gramly 1975 Gramly 1975

Lukenya 14

Rockshelter

Akira

1890 ± 110

N-1884

?Charcoal

Gramly 1975

Lukenya 22

Rockshelter

Narosura Akira Akira

2250 ± 50 1520 ± 50 1330 ± 100

UCLA-1709-C UCLA-1709-D N-1076

Collagen ?Charcoal Charcoal

Gramly 1975 Gramly 1975 Gramly 1975

Lukenya 44

Rockshelter

Nderit Narosura Akira Akira Narosura

3290 ± 145 2085 ± 135 2030 ± 125 1775 ± 150 1710 ± 135

GX-5348 GX-4160-A GX-4507-A GX-4507-G GX-4160-C

Charcoal Apatite Apatite Gelatin Collagen

Nelson & Kimengich 1984 Bower et al. 1977 Bower et al. 1977 Bower & Nelson 1978 Bower & Nelson 1978

23

Karega-Mũnene

Table 2.1 continued Site Name

Site type

Pottery

Date (BP)

Lab/specimen No.

Material

Reference

Lukenya 47

Rockshelter

Un-named Un-named

1340 ± 145 970 ± 130

GX-4161-A GX-4161-C

Apatite Collagen

Bower et al. 1977 Nelson & Kimengich 1984

Lukenya 48

Rockshelter

Narosura Narosura

1810 ± 135 1600 ± 130

GX-5347-G GX-5347-A

Gelatin Apatite

Nelson & Kimengich 1984 Nelson & Kimengich 1984

Lukenya 52

Rockshelter

Narosura

1840 ± 140

GX-5772-A

Apatite

Nelson & Kimengich 1984

Seronera

Rockshelter

Nderit

2020 ± 115

N-1067

Charcoal

Bower 1973a

Soit Nasera

Rockshelter

Kansyore

5400 ± 150

ISGS-444

Apatite

Mehlman 1979

Kansyore

4720 ± 150

ISGS-444

Collagen

Mehlman 1979

Nyang’oma

Rockshelter

?Kansyore

2640 ± 120

N-493

Charcoal

Soper & Golden 1969

Ngorongoro

Burial

Un-named

2260 ± 180

GX-1234

Bone

Sassoon 1968

Maua Farm

Open habitation

Nderit Nderit Nderit

4140 ± 200 2160 ± 190 1545 ± 140

GX-3346 GX-3347 GX-3348

Charcoal Charcoal Charcoal

Mturi 1986 Mturi 1986 Mturi 1986

Mumba-Hohle

Rockshelter

Kansyore Kansyore

4860 ± 100 4890 ± 70

UCLA-1913 FRA-1

Charcoal Charcoal

Mehlman 1979 Mehlman 1979

Elmenteitan, already in the literature and its replacement with a letter” (Collett and Robertshaw 1983a: 121, fn. 3). This implies that the names of the other four wares defined by Wandibba (1977, 1980) are neither acceptable nor in the literature. This is a misleading argument because the other names are just as acceptable and, as Bower (1991: 65) correctly oberves, "virtually universal in application, and are deeply entrenched in Pastoral Neolithic literature" as the Elmenteitan. (e.g., Ambrose 1982, 1984; Barthelme 1985; Bower 1984, 1991; Bower et al. 1977; Gifford-Gonzalez and Kimengich 1984; Hivernel 1978; KaregaMũnene 1993, 1996; Marean 1992; Marshall 1986, 1990, 1991; Mehlman 1977, 1979, 1989; Mturi 1986; Nelson 1995; Nelson and Kimengich 1984; Onyango-Abuje 1977a, 1977b, 1980; Wandibba 1977, 1980, 1990; cf. Robertshaw 1991, 1993; Schoenbrun 1990, 1993).

ware, which is ‘named after a single type-site’, has consistently been referred to as a ‘tradition’! These inconsistencies are confusing especially to those who are less familiar with East African archaeology. Instead of merely changing the names of the wares, thus muddling the fundamental issues, what was and still is required is a concerted study of all the ceramic assemblages from the region. This should not only involve appraisal of the existing definitions, but also examination of the suitability of the chronological and spatial schema that were originally devised using samples from a very small part of East Africa to the whole region. Attempts have also been made to examine the relationship between Neolithic sites in East Africa in terms of their chronology, geographic distribution, economy, pottery and stone artefacts. Consequently, the sites have been delineated into three groups which are considered to be broadly contemporaneous, namely, Savanna Pastoral Neolithic (SPN), Elmenteitan and Eburran. The SPN group consists of two sub-groups which are distinguished by dating, geographic distribution and elevation. The lowland SPN sites are dated to about 5,200 BP - 3,300 BP and are restricted to the low-lying areas of northern Kenya. In

The other reason that has been offered in favour of the proposed terminology is that “ a tradition cannot be recognized from a single type site” (Collett and Robertshaw 1983a: 121, fn. 3; Robertshaw et al. 1983: 34). Yet, in a recent publication where the proposed nomenclature has been effected (Robertshaw 1991), Urewe

24

Holocene Foragers, Fishers and Herders of Western Kenya

contrast, the highland SPN sites are found in central Kenya and northern Tanzania, where they are dated to about 3,300 BP - 1,300 BP (Ambrose 1982, 1984).

of a given ethnic or linguistic community, and not by the entire community. These issues will be discussed further in the closing chapters of this study.

The Eburran consists of five phases, all of which are represented in the Lakes Nakuru-Naivasha basin. Phases 1 - 4 are pre-Neolithic, dating to ca. 12,000 BP - 6,000 BP, while Phase 5 is Neolithic and dates ca. 2,900 BP - 1,900 BP. The Elmenteitan group is made up of sites that are restricted to the western side of the Rift Valley, the adjacent Mau Escarpment and the Loita plains to the west of the Mau (Ambrose 1982, 1984).

However, it must be emphasised here that the problem of defining and delineating Neolithic wares is outside the purview of the present study. Research is, however, encouraged in that direction for a better appreciation of the chronological and spatial patterning of Neolithic societies. The present study will, however, reexamine the existing interpretations of the wares, with specific reference to pottery from Gogo Falls. For this purpose, the present study will employ Wandibba’s (1977, 1980) definitions in the identification of the wares which are represented at the site.

More recent research, together with the present study, indicate that strict adherence to this classification could be misleading. The classification, for instance, places Kansyore pottery in the pre-Neolithic period rather than in the Neolithic period, where it is demonstrated to belong by the present study. Equally important is the fact that the classification is largely based on negative evidence. It is now clear, for example, that Elmenteitan sites are not necessarily restricted to the delineated areas; the Gogo Falls site which, as we have noted above, contains Elmenteitan material is well out of the outlined geographical zone (Karega-Mũnene 1986; Marshall 1986; Robertshaw 1985). Thus, the real extent of Neolithic sites in East Africa is unknown, a situation that is largely due to the confines of research.

Neolithic Research The above discussion gives the impression that Neolithic research in East Africa has been concentrated in the Rift Valley. This is a true reflection of the situation because very little is known about the Neolithic outside the narrow corridor defined by the Rift Valley. In the Lake Victoria basin, for instance, the little we know about that period comes from ceramic studies of the poorly understood Kansyore pottery. This pottery is named after the type-site, Kansyore (sometimes misspelt Kantsyore) Island in the River Kagera on the Uganda/Tanzania border (Figure 2.1). Excavations conducted there by Chapman (1967) yielded stone and iron artefacts, iron slag, oyster shells, human and animal bones, recent roulette-decorated pottery and Iron Age pottery, plus the pottery which she named Kansyore. The roulette-decorated pottery was comparable to contemporary pottery from Luoland in the Lake basin and was most abundant in the upper levels, whilst the Iron Age pottery was similar to Urewe pottery, which will be discussed at length below.

The Neolithic wares as initially defined by Wandibba (1977, 1980), may have been rigidly defined. If this is correct, the definition could have over-emphasised similarity at the expense of internal diversity. Therefore, there is need to re-study all the existing Neolithic pottery in order to appreciate the ordering of the Neolithic through time and space. There is also need to reexamine the interpretation of the Neolithic wares. Present evidence suggests that their correlation with specific subsistence activities and ethnic or linguistic groups is an inadequate explanation, largely because it ignores several of other possibilities. For example, it fails to recognise that the stylistic differences that have been used in defining the wares could be due to the function(s) of the vessels concerned, internal cultural diversity and/or exchange patterns between neighbouring groups belonging to the same linguistic or ethnic group. After all, pots are made by individuals who are only members

Unlike the Neolithic wares discussed above, Kansyore pottery was defined on the basis of its surface treatment and not on vessel types. That was because there was not a single reconstructable or complete vessel at the eponymous site. The pottery has a gritty surface and is poorly fired; its decoration consists of rounded shallow grooves, stab-and-drag grooves and parallel lines of impressions. Some of the sherds exhibited overall decoration of small impressions on the external surface and others 25

Karega-Mũnene

decoration of similar impressions on the inner side of the rim (Chapman 1967). The age of this pottery was difficult to determine because it was mixed with Urewe and the roulette-decorated pottery in the site’s stratigraphy. However, in the absence of radiometric dates for the site, relative ages of the pottery types represented at the site were established by comparing their respective weights through the site’s stratigraphy. Undoubtedly, this was a clumsy method of establishing the chronology of the pottery because it made the erroneous assumption that weight was proportional to frequency. The method also failed to recognise that the composition of the clays used in making the pots and the thickness of the sherds could have affected the weights.

however, Kansyore is thought to be the older of the two since most of it was associated with the charcoal sample dated to the third millennium BP (Table 2.1). Most of the faunal remains that were also found at these sites were counted and discarded in the field. The remaining few were identified to wild mammals and fish. The obvious bias in the resulting species list notwithstanding, it was asserted that the creators of Kansyore pottery at the sites were hunters and fishers (Bower 1973a; Soper and Golden 1969). Excavations at Mumba-Hohle yielded Kansyore pottery, Narosura ware and Iron Age Lelesu ware. Of the three pottery types, Kansyore was considered to be the oldest, followed by Narosura and Lelesu, in that order. This sequence was established on the basis of the distribution of the three pottery types in the site’s stratigraphy. Radiocarbon dates on charcoal and bone samples recovered from the human burials associated with Kansyore pottery at the site place Kansyore in the fifth millennium BP (Table 2.1). Excavations at Soit Nasera (Nasera Rock) also yielded Kansyore pottery and Narosura and Akira wares. According to the site’s stratigraphy, Kansyore is the oldest pottery; radiocarbon dates obtained place it between the fifth and sixth millennium BP (Mehlman 1977, 1979).

The weighing exercise resulted in a scheme whereby Kansyore pottery was considered to be older than Urewe because its sherds from the lower levels were heavier! A date of about 1000 BP was suggested for Kansyore (Chapman 1967). This date was literally borrowed from Nsongezi rockshelter which is located a few kilometres to the east of Kansyore Island. It must be added here that the Nsongezi radiocarbon date of 1025 ± 150 BP (M-1113) (Crane and Griffin 1962) was obtained on a sample that had no direct association with Kansyore pottery. Rather, the sample had been collected from the level immediately below which Kansyore and other pottery were found (Pearce and Posnansky 1963). Equally significant is the poor representation of Kansyore at Nsongezi – by a single sherd. Recent examination of the decoration on the sherd has raised doubts about the existence of Kansyore at Nsongezi because it indicates the sherd belongs to Urewe ware and not Kansyore (Wandibba 1990).

At the Seronera rockshelter Kansyore pottery was found together with Nderit and Akira wares and recent plain and rouletted pottery. Although three radiocarbon dates were obtained on charcoal samples from the site, none of them could be directly associated with Kansyore because of the site’s mixed stratigraphy. The relative age of the pottery was, however, established on the basis of calculations of its stratigraphic distribution in relation to the other pottery types. This resulted in a scheme which suggested Kansyore was older than Akira ware but younger than Nderit ware (Bower 1973a). Interestingly, the presence of Kansyore at the site is open to question since the sherds from which the pottery was originally identified are now attributed to Nderit ware (Collett and Robertshaw 1980; cf. Mehlman 1989: 456).

The reported presence of Kansyore pottery at Hippo Bay rockshelter and Karagwe in Uganda as well as at Kitulu in Tanzania (Soper and Golden 1969) may also be open to question. Although the status of the sherds has not been questioned heretofore, it is likely that the evidence is inadequate given the available evidence is also in the form of isolated sherds.

In central Kenya Kansyore pottery has been reported at Lukenya Hill GvJm 14 where it was associated with Akira ware and at Salasun (Bower et al. 1977; Bower and Nelson 1978; Gramly 1975). At the latter site, a single sherd that was stratified above Nderit ware and below Narosura ware represents the pottery (Bower et

Kansyore pottery has also been reported at Nyang’oma and Chole in the Tanzania Lake Victoria basin and further to the southeast of the lake at Mumba-Hohle and Soit Nasera. At Nyang’oma and Chole the pottery was found together with Iron Age pottery. At Nyang’oma, 26

Holocene Foragers, Fishers and Herders of Western Kenya

al. 1977; cf. Mehlman 1979: 91). Wandibba (1990) and Collett and Robertshaw (1980) have disputed the identification of Kansyore at both sites. They are of the opinion that the sherds in question were wrongly identified, but they fall short of correcting the misidentification.

1971a, 1971b), Gatung’ang’a (Siiriainen 1971) and Lanet pottery (Posnansky 1967a), plus other poorly known pottery types like Kisii Soft ware, Kisii Button-Necked pottery (Bower 1973b) and Deloraine pottery (Ambrose 1982; Cohen 1972). The latter are known only from preliminary work at one or just a few sites, all of which have been dated to a few centuries ago. Of these wares, we are concerned here with Urewe only because the geographical distribution of the other wares falls outside the purview of this study.

Other sites bearing Kansyore material in Kenya are Luanda, Kanjera West, Kanam East, Kanam Two and White Rock Point (Robertshaw et al. 1983); Ojolas rockshelter (Soper and Golden 1969); Ugunja (Mosley 1992); and Gogo Falls (Collett and Robertshaw 1980; Karega-Mũnene and Kahĩnju 1992; Robertshaw 1985, 1991). All these sites are located in the Lake Victoria basin. Small-scale excavations conducted at the first five sites yielded remains of birds, reptile, mammals and freshwater shells and fish, modified bones, stone artefacts, plus Kansyore and recent Luo pottery. Four radiocarbon determinations which fall between the sixth and early ninth millennium BP were obtained for Kansyore levels at Luanda, Kanjera West and White Rock Point sites (Robertshaw et al. 1983). At Gogo Falls Kansyore was found together with Akira, Elmenteitan and Iron Age wares. A mixed horizon in which Kansyore pottery was predominant has two radiocarbon dates falling in the sixth millennium and eighth millennium BP (Gowlett et al. 1987; Robertshaw et al. 1983).

Urewe ware Urewe ware is named after the type-site, Urewe, which is located near Ng’iya Mission in Siaya District in the Kenya Lake Victoria basin. Of the three ‘early’ Iron Age wares, it is the oldest and the most widespread. As mentioned earlier, it is found in the Lake Victoria basin; but its distribution stretches as far as Rwanda, Burundi and eastern Democratic Republic of Congo to the west, and the Victoria Nile in the Murchison Falls region of Uganda to the north (Figure 2.1). The ware was initially described as ‘dimplebased ware’ because it appeared to be characterised by what looked like a thumb imprint (or ‘dimple’) in the base of the vessels. Subsequently, however, the name ‘dimple-based ware’ was found to be inappropriate. Investigations in Uganda and in the Democratic Republic of Congo, for instance, revealed that the ‘dimpled’ bases did not occur on all the pottery attributed to the ware (Nenquin 1959; Posnansky 1961b). More importantly, it was reasoned that the name ‘dimple-based ware’ was descriptive and, therefore, not in accordance with the custom of naming cultures after their respective type-sites. Consequently, it was suggested that the pottery be renamed Urewe ware after the type-site (Posnansky 1961b, 1961c), a suggestion which was readily accepted.

Iron Age Research The Iron Age in the interlacustrine region of East and Central Africa has traditionally been divided into two phases: the ‘Early/early Iron Age’ and ‘later/Later Iron Age’ (Phillipson 1977, 1993; Soper 1982; Van Noten 1979). These phases are best known from pottery that was made and used during the Iron Age period and not from iron artefacts, as the term Iron Age seems to imply. The division between the two phases is arbitrary because pottery belonging to both phases is yet to be found at the same site. Moreover, the sites where ‘later Iron Age’ pottery has been found represent the earliest known iron-using communities in those areas. Therefore, using the term ‘later/Later Iron Age’ for those communities is misrepresenting the actual situation.

In addition to the ‘dimples’, the ware is characterised by multiple bevelled and often overhanging rims. Decoration is confined to the area immediately below the bevels on the external surface and sometimes on the body. The decoration motifs include bands of oblique or crosshatched incisions, scrolls and circles often combined with bands of straight or curvilinear grooves with occasional triangular-shaped kinks. Sometimes the bands of decoration are bordered by rows of dot impressions. Among the notable vessels are wide- and narrow-mouthed bowls,

The ‘early’ Iron Age consists of Urewe, Lelesu and Kwale wares (Soper 1967a, 1971a, 1971b, 1982) and is represented at sites dated to between ca. 2,500 BP and 600 BP (Table 2.2). The ‘later’ Iron Age consists of Maore (Soper 1967b; Odner 27

Karega-Mũnene

bowls with flared rims, globular pots with everted rims and elongated vessels (Leakey et al. 1948).

and Sesse Islands in Lake Victoria (Fagan and Lofgren1966a; Posnansky 1967b), Hippo Bay near Entebbe (Brachi 1960) and several other sites that have been identified through the study of museum collections (Posnansky 1961b). These latter collections were not made through systematic field investigations, but through uncontrolled surface collection principally by non-archaeologists who donated them to the museums. Therefore, their stratigraphic or cultural associations are uncertain.

At the type-site, Urewe ware was associated with iron slag and artefacts, fragments of tuyeres and patches of baked or burnt clay which were thought to be hearths. These associations also obtain at the nearby Yala Alego, Magari, Ng’iya, Aluala, Ulore I and Ulore II sites (Leakey et al. 1948; Soper 1969). Radiometric dates obtained at Ulore II place the ware between 1,700 BP and 1,500 BP and at Yala Alego about 1,550 BP (Soper 1969). Additional chronometric dates placing the ware between 2,400 BP and 1,300 BP have been obtained at Randhore, Rangong’ and Nyaidha (Gabel 1969) (see Table 2.2).

The other sites in the country where Urewe has been found are located in the Chobi area near the Rivers Nile-Chobi confluence (Fagan and Lofgren 1966b; Soper 1971c). The pottery was found in association with iron artifacts and slag, fragments of tuyeres, rings of burnt clay, grindstones and a previously unknown pottery (which was named Chobi) in deposits dated to ca. 1,640 BP. The Chobi pottery is characterised by hemispherical bowls and wide-mouthed pots with a slight neck whose rims are either vertical or slightly out-turned. The finger-marks from the moulding, especially on the upper part of the vessels were left unsmoothed, giving the impression that they were intended decoration. The association between the two types of pottery has been interpreted as an indication of their contemporaneity in that area (Soper 1971c). So far, the present author is not aware of any subsequent research in the area confirming or refuting this observation.

Some of the charcoal samples that provided the latter dates were recovered from archaeological deposits bearing grindstones, pestle rubbers and/or remains of domestic animals. These associations have been interpreted as indicating the sites belong to the Neolithic period as well (Onyango-Abuje 1977b), the underlying assumption being that the lithic artefacts and domestic fauna were peculiar to the Neolithic. This assumption is potentially misleading because those finds could also have belonged to the Iron Age period, as is the case at the Iron Age sites of Engaruka and Buhaya in Tanzania (Sassoon 1967, Schmidt 1978). In Tanzania, Urewe ware has been found at Rugomora Mahe (Katuruka) and Makongo to the southwest of Lake Victoria. These sites are probably the best-studied Urewe sites in the whole of East Africa, especially with regard to iron-working technology and cultural change during the Iron Age period. The finds made at the sites include Urewe and ‘later’ Iron Age pottery, grindstones, pestle rubbers, tuyeres, iron slag and iron smelting features. Several radiocarbon dates were obtained on charcoal samples recovered from the sites. These indicate Urewe pottery was used at the site between about 2,600 BP and 900 BP (Schmidt 1978), thus suggesting the sites are the oldest Iron Age sites in East and southern Africa.

It is important to note here that virtually everything we know about most of the sites discussed above is based on studies of surfacecollected artefacts. Indeed, very few Iron Age sites in East Africa have been excavated and, where this has been the case, the excavations have been limited to a few test trenches for purposes of obtaining charcoal samples for dating and pottery for correlation with pottery from other sites dating to the same period (Waane 1979). The only exception is the Buhaya sites where extensive excavations have been conducted. This state of affairs implies that our knowledge from the majority of the sites may be based on unrepresentative samples.

In Uganda Urewe has been reported at Nsongezi rockshelter and Kansyore Island (Chapman 1967; Pearce and Posnansky 1963) both of which are discussed above under Kansyore pottery. Other Urewe sites in the country include Lolui

Equally important is the fact that most of what we know about the ‘early’ Iron Age results from studies undertaken with the aim of ‘explaining’ the ‘migration’ of the Bantu-speaking peoples of

28

Holocene Foragers, Fishers and Herders of Western Kenya

Table 2.2 Chronometric dates of East African Iron Age Urewe sites.*

ancestors of modern Bantu-speaking peoples to East Africa. The testing itself has involved nothing more than correlating Iron Age pottery with the linguistic group although the association between the two is yet to demonstrated adequately (Waane 1979).

Site

Date (BP)

Lab. No.

Reference

Ulore II

1680 ± 110 1630 ± 110 1560 ± 95

N-435 N-436 GX-1186

Soper 1969 Soper 1969 Soper 1969

Yala Alego

1550 ± 235

N-437

Soper 1969

Randhore (?)

1310 ± 95

GX-1152

Gabel 1969

Rangong’ (?)

2315 ± 185

GX-1100

Gabel 1969

Nyaidha (?)

2230 ± 320

GX-1098

Gabel 1969

Chobi

1660 ± 125

N-125

Soper 1971c

Nsongezi

1025 ± 150

M-1113

Crane & Griffin 1962

Katuruka

2560 ± 100 2500 ± 115 2470 ± 110 2400 ± 115 1890 ± 115 1830 ± 110 1780 ± 150

RL-405 N-895 RL-406 N-890 N-891 N-892 N-898

Schmidt 1975 Schmidt 1975 Schmidt 1975 Schmidt 1975 Schmidt 1975 Schmidt 1975 Schmidt 1975

Buyozi

1610 ± 130

RL-1008

Mgomezulu 1981

Kemondo Bay II

2100 ± 230 1410 ± 110 1800 ± 110 1870 ± 130 1940 ± 150 2150 ± 210 1650 ± 140

RL-1009 RL-1014 RL-1010 RL-1011 RL-1012 RL-1013 RL-1015

Mgomezulu 1981 Mgomezulu 1981 Mgomezulu 1981 Mgomezulu 1981 Mgomezulu 1981 Mgomezulu 1981 Mgomezulu 1981

Makongo

1910 ± 100 1040 ± 100 965 ± 100

N-902 N-901 N-900

Schmidt 1975 Schmidt 1975 Schmidt 1975

Early Food Production Notably absent from virtually all of the Iron Age sites discussed above is direct evidence of cultivation and/or animal husbandry. In consequence, knowledge about the subsistence strategies of that period is adduced from historical linguistics. The necessary archaeological data are unavailable because researchers have been preoccupied with the subject of Bantu migration, hence the emphasis on the ubiquitous ceramics in the attempt to illustrate the migration. As we have noted above, the majority of the ceramics that have been studied for this purpose have been recovered through surface collections and/or very limited test excavations. In addition, only a handful of the faunal remains associated with those artifacts have been studied, but solely for taxonomic identifications (e.g., Soper and Golden 1969). Direct evidence of food production is also unavailable on virtually all the Neolithic sites investigated prior to the 1960s, the exception being Hyrax Hill. Consequently, food production was, as we have noted above, inferred from artefacts which were considered to be Neolithic fossiles directeurs. This state of affairs notwithstanding, research effort was expended solely on the reconstruction of the region’s cultural sequence, hence the accentuation on lithic artefacts and ceramics. In consequence, potential direct evidence for food production in the form of faunal remains was largely ignored:

*Randhore, Rangong’ and Nyaidha are rockshelters; the rest are open-air habitation sites. All the dates are on charcoal.

eastern and southern Africa.1 This is because researchers working on the Iron Age generally presume correspondence between the Bantu and the appearance of Iron Age cultures (e.g., Collett 1982; Ehret 1974; Huffman 1970, 1989; Nurse 1982; Phillipson 1977, 1984a; 1985; Posnansky 1961a, 1968; Soper 1971a, 1982; Van Noten 1979; Schoenbrun 1990, 1993).

Too often bones on a palaeolithic site are treated as sacrosanct and studied with care and attention as providing clues for relative dating and climatic indications whilst the humble domestic mammalia of a near modern site are dropped hastily in a bag and either forgotten or only briefly described. (Posnansky 1962: 273)

More often than not, archaeological research on the Iron Age has not specifically defined its own objectives. Instead, it appears to have been aimed at ‘testing’ the hypotheses proposed by historical linguists (e.g., Ehret 1972, 1982a, 1982b; Greenberg 1972; Guthrie 1962; Nurse 1982) about the ‘origins’ and ‘migration’ of the

Thus, the fact that archaeologists working in East Africa did not recognise the usefulness of faunal remains meant that earnest research on food production was not attempted until fairly recently. Although the situation has improved quite significantly during the last two decades or so, knowledge about the subject of food

1

The Iron Age studies under discussion were part and parcel of the Bantu Studies Project of the British Institute in Eastern Africa. The studies culminated in the publication of a series of articles on the Iron Age in a special number of Azania in 1971. 29

Karega-Mũnene

production in the region still remains scant. This is mainly because in the whole region it is only a few sites that have been investigated with the aim of addressing this problem. In addition, the geographic distribution of the sites is skewed infavour of the narrow corridor defined by the eastern Rift Valley (Figure 2.1), hence the lacunae in knowledge about the subject outside that corridor.

Further south, at Ngenyn in Baringo District, domestic sheep/goat and cattle are known from deposits dated to about 2000 BP. The deposits also yielded Akira, Narosura and probably Elmenteitan wares (Hivernel 1978). Maringishu, a site located near Lake Bogoria, yielded remains of domestic cattle from deposits dated to about 1,700 BP. The animals were associated with Maringishu and Narosura wares (GiffordGonzalez and Kimengich 1984; Bower et al. 1977). The faunal sample from the site was, however, very small (a total of 42 identifiable and non-identifiable specimens), for the absence of sheep/goat to be meaningful.

Among the extreme northern Kenya sites in the Rift Valley corridor are Ele Bor, Ileret ‘stone bowl site’ (FwJj 5), Dongodien (GaJi 4) and GaJi 2 which are located to the east of Lake Turkana. At Ele Bor, a site located close to the Ethiopia/Kenya border, domestic camel has been identified from two specimens (a tooth and foot bone) recovered from deposits dated between the fourth and sixth millennium BP. Other faunal remains from the site, with the exception of a probable sheep/goat tooth, belong to wild animals (Phillipson 1984b). At FwJj 5, Dongodien and GaJi 2 domestic sheep/goat and cattle are dated to about mid fifth millennium BP. Remains of these animals were found in association with Nderit ware at the latter two sites, whilst at FwJj 5 they were associated with stone bowls and previously unknown pottery, Ileret ware. Remains of wild bovids, fish and other aquatic animals were also found in the Neolithic deposits at the sites. This condition is interpreted to mean that fishing and hunting played a significant role, alongside animal husbandry, in the economy of the occupants of the sites (Barthelme 1984, 1985; Marshall et al. 1984).

In the Lakes Nakuru-Naivasha basin domestic fauna have been identified at several sites. At Prolonged Drift, for example, cattle and sheep/goat were found in association with Narosura ware in a horizon dated to about 2,500 BP. A range of wild animals was also represented in the very large faunal assemblage of about 165,000 specimens (Gifford et al. 1980; Gifford-Gonzalez and Kimengich 1984). Domestic animals have also been identified at Prospect Farm where they are associated with Elmenteitan and Narosura wares in deposits dated to about 2,900 BP - 2,600 BP (Cohen 1970). At Crescent Island remains of cattle and sheep/goat were found in association with Narosura ware in deposits with several radiocarbon dates, all of which lie in the middle of the third millennium BP. The nearby Causeway site also yielded remains of cattle and sheep/goat. These were associated with Narosura ware in a horizon dated to about 2,000 BP (Bower et al. 1977; Gifford-Gonzalez and Kimengich 1984; Onyango-Abuje 1977a, 1977b).

To the west of Lake Turkana, remains of domestic animals have been found at Namoratunga, Apeget I and Lopoy. At the Namoratunga sites dental remains of cattle and sheep/goat were associated with human burials dated to early third millennium BP. At Apeget I and Lopoy domestic animals were associated with hitherto unknown pottery, Turkwel ware, in deposits dated to the second millennium BP (Lynch and Robbins 1979; Robbins 1984). The geographical extent and chronology of Turkwel pottery is poorly understood because the pottery is known from only a few sites, located in one general area. Although the ceramic and chronometric evidence suggests both Apeget I and Lopoy belong to a more recent period, they have been classified as ‘Pastoral Neolithic’ sites (Ambrose 1982).

At the Naivasha Railway rockshelter, remains of cattle and sheep/goat were found together with Narosura ware in a horizon dated to ca. 2,000 BP. Maasai Gorge rockshelter also yielded remains of domestic stock from deposits dated to ca. 2,600 BP - 1,500 BP. These were associated with Elmenteitan ware. At Marula rockshelter remains of sheep/goat and cattle are dated to about 2,000 BP and at Ndabibi, in the same region, to about 2,200 BP - 1,400 BP (Ambrose 1985; Bower et al. 1977; Gifford-Gonzalez 1985; Gifford-Gonzalez and Kimengich 1984; Onyango-Abuje 1977b).

30

Holocene Foragers, Fishers and Herders of Western Kenya

To the south of Lake Naivasha cattle and sheep/goat are known from Salasun, near Mount Suswa. In the oldest deposits, with a radiocarbon determination of about 7,200 BP, the fauna are associated with Nderit ware, whilst in the deposits dated ca. 6,600 BP they are associated with the Kansyore sherd that was reportedly misidentified. In the younger deposits, dated to about 2,700 BP, the animals are associated with Narosura ware (Bower et al. 1977; GiffordGonzalez and Kimengich 1984). Cattle and sheep/goat have also been identified at Akira to the west of Salasun in deposits dated to about 2,000 BP - 1,200 BP. The deepest deposits at the site contained Akira and Elmenteitan wares, plus remains of cattle and sheep/goat. In contrast, the upper deposits contained Akira ware and were associated with remains of cattle only (Bower et al. 1977; Gifford-Gonzalez and Kimengich 1984; Nelson and Kimengich 1984).

in deposits dated to ca. 2,000 BP - 1,700 BP (Karega-Mũnene 1986, 1987; Marshall 1986, 1991). To the east of the Rift Valley, evidence of animal husbandry comes from Lukenya Hill, where remains of cattle and sheep/goat are dated to between the fifth millennium and early seventh millennium BP. At site GvJm 44, for example, the fauna is dated to between the second half of the second millennium and early fourth millennium BP. At GvJm 48, GvJm 52 and GvJm 184 domestic animals are dated to between the latter half of the second millennium and early third millennium BP. And at GvJm 47 they are dated to ca. 1,340 BP - 900 BP (Nelson and Kimengich 1984). To the south of Lukenya Hill cattle and sheep/goat have been identified at Maua Farm on the slopes of Kilimanjaro in northern Tanzania, where they are associated with Narosura ware in deposits dated to about 4,200 BP - 1,500 BP (Mturi 1986).

On the Mau Escarpment, to the west of the Lakes Naivasha-Nakuru basin, domestic animals have been identified at Enkapune ya Muto and Remnant site. At the former site, only sheep/goat are represented in the older deposits dated to about 4,000 BP; these are associated with Nderit ware (Marean 1992). Both sheep/goat and cattle have been identified at Remnant site, where they are associated with Elmenteitan ware in a horizon dated to about 2,300 BP - 1,300 BP.

It is worth noting here that the evidence outlined above does not lend strong support to the hypothesis that domestic animals were introduced into East Africa from the north. Indeed, as the available chronometric evidence suggests, there is no definite north-south trend in the appearance of domestic animals in the region. On the contrary, the older dates appear in the central Rift Valley and not in northern Kenya, as one would expect (Table 2.1).

To the west of the Mau, domestic fauna have been found at Narosura, Sambo Ngige, Ngamuriak and Lemek Northeast in southwestern Kenya and at Rangong’ and Gogo Falls in the Lake Victoria basin. At Narosura the fauna are associated with Narosura ware in deposits dating to the third millennium BP (Bower et al. 1977; Odner 1972). Radiocarbon dates from Ngamuriak place the appearance of fauna in the first half of the third millennium BP. At Sambo Ngige the fauna have been dated by the obsidian hydration method to the middle of the second millennium BP. The appearance of domestic stock at Lemek Northeast has been radiocarbon dated to early third millennium BP. The animals are associated with Narosura ware, whilst those from Sambo Ngige and Ngamuriak are associated with Elmenteitan ware (Marshall 1986, 1990). At Rangong’ domestic animals are associated with Kansyore pottery in deposits dated to about 2,300 BP (Gabel 1969). And at Gogo Falls remains of cattle and sheep/goat have hitherto been associated with Elmenteitan ware

Scepticism has been expressed with regard to the older central Rift dates because (a) “they contradict conventional wisdom about the spread of pastoralism through Africa” which suggests a north-south spread, and (b) they are on bone apatite which is susceptible to geochemical contamination (Collett and Robertshaw 1983b: 58). In contrast, the dates from northern Kenya are considered to be acceptable because they are on charcoal. While accepting this scepticism, it must be stated here that the claim that domestic animals were introduced into East Africa from the north has neither been archaeologically proven nor disproven. In fact, this claim is largely based on negative evidence. As such, it is best treated as an hypothesis rather than as an established fact. Attempts to segregate all the available chronometric dates according to the nature of the dated samples, that is, charcoal, apatite and collagen/gelatin also fail to reveal any useful

31

Karega-Mũnene

pattern in the dates. That is because the segregation leaves us with few random dates in each of the three sets, which are hardly adequate for the purpose, given the size of the area we are dealing with.

making a proposition for an autochthonous origin of domestication in East Africa before data to support such a proposition are available. On the contrary, what we are putting across is the fact that we should avoid stating hypotheses in such a manner that they appear to be proven facts. This is important because, as the above discussion demonstrates, knowledge about food production in East Africa is based on insufficient data. Furthermore, the representativeness of some of the faunal samples is open to question because of the manner in which they have been obtained. Indeed, except for a few cases, excavated earth has not been sieved to recover small faunal remains. In addition, little effort has been made to recover plant remains, which explains why direct evidence of agriculture is still lacking in the region.

This notwithstanding, the north-south trend in the ‘spread’ of domestic animals remains a persuasive hypothesis, largely because it conforms to the conventional wisdom about the ‘spread’ of animal husbandry in the Old World. While accepting the hypothesis as persuasive we must not overlook the fact that the archaeological data to support or to refute it are lacking. As such, the hypothesis needs rigorous testing through intensive research on pre-Neolithic and more Neolithic sites. This should involve obtaining more and reliable chronometric dates and a concerted search for and alayses of the faunal and floral remains as well as artifacts recovered during the excavation of those sites. Better methods of recovery and analysis as well as the need to obtain large samples that can be considered to be representative should also be emphasised.

Secondly, except for the analyses of the samples from Ngamuriak, Sambo Ngige, Lemek Northeast, Prolonged Drift, Dongodien and Gogo Falls, faunal data have generally been obtained from very small samples. These latter have largely been composed of specimens (usually teeth) selected during excavation. In some cases, animal husbandry has been inferred from taxonomic identifications of less than five specimens as at Ele Bor, Akira and Marula rockshelter.

In addition, there should be less reliance on ‘other lines of evidence’ to ‘explain’ archaeological phenomena for which we apparently lack archaeological explanation. Hitherto, when confronted with a situation similar to the one outlined above, the conventional approach in East African archaeology has been to resort to historical linguistics. In so doing, researchers have generally failed to recognise that the proposition by historical linguistics that domestic animals originated from the north of East Africa, is itself an hypothesis, not an explanation. This practice has resulted in a ‘vicious circle of hypotheses’, whereby archaeologists borrow major presumptions from historical linguistics and vice versa to support and to ‘explain’ their presumptions. Unless efforts are made to move away from this circularity, we are likely to remain in the situation we are in today, where shere repetition of hypotheses has, unfortunately, tended to elevate presumptions to facts.

Thirdly, very few pre-Neolithic and Neolithic sites have been excavated with the aim of investigating subsistence patterns. As a result, some of the data available about the economy for those periods are in the form of appendices to research reports on pottery and lithics. Lastly, the geographical distribution of the sites that have been investigated hitherto is, as we have noted above, skewed. It is, therefore, likely that the subsistence activities that have been reconstructed for those sites may not apply elsewhere. Moreover, sites where data on subsistence activities are available − their representativeness notwithstanding − are located far apart. The upshot of all this is that any conclusions about the origin and development of food production that may be drawn from the data can only be highly tentative.

It must be emphasised here that we are not

32

CHAPTER THREE

HOLOCENE ENVIRONMENTAL AND CLIMATIC SETTING

This chapter presents information about climatic and environmental conditions in East Africa during the Holocene, thus providing a general ecological framework within which the socioeconomic behaviour of the Holocene populations under study will be examined. The ecological framework is complementary to the evolutionary ecological model discussed in the introductory chapter. It is anticipated that the employment of both the ecological framework and the interpretive model will enable us to better understand the relationship between subsistence and the physical and biological environments of the human populations concerned through time.

by agricultural communities (Coetzee 1967; Griffiths 1962; Hamilton 1972, 1976; Lind and Morrison 1974; Morgan 1973). The amount of rainfall experienced in the region varies from one environmental/climatic zone to another. The rainfall cycle itself is directly related to the behaviour of the Inter-tropical Convergence Zone (ITCZ), the band or zone of low pressure that forms over the warmest land masses and waters in the tropics. This lowpressure zone is characterised by increased convection, cloudiness and precipitation and migrates to the warmest surface areas to the north and south of the equator seasonally. Its migration over the (African) continent has a fairly regular pattern, which corresponds with the north-south migration of the sun (Briggs and Smithson 1985; Hamilton 1982).

Modern Environmental and Climatic Conditions Present-day East Africa is characterised by highly varied environmental and climatic conditions although it lies astride the equator. These range from the hot arid and semi-arid conditions of northern Kenya to the moderatecold conditions of the highlands. Altogether several environmental-cum-climatic zones can be delineated in the region on the basis of the dominant type of vegetation in each of them (Figure 3.1). The vegetation has been used for this purpose because it is considered to be a reliable indicator of general environment and climate in the region (Hamilton 1982; Lind and Morrison 1974).

During March - May the ITCZ is on or close to the equator, between latitudes 3° S and 3° N, hence the long rains. From June onwards, it migrates to the north of the equator, where migration beyond the 9° N latitude is checked by the dry climatic conditions of the Sahara Desert. This results in a remarkable decline in rainfall between June and September. In October and November the ITCZ returns to the equator, hence the short light rains. These latter are succeeded by a dry spell in December that extends to the following January and February, when the ITCZ is to the south of the equator (Briggs and Smithson 1985; Hamilton 1982).

As we shall learn presently, the amount of rainfall as well as the temperatures experienced in the region are directly related to latitude, topography, distance from the Indian Ocean and exposure to rain-bringing winds and sunshine. In general the highlands receive more rainfall and have lower temperatures than the lower ground. The sides of the highlands and parts of the lowlying ground which are exposed to direct sunshine for a longer time have higher temperatures than those that are poorly exposed. These conditions explain why the highlands have been the most attractive to settlement, especially

The coastal region is characterised by coral vegetation with mangrove and remnants of lowland forests in some areas. This vegetation is fringed by wooded grassland in the immediate hinterlands of both Kenya and Tanzania. Further into the Kenya hinterland and in northern Tanzania the vegetation changes to grassland with bushes, thickets and occasional trees. The latter vegetation also covers most of eastern and northeastern Kenya as well as a substantial part of northeastern Tanzania. In southern Tanzania

33

Karega-Mũnene

Figure 3.1 East Africa: natural vegetation (after Hamilton 1982) 34

Holocene Foragers, Fishers and Herders of Western Kenya

the wooded grassland is fringed by woodland, which is locally known as miombo, after the deciduous trees which dominate the landscape. The miombo woodland covers most of western Tanzania as well, extending to the southern basin of Lake Victoria. The woodland is interrupted by moderately small patches of drier grassland vegetation with bushes and thickets and wooded grassland in a few places. In northern Tanzania the dominant vegetation is wooded grassland, that is, grassland with patches of scattered trees and thicket. This vegetation also extends to the southern basin of Lake Victoria and continues eastwards into western Kenya and westwards into Uganda (Lind and Morrison 1974; Morgan 1973).

highlands. The upper reaches of the mountains are characterised by afro-alpine and ericaceous vegetation, where the former consists of scrub, grassland and groundsel communities and the latter small-leaved trees and shrubs, especially of the Ericaceae family (Figure 3.2). The middle zone of the mountains is characterised by moist and dry montane forests and bamboo forest and the lower slopes by savanna vegetation and lowland forest (Hamilton 1982; Lind and Morrison 1974; Morgan 1973). The precise location of these vegetation zones on the mountains is determined by altitude and exposure to rain-bringing winds. The southern and eastern sides of Mount Kenya and Kilimanjaro, for instance, receive higher rainfall than the northern and western sides, which are in the rain shadow. Rainfall on Mount Kenya averages 2,000 mm per annum at altitudes of about 2,400 m, only to decrease quite rapidly beyond that point. The wetter sides of Kilimanjaro receive a mean annual rainfall of about 2,000 mm - 2,500 mm between the 1,500 m and 1,800 m altitudes. From just above the latter altitude, the amount of rainfall begins to decrease, falling to below 250 mm per annum above the 4,300 m altitude (Morgan 1973).

The vegetation and climate of the highlands − in southern and northeastern Tanzania, central and western Kenya as well as western and southwestern Uganda − are radically different from those of the plains. The highlands are approximately 2,000 m - 2,500 m above sea level, but the higher summits may be double that height or more. The most prevalent natural vegetation in these areas is montane forest; but bamboo occurs on many of the highlands at altitudes of about 2,200 m, while wetter savanna vegetation and lowland forest occur on the lower slopes of the highlands. Most of the forests in those areas have, however, been cleared to create room for settlement and farming. That is mainly because the cool climate and high rainfall (ca. 1,000 mm - 2,000 mm per annum), both of which are characteristic of those areas, have attracted fairly large human populations (Lind and Morrison 1974; Morgan 1973).

Unlike Mount Kenya and Kilimanjaro, the wettest sides of Mount Elgon are the west and southwest, receiving about 1,000 mm - 1,500 mm of rainfall per annum. Much of the natural forest on the flanks of the mountain has been cleared for farming to an altitude of about 2,100 m. The remaining forest, which is of the moist montane type with bamboo, is restricted to the areas above the 2,100 m altitude, particularly on the southwestern and western sides of the mountain. The vegetation changes to alpine moorland and heath above the 3,000 m altitude. Unlike the other East African mountains, Mount Elgon supports human settlement at the high altitude of 3,000 m - 3,300 m, where some of the natural vegetation has been replaced by grassland due to burning (Hamilton 1982; Lind and Morrison 1974; Morgan 1973).

Among the higher summits in the highlands are the snow-capped Mount Kenya (5,199 m above sea level), Nyandarwa Range (formerly known as the Aberdares Range) (3,999 m), Mau Range (3,098 m) and Cherangani Hills (3,370 m) and Mount Elgon (4,320 m) on the Kenya/Uganda border. In Tanzania they include the snowcapped Kilimanjaro (5,895 m), Mount Meru (4,566 m), Ngorongoro Hills (3,188 m) and Mount Hanang (3,418 m), and in Uganda the snow-capped Ruwenzori Mountains (5,109 m) in the west and Moroto (3,083 m) and Kadam (3,070 m) in the east.

The vegetation on the Ruwenzori Mountains consists of moist montane forest and alpine moorland, the lowest reach of the former being about 2,300 m, below which forest has been cleared for farming. The montane forest in the mountains is the wettest forest in East Africa, receiving as much as 2,500 mm of rainfall per

The vegetation and climate on these mountains and highlands are quite varied, but the mountains support a wider variety of vegetation than the

35

Karega-Mũnene

Figure 3.2 Vegetation zones on East African mountains (after Hamilton 1982) 36

Holocene Foragers, Fishers and Herders of Western Kenya

year. The amount of rainfall, however, decreases as the altitude increases from about the 3,000 m altitude (Hamilton 1982; Lind and Morrison 1974; Morgan 1973).

the Rift is unknown in most areas of western Kenya and in Uganda. Rainfall in the latter areas is fairly well distributed throughout the year, thus reducing or eliminating the intervening dry season in some areas (Lind and Morrison 1974; Morgan 1973).

The Rift Valley system is also a major geographical feature of East Africa. The system extends from Lake Malawi in the south into southern Tanzania, where it branches into the western and eastern Rifts. The western Rift stretches out through western Tanzania and western Uganda terminating in northern Uganda, whilst the eastern Rift extends through Tanzania and Kenya into Ethiopia and further north into the Red Sea and Israel. In East Africa both the eastern and western Rifts are home to virtually all the lakes in the region, the only exceptions being Lakes Victoria and Kyoga (Figure 3.3).

The climate within the Rift Valley is drier and warmer. The most common vegetation is wooded grassland in which Acacia trees predominate, thus rendering the vegetation more open than on either side of the Rift. In Tanzania the floor of the eastern Rift is covered by a mixture of bushland thicket, grassland and woodland vegetation and in southern and central Kenya Rifts by bushland thicket and wooded grassland. This vegetation is reduced to patches of dry thicket in the semi-arid and arid areas of northern Kenya.

The eastern Rift bisects the Kenya Highlands, separating the country into two distinctive climatic regimes. The eastern escarpment rises to about 500 m above the Rift Valley floor at Naivasha. The western side of the Rift rises smoothly to the western Kenya highlands, which include the Mau Range, Gusii Highlands and Cherangani Hills further west.

The vegetation on either side of the Kenya/Uganda border is savanna grassland with scattered bushes. The arid and semi-arid areas of northern Kenya receive an average annual rainfall of under 250 mm (Figure 3.4). The rainfall is unreliable and comes in one short season only, the rest of the year being dry. The area along the Kenya/Uganda border receives slightly higher rainfall, but there is also a rather long dry season in the region (Lind and Morrison 1974; Morgan 1973).

The climate to the east of the eastern Rift is influenced by the convergence between and within the southeasterly and northeasterly air systems, which are locally known as monsoons. The southeasterly monsoons occur in May to October and the northeasterlies from November to April. Because of the dual nature of the monsoons, rainfall in the region is bimodally distributed. The long and heavy rains are experienced between March and May and the short light rains from October to December. The rainy seasons are separated by two dry seasons of varied length and intensity (Lind and Morrison 1974; Morgan 1973). The June-September spell is generally less severe than the JanuaryFebruary dry spell, because the region is subjected to the relatively moist southeasterlies originating in the Indian Ocean during the former period and to the dry northeasterlies originating in the Arabian peninsula.

The Lake Victoria basin Lake Victoria is located between the eastern and western Rift Valleys, across the equator at an altitude of 1,170 m. Its surface area is estimated to be between 68,000 Km² and 75,000 Km², with a maximum length of about 290 Km and a maximum width of about 240 Km north-south and east-west, respectively. The lake is relatively shallow, with maximum and mean depths of about 80 m and 40 m, respectively. Its drainage basin extends as far as Mount Elgon and the Cherangani Hills to the northeast, where River Nzoia originates (Figure 3.5); Gusii Highlands and the Mau Range to the east, the origins of Rivers Sondu, Kuja/Gucha and Mara; and Rwanda and Kigezi Highlands to the west, the origin of River Kagera (Beadle 1974; Kendall 1969; Morgan 1973; Temple 1967). While these rivers bring significant amounts of moisture into the lake, most of the moisture comes from rainfall, which averages 1,450 mm per annum (Sene and Plinston 1994).

The climate to the west of the Rift is greatly influenced by the intersection of the moist westerlies (originating in the Atlantic Ocean) over Lake Victoria and the highlands of western Kenya. Since this occurs throughout the year, the seasonality experienced in the areas to the east of

37

Karega-Mũnene

Figure 3.3 The Great Rift Valley system (after Hamilton 1982) 38

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 3.4 East Africa: mean annual rainfall (after Hamilton 1982) 39

Karega-Mũnene

The lake basin extends for about 100 Km inland from the lake shores. The basin has fairly diverse vegetation, ranging from moist evergreen forest to swamps. The vegetation to the north and northwest of the lake consists of patches of moist evergreen forest fringed by a continuous semideciduous forest. The latter begins in the east of the source of River Nile and continues westwards along the northern and western shores across the Uganda/Tanzania border, where it fades out. The vegetation on most of the islands in the lake is also moist evergreen forest (Langdale-Brown et al. 1964; Lind and Morrison 1974; Trapnell and Langdale-Brown 1962).

southern basin. The mean annual rainfall in the higher rainfall zone – Kakamega and Gusii Highlands to the east, Tarime hills to the southeast of the lake, and the forest-covered areas to the north and west of the lake – ranges from 1,500 mm to over 2,000 mm per annum and is well distributed through the year. In contrast most of the low-lying lake margin and the entire southern basin receive about 750 mm - 1,000 mm per annum; the rainfall is double-peaked with a significant dry season (Kendall 1969; Langdale-Brown et al. 1964; Lind and Morrison 1974; Morgan 1973).

Palaeoenvironmental and Palaeoclimatic Setting

Part of the northern lake basin is covered by elephant grass (Pennisetum purpureum) which grows as high as 5 m. (The grass also occurs on the eastern escarpment of the western Rift in Uganda as post-cultivation vegetation prior to forest regeneration.) In the lake basin the grass occurs particularly on slopes of hills which rise to about 200 m above the lake. The valleys between the hills are covered in forests or swamps. The latter are widespread on the fringes of the lake and in the flood-plain of the rivers entering or leaving the lake (Langdale-Brown et al. 1964; Lind and Morrison 1974; Morgan 1973; Trapnell and Langdale-Brown 1962).

The reconstruction of the Holocene environmental and climatic conditions that follows has been attempted using data derived from published studies of fossil pollen, lake sediment, mineral content, micro-fossils like diatoms and algae, and former lake-levels. Information about these aspects is not available from all of the sites concerned. That is mainly because the nature of the samples from the lowland lakes, for example, has required use of several approaches, whereas the samples from the highland sites have required fewer approaches. This notwithstanding, it is generally agreed that the environmental and climatic changes which can be inferred from the sites that have been investigated have regional, rather than localised implications. As we shall presently learn, that is because there is an apparent agreement between evidence adduced at the highland sites on the one hand and evidence adduced at the lowland lake sites on the other hand. Additionally, the pattern of lake-level fluctuations observed in the East African lowland lakes is supported by evidence adduced at lakes located in Ethiopia for example. The few differences that exist in the fluctuation patterns are considered to be the effects of differences in altitude and local geology and hydrology (Gasse 1980; Gasse and Street 1978). The differences could also be a reflection of varied climatic and environmental conditions that might have prevailed at the sites concerned. Indeed, there is hardly any reason to assume that the environmental and climatic conditions of the Holocene were uniform throughout East Africa.

The dominant vegetation in the southwestern lake basin is savanna grassland with remnants of forest, whilst the southern basin is covered by a mix of miombo woodland and dry savanna grassland and the southeastern and eastern basins by wooded grassland. The vegetation in all these areas is characterised by broad-leaved trees of the Combretum species and tall grass (about 1.2 m) which is dominated by Hyparrhenia, Cymbopogon, Bothriocloa and Loudetia species. Large-leaved trees of Terminalia species also occur in the wetter areas while smaller-leaved trees of the same species and Acacia occur in the drier areas (Langdale-Brown et al. 1964; Lind and Morrison 1974; Morgan 1973; Trapnell and Langdale-Brown 1962). Rainfall in the low lying areas of the lake basin is bimodally distributed, the heavy rains falling between March and May and the short rains between October and December. The evaporation rate in the basin is fairly high, thus causing most of the rain to leave the lake and the low-lying areas through evaporation. The eastern, northern and western basins as well as the islands receive higher rainfall than the

It must be emphasised here, however, that in attempting a reconstruction of past environmental and climatic conditions, we are 40

Holocene Foragers, Fishers and Herders of Western Kenya

not proposing to present a precise account of the climate and environment, which prevailed during the past. Rather, we are only suggesting the kind of general environment and climate that may have prevailed. That is largely because the data we are dealing with are generally incomplete, a situation that has been brought about by several factors. To begin with, there is the problem of differential pollen production, dispersal, deposition, preservation and recovery. This problem is compounded by variations in the region’s vegetation, topography, humidity and soil conditions, all of which bear directly on the representativeness of the samples that have been studied.

include Lake Turkana which is located at 375 m above sea level in northern Kenya and Lakes Elmenteita (1,986 m), Nakuru (1,962 m) and Naivasha (1,894 m) in the central Rift (Coetzee 1967; Hamilton 1982; Mworia-Maitima 1991, 1999; Richardson 1972; Washbourn-Kamau 1975). In Tanzania there is only one highland site, Crater Lake, which is located at an altitude of 2,650 m in moist montane forest in Kilimanjaro (Coetzee 1964, 1967). Highland sites in Uganda include Lakes Mahoma, Kitandara and Bujuku, which are situated in bamboo forest in the Ruwenzori Mountains at altitudes of 2,960 m, 3,990 m and 3,920 m, respectively. Muchoya swamp is located at an altitude of 2,260 m, Butongo swamp at 2,025 m, Katenga swamp at 1,980 m and Lake Bunyonyi at 1,950 m, all in Rukiga Mountains. The Rukiga sites are located in deforested areas, whose present vegetation consists of grassland with a few remnants of forest. The areas around these sites had been covered by bamboo forest, which has now been destroyed by animals and humans (Hamilton 1972, 1976; Livingstone 1967; Morrison 1968; Morrison and Hamilton 1974). The lowland sites in Uganda are located in the centre of Lake Victoria and at Pilkington Bay near Buvuma Island in the northern portion of the lake (1,133 m) (Beuning 1999; Coetzee 1967; Kendall 1969; Ssemmanda and Vincens 1999).

There is also the problem of the number and location of the sites from which the relevant data have been obtained; these are few and located far apart in the region (Figure 3.6). Calibration of the data from those sites is also incomplete since only a few sites have been adequately dated. For example, the pollen samples from Butongo and Katenga swamps and Lake Bunyonyi in Uganda have no chronometric dates. Rather, the estimated dates for the samples have been extrapolated from other radiometrically dated sites.

Localities of Studies of Holocene Conditions The localities from which the data employed in the reconstruction of past environment and climate of East Africa are located in swamps and lakes in the highlands and in lakes located in the lowlands. In Kenya the highland sites include Sacred Lake1 and Lake Rutundu both in Mount Kenya at altitudes of about 2,400 m and 3,140 m, respectively; and Kaisungor swamp which is located at an altitude of 2,926 m in Cherangani Hills. Sacred Lake is located in moist montane forest, Lake Rutundu in the ericaceous belt and Kaisungor swamp in dry montane forest (Coetzee 1964, 1967). All the lowland sites in Kenya are located in the Rift Valley. They

Terminal Pleistocene Palynological studies suggest that East Africa, as a whole, was much more arid and colder than today during the 15,000 years or so preceding the Holocene, a period that falls within the Last Glacial Maximum. Pollen diagrams of samples from Sacred Lake, for example, indicate there was open vegetation at the altitude of the lake during that period. The vegetation consisted of much grassland mixed with small trees and shrubs. The existence of this type of vegetation has also been attested at Lake Rutundu, Kaisungor swamp, Lake Mahoma and Muchoya swamp. The composition of the vegetation at these sites has been viewed as indicating a dry and cold climate during the period in question (Coetzee 1964, 1967; Hamilton 1972, 1976; Morrison 1968; van Zinderen Bakker 1962, 1964).

1

The water at the centre of the lake was about 3.5 m deep when coring was done in the 1960s (Coetzee 1967: 65). A recent visit to the lake (October 1999) reveals the lake has almost dried up, primariry because of environmental degradation arising from deforestation, cultivation and grazing in its vicinity. Indeed, during the visit the present author walked about 10 m on hard surface towards the centre of the lake.

On the basis of the above palynological evidence, it has been estimated that the annual 41

Karega-Mũnene

Figure 3.5 Lake Victoria drainage basin (after Beadle 1974)

42

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 3.6 Localities of studies of Holocene conditions (after Hamilton 1982)

43

Holocene Foragers, Fishers and Herders of Western Kenya

mean temperature around Sacred Lake, Lake Rutundu and Kaisungor swamp during the terminal Pleistocene was 5.1°C - 8.8°C lower than at present. At Muchoya swamp and Lake Mahoma temperatures are estimated to have been about 5°C - 6°C lower than today during the same period (Coetzee 1964, 1967; Hamilton 1972, 1974, 1976; Livingstone 1967; Morrison 1968; van Zinderen Bakker 1962, 1964). These temperature estimates are in general agreement with each other and with an estimate of 5°C 6°C, arrived at on the basis of aspartic acid racemization on bone samples from Lukenya Hill in Kenya (Schroeder and Bada 1973).

Early Holocene (ca. 10,000 BP - 6,000 BP) The above environmental and climatic changes marked the transition to the Holocene environmental and climatic conditions. The Holocene itself is believed to have started between 12,500 BP and 10,000 BP (Hamilton 1972, 1976; Livingstone 1967; Roberts 1998). For the purposes of this study, however, the Holocene is viewed as having started about 10,000 BP (cf. Catt (1988). That is because by that date there is widespread evidence for the favourable conditions which were characteristic of most of the Holocene. Pollen diagrams from Lake Mahoma show a substantial increase of pollen from a variety of trees characteristic of a closed forest by 9,000 BP; this vegetation persisted for several millennia. Similar environmental conditions have been attested at Lake Kitandara and Muchoya swamp. Pollen diagrams from the former site for the period 7,500 BP - 6,000 BP indicate the existence of a closed forest at the altitude of the lake. And palynological evidence from Muchoya swamp indicates that the moist montane forest that appeared there towards the end of the Pleistocene persisted to about 6,000 BP (Hamilton 1972, 1976; Livingstone 1967; Morrison 1968).

The cold dry climate inferred from pollen diagrams of the highland sites has been corroborated by evidence adduced from lowland lakes. According to evidence from Lake Victoria, for example, the lake was shallow and closed; the level of the water being much lower than the present-day Nile outlet. Pollen diagrams from the lake indicate that the vegetation around the lake during that period was predominantly of an open type (Beuning 1999; Kendall 1969). Except for the period 17,000 BP - 15,000 BP when the central Rift Valley experienced a moderately wet climate, the region was generally dry as evidenced by pollen diagrams and shallow and closed lakes. The dry conditions prevailed until about 12,500 BP - 12,200 BP, when the climate became increasingly wet (Butzer et al. 1972; Mworia-Maitima 1991; Richardson 1966, 1972; Richardson and Richardson 1972; Perrot and Street-Perrot 1982).

Palynological evidence from Sacred Lake indicates there was a rapid expansion of the forest towards the lake between about 10,500 BP and 6,400 BP. Lake Rutundu pollen diagrams also indicate a marked expansion of the forest towards the lake from about 10,600 BP, reaching a maximum between 7,300 BP and 6,100 BP. These environmental changes are corroborated by evidence from Kaisungor swamp, which indicates there was a fairly rapid forest expansion during the first two millennia of the Holocene. This was, in turn, followed by a brief forest retreat and by further forest expansion at the swamp. These environmental changes have been interpreted to mean the climate became warmer and wetter during the first two millennia of the Holocene period and that both the humidity and temperature reached maxima during that period. These conditions were succeeded by a brief spell of a cold and dry climate (Coetzee 1964, 1967; van Zinderen Bakker 1962, 1964).

This transition is attested at several sites in the region. Pollen curves from Lake Mahoma, for instance, suggest that both the environment and climate were very similar to the present conditions around the lake (Hamilton 1972, 1976; Livingstone 1967). Pollen diagrams from Muchoya swamp indicate that trees characteristic of moist montane forest began to appear at the altitude of the swamp, replacing the vegetation of the arid phase (Morrison 1968). Evidence from Lake Victoria indicates there was a significant rise in the level of the lake and that a forest, which changed from evergreen to semideciduous through time, developed and expanded considerably in the lake basin between about 12,500 BP and 11,500 BP (Beuning 1999; Kendall 1969).

Further corroboration of these inferences comes from the Rift Valley lakes and Lake Victoria. The evidence indicates surface water became 44

Karega-Mũnene

abundant, resulting in the attainment of maximum levels in the lakes and the appearance of marshes around the lakes between about 10,000 BP and 8,000 BP. For example, sedimentological work conducted at Lake Turkana indicates that between about 10,000 BP and 9,500 BP the level of the lake rose to 80 m above its present level. By 8,900 BP the lake had fallen by 21 m, but it rose again by about the same margin 1,000 years later. At about 7,500 BP the lake-level fell to about its modern level; this was followed by further falls lowering the lake-level by 60 m about 6,600 years ago and a rise of about 70 m four hundred years later (Butzer et al. 1972; Street and Grove 1976).

Lake Nakuru, the drying up of Lake Elmenteita and expansion of dry montane forests in the region (Mworia-Maitima 1999). These changes have been attributed to high temperatures in the lakes, which are estimated to have been about 2°C - 3°C higher than at present (Butzer et al. 1972; Richardson 1966, 1972; Richardson and Richardson 1972).

Middle Holocene (ca. 6,000 BP - 3,000 BP) The favourable environmental and climatic conditions of the early Holocene persisted for some time during the middle Holocene before they started to deteriorate. Palynological evidence from Lake Mahoma, for example, indicates forest expansion around the lake reached a maximum before 4,600 BP. This suggests that the climate around the lake also reached a maximum during the same period. However, from about 4,600 BP forest reduction started to occur and the climate started to deteriorate (Hamilton 1972, 1976; Livingstone 1967).

Evidence from Lake Victoria indicates that from about 10,400 BP there was decreased humidity and a lowering of temperature in the lake basin for about a millennium. This caused a fall in the lake level by about half the previous rise, resulting in loss of the Nile outlet and retraction of the forest that had previously developed around the lake. It has been postulated that these changes were caused by decreased rainfall, but the rainfall shortage seems to have been less severe than that of the terminal Pleistocene period because the forest that had developed to the north of the lake did not disappear. At about 9,300 BP the lake rose rapidly to a level where it regained the Nile outlet. The evergreen forest also came back and developed more abundantly than previously; it also changed from evergreen to semi-deciduous. The lake remained open throughout the early Holocene and well into the mid-Holocene period; the down-cutting of the Nile outlet which had started during the early Holocene also persisted well into the midHolocene. The region witnessed the highest precipitation and expansion of the moist forest from about 7,700 BP - 6700 BP (Beuning 1999; Kendall 1969).

Pollen diagrams from Lakes Kitandara and Mahoma indicate that the forests in the vicinity of the lakes expanded to above the altitude of the lakes between 6,000 BP and 4,600 BP. Pollen curves from Kaisungor swamp suggest the favourable early Holocene conditions around the swamp improved further during the midHolocene. Between about 5,700 BP and 3,000 BP, for example, there was a steady forest expansion to the area around the swamp. During the same period, the forest became more and more closed and both temperature and humidity reached maxima (Coetzee 1967; Hamilton 1972, 1976; Livingstone 1967; van Zinderen Bakker 1962, 1964). The existence of a closed forest during the midHolocene has also been attested at Crater Lake, where palynological evidence indicates the lake was surrounded by a dense forest by about 4,600 BP. The evidence also indicates that both the forest and climate reached maxima about 3,200 BP. These environmental changes suggest that the climate was warmer and wetter at the sites in question during mid-Holocene than previously (Coetzee 1967; Hamilton 1972, 1976; Livingstone 1967; van Zinderen Bakker 1962, 1964).

Palaeolimnological evidence from Lakes Elmenteita, Naivasha and Nakuru indicates that the hitherto alkaline lakes became increasingly fresh and their levels rose considerably between about 10,000 BP and 8,500 BP (Butzer et al. 1972; Mworia-Maitima 1991, 1999; Richardson 1966, 1972; Richardson and Richardson 1972). It has been suggested that increased rainfall and relatively low rates of evaporation caused these changes. However, between about 8,500 BP and 6,400 BP the wetter climate changed to drier conditions, resulting in the drastic shrinking of

The improved conditions of the early Holocene period around Sacred Lake did not last for very long because the forest retreated back to just 45

Holocene Foragers, Fishers and Herders of Western Kenya

below the altitude of the lake after 6,400 BP. That means the climate around the lake became colder and drier than previously after the latter date. These conditions lasted for approximately one millennium after which there was a steady forest expansion, again to the altitude of the lake. Both the forest and climate reached maxima between about 5,000 BP and 2,800 BP (Coetzee 1964, 1967; Morrison 1968; Morrison and Hamilton 1974).

around the lake expanded quite significantly during that period. It also became semideciduous, probably in response to the onset of seasonal rainfall (Kendall 1969), which is characteristic of the area today. However, from about 3,000 BP the forest declined progressively, probably due to declining rainfall and/or human activity, the latter becoming more pronounced from ca. 2,000 BP (Mworia-Maitima 1997; Ssemmanda and Vincens 1999).

Pollen diagrams from Lake Rutundu also indicate that forest expansion and climatic conditions at the altitude of the lake reached maxima between 5,000 BP and 3,000 BP. Similar climatic and environmental changes have also been observed in the pollen diagrams from Butongo and Katenga swamps and Lake Bunyonyi. The only exception has been Muchoya swamp where the forest failed to recover from the previous period’s decline. Instead, the decline continued throughout the mid-Holocene period (Coetzee 1964, 1967; Morrison 1968; Morrison and Hamilton 1974).

Later Holocene (ca. 3,000 BP - present) Evidence from the majority of the sites under review indicates that the last 3,000 years have been characterised by a transition to the region’s present-day environmental and climatic conditions. Evidence from Lake Mahoma, for instance, suggests that the forest reduction, which had started during the preceding period, became more widespread from about 2,000 BP, only to be replaced by plants characteristic of disturbed areas. Somewhat similar changes in forest cover have been recorded at Muchoya, Kitandara, Katenga and Butongo swamps and at Lakes Bunyonyi, Kitandara and Bujuku. Pollen diagrams from Butongo swamp and Lake Bunyonyi, for example, show marked decline in forest tree pollen in the region from about 1,200 BP. This corresponds with the appearance of plants which are characteristic of disturbed areas in both places (Hamilton 1972, 1974, 1976; Livingstone 1967; Morrison 1968; Morrison and Hamilton 1974).

These inferences are supported by evidence adduced from the Rift lakes, which indicates the warmer and wetter climate of the early Holocene persisted for a few centuries only after 6,000 BP before becoming drier and colder than at present. Lakes Nakuru and Elmenteita, for example, experienced ecological changes involving desiccation and alkalinity from about 6,000 BP 5,000 BP onwards; but rapid and widespread contraction at the lakes did not occur until about 4,500 BP - 3,500 BP. Lake Naivasha remained large until about 5,600 BP when it shrank considerably in size. Between 3,400 BP and 3,000 BP the lake almost dried up before rejuvenating to its present level at about 1,000 BP (Butzer et al. 1972; Gasse 1980; MworiaMaitima 1991; Richardson and Dussinger 1986; Richardson and Richardson 1972). In contrast, evidence from Lake Turkana indicates the lake was unstable during the middle Holocene period, fluctuating between 60 m and 70 m above its present level before settling at the latter level about 3,250 BP (Butzer et al. 1972; Gasse 1980; Richardson 1966, 1972; Richardson and Richardson 1972).

Unlike at the above sites, palynological evidence from Kaisungor swamp indicates that the closed forest around the swamp expanded to above the swamp after 3,000 BP. The forest persisted to about 700 BP, when it started to decline as evidenced by remarkable increases in trees and shrubs characteristic of open forest in the site’s pollen diagrams. As a result of these changes, the climate became colder, mistier and drier (Coetzee 1967; van Zinderen Bakker 1962, 1964). The closed forest that had developed around Sacred Lake during the mid-Holocene persisted to about 2,400 BP when an open forest replaced it, thus marking the beginning of a colder and drier phase at the lake. The wetter and warmer climate of the mid-Holocene around Lake Rutundu prevailed to about 2,000 BP. An open forest appeared around Crater Lake, replacing the dense forest of the mid-Holocene. These

Evidence from Lake Victoria indicates the lake maintained a fairly high level throughout the period under review, its level being approximately 3 m higher than at present at about 3,700 BP. The forest that had developed 46

Karega-Mũnene

changes suggest the climate became cooler and drier than previously at these sites after 2,500 BP - 2,000 BP (Coetzee 1964, 1967).

forest cover in some areas. That is because, as Hamilton (1972: 64) observes, the ...distribution of [modern tree] species in highland Uganda is... mainly determined by the conditions of temperature, moisture availability and disturbance [by humans through clearance or grazing and by wild animals].

Evidence from the lowland lakes indicates that the level of the lakes had fallen to about their present levels by 3,000 BP and that the majority of the lakes had become closed and saline by that date. For the last 3,000 years the levels of Lakes Turkana, Nakuru, Naivasha and Elmenteita have been relatively low. In particular, the level of Lake Naivasha has remained low, fluctuating to near and below its present level. During that period there has been at least one wetter spell during which the lake has risen to a higher level than at present (Butzer et al. 1972; Gasse 1980; Richardson 1966; Richardson and Richardson 1972; Street and Grove 1976). The level of Lake Victoria has remained relatively stable for the last 3,000 years, but no significant inferences about the climatic conditions around the lake during this period can be made from the pollen diagrams or any other sources of evidence (Kendall 1969).

Faunal Distribution and Human Settlement The palaeoenvironmental and palaeoclimatic conditions discussed above had significant implications on the distribution and extent of prehistoric fauna and human settlement. Indeed, as Terasmae (1974: 16) observes, ...even smaller changes in temperature... and precipitation can have very substantial effects on the environmental conditions that govern our food and water resources and many other activities.

However, knowledge about the Terminal Pleistocene and early to mid-Holocene faunal communities in East Africa is quite scanty. That is mainly because researches that have been undertaken on sites dating to that period have been biased in favour of artefacts. The presence of Grevy’s zebra (Equus grevyi) and oryx (Oryx gazella) in faunal collections from sites located in central Kenya and northern Tanzania dating to that period supports the observation that the period was arid. Moreover, the faunal collections in question were not dominated by wildebeest, which dominates the fauna in the East African plains today. This condition has been interpreted to suggest that the moist grasslands which is the habitat preferred by wildebeest were absent in the region during that period. Thus, the presentday numerical superiority of wildebeest in the region’s fauna appears to have started with the onset of the Holocene conditions (Marean and Gifford-Gonzalez 1991).

It has been postulated that the fall in the Rift lake-levels was caused by the onset of relatively colder and drier conditions and that the small rises in the lake levels during the last 3,000 years are a reflection of sporadic increases in rainfall or of bimodally distributed rainfall. This, in turn, suggests the climate has been quite similar to the present during the last three millennia in the areas around the sites (Butzer et al. 1972; Gasse 1980; Kendall 1969; Richardson 1966; Richardson and Richardson 1972; Street and Grove 1976). However, it must be added here that caution is necessary when interpreting lake level changes because they are not necessarily caused by climatic change alone. They could also be caused by reduced drainage inflow or varied seasonal down-cutting of an outlet(s) (Temple 1967) or by differences in the geology and hydrology of the areas concerned (Gasse 1980; Gasse and Street 1978).

Evidence of this kind is lacking from Gogo Falls. We do not know, for instance, what animals existed in the area around the site during the preHolocene period. Similarly, we do not know whether or not the area was occupied during that period. In fact, the little evidence currently available suggests that the area has been occupied only for the past 7,000 years or so (Robertshaw et al. 1983; Robertshaw 1991). It is the present author’s contention that the area was occupied for a longer period than the 7,000 years; the evidence for this is unavailable

Unlike the lake level changes that have been attributed to climatic changes, the forest decline of the past 3,000 years has been attributed to human influence (e.g., Hamilton 1972; MworiaMaitima 1997). The attribution of the deforestation to humans is supported by the appearance of plants that grow in disturbed areas about the same time as forest decline occurred. Temperatures and moisture availability may also have been significant factors in the reduction of 47

Holocene Foragers, Fishers and Herders of Western Kenya

because, as we have noted above, very little research has been conducted there.

It is worth noting here that human groups who occupied East Africa are not considered to have had a major impact on the region’s environment until fairly recently. Indeed, as the above discussions have demonstrated, it is thought that humans affected the environment only after the advent of food production. The subsistence activities of the pre-Neolithic groups, namely, hunting, gathering and/or fishing, had minimum impact on the environment.

In contrast, longer periods of human occupation have been documented in other parts of East Africa. For example, archaeological evidence from Lukenya Hill in central Kenya indicates that the area has been inhabited for the last 40,000 years or so (Gramly 1975, 1976; Nelson and Kimengich 1984). In the Rift Valley region, human presence has been fairly well documented for the past 12,000 years (Ambrose 1984).

48

CHAPTER FOUR

GEOGRAPHICAL SETTING AND HUMAN SETTLEMENT

The purpose of this chapter is threefold. First, to describe the geographical setting of the Gogo Falls site, thus placing the site within the wider regional environment and climatic contexts discussed in the previous chapter. Second, to discuss, albeit briefly, the history of human settlement in the Kenya Lake Victoria basin within which the Gogo Falls site is located. Third, to discuss modern land-use patterns in the lake region. This information will be employed in our discussion of the archaeological data that will be explicated in the succeeding chapters.

represents 73.5% of the District, being land surface. The latter can be divided into three distinct physiographic regions: the eastern, central and lake shore regions (Republic of Kenya n.d.; Survey of Kenya 1970). The eastern region consists of the upper savanna belts and plateau that have an altitude of over 1,300 m (Republic of Kenya n.d.; Survey of Kenya 1970). This region extends into the Gusii Highlands in the neighbouring Kisii District, which was recently divided into three districts − Nyamira, Gucha and Kisii. This region lies within the western Kenya high rainfall zone, which was described at length in the preceding chapter. River Kuja and its major tributaries originate in the highlands located in that zone.

Geographical Setting of the Site The Gogo Falls archaeological site is located in the Kenya Lake Victoria basin in Nyatike (formerly Macalder) Division of Migori District. The district was created recently when South Nyanza District was divided into five districts, namely, Kuria, Suba, Migori, Homa Bay and Rachuonyo. However, since a map showing the new district boundaries is not available at the time of writing, the site’s geograhical setting will be described using the boundaries of the old South Nyanza District.

The major feature of the central region is the Olambwe Valley, a relatively narrow but longer depression bordered by Kanyamwa escarpment to the south and Gwassi hills to the north; the valley is home to Ruma National Park. This region has an altitude of about 1,220 m - 1,300 m. But there are a few areas, notably remnants of volcanoes, which are over 1,500 m high. The lakeshore region has an altitude of about 1,220 m or less (Republic of Kenya n.d.; Survey of Kenya 1970).

The site itself is situated on a gently sloping hillside on the west bank of River Kuja/Gucha (Figure 4.1) at an altitude of about 1,220 m and about 20 Km east of Lake Victoria. River Kuja originates from the higher rainfall region further east in the Gusii Highlands and flows into Lake Victoria. Across the river, close to where the site is located, is a power station which generates electricity using water from the dam next to the Gogo Falls, after which the site is named. The power station was originally built to generate power for the Macalder gold mines (Morgan 1973). Since the closure of the mines the station has been providing power to the urban areas and industries located in the region (Survey of Kenya 1970).

Rainfall and climate in the District fall within the general pattern for the Lake Victoria basin discussed in the preceding chapter. Thus, the District experiences two rainfall seasons: the long rains season which falls between March and May and the short rains season between October and November. The eastern part of the District receives higher rainfall, averaging 1,300 mm 1,600 mm per annum, and the rest of the District 760 mm - 1,270 mm per annum. The rainfall seasons are separated by two dry seasons from June to September and December to February. The effects of the dry weather during the latter seasons are more pronounced in the lakeshore region than in the central and eastern regions (Morgan 1973; Republic of Kenya n.d.; Survey of Kenya 1970).

South Nyanza District as a whole covers an area of 7,778 Km². A considerable portion of this area, that is, 2,064 Km² (26.5%) is covered by water, the remaining 5,714 Km², which

49

Figure 4.1 Location of Gogo Falls (after Robertshaw 1991).

Holocene Foragers, Fishers and Herders of Western Kenya

50

Karega-Mũnene

Climate in the District is greatly influenced by air currents from the lake and by altitude. The breeze from the lake, for instance, provides a cooling influence that considerably lowers the temperatures. Mean minimum and mean maximum annual temperatures in the lakeshore region and in parts of the central region average 17°C and 30°C, respectively. The eastern region has lower temperatures because of its higher altitude. These average 14°C and 25°C, for the mean minimum and mean maximum per annum, respectively (Lind and Morrison 1974; Republic of Kenya n.d.; Survey of Kenya 1970).

have settled en route. Thus, it was only the “more adventurous of them [who] continued with the trek” into the Kenya and Tanzania Lake Victoria basins (Ogot 1967: 82-83, emphasis added). The migrants reached modern Uganda during the first half of the 16th century A.D.; some of them remained in the country while the rest moved further south and into Kenya and Tanzania Lake Victoria basins. Those who remained in Uganda founded the Padhola, Acholi and Lang’o communities, whilst those who migrated to the Lake Victoria basin founded the present-day Luo communities of Kenya and Tanzania (Ogot 1967).

Human Settlement The present inhabitants of the Kenya Lake Victoria basin are the Luo who are classified by linguists as Western Nilotic-speakers. Their neighbours to the north are the Abaluyia, to the east the Kalenjin and the Abagusii and to the southeast the Abakuria (Figure 4.2). Linguistically, the Abaluyia, Abagusii and Abakuria belong to the Bantu-speaking cluster and the Kalenjin to the Southern Nilotic group (Greenberg 1966; Nurse 1982).

The first Luo migrants to enter western Kenya arrived there during the 16th century A.D. Their first settlement is believed to have been located in an ‘unoccupied’ area on the northern shores of Lake Victoria. Within less than a century of their settling in that area, their population as well as that of their livestock expanded rapidly. This, in turn, necessitated territorial expansion eastwards and southwards. Unlike in the previous case, the latter expansion occurred through military conquest of the Bantu-speaking groups who occupied the area (Ogot 1967).

According to evidence adduced from oral traditions, the Luo have lived in the Lake Victoria basin for a few centuries only. They are presumed to have originated from the Bahr-elGhazal region in present-day Sudan. Their migration from there to the Lake Victoria basin is believed to have been caused by over-stocking, over-population, external pressure, internal feuds, or a combination of these. Consequently, it has been argued that their migration was driven by a search for “empty lands” (Ogot 1967: 42).

The conquest itself was reportedly not undertaken by the entire Luo community, but by small Luo groups and ‘Luoised’ groups, the latter consisting of people of non-Luo origins who had been successfully assimilated into the Luo community. The non-Luo groups who could not be assimilated were compelled to seek refuge in the highlands farther north and east, away from the lake basin (Ogot 1967).

From their cradle-land in southeastern Sudan, oral traditions suggest that the Luo (or the socalled ‘pre-Luo migrants’) split into two groups. One of the groups moved northwards to central Sudan where they founded the Shilluk community and the other southwards into East Africa (Ogot 1967). Of the two groups, however, we are concerned with the latter because they are the ones who were responsible for the foundation of the region’s modern Luo community.

Subsequent expansions, again through military conquest, took the Luo to the present-day South Nyanza District and to northeastern Tanzania. The conquest of South Nyanza itself started during the first half of the 18th century A.D. and continued for the better part of the 19th century A.D. Further expansion beyond the present-day boundaries of the District was checked by the Maasai. The constant Maasai raids on Luo settlements forced the Luo to erect stone enclosures, which are known as ohingni (singular ohinga) in Dholuo, the local language. More effectively, Luo expansion was checked by the advent of British colonialism towards the end of the 19th century (Ogot 1967).

The migration of the ancestors of the Luo is thought to have taken “a complicated and confused pattern”, as the migrants followed neither one direction nor the same route (Ogot 1967: 57). Some of the migrants are believed to have moved backwards and forth and others to

51

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 4.2 Kenya: ethnic groups (after Barbour and Wandibba 1989)

52

Karega-Mũnene

Thus, in their expansion and settlement in western Kenya, the Luo seem to have preferred the lake basin to the wetter highlands bordering the basin. This preference has been attributed to two factors, namely, the basin’s close resemblance to the original home of the Luo in the Sudan and its suitability to the Luo nomadic pastoralism which was supplemented by fishing and hunting. These economic activities are believed to have remained unchanged until fairly recently (Butterman n.d.; Ogot 1967).

migration hypothesis to the realm of biology by suggesting that migration was not only a tradition of the Luo, but was also in their blood: But the Luo had not lost their migratory trait [due to the advent of British colonialism]. In dribs and drabs, and almost imperceptibly, they continued to move to South Nyanza and to Tanzania. Some traces of this ‘migratory vein’ may perhaps nowadays be seen in their movements to towns and outside employment which, apart from a primary economic drive, may also have this traditional basis. (Ogot 1967: 239, emphasis added)

Prior to their migration to the Kenyan side of the Lake Victoria basin the Luo are thought to have lacked knowledge of crop cultivation. They learned and adopted cultivation from their neighbouring Bantu-speaking communities whom, as we have noted above, are believed to have been predominantly cultivators with a few domestic animals. This knowledge eventually enabled the Luo to change their lifestyle from nomadic pastoralism to sedentary mixed farming; but fishing continued to be, and still is, an important economic activity (Butterman n.d.; Ogot 1967).

Soils and Land-Use Patterns Modern land-use patterns in South Nyanza District are influenced by a combination of factors. These include the nature and type of the soils, the amount of annual rainfall and the availability of natural resources like Lake Victoria and the rivers draining the District. These resources permit the practising of three major economic activities: cultivation, animal husbandry and fishing. On the basis of soil types, the District can be divided into two broad agro-ecological zones: the high and medium potential zones (Figure 4.3). The geographic distribution of these zones corresponds remarkably well with the physiographic regions described above. The high potential zone, for instance, corresponds with the whole of the eastern region and parts of the central region. These areas do not only receive higher rainfall, but also have rich loam and clay soils and are well-drained by the rivers originating in the Gusii Highlands. These conditions are conducive to the growth of a fairly wide variety of both cash and subsistence crops, namely, sugar cane, sunflower, coffee, tea, tobacco, maize, potatoes, sorghum, cassava, beans and bananas. The first five are the main cash crops in the region, while the rest are mainly for subsistence (Republic of Kenya n.d.; Personal observation).

The predecessors of the Luo in the lake basin are presumed, on the basis of oral traditions, to have been the ancestors of the Abaluyia, Abagusii and Abakuria (Ogot 1967; Were 1967). However, the oral traditions concerned are silent about who the predecessors of these Bantu-speaking communities in the region were. That is in spite of the fact that the archaeological evidence discussed in chapter two clearly indicates that human settlement in the region goes back to several millennia BP. Thus, if the correlation between the Bantu and the Iron Age Urewe ware that has been discussed in the preceding chapters is correct, it would appear that Bantu presence in the region dates as far back as ca. 2,500 BP. But this proposition is at variance with oral traditions, which, as the foregoing discussion indicates, suggest recent Bantu settlement in the region. This situation is hardly surprising since historians who use oral traditions (e.g., Muriuki 1974; Mwaniki 1973; Ogot 1967; Were 1967; Vansina 1985) are agreed that oral traditions have an extremely short time depth, usually a few centuries. In light of the currently available archaeological data, the supposition that the Luo and the Bantu are recent migrants in the region seems to be deficient. Equally deficient is the tendency by some historians like Ogot (1967) to extend the

The medium potential zone coincides with the low-lying parts of the central region and the entire lakeshore region. Sandy and clay soils are the most predominant soils in this zone, but rich alluvial soils do occur in highly localised areas. The main cash crops grown in the zone are cotton and tobacco, whilst subsistence crops comprise sorghum, millet, cassava, maize, groundnuts, sugar-cane, vegetable and fruit are also grown in 53

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 4.3 South Nyanza District: agro-ecological zones (after Republic of Kenya n.d.) 54

Karega-Mũnene

the zone. In addition, sisal is grown mainly for domestic use like making ropes and baskets (Republic of Kenya n.d.; Personal observation).

livestock) in the lakeshore region and along river courses especially in the northern half of the District (Republic of Kenya n.d.). The disease is spread by tsetse flies (Glossina sp.) which breed and live in thick bushes located in areas with moderate temperatures and high humidities at altitudes of less than 1,520 m. Unlike in West, Central and Southern Africa where tsetse flies and trypanosomiasis have been known for several centuries, their presence in East Africa is a recent phenomenon.

Fishing and stock-keeping are also important economic activities. Fishing is highly localised because fish is only found in Lake Victoria, artificial ponds, the Gogo Falls dam and rivers like the Kuja which drain into the lake. Several fishing methods are used in the exploitation of the resource. These include kek, a method which uses “a staked reed barrier across the river with gaps which lead into channels ending in one-way traps” (Beadle 1974: 199), plunge baskets which are used to scoop fish in shallow waters, fishing hooks and nets (Personal observation). The kek, plunge baskets and hooks are mainly used when the fish is intended for subsistence, whilst nets are used when fishing for the market. Some of the fish intended for the market is sold in markets located within the District, but most of the catch is sold in major urban areas like Nakuru and Nairobi and even in overseas markets (Republic of Kenya n.d.; Personal observation).

The flies were first noted in the Lake Victoria basin only during the first decade of the last century, following their invasion from Central Africa via river valleys in Uganda (Morris 1960; Soff 1969). Within a few years of invading the lake basin, the flies spread beyond the lakeshores into the interior of western Kenya following the courses of rivers like the Kuja, resulting in heavy losses of both human and domestic animal lives. To check the spread of the scourge, the colonial authorities ordered evacuation of infested areas and destruction of riverine and lakeshore vegetation. By 1910 the measures had ensured that the disease no longer threatened the region in epidemic dimensions (Soff 1969).

Unlike fishing, stock-keeping is fairly widespread in the District. The animals reared include cattle, sheep, goats and chicken. These animals are reared for such products as milk, ghee, eggs, meat and hide; some of these products are consumed locally, but others are sold outside the District. Oxen are highly valued because farmers use them as draught animals when ploughing the land. Dogs are also kept in some homesteads mainly for the purposes of security and occasional hunting (Republic of Kenya n.d.; Personal observation).

The fight against the disease and the flies continues to date. This involves the employment of a combination of various approaches which include bush clearing, ground spraying and trapping. In consequence, both the disease and the flies have been exterminated in the southern half of the District (Republic of Kenya n.d.). Indeed, during the recent excavations at the Gogo Falls site by the present author, not a single tsetse was sighted at either the site or the camp which was located at Thimlich Ohinga, approximately 4 Km to the northwest of the site.

It must be remarked here, however, that stockkeeping is seriously affected by trypanosomiasis (i.e., sleeping sickness in humans and nagana in

55

CHAPTER FIVE

ARCHAEOLOGICAL INVESTIGATIONS

The existence of the Gogo Falls site was first noted by an archaeologist in the 1960s during field survey of the numerous prehistoric stone structures, ohingni, in South Nyanza District (Lofgren 1967). In the course of that survey, the existence of several potsherds on the surface where the site is located was noted. The majority of those sherds were undecorated, but a few had regular parallel channels, stamped chevron decoration, or roulette decoration. In spite of that discovery, the site remained uninvestigated for over a decade.

Trench I measured 6 m x 2 m and had a depth of 125 cm. Both the topsoil and the uppermost levels contained a mixture of Urewe and Kansyore pottery, but the pottery in the lower levels was predominantly Kansyore. Also recovered from the same levels were stone artefacts, faunal remains and a human skeleton. A total of seven chronometric dates were obtained on obsidian, charcoal and bone samples from the trench. These range from about 18,000 BP to 2,000 BP (Table 5.1). Trench II measured 2 m x 2 m and was excavated to a depth of 225 cm. The trench was extended by another 2 m x 2 m trench which was only excavated to a depth of 65 cm. The upper levels in both the main trench and the extension consisted of ashy deposits from which iron artefacts, Elmenteitan and Urewe pottery and faunal remains were recovered. Large animal burrows were encountered at about 75 cm - 165 cm below the surface; the finds and soil from those burrows were discarded. Underlying the ashy deposits were clay loam deposits of brown/black colour in which faunal remains, Elmenteitan pottery and stone artefacts were found. These were, in turn underlain by Kansyore pottery, lithic artefacts and faunal remains. Two radiocarbon determinations, both lying in the second millennium BP, were obtained on charcoal samples from the trench (Table 5.1).

The first systematic field research was conducted at the site in 1980 when a limited surface collection of pottery was undertaken (Collett and Robertshaw 1980). Analysis of those finds indicated the presence of Kansyore and Urewe pottery at the site. In the following year a small ‘window’, measuring 75 cm x 45 cm, was excavated into a section that had been exposed at the site by repair work at the nearby dam, in order to test for in situ finds. More Kansyore and Urewe pottery, plus a few faunal remains and lithic artefacts were obtained from the ‘window’ (Robertshaw 1985, 1991). Two years later (1983) extensive investigations involving the excavation of five trenches that were designated Trenches I, II, III, IV and V (Figure 5.1) were undertaken at the site. Although the excavations have now been published (Robertshaw 1985, 1991), it is important that we describe them briefly here since part of the faunal remains recovered that year form part of this study.

Trench III also measured 2 m x 2 m and the maximum depth of deposits was 195 cm. The uppermost levels consisted of ashy deposits that yielded Elmenteitan and Urewe pottery, iron artefacts and faunal remains. Large animal burrows were encountered from about 85 cm below the ground surface. As in Trench II, the loose soil and all finds from the burrows were discarded. The lowermost deposits yielded faunal remains, lithics plus Elmenteitan and Akira pottery. Two charcoal samples from the trench provided determinations which place the

The excavation of each trench was preceded by the removal of the topsoil with picks and shovels to a depth of 15 cm. Excavation below the 15 cm level was undertaken using trowels and proceeded in arbitrary spits of 10 cm each. All the earth excavated below the 15 cm topsoil was screened through a sieve of 5 mm mesh.

56

Karega-Mũnene

Figure 5.1 Gogo Falls site plan 57

Holocene Foragers, Fishers and Herders of Western Kenya

Table 5.1 Chronometric dates from the 1983 excavations at Gogo Falls.* Dated material

Provenience

Deposits

Date (BP)

Laboratory/Specimen Number

Charcoal Obsidian Charred tooth Obsidian Obsidian Obsidian Obsidian Charcoal Charcoal Charcoal Charcoal

Trench I, level M3 Trench I, level M3 Trench I, level M4 Trench I, level M6 Trench I, level M6 Trench I, level M7 Trench I, level M9 Trench II, level A3 Trench II, level FA4 Trench III, level F2 Trench III, level F12

Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Sandy loam soil Ashy deposits Ashy deposits Ashy deposits Ashy deposits

3020 ± 100 2617 ± 194 7300 ± 500 3613 ± 115 2096 ± 93 2474 ± 65 18008 ± 383 1900 ± 70 1710 ± 70 1770 ± 80 1990 ± 80

HAR-6262 292-8 OxA-745 334-106 334-110 292-9 292-10 HAR-6260 HAR-6261 HAR-6252 HAR-6254

* Source: Robertshaw 1991.

concerned levels in the second millennium BP (Table 5.1).

1994; Stewart 1991), while the remainder and all the faunal remains from Trenches I, IV and V were analysed for this study.

The area covered by Trench IV measured 2 m x 2 m. Of the four 1 m x 1 m squares that formed the trench, only one was excavated to the maximum depth of 155 cm; the remaining 3 were excavated to a depth not exceeding 15 cm. The uppermost deposits in the trench consisted of hard sandy loam deposits that continued to a depth of about 85 cm. These were underlain by softer clay-like loam deposits. The finds from the trench included faunal remains and pottery that was predominantly Elmenteitan.

Recent Investigations Further excavations were undertaken at the site in 1989 by the present author. At the beginning of that year’s field research, the larger part of the site, further to the west of River Kuja, was a maize and sorghum inter-cropped farm and a small portion on the northern section of the farm was planted with tobacco and cabbage. The remaining part of the site, close to the Kuja, was covered by thorn brush and grass which measured over 1 m high, a typical characteristic of grasses in the lakeshore region.

Trench V measured 4 m x 2 m and was excavated to 135 cm below topsoil. The whole trench had ashy deposits that yielded Urewe, Kansyore and Elmenteitan pottery, the Elmenteitan being predominant throughout the deposits. Other finds from the trench included iron artefacts and slag, stone artefacts, faunal remains and an almost complete Urewe vessel.

Since working on the farm would have meant destroying crops for which the farmers concerned would have required compensation (Republic of Kenya 1984), initial research efforts were concentrated on the part of the site adjacent to the river, where it was hoped representative samples of archaeological material would be recovered. Indeed, it has been suggested that when one is investigating a multi-component site like Gogo Falls, efforts should be made to excavate close to and at the foot of slopes because these are the likely locations for rubbish middens (Alexander 1970).

The pottery, stone artefacts and plant remains that were recovered during the excavations described above have been published (Robertshaw 1991; Robertshaw and Wetterstrom 1989; Wetterstrom 1991); also published are obsidian pieces that have been studied in an attempt to identify the source of the raw material (Merrick and Brown 1984a, 1984b). The results of this latter exercise suggest that the obsidian used at the site came from Mount Eburu, Njorowa Gorge and Sonanchi Crater, all of which are located in the Lake Naivasha basin. In addition, portions of the faunal collections from Trenches II and III were analysed prior to the present study (Karega-Mũnene 1986, 1987; Marshall 1986, 1991; Marshall and Stewart

The first step involved physical examination of the vegetation cover by walking the area before it was cleared. This revealed the existence of a distinct patch with lush vegetation in the southern half of the area. The patch was marked out by driving long sticks into the ground (for easy identification) after which the whole area was cleared. This was followed by the

58

Karega-Mũnene

establishment of two baselines covering the entire length and breadth of the area at right angles: the east-west baseline was 20 m long and the south-north one 85 m. The point of intersection of the baselines was located to the extreme south of the area.

The first two spits from the surface were relatively hard because of soil compactness arising from treading by domestic animals that graze the area and from the network of roots of the grass and other plants that had been cleared away. The succeeding spits were much softer and consisted of ashy deposits (Figure 5.3), which were so identified because of their grey colour, looseness and the presence of charcoal flecks and burnt archaeological specimens. These factors are characteristic of deposits derived mainly from ashes (Courty et al. 1989).

Sampling and excavation The baselines were used to grid the area at 5 m intervals both eastwards and northwards. This yielded 52 squares, each measuring 5 m x 5 m. The surface of each square was then examined for pottery, lithics and faunal remains. These were collected and bagged separately for each square. Subsequent examination of the finds from all the 52 squares revealed that the highest concentration of finds occurred in two squares which were located in the patch that had been marked out prior to the clearing.

Some of the ashy deposits that were found between the boulders and small stones at the bottom of the squares were of a dark colour, probably because of mixing with what appeared to be immature black soil. Numerous stones with a maximum length of about 7 cm were also encountered in the ashy deposits. These had to be removed in order to allow for further excavation until the finds petered out and bedrock or boulders were reached.

In addition to yielding scanty finds, the majority of the squares in the northern half of the area were rocky. Attempts to excavate in two squares picked at random in that area indicated that any excavation there could only proceed with immense difficulties since the area was not only rocky, but the ground was also very hard. Thus, excavating in that area would not only have slowed down the excavation, but also involved expending the limited research resources in pursuit of meagre finds. It was, therefore, decided that excavation efforts should be concentrated on the southern half of the cleared area. Thereupon, the two squares with the highest concentration of surface finds were targeted for the season’s first excavations; this area was designated Area A (Figure 5.1).

In squares 1 and 2 bedrock was reached at 50 cm below the surface and in square 3 at 70 cm. The finds from these squares included faunal remains, stone and iron artefacts, ostrich eggshell beads, plus Akira, Elmenteitan, Kansyore, and Urewe pottery. The majority of these finds were recovered in situ, but some faunal remains, lithics and beads could only be recovered by sieving the earth through a 5 mm mesh because of their small size. Throughout the excavations, finds of similar nature, say, bones, lithics, iron artefacts or pottery from the same spit and square were bagged separately. The square and spit from which they came were identified by the East and North co-ordinates of the square and the depth of the spit concerned, which were labelled on each bag. In order to ensure that the finds collected through sieving were bagged in the appropriate bags, the earth excavated from every spit of each square was sieved separately.

The two squares in Area A were then gridded into 50 squares of 1 m x 1 m. Three alternate squares, that is, squares 1, 2 and 3 in the first row of the larger square next to the river bank, were selected for excavation (Figure 5.2). These were used to test the depth of deposit and whether there were any discernible natural layers that could be used for excavation. Unlike the 1983 excavations, the topsoil was not removed with shovels, but systematically excavated in arbitrary spits of 10 cm right from the surface to the bedrock or until no further finds were encountered.

Since the test excavations revealed only one layer of homogeneous deposits, further excavation in Area A proceeded in arbitrary spits of 10 cm each from the surface to bedrock or until the finds diminished. But efforts were made throughout to detect any other natural layers. Three squares (4, 5 and 6) were excavated in the second row, the first

The exposed sections in the squares revealed only one layer of homogeneous ashy deposits.

59

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 5.2 Gogo Falls: excavations at Area A

60

Karega-Mũnene

Figure 5.3 Gogo Falls: stratigraphy of square 20 61

Holocene Foragers, Fishers and Herders of Western Kenya

two to 60 cm below the ground surface and the third to a depth of 70 cm. This was followed by the excavation of three more squares in the third, fourth and fifth rows, respectively. This resulted in the exposure of a trench measuring 5 m x 1 m in the middle of one of the larger squares, plus two trenches measuring 3 m x 1 m, cutting across the long trench, and three 1 m x 1 m squares on either side of the long trench. On the whole, 15 squares, each measuring 1 m², were excavated.

light rains started, the surface survey conducted there was neither systematic nor intensive. This meant laying no grid, but walking the farm looking for distinct concentrations of archaeological finds on the surface. This task was made easier by the domestic animals which, while grazing on the farm, had kicked material around. Consequently, several concentrations were identified. Since it would not have been possible to investigate all of them, it was decided to excavate only two, which were designated Areas D and E.

The long trench was subsequently extended westwards by another five 1 m x 1 m squares (i.e., squares 16 through 20), thus cutting across the second larger square that had been targeted for excavation. The deposits in the trench were 90 cm - 100 cm deep. The finds from the trench and from the other 15 squares consisted of faunal remains, iron and lithic artefacts, ostrich eggshell beads, pieces of charcoal and Akira, Elmenteitan, Urewe and Kansyore pottery.

Area D was located to the west of Area C; the squares excavated there (i.e., 25, 26 and 27) formed a 3 m x 1 m trench which was aligned east-west. During the excavation, which was undertaken in arbitrary spits of 10 cm each as in the areas excavated earlier, efforts were made to locate natural strata. This revealed three distinct horizons besides the topsoil, the uppermost of which consisted of homogeneous ashy deposits to a depth of about 90 cm.

Two more areas, B and C, were selected at random in the cleared area and targeted for excavation. Both areas were located to the south of square 20, Area C being at the edge of the cleared area. In each of these areas a trench consisting of two 1 m x 1 m squares was excavated. The trench in Area B (squares 21 and 22) was aligned south-north and the one in Area C (squares 23 and 24) east-west.

These were underlain by deposits of dark brown/black loam soil (also known as black cotton soil) and clay-like deposits, in that order (Figure 5.5). The former deposits extended to a depth of about 140 cm and the latter to the bottom of the squares (i.e., 200 cm - 230 cm). While the brown/black loam deposits were relatively soft, the clay-like deposits became increasingly hard towards the bottom of the squares. The finds made from the squares included Akira, Elmenteitan, Kansyore and Urewe pottery, faunal remains, lithics, iron artefacts, ostrich eggshell beads and small pieces of charcoal.

Excavation in each of these squares also proceeded in arbitrary spits of 10 cm each. Although efforts were made to identify natural layers during excavation, only one layer of homogeneous ashy deposits was noted (Figure 5.4). The deposits graded into what appeared to be a very thin layer of immature black soil towards the bottom. The extent of the latter layer was difficult to record because in addition to mixing with the overlying ashy deposits, it was obscured by the presence of boulders and numerous small stones at the bottom of the trenches. The finds made in both trenches included pieces of charcoal, faunal remains, ostrich eggshell beads, stone and iron artifacts, plus Akira, Elmenteitan, Kansyore and Urewe pottery.

The last three squares to be excavated were located in Area E, to the northwest of Area D. These were designated 28, 29 and 30, respectively. Together, they formed a 3 m x 1 m trench which was aligned south-north. The deposits in the squares consisted of black loam (cotton) soil only and were 150 cm - 160 cm deep (Figure 5.6). Excavation in each of the squares also proceeded in arbitrary spits of 10 cm each. A large burrow was encountered in square 28 at about 60 cm - 100 cm below the ground surface (Figure 5.6). All the finds and soil from the burrow were discarded, but the portions of the square that had not been affected by burrowing were excavated systematically. The excavations yielded Akira, Elmenteitan, Kansyore and Urewe pottery, faunal remains,

When the crops were harvested research efforts were shifted to the farm in order to extend the spatial coverage of the excavations. When doing this, it was borne in mind that the 1983 excavations had been located there. Since the farm was to be prepared for planting before the 62

Karega-Mũnene

Figure 5.4 Gogo Falls: stratigraphy of square 23

63

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 5.5 Gogo Falls: stratigraphy of square 25

64

Karega-Mũnene

Figure 5.6 Gogo Falls: stratigraphy of square 28

65

Holocene Foragers, Fishers and Herders of Western Kenya

lithics, ostrich eggshell beads, pieces of charcoal and an iron artefact. A complete Elmenteitan vessel was recovered from square 29 at a depth of 60 cm - 78 cm below the surface.

in the Kuja may have given rise to overbank flooding, thus extending its west bank to the lower parts of the site, where most of the recent excavations were located. The clayey deposits excavated there attest to this.

Limited flotation for plant remains was also undertaken. This was done in collaboration with a palaeobotanist from Leiden University in the Netherlands, who was also to identify the recovered remains. Soil samples collected from ashy deposits in Areas A, B and C were wetsieved through three mesh sieves measuring 2 mm, 1 mm and 0.5 mm, respectively. Some plant remains were indeed recovered, but their identification is yet to be communicated to the author.

The subsequent retraction of the evergreen forest and the appearance of a deciduous forest in the region suggest there was considerable fluctuation in the amount of rainfall experienced in the lake basin. This may, in turn, have caused fluctuations in the river’s discharge. Thus, when the river was at its widest, it could have reached the lower part of the slope where it took away considerable quantities of soil and other material. Eventually, some of the area affected in this way may have become part of the river floor. The erosion may also have exposed the boulders that were encountered during the excavation of Areas A, B and C, if the assumption that they were eroded from the river floor and/or the banks is correct. That the boulders were fairly rounded and lacked sharp corners and protrusions suggests they had probably been worn away by friction with other eroded material, the riverbed and flowing water.

Site formation and stratigraphy It is clear from the above description of the excavations that the deposits in the 30 squares excavated in 1989 were either of one, two, or three kinds only. The squares which were located close to the river (i.e., squares 1 - 24) had ashy deposits only, whilst those farthest from the river (i.e., squares 28 - 30) had black loam deposits only. In contrast, the squares located in the middle (i.e., squares 25 - 27) had ashy deposits in the upper spits, brown/black loam deposits in the middle and clay-like deposits in the lowermost spits.

As is the case with all rivers (see Briggs and Smithson 1985; Waters 1992), the Kuja did not only carry materials eroded from its banks and floor. It must also have received and transported materials that had been eroded from its catchment area. Such materials would have included boulders, pebbles, sand, silt and clays. The movement of these materials would have been greatly influenced by their respective sizes and water flow velocity. Thus, other things being equal, the materials would have been transported over longer distances during periods of high water discharge than during periods of low discharge.

Given that the site is located on a hill-slope (Figure 5.1) and that the four pottery wares which were found in the same deposits at the site have previously been assigned to different periods, it is imperative that we attempt to evaluate the site’s stratigraphic integrity. This will indicate whether the excavated deposits were formed in place by human activity or derived from up-slope by processes like surface run-off or slumping. In order to do this effectively, our discussion will start in the early Holocene when the site was apparently unoccupied.

Transportation of eroded materials is, however, not always continuous. Some of the materials are carried only a short distance and deposited there, whilst others are carried over a long distance before they are deposited. Additionally, some of the deposited materials rest where they are initially deposited for a short time only before they are transported and deposited elsewhere (Briggs and Smithson 1985; Waters 1992).

On the basis of the past environmental and climatic evidence discussed in chapter three, it seems likely that the Kuja was at its widest during the early Holocene (ca. 10,000 BP - 8,000 BP) when the Lake Victoria basin experienced high rainfall. This latter resulted in a remarkable increase in surface water, hence high water levels in the rivers draining into the lake and the development of marshes and an evergreen forest around the lake. The increased volume of water

The rate and distance over which these materials are transported are dependent on the effectiveness of the water as a transporting agent and on the resistance of the eroded materials to

66

Karega-Mũnene

motion. The effectiveness of water as an agent of transportation, for instance, may be enhanced or reduced by water flow velocity and/or the topography of a given area. On the other hand, the resistance of material to transportation may be increased by vegetation cover as well as by the size, weight and shape of the material (Briggs and Smithson 1985; Waters 1992).

materials deposited under oxidising conditions, which could not have occurred when the area in question was under flowing water (C. French, personal communication, 1993). It was in this natural environment that the site seems to have been formed. The initial occupation of the site may have occurred when the river had receded, the west bank probably being located close to the present bank. The exposure of the area, which had previously been under water, seems to have facilitated soil formation there. This is attested by the presence of what appeared to be ‘immature’ black soil between the boulders and the small stones at the bottom of the squares located close to the present bank. (The ‘immature’ soil could have formed in situ or it could have been eroded and redeposited from up-slope.)

It has been observed, for instance, that coarse and heavier material like boulders, pebbles, gravel and large sand particles move along the river bed, whilst lighter material such as silt, clay and fine sand move as suspended load. Because of their weight, size and shape, boulders move only during heavy floods; pebbles and large sand particles are moved by floods and what may be regarded as normal-low discharge; and fine sand, silt and clay are moved by varied levels of discharge (including fairly low discharge). Because of these characteristics and because river discharge is never constant through time, eroded materials are continuously sorted and deposited by size, the coarsest being deposited first and finer sediments last (Briggs and Smithson 1985; Waters 1992).

Field and laboratory observations (e.g., Briggs and Smithson 1985; Gerrard 1981; Waters 1992) suggest there are several ways in which materials can be moved down or across a given slope. The most notable among these are wind action, rainsplash, surface run-off and mass movement. The effectiveness or otherwise of any of these processes is determined by such factors as the topography of the slope, erodibility of the parent material and climate. Since these factors vary through space and time, some areas are affected more by, say, mass movement and others by wind action, rain-splash, or surface run-off.

This may explain why boulders and pebbles generally accumulate as a ‘lag concentrate’ in areas like Gogo Falls, where the relevant parent materials are available. The deposition of these materials occurs especially when the discharge remains relatively low for a considerable period. In contrast, the finer sand, silt and clay are deposited on the floodplain and even on the riverbank and/or floor when the discharge is low or when the water recedes (Briggs and Smithson 1985; Waters 1992).

In areas like Gogo Falls which are located in dry savanna, material is moved across slopes mainly by rain-splash and surface run-off. The former process involves the detachment of individual particles by raindrops. On slopes like Gogo Falls this causes most of the detached particles to be splashed down-slope and a few up-slope. In contrast, movement of sediment on flat landscapes by rain-splash is minimal (Briggs and Smithson 1985).

Thus, the boulders that were found at the bottom of squares 1 through 24 may have been deposited in the manner outlined above given they lacked sharp edges and protrusions, both of which may have been worn away during transportation by the running water. It also appears likely that the clayey deposits that were excavated in squares 25 - 27 were deposited in the manner outlined above. Ponds could have formed when the river’s discharge was low and/or when the river receded, thus trapping the deposits and enabling the settling down of suspended silt and clay particles. This postulation is supported by the observation that the dark brown colour of the deposits suggests the accumulated material contained relatively organic, eroded soil

Surface run-off occurs when the amount of rain reaching the ground surface exceeds the infiltration capacity of the surface. The effects of this process as well as of rain-splash are less pronounced on vegetated slopes than on bare slopes. That is because vegetation improves soil structure and increases the rate of infiltration. In addition, it intercepts raindrops, thus reducing their impact on the soil as well as preventing the formation of a compacted layer which would

67

Holocene Foragers, Fishers and Herders of Western Kenya

otherwise facilitate erosion (Briggs and Smithson 1985; Daniels and Hammer 1992; Gerrard 1981).

example, could have contributed to the formation of the brown/black and black loam soils in Areas D and E, respectively. In addition, most of these soils could have been formed in situ, while some of them could have been derived from up-slope by rain-splash and/or surface run-off.

Thus, if the supposition that the site was covered by vegetation when the immature soil at the lower part of the slope was forming is correct, it is unlikely that all the immature soil found in the squares located in Areas A, B and C could have formed through erosion from up-slope. Some of it could have formed in situ, a process that would have been enhanced by the activities of soil microfauna. These latter play a very significant role in soil formation processes in areas characterised by high temperatures and high rainfall (Briggs and Smithson 1985; Bunting 1965; Limbrey 1975; Morgan 1973), conditions that obtained in the Lake Victoria basin during the early-middle Holocene.

The occupation of the site must have disturbed the ‘balance’ of the natural processes acting on the hill-slope. For example, in creating room for their settlement, the occupants may have stripped the vegetation. This, in turn, could have speeded up the accumulation of material from up-slope to the middle and lower parts of the site. Assuming the present climatic and environmental conditions in the area are a reliable guide to past conditions, it is likely that most of the material was transported down-slope by rain-splash and surface run-off mainly at the beginning of the rainy season. That is because a week or so of continuous rainfall in the area is enough to generate considerable vegetation (Pers. observ.), thereby slowing down the processes of erosion and transportation of material.

If the present climatic and environmental conditions in the region (see chapter three) are anything to go by, the high temperatures and rainfall would have given rise to a high turnover of organic matter, a factor that could have made a significant contribution to soil formation. The maceration of the organic matter by termites, beetles, earthworms, ants, slugs and snails could have facilitated its decay. This would, in turn, have facilitated the breakdown of the remaining matter by smaller organisms like bacteria and fungi and by chemical oxidation. These activities do not only result in the incorporation of the matter in the soil, but also in the modification of both the composition and structure of the soil. The turning over of the soil by the building operations of the organisms also modifies and sometimes destroys soil horizonation (Briggs and Smithson 1985; Bunting 1965; Limbrey 1975; Morgan 1973).

More importantly, human activity at the site generated the ashy deposits and the archaeological material that were excavated there. The ashy deposits appear to have been deliberately deposited in Areas A, B and C (and also in Trenches I, II, III and IV) by the site’s occupants. This action could have been caused by the occupants’ desire to increase the amount of land available for their use by dumping debris on the old riverbed and bank.

The speed with which this process takes place varies through space. In cool temperate areas, for instance, the incorporation in the soil of the dead leaves of a deciduous tree may take about a year; but the same process may take only a few weeks in warm tropical areas. The process is faster in the latter areas because of high temperatures; these tend to intensify the activities of the organisms in question as well as the chemical processes that cause the decay of the remaining matter (Briggs and Smithson 1985).

Robertshaw (1991: 163) has suggested that the ashy deposits “may represent the remains of burnt cattle dung”. It is also likely that some of the deposits may have come from hearths, while others may have been created by burning vegetation which had been cleared to increase the available land for settlement or by burning domestic refuse (including cattle dung). The ashes created in this way could have been raked off together with archaeological material and deposited down-slope. These suggestions are highly tentative; detailed microstratigraphic analysis of the deposits, an exercise that is beyond the scope of this study, is necessary if they are to be confirmed or refuted.

The soil formation process outlined above may not necessarily have been restricted to the lower parts of the site. Microfaunal activity, for

The fairly uniform texture and colour of the ashy deposits as well as the brown/black loam and black loam deposits excavated at the site suggest

68

Karega-Mũnene

they could have originated from the same source(s), respectively. This suggestion is supported by the fact that the differences observed in the texture and/or colour of those deposits occurred only between the strata, but not within. The homogeneity could also have arisen from transportation and deposition of each of the three types of deposits by the same processes. In contrast, deposits transported and deposited by different processes tend to have marked and sudden changes in both the vertical and horizontal profiles because the energy of the transporting and depositing agents varies between processes (Daniels and Hammer 1992).

brown/black loam deposits and in the black loam deposits could also have been incorporated in those deposits in a similar manner. This is not to say that none of the archaeological finds were moved by other agents after they had been discarded by humans. Indeed, it is likely that natural processes like surface run-off may have moved some of them. Like raking, these processes could have contributed to the apparent ‘mixing’ of cultural material of different ages. The looseness of the ashy deposits could also have contributed to the ‘mixing’. Needless to add, the contribution of the natural processes discussed above to the movement of archaeological materials would have been considerable if the settlement had been located up-slope and if the domestic refuse was discarded there. But evidence of residential units in the form of structural remains or features at the site is as yet not available. Moreover, the rubbish midden seems to have been located down-slope. Further research aimed at locating remains of residential units and/or any other useful features is needed, if this issue is to be resolved.

It will be recalled that numerous stones were found through most of the ashy deposits (particularly in the middle-lower parts of the site) and that such stones were not found in the other types of deposit. The stones were rough and angular, a condition that suggests one of two things. Firstly, if the stones were transported and deposited by flowing water, the transportation occurred over only a short distance before they were deposited because the water was not able to transport them further, or because they were trapped by vegetation or by the boulders. Secondly, the stones could have been raked off the ground surface together with the ashy deposits and archaeological material, as suggested above.

It must be emphasised here, however, that it is not possible to estimate the time over which either of the strata revealed by the excavations at the site was formed. Some or all of them could have been formed in a very short or a fairly long time. That is because time is a passive factor in the development of horizons in soil: “[it, i.e., time] is important... only to help establish a starting and stopping point and to compute process rate” (Daniels and Hammer 1992: 196). The most significant factors in horizonation include weathering intensity, type of parent material and depositional processes. Thus, we cannot assume that stratum X was formed over a longer period because it is thicker than stratum Z or vice versa.

Incorporation of archaeological material In sum, the incorporation of the archaeological finds made in the three types of deposit appears to have been largely due to human activity. As we have already suggested, the cultural and biological remains that were recovered from the ashy deposits could have been incorporated in the deposits when these were raked off the ground surface and redeposited down-slope. With time, this may have led to the formation of a rubbish midden in which domestic refuse may have been deliberately discarded.

Dating

The incorporation of archaeological finds in the clayey deposits seems to have occurred after the deposits had dried up since neither the cultural remains nor the faunal remains were encrusted in clay. This process could have been facilitated by the soil formation processes outlined above as well as by the mixing activities of soil microfauna. The finds that were made in the

Five of the charcoal samples that were collected in situ during the recent excavations were dated at the University of Oxford’s Radiocarbon Accelerator Unit. The context of those samples and the dates they provided are summarised in Table 5.2. The dates are uncalibrated in radiocarbon years, using the half-life of 5,568 years. Their significance and association with the

69

Holocene Foragers, Fishers and Herders of Western Kenya

archaeological material recovered from the site will be discussed in the next chapter. First, let us discuss, albeit briefly, the chronology of the site and the problems that may be associated with the dates.

sample to direct sunlight over long periods also has the same effect. Chemical variation within a given obsidian quarry can produce different ages on samples from the same archaeological contexts (Fleming 1976; Parkes 1986), whilst variation of the annual air/surface temperature wave in excess of 2°C - 3°C may result in unstable hydration environment, hence unreliable dates (Ridings 1996).

All the charcoal samples recovered from the site in 1989 were too small to be dated by conventional means. None of them exceeded 30 g, the minimum weight required for conventional dating (Parkes 1986: 13). Although bone could also have been dated, it was felt that the problems associated with such dates in East Africa outweighed any benefits that could have been obtained from them.

These shortcomings may explain why the obsidian chronology for the site of Aksum in Ethiopia (Michels 1986, cited in Munro-Hay [1989, 1991]), for example, is remarkably different from the chronologies established through other dating methods. In fact, while radiocarbon dates agree with the chronology devised from locally-minted coinage, ancient texts and architectural development in the region, the obsidian dates are either too young or too old (Munro-Hay 1989, 1991). Although Michels (1990) has suggested that this condition arises from sampling bias in the location of the excavation units at the site, it is probably advisable to treat the obsidian dates with caution. This view is supported by Parkes’ (1986) suggestion that obsidian hydration dating is probably best used as a relative rather than an absolute dating technique.

Thus, the charcoal samples were submitted for dating with the knowledge that samples of that size are not always directly associated with the level they come from. They can be derived from the level(s) above and/or below the level being dated. Consequently, they tend to provide average dates. However, the stratigraphy of the site and the nature of the deposits from which the samples came indicate they were associated with the artefacts and biological remains from the same levels. As such, the samples provided the best opportunity for dating. As we have already noted, several chronometric dates had also been obtained for the 1983 excavations. One of these was on bone, the rest being on charcoal or obsidian. Since we have already discussed the problems associated with bone dates in East Africa (see chapter two), it will suffice to briefly discuss only the problems associated with obsidian hydration dates. This dating method is based on the principle that when a fresh surface of obsidian is exposed by flaking, the flake or core absorbs water from its surroundings to form a hydration layer. The thickness of that layer increases through time and can be measured in the laboratory. Once it is known, it is correlated with known dates to obtain the age of the obsidian sample (Fleming 1976; Parkes 1986).

It is because of all this that caution is exercised in this study regarding the obsidian dates for Trench I at Gogo Falls. Inspection of the dates (Table 5.1) shows that two of the dated samples gave different readings – with a difference of about 1,500 years between them – even though they came from the same context (i.e., level M6). Secondly, the date for level M3, which was located 20 cm above M6, was about 500 years older than one of the dates for the latter level (i.e., 2096 ± 93 BP). Thirdly, the M3 date was about 140 years older than the date for level M7, which was located immediately below M6; and the M7 date was younger than one of the M6 dates (i.e., 3613 ± 115 BP) by about 1,100 years. Fourthly, the date of about 18,000 BP for level M9 is certainly too old for the associated archaeological material, a condition that also applies to the bone date of about 7,300 BP for level M4.

The main drawback of this method is that the rate of hydration is dependent on factors that are not necessarily constant. These include the chemistry and temperature of the area where the sample is buried, duration of exposure of the sample to direct sunlight and chemical composition of the sample. Deposits containing calcium compounds, for instance, tend to accelerate the rate of hydration; exposure of the

This situation could have arisen from incorrect dates or lack of stratigraphic consistency in the trench, probably due to previous disturbance. Although we cannot completely rule out the

70

Karega-Mũnene

Table 5.2 Chronometric dates on charcoal from the 1989 excavations at Gogo Falls. Provenience Sq. 26; 100-110 cm Sq. 26; 160-170 cm Sq. 27; 120-130 cm Sq. 28; 130-140 cm Sq. 29; 130-140 cm

Deposits

Date (BP)

Lab. No.

Brown/black loam Clayey deposits Brown/black loam Black loam Black loam

2030 ± 70 3170 ± 70 2000 ± 70 2030 ± 65 3480 ± 75

OxA-3493 OxA-3494 OxA-3495 OxA-3496 OxA-3497

Table 5.3 Frequency of artefacts and faunal remains by excavated areas.* Area

Deposits

Sherds

Iron

Bones

Total

(m³) A B C D E

14.6 1.8 1.5 6.4 4.6

29112 3675 4263 6176 5389

34 5 3 3 1

25553 3227 2917 20843 7705

54699 6907 7183 27022 13095

Total

28.9

48615

46

60245

108906

* The complete vessel is included among the sherds.

Table 5.4 Frequency of artefacts and faunal remains by excavated squares.* Square Depth (cm) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Total

50 50 70 60 60 70 60 60 60 70 70 70 70 90 80 90 90 90 100 100 90 90 70 80 200 210 230 160 150 150

Sherds

Iron

Bones

Total

626 872 844 525 1782 1045 1051 919 1023 1906 964 1527 1557 1949 1019 1607 2311 2499 2028 3058 1797 1878 2151 2112 1875 1938 2363 1726 1696 1967

1 2 0 0 2 2 3 1 0 0 2 0 1 0 0 1 3 13 2 1 4 1 0 3 2 0 1 1 0 0

264 593 459 586 1334 595 627 505 724 1148 894 848 1041 1664 889 1708 4074 1886 2611 3103 1615 1612 1246 1671 8476 4746 7621 2735 2280 2690

891 1467 1303 1111 3118 1642 1681 1425 1747 3054 1860 2375 2599 3613 1908 3316 6388 4398 4641 6162 3416 3491 3397 3786 10353 6684 9985 4462 3976 4657

48615

46

60245

108906

*The complete vessel is included among the sherds.

71

possibility that the trench had suffered previous disturbance, the former suggestion seems more likely because of the problems associated with obsidian hydration dates. Furthermore, evidence from other parts of the site suggests there was stratigraphic consistency. The charcoal dates from Trench III, for example, indicate there was stratigraphic consistency. The younger date of about 1,700 BP from there is for level F2 that was located above level F12, which is dated to about 1,900 BP. On the other hand, a casual glance at the dates from Trench II suggests stratigraphic inconsistency for that trench. That is because the older date of about 1,900 BP is for level A3, while the younger date of about 1,700 BP is for level FA4 which was located below A3. Statistically, however, the dates are in agreement. The younger dates of about 2,000 BP for squares 26 and 27 are for a level that was located above the level dated about 3,100 BP in the former square. The deepest deposits in square 29 date to about 3,400 BP, while those in square 28 date to about 2,000 BP (Table 5.2). It is quite likely, however, that the latter date was obtained on a sample derived from the overlying deposits. This likelihood is supported by the fact that a large animal burrow was encountered in the square at about 60 cm - 100 cm below ground surface. Thus, it appears that the clayey deposits in Area D (squares 25, 26 and 27) and the lowermost black loam deposits in Area E (square 28, 29 and 30) are the oldest deposits containing archaeological material. As such, they probably represent the earliest human occupation at the site, dating to about 3,400 BP - 3,100 BP. The brown/black loam deposits in Area D are the second oldest and the ashy deposits in Areas A D and in Trenches I, II, III and V the youngest. The former deposits date to about 2,000 BP and the latter to about 1,900 BP - 1,700 BP.

Sorting and cataloguing The excavation yielded several thousand artefacts and faunal remains. The frequency of the faunal remains and artefacts (ceramics and iron artefacts) that were analysed for this study is summarised in Tables 5.3 and 5.4. The former table summarises the frequency of the artefacts and faunal remains according to the excavated areas. The same information is presented in Table 5.4, but in relation to the maximum depth of deposits excavated in each square.

Holocene Foragers, Fishers and Herders of Western Kenya

As Table 5.3 shows, about half of the finds came from Area A where a total of 20 m² were excavated. Areas D and E also yielded considerable numbers of finds. In each of these areas only 3 m² were excavated, but the depth of deposits was greater than at any of the other areas (Table 5.4). The significance of the differences in the spatial distribution of the finds at the site (which is evident in both Tables 5.3 and 5.4) and also in their temporal distribution will be discussed at length in the succeeding chapters. First, we turn to the sorting and cataloguing of the finds.

each lot bagged separately. The sherds were also separated into two categories: (a) diagnostic specimens which consisted of decorated sherds, basal sherds, rims, handles, lugs, and spouts, and (b) non-diagnostic specimens consisting of undecorated body sherds. Further sorting of the faunal remains took place at the National Museum in Nairobi, where they were separated into identifiable and non-identifiable elements. The cataloguing of all the finds was also done at the National Museum. All ostrich eggshell beads, lithic and iron artefacts, diagnostic sherds, plus identifiable faunal remains were catalogued and bagged individually in order to facilitate detailed physical examination of each specimen. Nondiagnostic specimens of the same kind were catalogued together, if they came from the same spit and square. Thus, all undecorated body sherds from the same spit and square were packed in the same bag and given the same catalogue number. Similarly, all the nonidentifiable faunal remains from the same context were catalogued together. This exercise was followed by detailed analyses of the finds through the approaches discussed at length in the succeeding chapters.

Initial sorting of the finds from the site occurred during excavation when the faunal remains, ostrich eggshell beads, lithics, iron artefacts and pieces of charcoal that were collected in situ were bagged separately. The finds that were recovered during sieving were also packed in the bags holding finds of similar nature, but all the pieces of charcoal that were recovered in this manner were discarded. At the close of each day’s excavation the faunal remains, lithics and pottery were washed separately. Once dry, the faunal remains were separated into mammalian and fish bones and

72

CHAPTER SIX

THE FINDS AND APPROACHES TO ANALYSIS

Of the finds recovered at Gogo Falls during the 1989 excavations, only the faunal remains and ceramics were studied in detail. That is because the former provided information about subsistence activities, which is the concern of this study, whilst the ceramics were a potential source of information about the site’s chronology. That ceramics have hitherto been used as chronological markers in East Africa made it imperative that special attention be given not only to the identification of the sherds to specific wares, but also to their spatial and chronological distribution. The distribution of iron artefacts was equally important because, unlike the pottery wares by which the Iron Age is known, they provide incontrovertible evidence of use of iron at the site.

more than 10 pieces; the rest of the squares yielded 1 - 4 pieces each. Stratigraphically, the artefacts were generally restricted to the upper levels, 0 - 40 cm (Table 6.2) where, as we shall learn presently, they were associated with Urewe, Elmenteitan, Akira and Kansyore pottery. Ceramics The ceramic assemblage consists of 48,614 sherds, plus one complete vessel. The majority of the sherds, that is, 42,543 representing 87.5%

Table 6.1 Frequency of iron artefacts by squares. Square

Brief examination of the beads revealed they were made on ostrich eggshell and that most of them are complete. They comprise polished discs with a circular perforation at the centre that appears to have been drilled from both directions. Examination of the lithics revealed they are generally made on obsidian and that virtually all of the retouched implements are microliths, the dominant tool type being the crescent. Contrary to observations made at other Holocene sites in East Africa (e.g., Barthelme 1985; Hivernel 1978; Robertshaw 1990a; Robertshaw and Cable 1990), the examination revealed that potsherds exceeded stone artefacts by large margins throughout the deposits.

Frequency

%

1 2 5 6 7 8 11 13 16 17 18 19 20 21 22 24 25 27 28

1 2 2 2 3 1 2 1 1 3 13 2 1 4 1 3 2 1 1

2.17 5.35 4.35 4.35 6.52 2.17 4.35 2.17 2.17 6.52 28.26 4.35 2.17 8.70 2.17 6.52 4.35 2.17 2.17

Total

46

100.00

Iron Artefacts A total of 46 iron artefacts and fragments were recovered from the site during the recent excavations. The majority of those finds are fragments, the few complete specimens being heavily rusted arrowheads and various forms of rings. Attempts to conjoin the fragments were unsuccessful because of their rusty condition.

Table 6.2 Frequency of iron artefacts by levels. Level (cm) 0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80

The spatial distribution of those artefacts and fragments is worth noting here because 34 (73.9%) of them came from Area A (Table 6.1). The remaining 12 (26.1%) artefacts came from Areas B, C, D and E. Of the squares containing the artefacts, only one (i.e., square 18) produced

Total

73

Frequency

%

12 8 7 7 4 6 0 2

26.1 17.4 15.2 15.2 8.7 13.0 00.0 4.4

46

100.0

Holocene Foragers, Fishers and Herders of Western Kenya

of the entire assemblage, are plain body sherds. As such, they could not be identified to specific wares because the wares are defined exclusively on the basis of decoration and other features visible on the surface. The remaining 6,071 (12.5%) sherds, plus the complete vessel (Figure 6.1), made up the diagnostic category, which includes decorated sherds, rim sherds, basal sherds, lugs, handles and spouts.

similar pattern is evident from examination of the sample sizes and the volume of excavated deposits. In contrast, examination of the density of the ceramics versus sample size reveals a different pattern. The highest density occurs in Area C, followed by Areas B and A, and E and D in that order, which is consistent with the variation observed in the distribution of finds both between and within the excavated deposits during excavation (Appendix I). Thus, there appears to be a proportional relationship between the aerial extent of the excavations and the volume of excavated deposits on the one hand, and sample size on the other hand. But such a relationship is not evident between sample size, on the one hand, and density of the finds or depth of deposits, on the other hand.

Examination of the spatial distribution of the ceramics reveals that the largest portion of the assemblage, 29,112 sherds which represents 59.9% of the assemblage, came from Area A (Figure 6.2). The second largest collection consisting of 6,176 sherds (12.7%) came from Area D; Area E yielded 5,388 sherds plus the complete vessel (11.1%); Area C 4,263 sherds (8.8%); and Area B 3,675 (7.5%).

The frequency of both non-diagnostic and diagnostic sherds in each square is summarised in Figure 6.4 and Table 6.4. As both the figure and table demonstrate, the number of diagnostic sherds is much smaller than that of nondiagnostic sherds in each square. Additionally, the size of the diagnostic samples compares favourably with the size of the overall samples. For example, the largest sample of diagnostic sherds comes from square 18 which has the second largest overall collection, whilst squares 4 has the smallest sample of diagnostic sherds and the smallest overall collection. The second largest sample of diagnostic sherds comes from square 20, which also has the largest overall collection. The second smallest sample is from square 1, which corresponds with the size of the overall sample.

The spatial distribution of the collection from Area A also varied quite considerably between the squares. Of the 15 squares adjacent to the river, none yielded more than 2,000 sherds: five of the squares yielded 1,500 - 1,999 sherds each; four 1,000 - 1,499 sherds apiece; and the remaining six 500 - 999 sherds each (Figure 6.3). In contrast, only one of the squares in the adjacent 5 m x 1 m trench produced less than 2,000 sherds; three of the remaining squares yielded 2,000 - 2,999 sherds apiece and the fourth over 3,000 sherds. One or a combination of the following factors may explain the spatial patterning: the depth of the deposits excavated in each area; the lateral extent of the excavations; and/or sheer differences in the density of finds. To begin with, as we have already noted, the depth of deposits and the lateral extent of the excavations varied considerably. The deepest deposits (230 cm), for instance, were encountered in Area D and the shallowest (50 cm) in Area A. Yet in terms of sample size Area A yielded about four and a half times the sherds recovered from Area D.

Close examination of Table 6.4 reveals neither the number of diagnostic sherds nor that of nondiagnostic sherds is proportional to the maximum depth of excavated deposits. The highest number of diagnostic sherds comes from square 18, which was only 90 cm deep, and not from square 27 whose deposits were about two and a half times deeper. Similarly, the highest frequency of non-diagnostic sherds comes from square 20, which was only 100 cm deep. The largest overall collection comes from square 20, not from the square which had the deepest deposits. Overall, square 27, which had the deepest deposists (230 cm) ranks second (together with square 20) for diagnostic sherds, fourth for non-diagnostic sherds and third for the overall collection.

Excavations in Area A covered a total of 20 m², whilst those in Areas D and E covered 3 m² in each area and in Areas B and C 2 m² in each area, respectively. The clusters formed by the samples from the five areas – the largest of which is from Area A, followed by Areas D and E, and B and C in that order – corresponds with the areal extent of the excavations (Table 6.3). A

74

Figure 6.1 Gogo Falls: Elmenteitan vessel

Karega-Mũnene

75

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 6.2 Gogo Falls: spatial distribution of the ceramic assemblage 76

Karega-Mũnene

Figure 6.3 Gogo Falls: spatial distribution of ceramics in Area A

77

Holocene Foragers, Fishers and Herders of Western Kenya

78

Karega-Mũnene

of three Neolithic wares, namely, Kansyore, Elmenteitan and Akira and one Iron Age ware, Urewe (Figures 6.5 - 6.7). The number of sherds representing these wares vary from as few as 45 for Akira to 2,639 for Elmenteitan (Table 6.5). Altogether, the sherds identified to the four wares represent 98.3% of all the potentially diagnostic sherds. Although the remaining portion had been included in the identifiable category because the sherds concerned were decorated, they could not be positively attributed to any of the wares since they were either minuscule or heavily eroded.

Table 6.3 Excavated areas, sample size and density of the ceramics*. Area

Excavated area

A B C D E

Volume of deposits

20 m² 2 m² 2 m² 3 m² 3 m²

14.6 m³ 1.8 m³ 1.5 m³ 6.4 m³ 4.6 m³

Sample

Density

29112 3675 4263 6176 5389

1993.97 2041.67 2842.00 965.00 1171.52

*The complete vessel is included in the Area E sample.

Table 6.4 Frequency of the ceramic assemblage by squares. Square

Depth (cm)

Diagnostic

Non-diagnostic

Total

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Total

50 50 70 60 60 70 60 60 60 70 70 70 70 90 80 90 90 90 100 100 90 90 70 80 200 210 230 160 150 150

78 96 99 48 234 150 128 152 157 217 124 154 193 200 131 184 265 376 201 327 152 207 197 189 309 309 327 246 241 279 5970

548 776 745 477 1548 895 923 767 866 1689 840 1373 1364 1749 888 1423 2046 2123 1827 2731 1645 1671 1954 1923 1566 1629 2036 1480 1455 1688 42645

626 872 844 525 1782 1045 1051 919 1023 1906 964 1527 1557 1949 1019 1607 2311 2499 2028 3058 1797 1878 2151 2112 1875 1938 2363 1726 1696 1967 48615

Spatial representation The frequency of the four wares through the excavated areas is summarised in Tables 6.6 and 6.7. Examination of both tables shows that the majority of the sherds (58.9%) that were identified to the wares came from Area A, followed by the samples from Areas D (15.8%), E (12.8%), C (6.5%) and B (6%). The representation of Kansyore and Elmenteitan wares in the five areas follows the same pattern, but that of Akira and Urewe wares is somewhat different. Area A yielded the largest sample, followed by Areas E, D, B and C, respectively. Interestingly, Elmenteitan sherds predominated in every area. The representation of Kansyore was also better than that of Urewe and Akira, except in Areas B and C where Urewe is better represented. The number of sherds belonging to each of the wares vary from square to square (Table 6.7). Akira, for instance, occurrs in 20 squares only, and, in each of those squares, it is represented by 5 or fewer sherds. In squares 11 and 22, for example, it is represented by two sherds apiece; in each of squares 3, 10, 18, 26 and 27 by three sherds; in squares 17, 21 and 29 by four sherds apiece; and in square 30 by five sherds. Its representation in the remaining nine squares is by isolated sherds.

Table 6.5 Composition of the ceramic assemblage. Identification

Number of sherds

%

Akira ware Elmenteitan ware Kansyore ware Urewe ware Nondescript decorated shreds Non-diagnostic plain sherds

45 2639 1676 1610 102 42543

0.1 5.4 3.5 3.3 0.2 87.5

Total

48615

100.0

Unlike Akira ware, Elmenteitan, Kansyore and Urewe are fairly well represented in all the 30 squares. The number of Elmenteitan sherds in each square varies from 12 to 183, Kansyore sherds from 21 to 111 per square and Urewe from 13 to 108 per square. Additionally, as Table 6.7 clearly demonstrates, these wares were found in every square and type of deposit. The only exception is the absence of Urewe sherds in the clayey deposits of square 26.

Composition of the ceramic assemblage The diagnostic sherds were identified to specific wares using the criteria defined by Wandibba (1977, 1980). This resulted in the identification 79

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 6.5 Gogo Falls: Kansyore pottery 80

Karega-Mũnene

Figure 6.6 Gogo Falls: Elmenteitan pottery 81

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 6.7 Gogo Falls: Urewe pottery 82

Karega-Mũnene Table 6.6 Frequency of the wares by areas. Area A B C D E Total

Akira

Elmenteitan

Kansyore

Urewe

No.

%

No.

%

No.

%

22 6 1 7 9

48.9 13.3 2.2 15.6 20.0

1389 183 213 505 349

52.6 7.0 8.1 19.1 13.2

1056 60 65 266 229

63.0 3.5 3.9 15.9 13.7

45

100.0

2639

100.0

1676

100.0

No.

Total %

No.

%

1047 110 107 167 179

65.0 6.8 6.7 10.4 11.1

3514 359 386 945 766

58.9 6.0 6.5 15.8 12.8

1610

100.0

5970

100.0

Table 6.7 Frequency of the wares by squares and type of deposit. Square 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Type of deposit Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Brown/black loam soil Clayey deposits Ashy deposits Brown/black loam soil Clayey deposits Ashy deposits Brown/black loam soil Clayey deposits Black loam soil Black loam soil Black loam soil

Total

Akira

Elmenteitan

Kansyore

Urewe

Total

1 0 3 1 0 0 1 1 0 3 2 1 1 0 0 0 4 3 0 1 4 2 1 0 0 1 0 3 0 0 3 0 0 0 4 5

33 47 35 12 101 63 54 59 57 77 47 35 76 76 32 86 119 156 88 136 75 108 108 105 100 57 2 127 31 5 136 34 13 128 116 105

21 21 38 22 63 50 40 52 64 58 30 60 47 58 66 50 73 111 50 82 23 37 28 37 9 38 52 26 15 42 10 11 63 62 73 94

23 28 23 13 70 37 33 40 36 79 45 58 69 66 33 48 69 106 63 108 50 60 60 47 34 10 6 50 10 0 42 4 11 56 48 75

78 96 99 48 234 150 128 152 157 217 124 154 193 200 131 184 265 376 201 327 152 207 197 189 143 106 60 206 56 47 191 49 87 246 241 279

45

2639

1676

1610

5970

concerned spits, or (ii) the stratigraphic integrity of the deposits may be open to question owing to their loose nature.

Chronological representation The stratigraphic distribution of the wares is also significant. Akira, for example, was restricted to the upper levels, where it is associated with Urewe, Elmenteitan and Kansyore (Appendix I). The few exceptions, which were observed in some spits of squares 4, 21, 27 and 29, are inconsequential because: (i) the four wares were associated in the deposits underlying the

The levels that contained Akira sherds were mainly of ashy deposits, the exceptions being squares 25, 29 and 30 where the sherds were recovered from loam deposits (Table 6.7). In addition, Akira was not represented in any of the dated levels; but the association between Urewe, 83

Holocene Foragers, Fishers and Herders of Western Kenya

Elmenteitan and Kansyore was dated to about 2,000 BP in squares 26, 27 and 28.

specimens. By definition, the diagnostic category included all specimens that could be identified to specific anatomical elements and taxa. These numbered 26,561, representing 35.5% of the entire assemblage; the majority of these (i.e., 13,611 elements) are from the 1983 excavations and the remaining 12,950 elements from the 1989 excavations (Figure 6.8).

Two dates were, however, obtained for Kansyore, the only ware found exclusively on its own. This condition existed in four squares: square 26 at 150 cm - 210 cm below the ground surface; square 27 at 190 cm - 230 cm; square 29 at 110 cm - 150 cm; and square 30 at 120 cm 150 cm (Appendix I). A charcoal sample from spit 160 cm - 170 cm of square 26 gave a reading of 3170 ± 70 BP (OxA-3494) and another sample from spit 130 cm - 140 cm of square 29 yielded a date of 3480 ± 75 BP (OxA-3497).

The elements belonging to the non-diagnostic category totalled 48,345, a figure that represents 64.5% of the entire assemblage. The majority of those specimens (47,295) were excavated in 1989 and the remaining 1,050 in 1983. The reasons for the disparity in the composition of the faunal collections that is evident in Figure 6.8 will be discussed at length in the next chapter. First, let us discuss the spatial configuration of the 1989 collection.

These dates are firmly associated with Kansyore pottery. The sample from square 26, for instance, was recovered in situ from below two other spits whose deposits were hard and clay-like and in which only Kansyore pottery was found. The underlying deposits were also clay-like and contained Kansyore pottery only. The sample from square 29 was also recovered in situ from hard brown/black loam deposits containing Kansyore pottery only. The pottery also occurred exclusively on its own in the deposits underlying the spit that yielded the dated sample.

The spatial distribution of the faunal remains from the 1989 excavations is summarised in Figures 6.9 and 6.10. It is evident from the former figure that Area A yielded the largest portion of the collection (i.e., 25,553 elements), representing 42.4% of the whole assemblage. The second largest sample, consisting of 20,843 elements (34.6%) comes from Area D followed by Areas E, B and C, which have 7,705 (12.8%), 3,227 (5.4%) and 2,917 (4.8%) elements, respectively. The clustering formed by this patterning is similar to that of the ceramic assemblage, the only difference being the reversal of the positions of Areas B and C in the sequence.

Three more dates were obtained on charcoal samples, which were also recovered in situ from squares 26, 27 and 28, respectively. The sample from the first square was recovered from the 100 cm - 110 cm spit where it was associated with Elmenteitan, Kansyore and Urewe pottery; it yielded a date of 2030 ± 70 BP (OxA-3493). The sample from square 27 came from the 120 cm 130 cm spit, where it was also associated with Elmenteitan, Kansyore and Urewe pottery; it gave a reading of 2000 ± 70 BP (OxA-3495). The third sample, with a reading of 2030 ± 65 BP (OxA-3496), was obtained from spit 130 cm - 140 cm of square 28, where it was associated with Kansyore and Elmenteitan pottery and a single Urewe sherd. The significance of these associations will be discussed at length in the succeeding chapters.

Close examination of Figure 6.10 shows that the 15 squares located close to the river yielded smaller samples than any of the five squares in the adjacent 5 m x 1 m trench. Of the former squares, two yielded 250 - 499 specimens each; nine squares 500 - 999 elements apiece; three squares 1,000 - 1,499 elements each; and one square 1,500 - 1,999 specimens. In contrast, two of the squares in the trench produced 1,500 1,999 elements apiece; another two 2,500 - 2,999 elements and 3,000 - 3,999 elements, respectively; and the remaining one over 4,000 elements. Interestingly, the largest faunal sample (over 4,000 elements) comes from square 17 rather than square 20, which produced the largest ceramic sample in Area A. Equally significant is the fact that, as with the ceramic samples, the second largest faunal

Faunal Remains The faunal assemblage that was analysed for this study consisted of 74,906 specimens. The bulk of these, 60,245 (80.4%), were excavated in 1989 and the remaining 14,661 (19.6%) in 1983. Like the ceramic assemblage, the faunal collection comprised both diagnostic and non-diagnostic

84

Karega-Mũnene

1989 collec 1983 collec

Whole assemblage Whole assemblage 60,245 14,661

14,661

1989 collection 1983 collection

1989 faunal collection Identifiable 12,950 47,295 Nonidentifia

60,245

1989 collection

12,950

1983 faunal collection Identifiable 13,611 Nonidentifia 1,050

Identifiable Nonidentifiable

47,295

1983 collection

1,050

Identifiable Nonidentifiable

13,611

Figure 6.8 Composition of the faunal assemblage

85

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 6.9 Gogo Falls: spatial distribution of the 1989 faunal collection

86

Karega-Mũnene

samples came from the trench.

Table 6.8 Frequency of the 1989 faunal collection by squares. Square

Depth (cm)

Identifiable

Unidentifiable

Total

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

50 50 70 60 60 70 60 60 60 70 70 70 70 90 80 90 90 90 100 100 90 90 70 80 200 210 230 160 150 150

33 33 7 147 169 97 104 79 233 131 122 49 242 221 72 68 308 270 217 319 172 142 220 271 4433 1074 1449 378 616 1274

231 560 452 439 1165 498 523 426 491 1017 772 799 799 1443 817 1640 3766 1616 2394 2784 1443 1470 1026 1400 4043 3672 6172 2357 1664 1416

264 593 459 586 1334 595 627 505 724 1148 894 848 1041 1664 889 1708 4074 1886 2611 3103 1615 1612 1246 1671 8476 4746 7621 2735 2280 2690

12950

47295

60245

Total

Examination of the size of the faunal samples from the 1989 excavations vis-à-vis the depth of deposits reveals there is a fairly significant relationship between the two (Tables 6.8). To take a few examples, the three squares (i.e., 25, 26 and 27) with the deepest deposits yielded the three largest largest overall collections and samples of identifiable and non-identifiable specimens. In contrast, the relationship between sample size and the volume of excavated deposits is remarkably different from that of the ceramics. Unlike in the latter case, where there are only three clusters, the faunal samples form four clusters, the largest of which is from Area A followed by the samples from Areas D, E, and C and B, respectively (Table 6.9). Further, unlike the ceramics, the density of the faunal remains was fairly even across the excavated areas, the aberration being the density of Area D which was about one and a half times that of the other areas. The frequency of both the non-identifiable and identifiable specimens in each square is summarised in Table 6.8 and Figure 6.11, whilst the respective frequencies for the trenches are summarised in Table 6.10 and Figure 6.12. Inspection of Table 6.8 and Figure 6.11 clearly indicates that the identifiable specimens from each square were fewer than the non-identifiable elements from the same square. The only exception was square 25, where identifiable specimens surpassed non-identifiable ones by a considerable margin. Equally significant is the fact that square 3 had, unlike the rest of the squares, a very small sample of identifiable specimens. As we shall learn presently, this patterning is in general agreement with the composition of faunal assemblages from other sites in East Africa.

Table 6.9 Excavated areas, sample size and density of the 1989 faunal collection. Area A B C D E

Excavated area 20 m² 2 m² 2 m² 3 m² 3 m²

Volume of deposits 14.6 m³ 1.8 m³ 1.5 m³ 6.4 m³ 4.6 m³

Sample

Density

25553 3227 2917 20843 7705

1750.21 1792.78 1944.67 3256.72 1675.00

In contrast, the number of identifiable specimens was much greater than that of non-identifiable specimens in the 1983 faunal collection (Table 6.10 and Figure 6.12). In addition, the samples from Trenches III and IV were much smaller than those from the other trenches. The reason for the small sample in the former trench is that a total of 36,282 specimens from there had been analysed prior to this study (Appendix II). Since none of the remains from Trench IV had hitherto been studied, the small sample must have been the result of the limited excavations which, as we

Table 6.10 Frequency of the 1983 faunal collection by trenches. Trench I II III IV V Total

Depth (cm) 110 230 190 140 130

Identifiable Nonidentifiable

Total

1657 5355 184 427 5988

114 433 2 5 496

1771 5788 186 432 6484

13611

1050

14661

87

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 6.10 Gogo Falls: spatial distribution of the 1989 faunal collection in Area A

88

Karega-Mũnene

89

Holocene Foragers, Fishers and Herders of Western Kenya

90

Karega-Mũnene Table 6.11 Frequency of the 1989 faunal collection by squares and type of deposit. Square

Identifiable

Non-identifiable

Total

33 33 7 147 169 97 104 79 233 131 122 49 242 221 72 68 308 270 217 319 172 142 220 271 3592 769 72 758 212 104 776 248 425 378 616 1274

231 560 452 439 1165 498 523 426 491 1017 772 799 799 1443 817 1640 3766 1616 2394 2784 1443 1470 1026 1400 1532 1928 583 2803 489 380 2755 1734 1683 2357 1664 1416

264 593 459 586 1334 595 627 505 724 1148 894 848 1041 1664 889 1708 4074 1886 2611 3103 1615 1612 1246 1671 5124 2697 655 3561 701 484 3531 1982 2108 2735 2280 2690

Total

12950

47295

60245

have noted in the previous chapter, were conducted in the trench.

domestic and wild fauna present in East Africa. The collection aided both the anatomical and taxonomic identifications of all the specimens.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Type of deposit Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Brown/black loam soil Clayey deposits Ashy deposits Brown/black loam soil Clayey deposits Ashy deposits Brown/black loam soil Clayey deposits Black loam soil Black loam soil Black loam soil

Further examination of Tables 6.8 and 6.9 reveals there was a fairly even chance that samples from each square and trench would yield information about the subsistence economies of the human groups who occupied the site. That is mainly because identifiable remains were recovered from every square and trench. Moreover, although the sample size of the identifiable specimens was quite varied, ranging from 7 in square 3 to over 5,900 in Trench V (Tables 6.8 and 6.10), the remains occurred in every type of deposits (Table 6.11, see also Appendices III and IV).

The identification process depended on the completeness and/or the existence of diagnostic features on the specimens. The process started with the identification of every specimen to a particular anatomical element. This was followed by the identification of the specimen to a specific taxon, which was usually the species. Where the specimens were too fragmentary to be identified to a given species, they were identified to a taxonomic level higher than the species, namely, the genus, sub-family, family, order or class. To take one example, if a given specimen could positively be attributed to sheep it was classed as a sheep (Ovis aries) specimen. But if it was attributable to sheep/goat and not specifically to either of the two, it was classed as a sheep/goat (Ovis/Capra or caprini) specimen. If the specimen could not be attributed to sheep/goat

Identification As indicated in the previous chapter, detailed study of the faunal collection was undertaken at the National Museum in Nairobi. The institution houses a comprehensive collection of comparative skeletal material of both the 91

Holocene Foragers, Fishers and Herders of Western Kenya

but was definitely from an animal of the same size and belonging to the same family it was classed as a bovid specimen.

only, namely, juvenile and mature. That is because the faunal remains may be too fragmentary to facilitate the recognition of the sub-adult category. The postcranial specimens under study fell within this ambit.

The bovid family was represented by a fairly wide range of animals in the assemblage, thereby necessitating their grouping into several categories on the basis of body size. In consequence, four classes, viz., bovid 1 for small animals of oribi size, bovid 2 for sheep/goatsized animals, bovid 3 for topi-sized animals and bovid 4 for cow-sized animals were established. The other family categories identified in the assemblage are suid and equid. Animals belonging to each of these families are generally of the same size; therefore, grouping them into smaller categories could not be justified. The specimens which could not be identified to either the species, genera, or family level were identified to the order or class level.

The dental eruption and wear technique yields more age categories than the epiphyseal fusion approach. The categories established by the former approach are also relatively more precise than those devised by the latter method. That is because dental wear, unlike the fusion of the epiphyses of postcranial bones, is more progressive and continues into old age. As such, it is possible to delineate several stages of dental attrition, all representing different age categories (Hillson 1986). The dental eruption and wear approach has been used to establish as many as nine or more age categories for which specific age brackets have been suggested (e.g., Levine 1982; Payne 1973). This procedure is most effective where the majority of the teeth are intact, which is not the case with the assemblage under study. Consequently, only six relative categories could be positively recognised. These compare favourably with those used by Marshall (1986): -

All the specimens that could not be identified to specific skeletal elements and taxa were treated as a potential source for information crucial to the evaluation of the taphonomic history of the assemblage. They were, therefore, studied in detail to obtain evidence of modification by humans and/or other animals. Such evidence was manifest in the surface condition of the specimens, cut-marks, gnaw-marks, burning and patterns of fragmentation. Examination of this evidence was done with the naked eye, but, when in doubt about given cut-marks or gnaw-marks, the specimens concerned were examined under a hand lens or microscope.

(a) unworn deciduous teeth – for neonates, (b) worn deciduous teeth – juveniles, (c) unworn permanent teeth – older juveniles, (d) slightly worn permanent teeth – young adults, (e) moderately worn permanent teeth – adults, (f) heavily worn permanent teeth – older adults

Age estimation Two approaches were used to establish relative ages of the specimens concerned, namely, the epiphyseal fusion technique for postcranial elements and the dental eruption and wear technique for cranial elements. The age categories devised by both methods will be useful in establishing the kill-off patterns of the animals concerned. They will also serve the additional purpose of elucidating the taphonomic history of the entire assemblage, as we shall see presently.

Quantification The abundance of the anatomical elements and animals represented in the faunal assemblage was estimated by two techniques: the number of identified specimens (NISP) and the minimum number of individuals (MNI). The NISP approach is a raw quantification method involving only the counting of all the specimens assigned to the animals represented within a given faunal sample (Grayson 1979). Both fragments and complete elements are treated in the same way, regardless of whether the fragments belong to the same bone or not. For this reason, if the size of the faunal assemblage concerned is reduced or increased, the new NISP value is obtained not through complicated statistical manipulation of the original NISP value, but by the addition or subtraction of the

Ideally, the epiphyseal fusion approach can be used to establish juvenile, sub-adult and adult age categories only. That is because, in general, bones are either unfused (for the juvenile category), fusing (for the sub-adults), or fused (for the adults) (Chaplin 1971). In most cases, however, the approach yields two categories 92

Karega-Mũnene

number of specimens added or taken away (Klein and Cruz-Uribe 1984).

Grayson 1977; Chaplin 1971; Grayson 1978, 1979, 1984; Klein and Cruz-Uribe 1984; Marshall and Pilgram 1993; Payne 1972). Therefore, they need not detain us here. Suffice it to observe that they are complementary. That is because one of them, the NISP approach, provides the maximum number of the anatomical elements and animals that may be represented in a given sample, whilst the MNI approach provides the minimum values for both the anatomical elements and animals that may be represented in the sample. It is quite likely that the actual number of anatomical elements or animals utilised at a given site lies within the spectrum provided by the two approaches. The results of this exercise as well as those of the identification and ageing of the animals represented in the assemblage under study will be discussed at length in the next chapter.

The MNI approach involves the separation of all the identified specimens assigned to different taxa into separate body parts and then into those from the right and left sides of the body for each taxon. These are then counted and the larger of the two values obtained for a specific anatomical element from either the right or left side is taken to be the MNI for that particular taxon. Thus, if a given animal is represented by different anatomical elements, MNI values are obtained for each of those elements and the largest figure is taken to be the MNI for that animal (Chaplin 1971; Klein and Cruz-Uribe 1984). The merits and demerits of each of these approaches have been discussed in detail by various authors (e.g., Casteel 1977; Casteel and

93

CHAPTER SEVEN

PATTERNING OF THE FAUNAL ASSEMBLAGE

It will be recalled from the previous chapter that the faunal assemblage under study consisted of all the remains that were excavated in 1989, plus a sizeable portion of the remains from the 1983 excavations. Each of those collections consisted of nonidentifiable and identifiable remains, the latter category comprising both maximally or minimally identifiable specimens. The maximally identifiable specimens were identifiable to the species, genera, or family level while the minimally identifiable were identifiable only to order or class. The specimens in the non-identifiable category could not be positively assigned to specific taxa.

collection was quite similar to that of the 1989 collection.

Taxonomic Representation A fairly large number of animal species were identified in the whole assemblage. The complete list of the species, including their common and scientific names, where applicable, follows in alphabetical order: 1. Aardvark (Orycteropus affer) 2. Barbus fish (Barbus sp.) 3. Bird (Aves sp.) 4. Bohor reedbuck (Redunca redunca) 5. Buffalo (Syncerus caffer) 6. Burchell’s zebra (Equus burchelli) 7. Bushbuck (Tragelaphus scriptus) 8. Bush duiker (Sylvicapra grimmia) 9. Bush pig (Potamochoerus porcus) 10. Cane rat (Thryonomys sp.) 11. Catfish (Clarias sp.) 12. Chanler’s reedbuck (Redunca fulvorufula) 13. Eland (Taurotragus oryx) 14. Cow (Bos taurus) 15. Crocodile (Crocodylus sp.) 16. Waterbuck (Kobus defassa/ellipsiprymnus) 17. Dog (Canis familiaris) 18. Donkey (Equus asinus) 19. Giraffe (Giraffa sp.) 20. Goat (Capra hircus) 21. Grant’s gazelle (Gazella granti) 22. Grevy’s zebra (Equus grevyi) 23. Hartebeest (Alcelaphus buselaphus) 24. Hippopotamus (Hippopotamus amphibius) 25. Human (Homo sapiens) 26. Impala (Aepyceros melampus) 27. Klipspringer (Oreotragus oreotragus) 28. Lesser kudu (Tragelaphus imberbis) 29. Monkey (Cercopithecus sp.) 30. Oribi (Ourebia ourebi) 31. Rhino (Diceros/Ceratotherium sp.) 32. Roan antelope (Hippotragus equinus) 33. Sheep (Ovis aries) 34. Snake 35. Thomson’s gazelle (Gazella thomsoni) 36. Topi (Damaliscus lunatus)

The frequency of the specimens according to their respective levels of identifiability is summarised in Figure 7.1. As the figure clearly shows, the majority of the remains excavated in 1989 were non-identifiable. In fact, of the 60,245 remains excavated that year only 21.5% were identifiable. This composition is quite similar to that of other East African Holocene faunal assemblages analysed by various researchers (e.g., Gifford et al. 1980; Gifford-Gonzalez and Kimengich 1984; Marean 1992; Marshall 1986). In contrast, the composition of the collection excavated in 1983 was quite unusual. Indeed, a very large portion of that collection – 13,611 elements, representing 92.8% of the collection – was identifiable. This condition may be explained by the fact that a very large portion of the 37,783 specimens from the original assemblage that had been studied previously was not identifiable. These latter numbered 33,397, representing 88.4% of the sample (Appendix II). Addition of the previously studied sample to the 1983 collection that was analysed for this study radically alters its composition. The number of non-identifiable specimens in the collection rises to 34,447 and that of identifiable specimens to 17,997, representing 65.7% and 34.3% of the original collection, respectively (Figure 7.2). In sum, therefore, the composition of the original

94

Karega-Mũnene

1983 collection

1,520 11,430

47,295 Maximally identifiable Minimally identifiable Non-identifiable

1989 collection

10,831

15,730 48,345

Maximally identifiable Minimally identifiable Non-identifiable

Figure 7.1 Composition of the faunal assemblage

95

Holocene Foragers, Fishers and Herders of Western Kenya

96

Karega-Mũnene

Spatial representation

37. Warthog (Phacochoerus aethiopicus) 37. Wildebeest (Connochaetes taurinus).

The spatial representation of the animal species identified in the 1989 and 1983 collections is summarised in Tables 7.2 and 7.3, respectively. As the tables demonstrate, the majority of the animals are not represented in all the excavated squares or trenches. For example, in the 1989 collection the two best represented species, cow and barbus fish, appear only in 20 and 10 squares, respectively. Except for human whose remains appear in 6 squares, each of the other species identified in the collection is represented in 4 or fewer squares. The other notable exceptions are the taxonomic categories that may represent more than one species, namely, alcelaphini, bovid, suid, equid, fish and mammal.

Virtually all the 38 animals listed above are represented in the 1983 sample. The only exceptions are aardvark, crocodile and barbus fish which are represented only in the 1989 sample (Table 7.1). The representation of the animals in the latter sample is, however, rather poor since only 18 of them are represented. This condition could be due to one or a combination of three factors: (a) differences in the chronological and/or spatial distribution of the faunal remains at the site; (b) representation of the missing animals by non-identifiable specimens; or (c) absence of the animals concerned from the sample. We shall be exploring these factors further below.

The squares in which the animals are represented are fairly well distributed in the five areas excavated at the site. Consequently, there appears to have been no definite pattern in the spatial distribution of the remains of taxonomic categories like domestic animals, fish and/or wild animals. The majority of the taxa, however, are best represented in squares 25 through 30. These squares, it will be recalled from the preceding chapters, had fairly deep sequence of deposits and also yielded relatively large faunal samples.

Table 7.1 Taxonomic representation in the faunal assemblage.* Taxon Aardvark Barbus fish Bird Bohor reedbuck Buffalo Burchell’s zebra Bushbuck Bush duiker Bush pig Cane rat Catfish Chanler’s reedbuck Cow Crocodile Dog Donkey Eland Giraffe Goat Grant’s gazelle Grevy’s gazelle Hartebeest Hippopotamus Human Impala Klipspringer Lesser kudu Monkey Oribi Rhino Roan antelope Sheep Snake Thomson’s gazelle Topi Warthog Waterbuck Wildebeest *x = present, - = absent.

1983 Collection

1989 Collection

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x

The most striking thing about the spatial distribution of the taxa identified in the 1983 collection is their poor representation in Trenches III and IV. This situation must have arisen from the comparatively small samples from those trenches that were available for the present study, an observation that has been discussed at length in the preceding chapter. The actual taxonomic representation in Trench III was certainly better than is the case here, when the data from the samples studied heretofore are taken into account. Indeed, domestic and wild animals and fish are represented reasonably well in these latter samples (Karega-Mũnene 1986, 1987; Marshall 1986, 1991; Stewart 1991). Unlike in Trench III, the poor taxonomic representation in Trench IV may be a consequence of the small sample arising from partial excavation of the trench. As in the 1989 collection, cow is the best represented species in the 1983 collection, its remains having been recovered from all the trenches. Oribi, roan antelope, Burchell’s zebra

97

Holocene Foragers, Fishers and Herders of Western Kenya

98

Karega-Mũnene

It must be emphasised here that the apparent absence of some or most of the identified species in some squares and/or trenches may not be real since the bulk of the faunal remains were either minimally identifiable or not identifiable. Therefore, elements belonging to some or most of species may have been attributed to genera, family, order or class. If correct, this may, in turn, explain why the equid, suid, caprini, alcelaphini, fish and mammal categories plus the four bovid categories are represented in virtually all the 30 squares and 5 trenches.

Table 7.3 Taxonomic representation in the trenches excavated in 1983.* Taxon Sheep Goat Caprini Cow Bushbuck Bush duiker Buffalo Eland Chanler’s reedbuck Waterbuck Grant’s gazelle Hartebeest Impala Klipspringer Lesser kudu Oribi Roan antelope Reedbuck Topi Thomson’s gazelle Wildebeest Bohor reedbuck Alcelaphini Bovid 1 Bovid 2 Bovid 3 Bovid 4 Bush pig Warthog Suid Burchell’s zebra Donkey Grevy’s zebra Equid Giraffe Rhino Hippopotamus Catfish Fish Cane rat Rodent Snake Bird Dog Monkey Human Mammal

Trench I

II

III

IV

V

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x

x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

Chronological representation The chronological representation of the animals identified in the faunal collections is of great significance. That is because it does not only yield information about the subsistence strategies practised at Gogo Falls through time, but also sheds light on the appearance of domestic animals. However, it must be borne in mind that the usefulness of such information depends on reliable chronometric dating. For this purpose, the following discussion will focus on the trenches and squares that have been dated, namely, Trenches II and III and squares 26, 27, 28 and 29. Faunal and chronometric evidence from the trenches suggests fish, domestic animals and wild animals were being exploited at the site by about 1,900 BP - 1,700 BP (Karega-Mũnene 1986, 1987; Marshall 1986, 1991; Marshall and Stewart 1994; Stewart 1991). This observation is supported by evidence from square 26, where a charcoal sample associated with caprini, wild animals and Elmenteitan, Kansyore and Urewe pottery at 100 cm -110 cm below the ground surface is dated to ca. 2,000 BP. Although fish is absent from the dated level, it is represented in the spits above and below the dated sample (Table 7.4).

*x = present, - = absent.

and cane rat are represented in 4 trenches and the rest of the species in 3 or fewer trenches, respectively (Table 7.3). Further, as is the case with the 1989 collection, the exceptions are taxonomic categories like caprini, alcelaphini, bovid, equid, fish and mammal, which may represent several species. It is also noteworthy that there is no definite pattern in the spatial representation of the animals identified in the sample. Indeed, domestic and wild animals as well as fish are represented in virtually all the trenches.

The older date of 3170 ± 70 BP (OxA-3494) on a charcoal sample from 160 cm - 170 cm below the ground surface in the same square is associated with Kansyore pottery, cow and taxa that could only be identified to the bovid category (Table 7.4 and Figure 7.3). Remains of wild animals are represented in the spits below and immediately above the dated level, but fish is absent. If correct, this suggests exploitation of wild and domestic animals may pre-date that of fish.

99

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 7.3 Gogo Falls: stratigraphic location of dated sample, cow elements and pottery in Square 26

100

Karega-Mũnene

The date of 2000 ± 70 BP (OxA-3495) from 120 cm - 130 cm below the ground surface in square 27 is associated with remains of sheep/goat, wild animals and fish as well as Elmenteitan, Kansyore and Urewe pottery. These animals as well as cow are also represented in the deposits above and below the dated spit. The date of 2030 ± 65 BP (OxA-3496) from 130 cm - 140 cm below the ground surface in square 28 is associated with sheep/goat and fish, plus Elmenteitan, Urewe and Kansyore pottery. Although there are no wild animals in the dated spit, they are represented in the underlying deposits.

assemblage. Information about the minimum number of individuals (MNIs) obtained from postcranial and cranial elements in the overall assemblage is summarised in Table 7.8. It will be noted that MNIs were not obtained for the animals represented in each square or trench separately. That was because very few elements could be positively aged and separated into either the right or left side of the animals concerned in order to be of use in obtaining MNIs for each trench or square. Inevitably, the MNIs obtained for individual squares and trenches proved to be of little value in terms of overall comparisons with the respective NISP counts.

The oldest date of 3480 ± 75 BP (OxA-3497), which was obtained on a charcoal sample from 130 cm - 140 cm below the surface in square 29, is not directly associated with sheep/goat. Rather, this latter is represented in the underlying deposits where it is associated with Kansyore pottery. Domestic animals also occur in the overlying deposits where they are associated with wild animals, fish and Kansyore, Elmenteitan, Akira and Urewe pottery (Table 7.4 and Figure 7.4). Since both the dated sample and associated remains of sheep/goat were recovered from a sealed horizon and the spits immediately above the dated one were also sealed, it is reasonable to posit that domestic animals were present at the site by about 3,400 BP.

NISP counts Examination of Tables 7.5 and 7.6 reveals that the NISPs for the majority of the species represented in the 1989 collection are significantly lower than those of the animals identified in the 1983 collection. The figures for the former collection range from 1 for bushbuck and buffalo in Square 29, to 21 for cow in Square 25. On the other hand, the figures for the 1983 collection range from 1 for klipspringer and Chanler’s reedbuck in Trench I to 259 for cow in Trench V. In comparison taxonomic categories like caprini, equid, alcelaphini, fish, suid, bovid and mammal have higher NISps in both collections. Close inspection of Table 7.5 reveals that virtually all of the species represented in the 1989 assemblage have NISP values of 5 or less in virtually all of the squares. In fact, only cow and barbus fish have higher NISPs as evidenced in Squares 25 - 30. In contrast, 15 of the species identified in the 1983 collection have NISPs of 6 or more per trench in one or more trenches. Cow, for instance, has 47, 132 and 259 NISPs in Trenches I, II and V, respectively, while bohor reedbuck has 6, 12 and 13 in the same trenches (Table 7.6).

Two of the dated squares (i.e., 27 and 28) also yielded isolated iron artefacts. These come from the loose ashy deposits at 30 cm - 40 cm below the surface in both squares. That the artifacts are extremely few and are not associated with any of the available radiocarbon dates suggests they cannot be meaningfully used as chronological markers (Table 7.4). This notwithstanding, it can be reasonably inferred from the evidence currently available from the site that the artefacts may be about 2,000 years old and that their use at the site was preceded by an economy that was based on herding, hunting and probably fishing.

It will, however, be observed that taxonomic abundance in Squares 1 - 24 and in Trenches III and IV is significantly lower than in the rest of the squares and trenches (Tables 7.5 and 7.6). This condition may be explained by one of two factors: (a) the small samples from those squares and trenches and (b) the large samples of specimens that could not be identified to species level. The latter suggestion seems to be more likely since the lower NISPs are mainly associated with animals

Taxonomic Abundance Information about the number of identified specimens (NISPs) of the animals represented in the 1989 and 1983 faunal collections is summarised in Tables 7.5 - 7.7. Whereas the first two tables present information of the taxa represented in all the excavated squares and trenches, the NISPs in Table 7.7 are derived from cranial and postcranial elements in the overall 101

Holocene Foragers, Fishers and Herders of Western Kenya

Figure 7.4 Gogo Falls: stratigraphic location of dated sample, caprini elements and pottery in Square 29

102

Karega-Mũnene

Table 7.4 Stratigraphic representation of taxa and artefacts in selected squares.* Akira

Elmenteitan

Kansyore

Urewe

Cow

Caprini

Wild

Fish

Date (BP)

Square 26, Spit:

x x -

x x x x x x x x x x x x x x x -

x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x -

-

x x x x x x x x x x -

x x x x x x -

x x x x x x x x x x x x x x x -

x x x x x x x -

2030±70 3170±70 -

Square 27, Spit :

x x -

x x x x x x x x x x x x x x x -

x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x -

x -

x x x x x x x x x

x x x x x x x -

x x x x x x x x x x x x x x x x x x

x x x x x x x x x

2000±70 -

-

x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x

x x x x x x x x x x x x -

x -

x x x x x x x -

x x x x x -

x x x x x x x x x x x

x x x x -

2030±65 -

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210 210-220 220-230

Square 28, Spit :

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160

103

Iron

Holocene Foragers, Fishers and Herders of Western Kenya

Table 7.4 continued Square 29, Spit :

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150

Akira

Elmenteitan

Kansyore

Urewe

x x x x -

x x x x x x x x x x x -

x x x x x x x x x x x x x x x

x x x x x x x x x -

Iron -

Cow

Caprini

Wild

Fish

Date (BP)

x x x x x -

x x x x

x x x x x x -

x x -

3480±75 -

* x = present, - = absent

that have been identified to species level. Conversely, the higher NISPs are associated with caprini, equid, alcelaphini, fish, suid, bovid and mammal (Table 7.7).

Although the MNI counts vary widely, from 1 to 45 as Table 7.8 shows, the values in the lower range (i.e., 5 or less) are associated with animals that have been identified to species level. The only exception is cow whose respective counts under postcranial and cranial elements are 11 and 44. The rest of the MNIs in the upper range belong to the alcelaphini, equid, caprini and bovid categories. The table also shows that in certain cases MNIs derived from postcranial elements exceeded cranial MNIs and vice versa. Indeed, of the 49 taxa listed in the table, the only MNIs that are equal under both categories of elements are those of eland, giraffe, hippopotamus, roan antelope and cane rat.

It will also be noted from Table 7.7 that animals like topi, waterbuck and crocodile are represented by cranial elements only, whilst goat, aardvark, bird, crocodile, catfish, waterbuck, klipspringer, Grevy’s zebra, lesser kudu, monkey, sheep, snake and topi are represented by postcranial elements only. Equally significant is the fact that the NISPs under cranial and postcranial elements vary remarkably. In fact, it is only Chanler’s reedbuck, donkey and giraffe that have equal numbers of cranial and postcranial elements. In general, however, the higher NISPs are associated with postcranial elements mainly because these latter are overwhelming in the assemblage.

The corresponding counts for the rest of the animals differ by between 1 and 40. The postcranial MNIs for bovid 2 and 3, for example, exceed the cranial MNIs by 35 and 40, respectively. The converse is true of the MNIs for cow and caprini: the cranial MNIs exceed the respective postcranial MNIs by 33 and 25. In addition, as the table clearly demonstrates, postcranial MNIs are relatively higher than the corresponding cranial MNIs, the converse being true in 9 instances only. This condition may be explained by the predominance of postcranial elements in the faunal assemblage.

MNI counts MNI counts were obtained using postcranial and cranial elements. Examination of Table 7.8, in which the counts are summarised, reveals that MNIs for such animals as fish and mammals are missing and that some of the animals have MNIs under either the postcranial or cranial elements only. The reason for this is that the specimens from which those animals are known could neither be aged nor assigned to the left or right side of the animals, both of which were prerequisite to the application of the approach. Teeth, vertebrae and bone fragments, all of which lack the necessary diagnostic features, for example, represent fish.

Comparison of the MNIs and the corresponding NISPs under cranial and postcranial elements reveals considerable differences between them for virtually all the taxa (Table 7.9). Indeed, only giraffe and lesser kudu have equal MNIs and NISPs under both categories of elements. The

104

Karega-Mũnene

105

Holocene Foragers, Fishers and Herders of Western Kenya

Table 7.6 Taxonomic quantification of the 1983 faunal collection by NISP. Taxon Sheep Goat Caprini Cow Bushbuck Bush duiker Buffalo Eland Impala Oribi Topi Wildebeest Waterbuck Chanler’s reedbuck

Grant’s gazelle Hartebeest Klipspringer Lesser kudu Roan antelope Thomson’s gazelle

Bohor reedbuck Alcelaphini Bovid 1 Bovid 2 Bovid 3 Bovid 4 Bush pig Warthog Suid Dog Donkey Burchell’s zebra Grevy’s zebra Equid Giraffe Rhino Hippopotamus Catfish Fish Cane rat Rodent Bird Snake Monkey Human Mammal Total

Table 7.7 Overall taxonomic quantification by NISP. Taxon

Trench I

II

III

IV

V

Total

0 0 3 47 0 3 8 0 0 1 0 7 0 1 0 9 1 1 1 0 6 24 83 244 291 322 0 3 21 4 0 1 0 45 0 1 5 0 31 1 20 8 3 0 24 438

1 2 103 132 3 5 18 4 4 7 5 16 2 2 1 27 1 0 2 4 12 99 533 1067 1055 740 4 13 35 1 0 11 10 149 1 6 7 0 97 2 6 1 1 1 3 1162

0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 6 2 53 7 19 0 0 0 0 0 0 0 3 0 0 0 0 1 0 0 0 0 0 0 90

0 0 20 9 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 3 34 66 33 52 0 0 0 0 0 1 0 8 0 0 0 0 3 0 0 0 1 0 0 194

5 5 129 259 1 5 3 0 1 5 1 8 0 1 2 3 1 0 1 12 13 48 525 910 1059 581 2 2 30 0 4 4 1 98 1 2 4 1 167 0 9 2 0 0 1 2082

6 7 256 448 6 13 29 4 5 14 6 31 2 4 3 39 3 1 5 16 31 180 1177 2340 2445 1714 6 18 86 5 4 17 11 303 2 9 16 1 299 3 35 11 5 1 28 3966

1657

5355

184

427

5988

13611

values for the rest of the taxa are equal under either the cranial or postcranial elements or significantly different. Under the postcranial elements, for example, bush pig, catfish, aardvark, and monkey have equal MNIs and NISPs, whilst Chanler’s reedbuck, donkey and crocodile have equal counts under cranial elements. Interestingly, neither the NISPs nor the MNIs of any of these taxa exceeded 2.

Sheep Goat Caprini Cow Bushbuck Bush duiker Buffalo Eland Impala Oribi Topi Wildebeest Waterbuck Chanler’s reedbuck Grant’s gazelle Hartebeest Klipspringer Lesser kudu Roan antelope Thomson’s gazelle Bohor reedbuck Alcelaphini Bovid 1 Bovid 2 Bovid 3 Bovid 4 Bush pig Warthog Suid Dog Donkey Burchell’s zebra Grevy’s zebra Equid Giraffe Rhino Hippopotamus Barbus fish Catfish Fish Crocodile Cane rat Rodent Bird Snake Aardvark Monkey Human Mammal Total

Postcranial

Cranial

Total

6 7 48 78 5 3 15 1 2 11 0 22 0 2 2 33 3 1 2 12 9 50 999 1450 1935 885 2 14 60 2 2 5 14 230 1 8 4 203 1 1 0 2 10 11 10 1 1 12 9026

0 0 252 486 8 10 22 6 5 6 7 11 2 2 1 6 0 0 3 4 22 161 199 1235 707 1198 4 12 37 3 2 12 0 184 1 3 17 8 0 605 1 4 30 0 0 0 0 26 3008

6 7 300 564 13 13 37 7 7 17 7 33 2 4 3 39 3 1 5 16 31 211 1198 2685 2642 2083 6 26 97 5 4 17 14 414 2 11 21 211 1 2656 1 6 40 11 10 1 1 38 12034

18251

8310

26561

MNIs and caprini 48 NISPs and 11 MNIs under postcranial elements. Under the cranial elements Bovid 2 has 1,235, cow 486 and caprini 252 NISPs respectively; the corresponding MNIs are 5, 44 and 36. Thus, it is clear that the taxa with high NISPs did not necessarily have high MNI counts. This observation may have significant implications for the taphonomic history of the faunal remains, an issue to which we shall be returning in the next chapter.

In contrast, Bovid 3 has 1,935 NISPs and 45

106

Karega-Mũnene Table 7.8 Overall taxonomic quantification by MNI. Taxon Sheep Goat Caprini Cow Bushbuck Bush duiker Buffalo Eland Impala Oribi Topi Wildebeest Waterbuck Chanler’s reedbuck Grant’s gazelle Hartebeest Klipspringer Lesser kudu Roan antelope Thomson’s gazelle Bohor reedbuck Alcelaphini Bovid 1 Bovid 2 Bovid 3 Bovid 4 Bush pig Warthog Suid Dog Donkey Burchell’s zebra Grevy’s zebra Equid Giraffe Rhino Hippopotamus Barbus fish Catfish Fish Crocodile Cane rat Rodent Snake Bird Aardvark Monkey Human Mammal Total

Postcranial

Cranial

3 2 11 11 1 1 5 1 1 3 0 4 0 0 1 5 1 1 1 3 2 9 31 40 45 28 2 4 3 0 1 1 2 9 1 2 1 0 1 0 0 2 3 0 1 1 1 2 0

0 0 36 44 3 4 4 1 3 2 1 3 1 2 0 1 0 0 1 1 5 18 3 5 5 8 1 1 0 2 2 3 0 7 1 0 1 0 0 0 1 2 0 0 0 0 0 1 0

246

173

Table 7.9 Comparison of overall NISP and MNI values. Taxon Sheep Goat Caprini Cow Bushbuck Bush duiker Buffalo Eland Impala Oribi Topi Wildebeest Waterbuck Chanler’s reedbuck Grant’s gazelle Hartebeest Klipspringer Lesser kudu Roan antelope Thomson’s gazelle Bohor reedbuck Alcelaphini Bovid 1 Bovid 2 Bovid 3 Bovid 4 Bush pig Warthog Suid Dog Donkey Burchell’s zebra Grevy’s zebra Equid Giraffe Rhino Hippopotamus Barbus fish Catfish Fish Crocodile Cane rat Rodent Bird Snake Aardvark Monkey Human Mammal Total

Postcranial

Cranial

NISP

MNI

NISP

MNI

6 7 48 78 5 3 15 1 2 11 0 22 0 2 2 33 3 1 2 12 9 50 999 1450 1935 885 2 14 60 2 2 5 14 230 1 8 4 203 1 1 0 2 10 11 10 1 1 12 9026

3 2 11 11 1 1 5 1 1 3 0 4 0 0 1 5 1 1 1 3 2 9 31 40 45 28 2 4 3 0 1 1 2 9 1 2 1 0 1 0 0 2 3 1 0 1 1 2 0

0 0 252 486 8 10 22 6 5 6 7 11 2 2 1 6 0 0 3 4 22 161 199 1235 707 1198 4 12 37 3 2 12 0 184 1 3 17 8 0 605 1 4 30 0 0 0 0 26 3008

0 0 36 44 3 4 4 1 3 2 1 3 1 2 0 1 0 0 1 1 5 18 3 5 5 8 1 1 0 2 2 3 0 7 1 0 1 0 0 0 1 2 0 0 0 0 0 1 0

18251

246

8310

173

to the cranial elements, yielded six categories. The criteria employed in identifying the age categories have been discussed at length in the previous chapter; therefore, they need not be repeated here.

Age Categories As indicated in the preceding chapter, the animals identified in the faunal assemblage were aged by either of two techniques: the epiphyseal fusion approach or the dental eruption and wear approach. For the postcranial elements, all of which were aged by the former approach, only three age categories – juvenile, sub-adult and adult – were established. In contrast, the dental eruption and wear approach, which was applied

Thus, for the animals that are represented by both cranial and postcranial specimens two sets of relative ages were obtained, while a single set was obtained for the taxa that are represented by only one of these. The results of this exercise are 107

Holocene Foragers, Fishers and Herders of Western Kenya

summarised in Tables 7.10 and 7.11, the former table presenting information about the categories derived from postcranial elements and the latter categories obtained from cranial elements.

Table 7.10 Age categories represented by postcranial elements. Taxon Sheep Goat Caprini Cow Bushbuck Bush duiker Buffalo Eland Impala Oribi Wildebeest Grant’s gazelle Hartebeest Lesser kudu Roan antelope Thomson’s gazelle Bohor reedbuck Alcelaphini Bovid 1 Bovid 2 Bovid 3 Bovid 4 Bush pig Warthog Suid Donkey Burchell’s zebra Grevy’s zebra Equid Rhino Cane rat Rodent Monkey Human

As both tables clearly demonstrate, not all the animals identified in the assemblage are represented in each of the age categories. Aardvark, bird, crocodile, snake, fish and mammal, for instance, are not listed in the tables because the specimens representing them could not be aged by either the epiphyseal fusion or the dental eruption and wear method. In addition, it will be observed from the tables that cane rat, Grant’s gazelle, Grevy’s zebra, lesser kudu, monkey and rhino are listed only in Table 7.10, whilst Chanler’s reedbuck, waterbuck, dog and topi appear only in Table 7.11. That is because only one of the ageing methods could be employed in ageing the specimens belonging to these animals. Examination of the representation of the age categories devised from the epiphyseal fusion approach shows that the best represented category is that of the adults. In fact, except for impala and rhino, all the animals aged by the approach are represented in that group by between 1 and 26 individuals. In contrast, each of the animals in the sub-adult and juvenile categories is represented by only 3 or fewer individuals, the exceptions being the bovid categories. In addition, out of the 34 taxa listed in Table 7.10, only about half of them are represented in the sub-adult group and a mere 8 in the juvenile group. One or a combination of factors that will be examined at length in the succeeding chapter may explain this condition.

Total

Juvenile

Sub-adult

Adult

Total

0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 3 8 6 3 0 0 0 0 0 0 2 1 0 0 0

0 0 3 2 0 0 1 0 1 0 0 0 1 0 0 1 0 2 10 8 13 11 1 2 1 0 0 1 2 1 1 2 0 1

3 2 7 9 1 1 4 1 0 3 4 1 4 1 1 2 2 6 18 24 26 14 1 2 2 1 1 1 5 0 1 1 1 1

3 2 11 11 1 1 5 1 1 3 4 1 5 1 1 3 2 9 31 40 45 28 2 4 3 1 1 2 9 2 2 3 1 2

25

65

151

241

The variation in the representation of the animals in and between the age categories may be indicative of the actual animal populations exploited at the site or of the effects that various taphonomic processes have had on the faunal assemblage. These postulations will be discussed at length below. First, let us examine the representation of the various anatomical elements in the faunal collections and the modification patterns observed therein.

A somewhat similar pattern is evident in Table 7.11 where the adult category is also the best represented among the animals aged by the dental eruption and wear approach. The majority of the animals, however, are represented by 3 or fewer individuals, the exceptions being caprini, cow and alcelaphini which are represented by 12, 22 and 26 individuals, respectively. Taxonomic represe-ntation in the other age categories is rather poor: in the neonate and very old adult categories, for example, cow is the only animal represented though only by 2 and 1 individuals, respectively. In the juvenile group only caprini, cow and bush duiker are represented, whilst in the sub-adult and old adult categories taxonomic representation rises to 15 and 11, respectively.

Skeletal Element Representation The overall skeletal element (i.e., body part) representation is summarised in Table 7.12. As the table shows, the largest portion of cranial elements consists of loose teeth followed by indeterminate cranial specimens, jaw bones and horn cores, in that order. Of the postcranial elements, the largest portion consists of fragments of vertebrae, long bones,

108

Karega-Mũnene Table 7.11 Age categories represented by teeth. Taxon

Neonate

Juvenile

Sub-adult

Adult

Old adult

Very old adult

Total

Caprini Cow Bushbuck Bush duiker Buffalo Eland Impala Oribi Topi Wildebeest Waterbuck Chanler’s reedbuck Hartebeest Roan antelope Thomson’s gazelle Bohor reedbuck Alcelaphini Bovid 1 Bovid 2 Bovid 3 Bovid 4 Bush pig Warthog Dog Donkey Burchell’s zebra Equid Human

0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

3 1 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

4 5 2 0 1 0 1 1 0 0 0 1 0 0 0 1 4 1 2 2 4 0 0 0 0 1 1 0

26 22 1 2 2 1 1 1 1 3 1 1 1 1 1 3 12 2 2 2 3 1 1 2 1 2 5 1

3 13 0 0 1 0 1 0 0 0 0 0 0 0 0 1 2 0 1 1 1 0 0 0 1 0 1 0

0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

36 44 3 4 4 1 3 2 1 3 1 2 1 1 1 5 18 3 5 5 8 1 1 2 2 3 7 1

Total

2

6

31

102

26

1

168

ribs and carpal and tarsal bones. Interestingly, the majority of the specimens in all these cases are fairly small fragments that could not be identified to species level. The only skeletal elements with considerable numbers of complete specimens are teeth and carpal and tarsal bones. Complete specimens of long bones are also unrepresented, the only exception being two cane rat femora. Also absent are complete specimens of ribs, vertebrae, crania and maxillae.

Table 7.12 Overall skeletal element representation. Skeletal element Horn Cranial bones Maxilla Mandible Teeth Vertebra Scapula Humerus Radius and ulna Rib Pelvis Femur Patella Tibia and fibula Metacarpal Carpal Metatarsal Tarsal Sesamoid Metapodial 1st phalanx 2nd phalanx 3rd phalanx Indeterminate long bone fragments Total

Number

%

36 428 92 211 7543 5367 120 139 282 1821 389 133 47 244 90 581 73 797 661 866 929 505 429 4778

0.1 1.6 0.3 0.8 28.4 20.2 0.4 0.5 1.1 6.9 1.5 0.5 0.2 0.9 0.3 2.2 0.3 3.0 2.5 3.3 3.5 1.9 1.6 18.0

26561

100.0

The fragmentation of the anatomical elements is in general accord with observations made on ethnoarchaeological faunal assemblages. These indicate that low density bones like sternum, ribs and vertebrae are prone to fragmentation to high degrees of unidentifiability by trampling, gnawing and other taphonomic processes (e.g., Brain 1981; Gifford 1978, 1980). Crania, maxillae and mandibles are also susceptible to fragmentation because of their low density, the susceptibility accelerating considerably when teeth fall off. This appears to have been the case at Gogo Falls where maxillae with intact teeth were unrepresented. It is also likely that some of the fragmentation may have arisen during

109

Holocene Foragers, Fishers and Herders of Western Kenya

Table 7.13 Incidence of modification on anatomical elements.* Skeletal element

Cut-marked

Burned

Rodent-gnawed

Carnivore-gnawed

Total

Horn Cranial bones Maxilla Mandible Loose teeth Femur Tibia /fibula Humerus Radial/ulna Metacarpal Metatarsal Metapodial Long bone fragment Patella Astragalus Pisiform Calcaneum Semi-lunar Scaphoid Cuneiform Magnum Navicular-cuboid Sesamoid Trapezoid Unciform 1st phalanx 2nd phalanx 3rd phalanx Scapula Vertebra Rib Pelvis Indeterminate fragments

0 2 0 12 0 7 10 15 13 4 4 34 164 1 24 0 4 3 4 9 4 3 3 0 6 25 8 1 9 36 57 6 10

4 134 8 31 112 23 47 28 66 19 10 163 564 11 55 8 35 21 30 33 30 22 112 1 25 148 79 35 29 309 157 43 1496

2 3 0 7 1 5 13 6 12 6 6 38 117 0 30 1 7 1 1 5 4 9 9 0 2 26 14 5 14 27 25 13 111

0 0 0 0 0 3 1 3 3 1 0 3 5 0 2 0 3 0 1 0 0 0 0 0 0 1 0 0 0 1 3 3 2

6 139 8 50 113 38 71 52 94 30 20 238 850 12 111 9 49 25 36 47 38 34 124 1 33 200 101 41 52 373 242 65 1619

Total

478

3888

520

35

4921

butchery, meat processing and/or extraction of marrow and brain. These postulations will be pursued further in the next chapter; first let us briefly examine the nature and form of the modification patterns in the faunal assemblage.

A total of 4,921 elements, which represented 6.6% of the whole assemblage, were modified. Of these, 3,302 are identifiable and the remaining 1,619 non-identifiable (Figure 7.5). The former category represents 12.4% of all the identifiable elements and the latter a mere 3.3% of all the non-identifiable elements. It is also evident from the illustrations that the majority of the modified identifiable elements are postcranial. These number 2,986, whilst the number of modified cranial elements is a mere 316 (Figure 7.6). These figures represent 16.4% and 3.8% of all the identifiable postcranial and cranial elements, respectively.

Modification Patterns Four types of modification – cut-marks, burning and rodent and carnivore gnaw-marks – were observed on a considerable number of specimens. Information about the occurrence of these on the specimens concerned is summarised in Figures 7.5 and 7.6 and in Table 7.13. Figure 7.5 compares the proportion of modified specimens to identifiable and non-identifiable portions of the faunal assemblage. Figure 7.6 and Table 7.13 present similar information, but for the affected cranial, postcranial and nonidentifiable elements and for the concerned anatomical elements, respectively.

The most common type of modification in the assemblage is burning, which affects 91.5% and 70% of the modified cranial and postcranial elements, respectively. Cut-marking and rodent gnawing also affect considerable numbers of postcranial and cranial specimens, as opposed to carnivore gnawing, which affects only 1.1% of

110

Karega-Mũnene

Modified Unmodified

Identifiable 3302 23259 Identifiable elements

3,302

23,259

Modified Unmodified

Modified Unmodified

Non-identifiable 1619 46726

Non-identifiable elements

1,619

46,726 Modified Unmodified

Figure 7.5 Incidence of modified specimens in the faunal assemblage.

111

Holocene Foragers, Fishers and Herders of Western Kenya

Cranial specimens

Cranial Cut-marking Burning Rodent-gnawing

14 289 13

13

14

Cut-marking Burning Rodent-gnawing

289

Postcranial specimens

Postcranial 396 Cut-marking 454 Burning 2103 Rodent-gnawing 396 Carnivore-gnawin 33

33

454 Cut-marking Burning Rodent-gnawing Carnivore-gnawing

2,103

Non-identifiable specimens

1,496

Non-identifiable Cut-markin 10 1400 Burning 1496 1200 Rodent-gna 111 1000 Carnivore-g 2 1600

800 600 400

111

10

2

200 0

Cut-marking

Burning

Rodent-gnawing

Carnivoregnawing

Figure 7.6 Incidence of specimen modification in the faunal assemblage.

112

Karega-Mũnene

Cranial 49 Slightly wea 5 Moderately 2 Heavily wea 5

Cranial elements 2

Slightly weathered Moderately weathered Heavily weathered

49

Postcranial elements

Postcranial 1172 Slightly wea 191 Moderately 191 47 Heavily wea

47

Slightly weathered Moderately weathered Heavily weathered

Non-identifiable 23 Slightly wea 192 Moderately 7 Heavily wea

1,172

Non-identifiable elements

7

23

Slightly weathered Moderately weathered Heavily weathered

192

Figure 7.7 Incidence of weathered specimens in the faunal assemblage.

113

Holocene Foragers, Fishers and Herders of Western Kenya

postcranial elements (Figure 7.6). The patterning of modification on nonidentifable elements is somewhat similar: burning affects 92.4%, rodent gnawing 6.9%, cut-marking 0.6% and carnivore gnawing 0.1%.

contain neither meat nor marrow. Interestingly, however, a considerable number of loose teeth are affected by burning, which probably occurred during the preparation of mandibular and/or maxillary elements for consumption (Table 7.13). The burnt cranial bones in the assemblage may also have been affected during preparation of meat and/or consumption.

Interestingly, however, rodents gnawed all the cranial specimens affected by gnawing. Rodentgnawing is also quite significant among the postcranial elements, where it affects 396 specimens out of the 429 gnawed specimens, the remaining 33 having been gnawed by carnivore. The number of specimens affected by rodent gnawing among the non-identifiable specimens is overwhelming when compared to that of carnivore gnawed specimens: these number 111 and 2 specimens, respectively.

Close examination of Table 7.13 also shows that long bones make up the largest portion of modified postcranial elements, representing about 28% of all the modified specimens. Considerable numbers of scapulae, vertebrae, ribs and pelves were also affected by burning; together they number 732 (14.9%). A total of 507 (10.3%) carpal and tarsal bones, plus 342 (6.9%) phalanx bones were also burned. The relatively high frequency of burned long bones and axial elements, as reflected by these figures, supports the hypothesis that their high food value may have rendered them susceptible to burning and cut marking during processing.

The number of specimens affected by the modification pattern outlined above bears little relationship to the respective proportions of postcranial, cranial and non-identifiable elements in the entire assemblage. As we have noted above, the largest portion of the assemblage is non-identifiable, whilst the largest portion of identifiable elements consists of postcranials.

However, it must be remarked here that it is likely that the proportion of modified bones in the whole assemblage may have been higher than the above figures suggest. Specimen weathering could have obscured cutmarks, a suggestion that is supported by the relatively high frequency of weathered specimens in the assemblage (Figure 7.7). Weathering itself is classified in three relative groups: slight, moderate and heavy. Although it is likely that weathering may not have obscured modification on a significant part of the overall assemblage, its effect on the postcranial elements may have been significant given the largest number of weathered specimens are postcranial.

The relatively high frequency of postcranial specimens among the modified elements may be explained in one of two ways. It could be the direct consequence of their predominance in the faunal assemblage. It could also be explained by the fact that postcranial elements contain more food value than cranial elements; therefore, they are more likely to be modified during butchery, meat preparation and consumption. This postulation is supported by the fact that the majority of cranial elements in the overall assemblage are loose teeth, elements that are unlikely to be cut-marked or gnawed since they

114

CHAPTER EIGHT

SUBSISTENCE STRATEGIES

The subsistence strategies of the inhabitants of the Gogo Falls site are reconstructed below using the faunal data explicated in the previous chapters. This is done without specific reference to the activities of either the Neolithic or Iron Age periods mainly because, as we have noted above, it is not possible to separate the Neolithic occupation from that of the Iron Age period. Both periods are known largely from material culture, yet pottery wares, which are the key markers of the periods, were found together in the same deposits.

retrieval employed during excavation were quite effective, hence recovery of considerable numbers of small-sized fragments. The implications of these suggestions on the interpretation of the faunal patterning discussed above are considerable. Therefore, each of them will be evaluated in considerable detail here. To begin with, it will be recalled from chapter five that all excavated earth was sieved. The reason for this was to facilitate the recovery of any remains of small-sized animals as well as small specimens of larger animals that may not have been recovered in situ during excavation. It can, therefore, be safely argued that, other things being equal, the sieving procedure enhanced the recovery of representative samples from the site. This observation is supported by the fact that the majority of the specimens in the assemblage are indeterminate small bone and teeth fragments, some of which had a maximum length of less than 0.5 cm.

Taphonomic History The taphonomic history of any given faunal assemblage is crucial to the interpretation of the patterning observed therein. This not only enables us to establish the agents responsible for accumulating the faunal remains at a given site, but also the effects that various taphonomic processes may have had on the remains prior to their excavation. This procedure is necessary because it is generally agreed among researchers that archaeological faunal remains are the result of several accumulative and destructive processes from the time a given animal is killed until its remains are excavated and studied (e.g., Binford 1984; Gifford et al. 1980; KaregaMũnene 1991; Klein and Cruz-Uribe 1984; Marean 1992; Marshall 1986). As the present author has argued elsewhere, the “importance of this procedure is underlined by the dictum that differential representation, preservation and eventual recovery of animal bones from archaeological sites tends to be the rule rather than the exception” (Karega-Mũnene 1991: 13).

The actual preservation of the remains prior to their excavation seems to have been good because they were not affected by such factors as leaching or uncontrolled burning. The evidence for uncontrolled burning on the faunal remains would have been a higher frequency of burning than is evident in the assemblage. The specimens affected in this way would also have been confined to close to the ground surface at the time of excavation. In addition, the burning would have covered most of the surface of the affected specimens, which is not the case. Marean (1992) has also noted that uncontrolled burning causes warping and/or cracking of the affected specimens, but neither of these forms of evidence was observed in the assemblage.

It will be recalled from the previous chapter that the majority of the faunal remains under study were non-identifiable. This condition suggests one of three things. First, that prior to their excavation, the remains were poorly preserved. Second, that the remains may have been subjected to destruction during butchery, food preparation, consumption, deposition, excavation and/or subsequent transportation from the site to the National Museum, Nairobi, where they were analysed. Third, that the methods of specimen

A small proportion of the fragmentary specimens had fresh breaks; these numbered 910, which represent a mere 1.2% of the whole assemblage. The breakage of those specimens may have occurred during excavation and/or transportation to the National Museum. The fact that only a tiny portion of the whole assemblage was affected in this way, strongly suggests that destruction of the remains during and after excavation was

115

Holocene Foragers, Fishers and Herders of Western Kenya

minimal. The fragmentation of the majority of the remains must, therefore, have occurred prior to excavation, as evidenced by ‘old’ breakage marks.

represents a mere 0.05% of the assemblage. This condition suggests carnivores were not a major agent in the accumulation and/or destruction of faunal remains at the site. That was probably because domestic dog, the only carnivore identified in the assemblage, may not have been popular at the site given it is represented by only a few specimens. Alternatively, the animal could have been popular but its remains were disposed far away from the site because it was not a food resource.

The long bones and crania, for instance, could have been broken to enable extraction of marrow or brain, a practice that has been observed in modern pastoralist and hunting societies (e.g., Binford 1978, 1981; Brain 1981; Mbae 1986). This likelihood is supported by the fact that except for two cane rat femora, complete long bones are not represented in the assemblage. However, the same suggestion cannot be offered to explain the fragmentation of such elements as teeth, ribs, ulnae, vertebrae and pelves. Since none of these elements contains marrow, it is highly likely that their breakage was caused by other factors.

This is not to say that wild carnivores like jackal, hyaena and leopard did not exist in the vicinity of the site. On the contrary, the environmental and climatic conditions that prevailed in the Lake Victoria basin at the time were not only conducive to the existence of wild ungulates, but also of wild carnivores. The contribution of the carnivores to the accumulation and/or destruction of faunal remains at the site may have been minimised by plentiful food supplies in the area around the site, a condition that is attested by taxonomic diversity in the assemblage. Alternatively, carnivores could have been prevented from foraging at the site by the continual presence of humans given the nature of its occupation.

As the present author has argued elsewhere, bones need not be broken only to obtain marrow or brain. In societies that possess cooking vessels, for instance, bones are broken mainly to facilitate cooking in the vessels (Karega-Mũnene 1991). There is no reason to suggest this may not have been the case at Gogo Falls since, as we have already noted, the site’s occupants possessed pots. In fact, as the patterns of specimen modification discussed in the preceding chapter clearly show, the incidence of burning in the assemblage was quite low. This condition probably suggests that most of the meat may have been prepared for consumption by cooking it in the pots rather than by roasting.

Marean (1992) has suggested that carnivore gnawing may be minimised by removal of the grease sought by the animals through boiling. Although the extent of the boiling which is required to remove the grease is not specified, observations made by the present author in instances where meat is cooked by boiling in East Africa, indicates that boiling bones just once does not remove the grease. Thus, in order to significantly reduce the incidence of gnawing by dogs and other carnivores by boiling, bones must be boiled several times over, a practice that is quite common in societies which boil bones for soup. If such a practice obtained at Gogo Falls, it could also explain the low incidence of carnivore gnawing.

It has also been observed that less dense elements like ribs, ulnae, vertebrae, pelves and sternum may be fragmented by trampling and other taphonomic processes (Gifford 1978, 1980). As indicated in the previous chapter, the first four body parts are represented by considerable numbers of specimens, all of which are fragments. But sternum is not represented at all, probably because it had been destroyed by humans and/or gnawing animals given that it is softer than the rest of the skeletal elements.

Sediment pressure and/or leaching could also have contributed to the fragmentation of the faunal remains. In this regard, it will be recalled from chapter five that the deposits at the site were of three types: ashy, clayey and loamy deposits. The deposits were generally dry during excavation, a condition that is not conducive to bone leaching; discoloured specimens, which are generally considered to be evidence of leaching, were also absent. Moreover, the ashy deposits,

As indicated in the previous chapter, rodent gnawing was minimal; it affected 520 specimens, which represents 0.69% of the whole assemblage. The rodent gnaw-marks observed on the majority of the specimens are of recent origin, thus suggesting gnawing occurred after the specimens were buried in the ground. Carnivore gnawing affected a much smaller portion of the assemblage, 35 elements, which 116

Karega-Mũnene

which were generally confined to the upper levels, would not have exerted enough sediment pressure on the faunal remains to fragment them. Rather, such pressure may have come from the small stones that were encountered in most of those deposits, a condition that seems quite likely given that the majority of the faunal remains from the squares with considerable ashy deposits were fragmentary.

handful of the faunal remains and the weathered specimens lack cracks which are common in situations where weathering is caused by rain and intense sunshine (Behrensmeyer 1978). Rather, the weathering seemed to have been caused mainly by root action, probably after the remains had been covered by sediment. Be that as it may, we have no means of knowing what the actual rate of sedimentation at the site was. As such, we should not attach much meaning to either of the propositions. Both of them are highly tentative because the processes of weathering, erosion, transportation and deposition of material are neither constant nor continuous through time and space. Strictly speaking, therefore, the 1,140 radiocarbon years within which the 60 cm - 70 cm profile of the deposits in square 26 formed tells us nothing about the actual rate of sedimentation.

Exposure of the faunal remains to vagaries of the weather and other destructive agents over long periods could also have been fragmented them. This suggestion seems likely given that the rate of sedimentation at the site appears to have been quite slow. In square 26, for example, which yielded two radiocarbon dates, it took 1,140 radiocarbon years for 60 cm - 70 cm of deposit to accumulate. Assuming that the processes of weathering, erosion, transportation and deposition of material were constant through time, we can tentatively suggest that the rate of deposition may have averaged less than 0.1 cm of sediment per radiocarbon year.

In view of the evidence discussed above, it appears that the fragmentation of the faunal remains occurred mainly during butchery, food processing and consumption and that the remains were primarily accumulated by humans. The extremely low incidence of carnivore gnawing is adequate evidence that carnivores were not a major accumulative and/or destructive agent. Similarly, the patterns of rodent gnawing indicate that rodents could not have been responsible for either the accumulation or fragmentation of a significant part of the assemblage.

This is in contrast to Robertshaw’s (1991) inference that the rate of sedimentation especially in Trench I was rapid. Although the rate of sedimentation has not been computed, the dating evidence from the trench does not support this inference, largely because the dates in question lack stratigraphic consistency. One of the dated levels in the trench, M3, for instance, has an obsidian hydration date of 2617 ± 194 BP (Specimen No. 292-8) and a charcoal date of 3020 ± 100 BP (HAR-6262). Level M6 which was located 30 cm below M3 has two obsidian dates of 3613 ± 115 BP (Specimen No. 334-106) and 2096 ± 93 BP (Specimen No. 334-110), whilst the underlying level M7 has an obsidian date of 2096 ± 93 (Specimen No. 292-9).

Subsistence Strategies The subsistence strategies which were adopted by the occupants of the Gogo Falls site ranged from hunting to fishing, animal husbandry, fowling and exploitation of plants. The evidence for the first four activities is derived from the faunal data explicated above, while that of exploitation of plants is drawn from the plant remains that were recovered from the site in 1983 and are now published. This combination of subsistence activities has been observed only at Ngenyn (Hivernel 1978) and, to a certain extent, at GaJi 2 and GaJi 4 sites on the eastern shores of Lake Turkana (Barthelme 1985; Marshall et al. 1984). Detailed comparison between these sites is, however, rendered difficult by the fact that the biological data from Ngenyn and the Lake Turkana sites are not as detailed as the Gogo Falls data. Although the discussions which follow will specifically be on

Nonetheless, if the supposition that the rate of sedimentation was slow is correct, the remains could have remained on the ground surface for some considerable time before they were covered by sediment. Naturally, this could have exposed them to gnawing, trampling, root action, rain and sunshine over fairly long periods. Given that Gogo Falls is an open-air site, these factors could have contributed significantly to the weathering and fragmentation of the specimens. However, the gnawing and weathering patterns discussed above do not support the inference about a slow rate of sedimentation. Indeed, as we have already noted, gnawing and weathering affected only a 117

Holocene Foragers, Fishers and Herders of Western Kenya

the subsistence strategies that were adopted at Gogo Falls, reference will be made to evidence from these and other East African sites dating to the same general period whenever necessary.

would have grown in the wild or been cultivated along the banks of River Kuja, an area that provides suitable conditions for their growth, as evidenced by the flourishing cassava crop in the area (Pers. observ.). The absence of tuber crop remains may be explained by the fact that their nature may have rendered their preservation unlikely (Robertshaw and Wetterstrom 1989; Wetterstrom 1991).

Gathering and cultivation It will be recalled from chapter five that some floral remains were recovered during both the 1983 and 1989 excavations. The remains from the former excavations have been published (Lange 1991; Robertshaw and Wetterstrom 1989; Wetterstrom 1991); but the identification of the remains that were recovered during the recent excavations is yet to be communicated to the present author. The brief discussion that follows is, therefore, based solely on the published findings of other investigators.

It has also been suggested that some of the plants, notably the woody Rubiaceae, may have been used for fuel and that plants belonging to the Solanaceae family may have been used for medicinal purposes (Robertshaw and Wetterstrom 1989; Wetterstrom 1991). Other uses for the various kinds of plants represented at the site could have been construction of shelters and fences. Although direct evidence for either of these uses is as yet unavailable, there is no reason to assume that the occupants of the site did not build shelters or fence in their animals and/or homesteads. This may have been the case especially because the cultural and biological remains recovered from there suggest permanent rather than seasonal occupation.

Work on the floral remains in question reveals no direct evidence of cereal cultivation since remains of domestic plants like “spikelet fragments or other waste generated in cereal processing... [or] any unequivocal weeds of cultivation” were not represented (Robertshaw and Wetterstrom 1989: 27). Indeed, all the remains belong to wild plants. Recently, however, one carbonised seed from the site has been identified to Eleusine coracana africana, the wild progenitor of finger millet. This plant is known in its wild form from most of eastern and southern Africa, where it flourishes at altitudes of 1,000 m - 1,200 m and breeds freely with finger millet. According to Lange (1991: 192), both the wild and domestic forms of Eleusine coracana are “indistinguishable for a long time during their life cycle”.

Other plants like Zaleya pentandra and Abutilon sp. may have served neither of the purposes discussed above. Rather, they may have been weeds resulting from farming and/or herding. This inference is based on the fact that although both Zaleya and Abutilon grow in open spaces in a variety of habitats, they are especially found in disturbed areas, as is the case with Chenopodiaceae, Amaranthaceae and Gramineae at Ngenyn (Hivernel 1978). Due to the unavailability of direct evidence for cultivation at the site and also to the fact that none of these plants is specifically a weed of cultivation, it has been suggested that they grew as weeds in the enclosures of domestic animals (Robertshaw and Wetterstrom 1989; Wetterstrom 1991). While this suggestion is plausible, it must be pointed out here that the plants in question are not specifically weeds of herding either. Therefore, the supposition that they resulted from herding is by no means conclusive because it is premised on negative evidence.

In spite of the unavailability of direct evidence for cultivation at the site, it is evident that plant foods were exploited there. Indeed, it is conceivable that the site’s inhabitants may have supplemented their meat diet by a variety of plants. These may have included cereals like Eleusine coracana africana and fruits from plants like Abutilon, Ziziphus and Cordia, all of which are represented at the site (Lange 1991; Robertshaw and Wetterstrom 1989; Wetterstrom 1991). Remains of Ziziphus have also been recovered from Ngenyn, where ethnographic evidence from the area around the site indicates the fruit is consumed by modern Tugen and Ilchamus (Hivernel 1978).

It must be emphasised here that the absence of direct evidence of cultivation at Gogo Falls provides no real proof that domestic plants were unknown at the site. This is mainly because the negative evidence itself rests on samples that have been recovered from only a limited part of

Although tubers and vegetables are not represented in the floral remains, it is likely that they could also have been exploited. Such plants 118

Karega-Mũnene

the site. As such, there is no reason to firmly assume that remains of domestic plants may not be recovered from parts of the site that have not been investigated in the future. The direction that such investigations could take may be made clearer by the identification of the remains that were recovered from the site recently.

bird and fish by the MNI approach. In order to minimise the biases that either of the approaches may have introduced in the taxonomic groupings, we shall employ the percentages derived from both approaches in the assessment of the importance of the subsistence strategies which follows. To begin with, there is need to emphasise here that the inference about fowling is based on equivocal data. These are derived from 11 bird specimens only, none of which could be identified to species level. Although none of those specimens bore any evidence of gnawing, cut-marking or burning, this in itself was not sufficient evidence to refute or to support the fowling supposition. More field research is, therefore, necessary if this issue is to be resolved.

Fowling On the basis of NISP counts, the bird specimens accounted for a mere 0.2% of the specimens which were identified to species and genera levels. In contrast, terrestrial wild ungulates accounted for 18.5%, domestic animals for 15.1% and fish and other aquatic animals for 66.2% (Figure 8.1). Interestingly, these figures bear little relationship to the percentages calculated from MNI counts of the taxonomic groups concerned. For example, fish which was the most abundant resource under NISP counts, accounted for the same percentage (0.5%) of the specimens as bird under the MNI approach. And wild ungulates which accounted for 18.5% of the resources under NISP counts, accounted for 50.5% of the specimens under MNI counts.

Fishing

This is in contrast to the figures obtained from the 1983 samples that had been studied hitherto. In the dental sample, for instance, wild animals accounted for about 45.4% and 57.8% by NISP and MNI, respectively (Marshall 1986, 1991). But in the other sample they accounted for 85.9% and 83.3% by NISP and MNI, respectively (Karega-Mũnene 1986, 1987). Bird was not represented in the latter sample.

The fairly large portion of fish bones and teeth in the assemblage (over 12,000 specimens) suggests fish was the most important aquatic food resource at Gogo Falls. In contrast, remains of other aquatic animals like hippopotamus and crocodile are extremely few: the latter is represented by an isolated tooth and the former by 21 specimens. As Figure 8.1 clearly shows, aquatic resources account for about 65% of the fauna represented in the assemblage. Unlike the NISP counts, the MNI counts for aquatic resources, especially fish, are extremely low largely because the majority of the specimens could not be identified to species level, separated into left and right sides of the body or aged.

These differences may have been caused by one of four factors. First, the absence of specimens belonging to some of the taxa, notably bird. Second, the tendency of the MNI approach to inflate the importance of animals represented by fewer specimens, like bird. Third, the tendency of the NISP approach to exaggerate the abundance of animals represented by fragmentary specimens like fish. Fourth, the inability to quantify the relative abundance of

As indicated in the previous chapter, only two fish species, barbus (Barbus sp.) and catfish (Clarias sp.), were identified. These species are also represented in the samples from the 1983 excavations that had been studied hitherto (Karega-Mũnene 1986, 1987; Stewart 1991). In addition to these, other fish represented in one of the 1983 samples are Mormyrid, Bagrus, Synodontis, Cichlid and Labeo (Stewart 1991). It is quite likely that some or all of these fish are

119

Holocene Foragers, Fishers and Herders of Western Kenya

NISP Counts

NISP 15.10% Domestic 70.00% a Wild ungula 18.50% 60.00% Bird 0.20% 50.00% Fish 65.80% 40.00% 0.40% Other aqua 15.10% 30.00%

65.80%

18.50% 0.40%

0.20%

20.00% 10.00% 0.00%

Domestic Wild ungulates MNI Domestic a animals 47.40% Wild ungula 50.50% Bird 0.50% Fish 0.50% 1.10% Other aqua

Bird

Fish

Other aquatic animals

MNI Counts

60.00%

47.40%

50.50%

50.00% 40.00% 30.00% 0.50%

20.00%

0.50%

1.10%

10.00% 0.00%

Domestic animals

Wild ungulates

Bird

Fish

Other aquatic animals

Figure 8.1 Relative importance of taxonomic groups represented in the faunal assemblage.

120

Karega-Mũnene

also represented in the faunal collection excavated in 1989, but could not be identified from the available specimens.

in chapter four. It is also likely that some fish could have been caught by spearing. If the latter supposition is correct, the spearing could have been done using iron arrowheads and/or bone points both of which were represented at the site (Karega-Mũnene 1986; Robertshaw 1991). The bone harpoons associated with fishing communities in Lake Turkana basin (Bartheleme 1985) are absent.

It will be remembered from the previous chapter, however, that remains of barbus fish exceed those of catfish by a relatively large margin. The apparent abundance of barbus is also evident in the previously studied fish remains, a condition that may be indicative of exploitation of runs of the fish as they spawned in River Kuja (Stewart 1991). That none of the fish remains in the faunal assemblage was cut-marked suggests one of four things. Firstly, that fish were prepared for consumption by roasting, hence the high frequency of burned specimens in the assemblage. Secondly, it could have been cooked by boiling it in pots. Thirdly, fish may not have been cut up before cooking; instead, it could have been cooked whole. Fourthly, fish could have been cut up before cooking, but the burning that affected a considerably large number of the elements obscured the cut-marks.

It must be added here that evidence for fishing in Holocene East Africa is available only from a few sites. Notable among these are the shell midden sites in the Lake Victoria basin; Ileret, GaJi 2, Dongodien, Apeget I and Lopoy in the Lake Turkana basin; and Ngenyn in north-central Kenya. At the shell middens sites, for instance, perch-like fish (Percoidea sp.), lungfish (Protopterus aethiopicus) and catfish were exploited together with shellfish and terrestrial wild ungulates. The remains of these animals are associated with Kansyore and modern Luo pottery (Robertshaw et al. 1983).

Examination of the stratigraphic distribution of the fish remains indicates that the only pottery with which they are exclusively associated is Kansyore. This association occurred in the deepest deposits of squares 27 and 29. Although both squares have been dated, none of the dated samples was obtained from the levels where the association occurred. The association in square 27, for instance, occurred below the level dated to about 2,000 BP. The fish remains from the dated level in square 27 are associated with Kansyore, Elmenteitan, Akira and Urewe pottery. In square 29 the association between fish remains and Kansyore ware occurrs immediately above the level dated to about 3,400 BP. Thus, in light of the evidence currently available we can surmise that fish was an important food resource at the site by 3,400 BP. This observation is supported by chronometric and faunal data from the 1983 excavations where fish is associated with Kansyore deposits underlying a level dated to about 1,900 BP (Robertshaw 1991).

At Ngenyn fishing was practised together with herding, hunting and probably gathering. The fish that were exploited there include catfish and barbus. These are associated with Turkwel pottery in deposits dated to about 2,000 BP (Hivernel 1978). At GaJi 2 and Dongodien fish remains are associated with remains of other aquatic animals, terrestrial wild ungulates and domestic animals and Nderit pottery in deposits dated to about 4,600 BP - 3,400 BP (Barthelme 1985; Marshall et al. 1984).

Hunting A wide range of wild animals appears to have been hunted by the occupants of the Gogo Falls site. This is attested by the fact that the majority of the species identified in the faunal remains are wild animals. Out of the 38 species represented in the remains, 28 are terrestrial wild animals; the remaining taxa consist of 4 aquatic animals and 5 domestic animals. The majority of the species identified in the faunal assemblage under study are also represented in the 1983 samples that had been studied previously (KaregaMũnene 1986, 1987; Marshall 1986, 1991).

Although evidence of the fishing methods used by the occupants of the site is lacking, we can tentatively suggest that these may not have been very different from the methods used today by the Luo when fishing for domestic consumption. If correct, this suggestion would imply that the methods would have included the artificial barrier method which is known as kek and fishing hooks both of which have been discussed

Of the wild animals in question, giraffe and rhino account for 1.3%, equids 43.4% and suids 12.4%. Animals belonging to the bovid 4 group account for 31.6%, bovid 3 category 1.9%, bovid 2 category 6.2% and bovid 1 group 3.2% on the basis of NISP counts. The general pattern 121

Holocene Foragers, Fishers and Herders of Western Kenya

implied by these figures is also evident from the MNI counts, thus confirming that the largest portion of the wild animal remains was made up of medium- to large-sized ungulates. This situation is by no means unique to Gogo Falls. Indeed, somewhat similar observations have been made at Maasai Gorge rockshelter, Prolonged Drift and Sambo Ngige and Ngamuriak, where the wild fauna represented were also dominated by medium- and large-sized ungulates like zebra and Grant’s gazelle (Gifford et al. 1980; Gifford-Gonzalez 1985; Marshall 1986, 1990). Thus, it seems that mid-late Holocene East African hunters may have deliberately hunted larger ungulates probably because (a) the animals yielded more meat; (b) the animals were present in large numbers in the region; or the human groups concerned were specialised hunters.

The evidence presented in the previous chapter bears out some of these suggestions. For example, a significant part of the assemblage could only be identified to either of the four bovid classes (32.4%) or to the general fish (13.8%) or mammal (45.4%) categories. Secondly, some of the animals identified in the faunal assemblage, notably buffalo and rhino, which are represented by 37 and 11 elements, respectively, may have been avoided by the hunters because of their fierce nature. If correct, meat from these animals may have been obtained through scavenging rather than hunting. It is also likely that some of the animals, namely, aardvark, monkey, lesser kudu, crocodile and catfish may not have been exploited for food, especially because they are represented by isolated elements. This notwithstanding, it is quite likely that elements belonging to some of these animals may, as we have noted in the previous chapter, have been identified to taxonomic levels other than the species or genera. Elements belonging to lesser kudu, for instance, could have been identified to one of the four bovid classes or to the mammal category. Similarly, elements belonging to catfish and barbus fish could have been assigned to the general category of fish and those of aardvark and monkey to the mammal category only.

It will, however, be recalled from the previous chapter, that a considerable number of the wild animals are represented by few specimens only. Notable among these are lesser kudu, which is represented by a single specimen; giraffe and waterbuck by 2 specimens each; and Grant’s gazelle and klipspringer by 3 elements each. Chanler’s reedbuck is represented by 4 elements; roan antelope by 5 elements; bush pig by 6 specimens; and impala, eland and topi by 7 elements apiece. Interestingly, all the elements belonging to klipspringer and lesser kudu are postcranial, while waterbuck and topi are represented by cranial elements.

There is little likelihood, however, that specimens belonging to crocodile could have been attributed to any other taxon. That the animal is also represented by isolated elements in the assamblage under study and in one of the samples that had been studied previously (Karega-Mũnene 1986, 1987) strongly suggests it was not a food resource. Rather, its elements may have been intrusive, a condition that also appears to apply to the elements representing snake, rodent and human. Unlike the rest of the faunal assemblage, elements belonging to these latter seem to be of recent origin since none of them is modified, comminuted or weathered.

The poor taxonomic representation or underrepresentation suggested by these figures may indicate one of several things. First, the specimens belonging to the animals concerned could not be identified to species or genera levels. Second, the hunting methods employed by the occupants of the site may not have been efficient enough to enable the hunters to kill more of those animals. Third, the hunters may have deliberately avoided those animals because they were difficult to hunt or because they were fierce. Fourth, the animals were generally absent from the site’s hunting range. Fifth, the animals may have been butchered far afield, as a result of which, only selected body parts were carried back to the site. Sixth, some of the animals identified in the assemblage may not have been exploited for food; therefore, elements representing them are intrusive.

The suggestion that some of the animals represented at the site may have been butchered away from the site and that only certain parts of the carcasses were transported to the site is not borne out by the evidence presented in the previous chapter. If this were the case, specific patterns would have emerged in the skeletal representation of the animals. The small-sized animals, for instance, could have been represented by virtually all of their skeletal 122

Karega-Mũnene

elements because their carcasses would have been transported more or less intact to the site. As for the larger animals, only a few elements, specifically those that contain high meat yields, would have been transported to the site. This latter would have translated into a preponderance of elements like femora, humerii, scapulae and pelves in the assemblage and correspondingly low frequencies of teeth and other cranial elements; but this pattern is not evident in the assemblage.

The number of cattle represented at the site seems to have been greater than that of caprini. On the basis of the NISP counts, for instance, cattle account for 63.7% of all the domestic animals, caprini for 35.3%, while dog and donkey account for 0.6% and 0.4%, respectively. In comparison, cattle account for 52.4%, caprini 42.8% and dog and donkey for 2.4% apiece by MNI counts. Thus, both the NISP and MNI counts strongly suggest that the occupants of Gogo Falls had a preference for cattle. Interestingly, this suggestion is at variance with the findings of previous work on samples from the 1983 excavations. Marshall’s (1986, 1991) work, for instance, suggests that the caprini herd was about twice the size of the cattle herd by both NISP and MNI counts. Caprini numbered 223 (68%) and cattle 105 (32%) by NISP and 17 (63%) and 10 (37%) by MNI, respectively. This composition compares quite favourably with the findings made on the other sample, the only difference being that caprini exceeded cattle by a smaller margin. In that sample caprini numbered 91 (57%) and cattle 70 (43%) by NISP; the corresponding MNI values were 9 (75%) and 3 (25%) (Karega-Mũnene 1986).

The hunting strategies adopted at the site seem to have favoured the killing of mature animals. This is borne out by the fact that sub-adults and adults dominate the age categories devised using postcranial elements. The only exceptions are alcelaphini and rhino both of which are represented in the juvenile category. This patterning is confirmed by the mortality profile obtained from cranial elements, the only difference being the representation of bush duiker in the juvenile category. The preponderance of mature animals is also evident in the previously studied samples (KaregaMũnene 1986; Marshall 1986, 1991). Thus, it appears that the site’s occupants had a preference for adult and sub-adult ungulates. As hinted above, this condition may have arisen from the fact that the animals had greater food value than younger ones. More importantly, however, the condition suggests the site’s inhabitants were not opportunistic but highly skilled or specialised hunters, a suggestion that is supported by the absence of very young and very old animals which are easier to kill. The younger animals could have been spared to ensure future supplies whereas, the very old ones may have been avoided because of their tough meat or fear of diseases. Alternatively, the animals may not have been represented in the herds that were targeted by the hunters.

The differences in the ratio of cattle to caprini between the previously studied samples and the assemblage which was analysed for this study, seems to be a function of the size and identifiability of the samples. As we have noted in the previous chapter, the samples varied from as few as 612 elements for Marshall’s (1986, 1991) sample to 74,906 elements for this study. Given that larger samples are more likely to be representative than small ones, it is arguable that the sample currently under study reflects the most likely proportions of cattle and caprini at the site. The apparent predominance of cattle at Gogo Falls is, however, not unique to the site. At Prolonged Drift, for example, the ratio of cattle to caprini is 252:46 (85%:15%) by NISP and 22:5 (81%:19%) by MNI. At Crescent Island there are about three times as many cattle as there are caprini under both MNI and NISP, the ratios being 303:92 (77%:23%) and 15:6 (71%:29), respectively. Cattle also predominate at Hyrax Hill where they number 70 (56%) by NISP and caprini 54 (44%) by the same approach (Gifford et al. 1980; Onyango-Abuje 1977b).

Herding The domestic animals represented in the faunal assemblage include cow, sheep, goat, dog and donkey. Together, these animals account for a considerable part of the fauna, 15.1% by NISP and 47.4% by MNI counts (Figure 8.1). If the NISP counts are correct, then it follows that herding was less important than fishing and hunting. But, if the MNI counts are correct, herding may have been as important as hunting.

In comparison, evidence from other Holocene sites in the region indicates that caprini herds are 123

Holocene Foragers, Fishers and Herders of Western Kenya

larger than cattle herds. To take a few examples, the caprini population at Maasai Gorge is five times as high as that of cattle by NISP and twice as high under MNI (Gifford-Gonzalez 1985). At Ngamuriak caprini population is twice as high as that of cattle under both NISP and MNI. At Sambo Ngige caprini population exceeds that of cattle by a factor of three under both NISP and MNI and at Lemek Northeast by a factor of seven under NISP and a factor of five under MNI (Marshall 1986). Similarly, at Naivasha Railway rockshelter caprini population is six times as high as that of cattle under NISP and twice as high under MNI.

of adults, followed by old adults and sub-adults. This profile compares quite favourably with the kill-off patterns reconstructed from the previously studied samples, where sub-adult and adult age groups are the best represented (Karega-Mũnene 1986; Marshall 1986, 1991). That cattle were not killed at the site until they reached maturity may be a reflection of one of several things. For example, mature animals may have been over-represented at the expense of immature ones because their remains suffered less destruction by the taphonomic processes discussed above. Second, immature animals may have been under-represented or not represented at all because the faunal assemblage was not representative. Third, the animal management strategies adopted by the occupants of the site may have militated against the killing of younger animals.

The predominance of either cattle or caprini at any of the sites could be explained in one of three ways: it could be a function of the size of the samples; the result of sampling bias; or it could be real. Given the fairly large samples from Gogo Falls, Prolonged Drift and Ngamuriak, it is rather unlikely that the ratios of cattle to sheep/goat in the respective faunal assemblages could have been affected seriously by sampling bias or by sample size. Rather, it appears that the actual composition of domestic herds may have varied between the sites. The caprini herd at Gogo Falls consisted of goat and sheep, the former being represented in one of the 1983 samples as well (Karega-Mũnene 1986). Elsewhere in East Africa goat has been identified at Akira and sheep and goat are known from Ngenyn, Ndabibi and Narosura (GiffordGonzalez and Kimengich 1984; Gramly 1972; Hivernel 1978). The representation of the animals at these sites suggests there were more goats than sheep in the region. The sheep and goat population at Gogo Falls appears to have been about equal, assuming the 7 goat elements and 6 sheep elements are representative of the caprini herd.

The first two suggestions seem rather unlikely because younger animals, neonates included, are represented in the assemblage. Moreover, elements like vertebrae, ulnae, pelves and ribs that would have been affected seriously by the taphonomic processes in question, are fairly well represented. The third suggestion seems to be the most likely. That is mainly because the mortality profile outlined above is similar to profiles that are created by long-term culling of domestic animals by modern pastoralists like the Maasai. These tend to be dominated by mature animals since the primary aim of keeping livestock is provision of milk and blood, rather than meat (Dyson-Hudson and Dyson-Hudson 1982; Mbae 1986). Similar profiles are also evident at Prolonged Drift, Ngamuriak and Lemek Northeast (Gifford et al. 1980; Marshall 1986, 1990; Onyango-Abuje 1977a), thus reinforcing the supposition about controlled herd management in Holocene East Africa.

The apparent predominance of cattle at Gogo Falls suggests that cattle played a significant role in the economy of the site’s inhabitants, a postulation that is supported by the animals’ mortality profiles. As we have noted in the previous chapter, the profile reconstructed from postcranial elements indicates the animals were killed when they reached sub-adult and adult ages, the latter category being bettter represented. Although the mortality profile reconstructed from dental elements shows the herd included individuals from the entire age spectra, the best represented categories are those

The Gogo Falls cattle could have been allowed to attain maturity or old age because they were more useful when alive than when they became meat. This condition could have been given rise to by the social and/economic role(s) that the animals played. For example, the animals could have been primarily reared for the provision of food products like milk, ghee and blood and not for meat and/or hide. They could also have provided dung which may have served as a raw material in the construction of shelters and/or for the fertilisation of fields, if the people concerned practised cultivation and/or tended wild plants. 124

Karega-Mũnene

In addition, the animals may have had symbolic significance to their owners, say, by being status symbol or by being offered as sacrifices during important ceremonies. If correct, this would mean that meat and hide might have been obtained mainly from wild ungulates, hence the predominance of their remains in the assemblage.

Given the impossibility of gleaning such information from skeletal remains, we can only offer informed hypotheses. Thus, assuming that a relationship similar to the one outlined above may have prevailed at Gogo Falls, domestic animals could have served an important social role. They could, for instance, have enhanced the owner’s status and played a significant role in ritual. They could also have provided the means by which relationships involving relatives and friends were bonded, say, by using them for the ‘payment’ of bridewealth or assisting each other in times of need (e.g., Ryan et al. 1991).

The mortality profile of the caprini herd indicates the animals were also allowed to attain maturity before they were slaughtered. This was also the case in the caprini herd represented in the previously studied 1983 samples (Marshall 1986, 1991; Karega-Mũnene 1986) as well as at Prolonged Drift, Lemek Northeast, Sambo Ngige and Ngamuriak (Gifford et al. 1980; Marshall 1986, 1990). This suggests that the animals were also subjected to systematic culling for reasons that may have been similar to those offered above to explain the kill-off patterns of cattle. But it is also quite likely that caprini may have been less useful than cattle in the provision of products like milk, but more useful in providing meat since their rates of reproduction and growth are much higher than those of cattle.

So far, we have not touched on the question of how domestic animals were introduced at Gogo Falls. That is mainly because the data explicated in this study have not shed adequate light on this issue. This notwithstanding, there is need to evaluate three suggestions that have hitherto been offered to explain how domestic animals may have reached the site. To begin with, it has been suggested that the animals were introduced to the site by “an immigration of new groups of people [from the Loita-Mara and Rift Valley regions]” (Robertshaw 1991: 170). This suggestion is based on three main assumptions. One, that Elmenteitan ware was created by people who owned domestic animals; therefore, its appearance at Gogo Falls signifies the occupation of that site by pastoralists. Two, that Gogo Falls had been occupied by people who lived by hunting, gathering and fishing prior to the immigration of the so-called Elmenteitan population. Three, that the associations observed at the site between Elmenteitan, Kansyore, Akira and Urewe wares was the result of disturbance and not coexistence.

Nevertheless, it must be pointed out here that it is likely that by discussing domestic animals in purely economic terms, as we have done, we might be telling only half the story. The animals could have played a significant social role besides providing food and other products. That is, of course, assuming that observations made in extant pastoralist communities are a reliable guide to past lifeways. To take one example, modern Maasai attach great social significance to their cattle, a condition that is underscored by the fact that the animals have individual rather than generalised identities. As a matter of fact, each of the animals in a given herd is given a name by members of the household concerned:

To begin with, if the site had previously been occupied by people who lived by hunting and gathering, and if the envisaged migration and displacement or absorption of the huntergatherers did really occur, evidence for a preNeolithic presence at the site would have been uncovered. Save for tentative suggestion that such a horizon existed (Robertshaw 1991), the necessary evidence is lacking. Also unavailable is evidence for the occurrence of a transformation from a food-collecting economy to a food-producing economy. In short, neither the evidence adduced from the 1983 samples studied previously nor the evidence explicated in this study support the migration proposition.

Men tend to refer to an animal by its ‘line’ name, remembering the transaction of acquisition such as Sotua (gift from a friend) and Nesotua (from [or offspring of] Sotua). Women and children tend to refer to animals by their individual names such as Nkeyi (speckled) and Nenkeyi (offspring of Nkeyi). Since women and children are more closely involved with individual animals through milking and herding, the names given by them to animals are used to distinguish herd members and include descriptives such as colour, horn shape, behavior, milking capacity and age and gender. (Ryan et al. 1991: 95)

125

Holocene Foragers, Fishers and Herders of Western Kenya

Secondly, it has been suggested that the occupants of the site were essentially pastoralists who were forced to exploit wild food resources because they had lost their livestock to trypanosomiasis (Robertshaw 1989). Although this suggestion is quite plausible, there is no evidence for the existence of a purely pastoral mode of production at the site. Moreover, contrary to the presumed antiquity of tsetse flies in the region, trypanosomiasis is barely a century old in the region (Morris 1960; Soff 1969).

Butchery Although only a small portion of the assemblage is cut-marked, the distribution of the cut-marks on the skeletal elements strongly suggests that they were inflicted during butchery. To take a few examples, the cut-marks occurring on mandibles are generally restricted to the mandibular condyle. On the long bones, cutmarks are generally restricted to close to the articular ends; on ribs they occurr just below the rib head, that is, close to the rib-vertebrae articulation; and on vertebrae they occurr mainly on the centrum. That the frequency of cut marking on specific body parts of each taxon is extremely low, explains why meaningful comparisons cannot be made between the butchery of different animals.

Thirdly, it has been postulated that the site may have been occupied by “pastoralists together with a group of client hunters who exchanged meat and skins for livestock and milk” (Robertshaw 1991: 168). Like the first proposition, this suggestion is predicated on the assumption that past economies can be reconstructed from material culture. Thus, Elmenteitan ware is equated with pastoralists and Akira ware with hunter-gatherers. The validity of these associations will be discussed further in the next chapter.

Palaeoenvironmental Implications Before making inferences about past climatic and environmental conditions at Gogo Falls from the faunal data, it must be pointed out here that faunal data are not very reliable indicators of environmental conditions of small geographical entities like archaeological sites. Rather, they provide general information about the conditions that may have prevailed within a much wider geographic area either over long or short periods. This limitation can be explained in one of two ways. Firstly, the wild animals represented at a given site could have been killed far afield, depending on the preferences of the hunters concerned, the presence or absence of the animals and/or the extent of the hunting range. Secondly, some of the animals might be migratory in character. As such, the environmental implications drawn from their presence or absence could only have obtained for a brief season only, usually a few months.

It will, therefore, suffice here to note that once a given human group has acquired domestic animals, they must devise ways of replenishing their stock from time to time. This is crucial to the existence of a herding community because raids, disease or drought can wipe out whole herds. Indeed, such occurrences are not uncommon in historic East Africa, as OdegiAwuondo’s (1990) work on the Turkana of northwestern Kenya demonstrates. For the purposes of replenishing their herds, modern East African herding communities adopt a combination of several strategies. For example, they obtain animals through an elaborate and reciprocal gift system that involves friends and relations as well as through trade and payment and receipt of bridewealth. Rustling which is now illegal in the region also played a significant role in the past. These activities enable those who lose their herds to rebuild them (OdegiAwuondo 1990; Ryan et al. 1991). Any of these methods or a combination of some or all of them could have been adopted by the occupants of Gogo Falls, if the animals in question were as important as the data explicated in this study suggest.

Thus, the environmental conditions represented by faunal data are likely to encompass a mosaic of habitats. Indeed, this seems to have been the case at Gogo Falls, where species characteristic of varied habitats were identified. To take a few examples, animals characteristic of open grassland like warthog, wildebeest, bush duiker, zebra, oribi and Thomson’s gazelle are represented (Table 8.1). Also represented are buffalo, bushbuck, bush pig, bush duiker,

126

Karega-Mũnene

Table 8.1 Environmental characterisations of the Gogo Falls fauna. Taxa Aardvark Bohor reedbuck Buffalo Burchell’s zebra Bushbuck Bush duiker Bush pig Chanler’s reedbuck Eland Giraffe Grant’s gazelle Grevy’s zebra Hartebeest Impala Klipspringer Lesser kudu Oribi Roan antelope Thomson’s gazelle Topi Warthog Waterbuck Wildebeest

Grassland

Wooded savannah

Light bush

Thick bush

Forest

x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x -

x x x x x x x x x x x x -

x x x x x x -

x x x x x -

x = present in named habitat - = absent in named habitat

hartebeest and eland all of which are found in montane forest and sometimes in open habitats. Roan antelope, a wooded grassland dweller; lesser kudu and klipspringer both of which are characteristic of dry bush; hippopotamus, waterbuck and bohor reedbuck which are found near water (Estes 1991); and crocodile and fish both of which live specifically in water are also represented.

shows that the taxa are dominated by species that live in light bush, grassland and grassed woodland. These vegetation types are quite similar to those that appear to have prevailed in the Lake Victoria basin during the mid-late Holocene. Interestingly, some of the animals represented in the assemblage are no longer found in the region, thanks to extensive human settlement and hunting by present-day inhabitants. The roan, for example, is an endangered antelope found only in Ruma National Park in the Olambwe Valley on the shores of the lake. Previously it was found in habitats stretching from the Maasai Mara in the south to Mt. Elgon in the north. It has been hunted to near extinction for its meat, beautiful hide for ritual purposes and long horns for making musical instruments.

Thus, the picture that emerges from the habitat characterisations of the taxa represented in the faunal assemblage is that of an environment that may have been significantly similar to presentday’s in the eastern Lake Victoria basin, namely, grassland and wooded savannah with light-thick bush and marshes in some places. A close look at Table 8.1, which summarises the habitat preferences of the animals concerned, clearly

127

CHAPTER NINE

SUBSISTENCE STRATEGIES AND TECHNOLOGY

The purpose of this chapter is twofold. One, to evaluate the validity or otherwise of making inferences about the subsistence activities and the linguistic and/or ethnic identity of the people who occupied Holocene East Africa from pottery wares. Two, to assess whether the wares in question can be used to build a reliable chronological framework for the region. Although these evaluations are made with specific reference to Gogo Falls, references will be made to other East African sites because the discussions have significant implications for the whole region. The evaluations will be made using the chronometric dates and the faunal and ceramic data explicated in the preceding chapters.

excavations (Robertshaw 1991). This date is in general agreement with the radiocarbon date of ca. 3,400 BP that was obtained for this study. Together, the dates suggest the site was occupied by mid fourth millennium BP. If correct, the occupation coincided with the progressive decline of the semi-deciduous forest that had developed around Lake Victoria in midHolocene, an occurrence that has been attributed to human activity and climatic changes. Although it is not clear when the occupation ceased, it is evident that the site was still occupied by about 1,700 BP. The linguistic and/or ethnic identity of the occupants of the site during that period remains a matter of conjecture. The oral traditions discussed in chapter four, for instance, suggest the area within which the site is located was occupied by Bantu-speaking peoples prior to its conquest by the Luo in the 18th century A.D. But the traditions are silent about the antiquity of Bantu presence in the region. The little archaeological evidence available in the area on this period comes from surface collections and test-excavated materials, rather than systematic investigation of the Iron Age phenomenon. Borrowing hypotheses on Bantu migration and settlement from historical linguistics and oral traditions to explain occupation of the site will only result in a circularity of assumptions, which cannot be viewed as an explanation of the archaeological phenomenon.

Occupation of Gogo Falls The oldest date that is currently available from Gogo Falls suggests that the site was occupied from about 18,000 BP. This date is, however, considered to be unacceptable because the dated obsidian sample was recovered from an uncertain context. In the words of the excavator, the date ...may [be] ignore[d]... as relating to a much older artefact incorporated into later deposits or to an earlier ephemeral occupation on the older land surface. (Robertshaw 1991: 73) Another date obtained on a faunal sample from Trench I suggests the site was already occupied by about 7,300 BP. An additional bone date for the test trench that was made at the site in 1981 suggests the site was occupied by about 5,800 BP (Gowlett et al. 1987; Robertshaw et al. 1983). Scepticism has been expressed with regard to these dates mainly because of the uncertain contexts and associations of the dated samples and because of the general lack of reliability of bone dates (Collett and Robertshaw 1983b; Robertshaw 1991).

Pottery, Chronology and Subsistence The spatial and chronological representation of the Iron Age and Neolithic pottery wares represented at Gogo Falls and the main subsistence strategies with which they were associated is summarised in Table 9.1. This is done with specific reference to the excavated squares and types of deposit that were encountered in each of them, the latter being indicative of the natural layers. Information about the representation of the wares and faunal resources through the squares and spits that were followed during excavation is presented in Appendix V.

Thus, as we have hinted in chapter five, the only acceptable dates for the site are those that fall between the fourth and second millennia BP. The oldest of these is an obsidian hydration date of about 3,600 BP that was obtained for the 1983 128

Karega-Mũnene

Examination of Table 9.1 reveals that the pottery and subsistence activities practised at the site are represented in each of the five areas that were excavated recently. That is Areas A (squares 1 20); B (squares 21 and 22); C (squares 23 and 24); D (squares 25, 26 and 27); and E (squares 28, 29 and 30. However, examination of the representations of the same through the 30 squares reveals that pottery was better represented both spatially and temporally than the faunal resources. Indeed, except for Akira pottery which, as we have already noted, is represented in 20 squares − 17 of which have ashy deposits and 3 loamy deposits − the other wares occur in every square and type of deposit. This is in contrast to the representation of the skeletal remains from which the subsistence activities are inferred. Although the remains representing each of these activities were obtained basically from all types of deposits in Areas A through E, each of them is represented in 20 or fewer squares.

own, none of the subsistence activities is represented. Unlike in the lowermost deposits, the associations in the upper levels of the three squares involve all the wares and subsistence strategies. And in the ashy deposits in squares 1 and 13 the wares are associated with herding only; in squares 8 and 17 the wares are associated with herding and fishing; and in squares 20, 21, 22 and 26 with herding, hunting and fishing.

Close inspection of the table reveals there is no specific association between either of the wares and any of the subsistence activities and/or type of deposit which represent natural layers. There also seems to be no specific association between the wares and the levels of occupation which may be represented by the natural layers. For example, in the clayey deposits of square 25, Elmenteitan, Kansyore and Urewe wares are not associated with any of the subsistence activities, probably because the associated faunal remains were not identifiable to species or genera level. In the deepest clayey deposits in square 26, Kansyore is associated with hunting only, while in the upper levels of the same deposits it is associated with hunting and herding. In the lowermost clayey deposits in square 27 Kansyore is associated with the three subsistence activities. In the upper levels of the same deposits, Elmenteitan, Urewe and Kansyore are associated with hunting, fishing and herding. The associations in the black loam deposits of the three squares involve all the subsistence activities and wares.

The first suggestion has been offered to explain the association of Akira, Elmenteitan, Kansyore and Urewe wares in the trenches that were excavated in 1983. This latter phenomenon is considered to be "the result of mechanical mixing of what are often loose midden deposits, rather than... contemporaneity of different traditions at the site" (Robertshaw 1991: 113). Although the looseness of the ashy deposits may account for some of the ‘mixing’, as we have also suggested in chapter five, it cannot explain all the ‘mixing’ since some of the deposits were hard and compact. Other factors like burrowing animals, erosion and human activity may also have contributed to the ‘mixing’. For example, the animal burrows encountered during the 1983 and 1989 excavations may explain disturbance in Trenches II and III and square 28 in Area E. Clearing of the site by burning and deposition of domestic refuse by the site’s occupants may, as we have reasoned above, explain mixing in the ashy deposits only. The process of erosion may also explain mixing in middle-lower parts of the site, but not in the areas upslope.

The associations in the lowermost black loam deposits in square 28 involve Elmenteitan, Kansyore and hunting. In contrast, the three subsistence activities are evident in the deepest deposits in square 29, where Kansyore occurs exclusively on its own. In the deepest deposits in square 30, where Kansyore also occurs on its

While these suggestions are quite plausible, chronometric evidence from the site suggests consistency in the site’s stratigraphy in some areas. Indeed, as we have observed in chapter five, stratigraphic inconsistency is clearly evident only in Trench I. It is, therefore, likely that the occurrence of some of the wares in the same

This condition suggests one of two things. First, that the archaeological deposits had been disturbed prior to excavation, hence the occurrence of the four wares and the skeletal remains representing the subsistence activities in question in the same deposits. Second, that the wares are not discrete cultural entities and, therfore, they cannot be used to construct a chronological framework for the site or to make reliable inferences about subsistence and/or the identity of the human groups who created or used the pottery.

129

Holocene Foragers, Fishers and Herders of Western Kenya

deposits may attest to coexistence. If correct, this condition would be at variance with the chronology of the wares that has been proposed for the site. This latter suggests that the oldest pottery is Kansyore, followed by Elmenteitan and Akira (both of which are contemporaneous) and Urewe, in that order (Robertshaw 1991).

excludes Elmenteitan. The third views Kansyore to be the oldest, followed by Nderit and Narosura (both of which are contemporaneous), Elmenteitan, Turkwel and the coeval Maringishu and Akira wares, respectively (Ambrose 1982, 1984). In each of these schemes, Urewe ware is considered to be the youngest.

It will be recalled from the introductory chapters that three chronological schemes have been proposed for the East African Holocene wares. The first of these considers Nderit to be the oldest followed by Narosura, Remnant (Elmenteitan), and the contemporaneous Maringishu and Akira, respectively (Wandibba 1977, 1980). The second also considers Nderit to be the oldest ware, followed by Kansyore, Narosura, Akira and Maringishu, in that order (Bower et al. 1977). Interestingly, this scheme

Of the wares represented at Gogo Falls, Kansyore is excluded in the first scheme and Elmenteitan in the second. Therefore, both schemes cannot be applied at the site in their original form. The third scheme includes all the wares represented at the site, but the sequence it entails − Kansyore, Elmenteitan, Akira and Urewe − is not attested to at the site. Indeed, apart from Kansyore which, as we have already noted, was found on its own in a

Table 9.1 Pottery wares and subsistence strategies by squares and type of deposit.* Square 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Type of deposit Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Ashy deposits Brown/black loam soil Clayey deposits Ashy deposits Brown/black loam soil Clayey deposits Ashy deposits Brown/black loam soil Clayey deposits Black loam soil Black loam soil Black loam soil

Pottery wares

Subsistence strategies

Akira

Elmenteitan

Kansyore

Urewe

Herding

Hunting

Fishing

x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x x

*x = Present, - = absent.

130

Karega-Mũnene

few squares, the rest of the wares were found together and also in association with Kansyore.

The only sites bearing Kansyore pottery that have yielded considerably large faunal samples are Mumba-Hohle in Tanzania and Gogo Falls. Unlike at the latter site, however, domestic animals are not associated with Kanyore at Mumba-Hohle. Rather, they are associated with Nderit, Narosura and Akira pottery (Mehlman 1989). Thus, given that only a few faunal samples associated with Kansyore pottery have so far been studied, further systematic research on the question of subsistence is called for.

There is hardly any doubt that the chronological sequence that has been proposed for Gogo Falls – Kansyore, Elmenteitan, Akira and Urewe – is premised on the assumption that the wares were used by different linguistic/ethnic groups who also adopted different subsistence strategies. Kansyore pottery, for instance, has been viewed as an invention of autochthonous huntergatherers who may have belonged to an unnamed “single ethnic group” (Robertshaw 1991: 158). In this regard, it has been reasoned that we

Unlike with Kansyore, the origin of the other wares represented at Gogo Falls has been explained in terms of large-scale human migrations. Urewe, for example, has been assigned to immigrant Bantu-speaking peoples whose economic activities encompassed cultivation and animal husbandry. Interestingly, although several Bantu-speaking communities inhabit Lake Victoria basin today, none of them produces Urewe ware or its derivative. In consequence, several questions emerge: could the extinction of Urewe pottery have resulted from cultural encounter with new immigrants? If the answer is yes, who were those immigrants? If the culture of those immigrants was so overwhelming to result in the extinction of Urewe, why did they not absorb or displace the Bantu from the lake basin?

...need not envisage an immigration of people into South Nyanza region to explain the origins of... Kansyore.... Although the evidence is scanty, it seems that pottery may be the only new element to be added to the repertoire of the indigenous LSA [i.e., Later Stone Age] inhabitants of the region. (Robertshaw 1991: 170)

The evidence employed to support this proposition is twofold. One, that the populations who created Kansyore pottery were huntergatherers as evidenced by the findings of a preliminary study of the mammalian teeth from Trench I in which domestic stock were not represented (Robertshaw 1991: 161). Two, that Kansyore is associated with fishing and hunting at Kanam East, Luanda, Kanam Two, Kanjera West, White Rock Point and Nyang’oma, (Robertshaw et al. 1983). Nevertheless, it will be recalled from chapter two, that representativeness of the faunal remains from some of these sites is open to question because of two reasons: (a) sample size and (b) the manner in which they were obtained.

Like Urewe, Elmenteitan ware has been attributed to large scale migration, the only difference being that the economy was based on pastoralism: It is difficult to envisage the onset of... [the] Elmenteitan occupation as anything other than an immigration of new groups of people into South Nyanza, who brought with them the first domestic animals and radically new styles of pottery... the cultural connections of these pioneer Neolithic inhabitants of Gogo Falls lay to the east towards the Loita-Mara and Rift Valley regions, and one need not necessarily look any further afield for their origins. (Robertshaw 1991: 170)

Although sheep/goat are represented in the sample from the Kanjera West midden, they are associated with recent Luo pottery, not with Kansyore. Domestic animals are also not represented in any of the samples from the levels containing Kansyore pottery at the other middens (Robertshaw et al. 1983). Be that as it may, it must be pointed out here that the samples in question are quite small: Kanam Two, White Rock Point and Kanam East have less than 50 potentially identifiable mammalian, bird and reptile elements apiece; and Kanjera West and Luanda about 160 and 320 specimens, respectively. Thus, we cannot rule out the possibility that some of the samples may not be representative.

But it is not clear whether the Elmenteitan immigrants found the site occupied or not: Since the dating of... [Kansyore] in South Nyanza is rather vague, it is not known whether Elmenteitan pastoralists replaced or assimilated indigenous hunter-gatherers or merely occupied a vacant region or niche. The continued importance of foraging – hunting, gathering and fishing – during the Elmenteitan

131

Holocene Foragers, Fishers and Herders of Western Kenya occupation is suggestive of continuity with earlier settlement. (Robertshaw 1991: 170)

Urewe wares in deposits dating to about 2,000 years ago, begs appraisal, rather than off-hand dismisal. As reasoned above, the associations may not have been caused by previous disturbance for, the likely causes of disturbance seem to have been rather localised. It is, therefore, quite likely that some or all of the wares represented at the site may have been contemporaneous at some point during their lifespan

The presence of Akira ware at the site has been viewed as “particularly intriguing” because it suggests the existence of hunter-gatherers at a time when the site was occupied by Elmenteitan pastoralists (Robertshaw 1991: 168). Another postulation has it that the pottery may have been made by hunter-gatherers in the Serengeti region, where it was first identified. From there it could have reached Gogo Falls through exchange with domestic animal products from Elmenteitan pastoralists (Robertshaw 1990a: 200). The pottery could also have been made by huntergatherers who lived in the vicinity of Gogo Falls, where they had a symbiotic relationship with the pastoralists in question (Robertshaw 1991: 168). Interestingly, these propositions are silent on whether Kansyore and Akira were created or used by the same group(s) or not.

This condition is not unique to Gogo Falls for, as we have noted in chapter two, wares which are thought to have been created or used at different times have been found in association elsewhere in East Africa. To take a few examples, Kansyore is associated with Nderit and Akira at Seronera, whilst at Nyang’oma and Chole it is associated with Iron Age pottery. Narosura ware is associated with Akira at Salasun and with Maringishu ware at Maringishu. The likelihood that these wares may have coexisted, albeit for short periods, is supported by the chronometric evidence presented in chapter two. This latter clearly demonstrates overlap between dates of different Neolithic wares and, in some instances, between Neolithic wares and the Iron Age Urewe ware.

It will be recalled from chapter two that the association between Elmenteitan ware and domestic animals has been observed at other sites in East Africa. But the same cannot be said about the association between Akira and a foraging economy since the relevant faunal evidence is generally lacking. As such, the presumed association is based on negative evidence.

Critical evaluation of the dates associated with Urewe ware indicates the ware lasted from about 2,500 BP to 1,300 BP (Clist 1987). Although the dates for the Neolithic wares have not been subjected to a similar appraisal, they show considerable overlap between the wares and even with those of Urewe ware. While Kansyore is the earliest known Neolithic ware, it is evident that Nderit, Ileret, Narosura and Elmenteitan wares were created during its lifetime. Chronometric evidence also suggests that Kansyore had the longest life span, followed by Nderit and Elmenteitan, in that order. But the life spans of Ileret and Maringishu wares remain unknown since each of them is dated at a single site. Additionally, some of the Neolithic wares, namely, Akira, Maringishu and Turkwel appear to have been created during the lifetime of Urewe ware. In sum, there is no clear-cut succession between any of the wares in any defined geographical space. Therefore, we cannot simply assume that Kansyore is older than Elmenteitan or Nderit wherever they occur together since their appearance may have varied through time and space.

It is also interesting to note that Akira is the only ware that is thought to have spread in East Africa through exchange. This suggestion is based on the assumption that its thin-walled vessels were used as “serving or eating bowls or drinking cups (perhaps analogous to the ‘best’ china sets of European households” (Robertshaw 1990: 200). As far as the conventional archaeological wisdom is concerned, the rest of the wares spread through the migration of the people who presumably created them. Similarly, the spread of domestic animals in the region is also attributed to large-scale migrations. Three major assumptions are evident here: (a) that pottery produced by a given ethnic/linguistic group is identical through time and space; (b) that a given ethnic/linguistic group only uses the pottery it produces; and (c) that the last 4,000 years or so have been characterised by large-scale population movements and absorption or displacement of weaker groups by stronger ones. Granted that Kansyore is the oldest pottery at Gogo Falls since it was in use by about 3,400 BP, its association with Elmenteitan, Akira and 132

Karega-Mũnene

If the suggestion that the wares represented at Gogo Falls co-existed at some point, then it follows that they cannot be treated as discrete cultural entities for purposes of constructing a reliable chronological framework for the site. As such, they cannot be used to generate reliable inferences about the linguistic or ethnic identity of the people who created or used them. Similarly, they cannot be used to make inferences about the subsistence activities of the people concerned. If correct, the interpretive model that has hitherto been predominantly used in East African Holocene archaeology may be invalid not only for Gogo Falls, but also for other sites in the region. The model, it will be recalled, makes specific associations between the wares, ethnic and/or linguistic groups and their subsistence economies.

identity of the immigrants and the routes that brought them to the region, archaeologists have relied on the ubiquitous pottery for the same purpose. The literal borrowing of hypotheses especially from historical linguistics to explain archaeological phenomena has hampered, rather than advanced archaeological inquiry in the region. That is because it has resulted in oversimplification of the region’s Holocene archaeology, whereby it appears that the placement of the wares and the ethnic and/or linguistic groups associated with them is nothing more than a jigsaw puzzle, in which everything will eventually fall in place. The only prerequisite in the exercise is the identification of the wares since, once these are known, the identity of the people who occupied a given site and their economy automatically follow!

According to the model in question, several linguistic and ethnic groups would have occupied Gogo Falls. These would have included antecedents of modern Southern Nilotes (e.g., Tugen, Nandi, Kipsigis and Dadog) who are associated with Elmenteitan ware. Secondly, predecessors of Southern Cushites (e.g., Asa, Burungi, and Dahalo), whose presence is signified by Akira ware. Thirdly, antecedents of modern Khoisan of southern Africa, whose presence is inferred from Kansyore pottery. Fourthly, ancestors of modern Bantu-speaking peoples like the Gusii, Luyia and Kuria (in the Lake Victoria basin) who are associated with Urewe pottery (Ambrose 1982, 1984). Given the remarkable differences in the languages spoken by these groups, the site would have been akin to the biblical tower of Babel if there was overlap in their occupation of the site. In addition, it would have been difficult, if not impossible, to reduce conflict arising from differences and probable incompatibility in the cultural practices and economic activities of the groups in question.

Close examination of the concerned archaeological literature clearly demonstrates that this approach has been predominantly employed in East African archaeology because it is in general agreement with pre-colonial ethnic histories that were reconstructed in the 1960s and 1970s when such undertakings were in vogue (e.g., Muriuki 1974; Mwaniki 1973; Ogot 1967; Were 1967). As the discussion of human settlement in South Nyanza District in chapter four demonstrates, the histories in question have been reconstructed largely from oral traditions, with occasional references to archaeological and/or linguistic ‘evidence’ where this was available. Interestingly, the only link between the ‘evidence’ adduced by the three disciplines (i.e., history, historical linguistics and archaeology) is a commonality of assumptions. Inevitably, the inter-borrowing of assumptions between these disciplines has not only resulted in an endless circle of hypotheses, but also elevated what were originally, and still are, hypotheses to established facts. Outstanding examples of this are the muchcelebrated Cushitic and Bantu migrations. It is the contention of the present author that the routes through which these latter groups are presumed to have reached East Africa may be exchange/trade routes used by Holocene populations in their quest to dispose surpluses and to acquire what they did not produce or possess. That systematic investigation of interactions in the region’s Holocene archaeology is yet to be carried out strongly

Undoubtedly, the associations, which are the building blocks of the model, have been made in an attempt to support the hypothesis that peopling of East Africa throughout the Holocene occurred through large-scale migrations. As we have noted in the introductory chapters, this hypothesis is literally borrowed by archaeologists from historical linguistics and history. While historical linguists and historians have used linguistic data and oral traditions, respectively, to reconstruct the linguistic/ethnic

133

Holocene Foragers, Fishers and Herders of Western Kenya

suggests the linear chronological model that has hitherto been in vogue may be out-dated.

sufficient in the production of that commodity or did they acquire some of the vessels from other potters through trade or exchange?

Interpretive Model

While answering these questions, the roles played by the individual and the community or the corporate society must be accorded the emphasis they deserve. The individual potter, for instance, may have decided what decorative style or form a given pot should take. But the ‘norms’ established by the community in which the potter operated must have determined whether the pot was acceptable to the intended consumers or not. This would, in turn, have determined whether more pots with the same style and form were made or not and for how long. Similarly, individual hunters may have deliberately avoided killing a given animal because it was fierce. That notwithstanding, the social prestige accorded to any hunter who killed such an animal by the community could have influenced the hunter to take the risk during his/her next hunt. Alternatively, the community could have been influenced by the attitude of individual hunters to formalise the view that the animal was fierce, thus making its avoidance socially acceptable.

The problem in East African Holocene archaeology probably lies with the manner in which the wares have been defined and the significance that we continue to attach to those wares (Karega-Mũnene 1996). It will be recalled from chapter two that the wares have been defined on the basis of gross similarities in decorative style and form. The significance of these characteristics to the individuals who made and/or used the pottery, which has been classified into the wares, may have been radically different from what we read in them today. While decoration and other physical attributes may be useful as cultural identifiers ceteris paribus, it must be appreciated that the attributes could also be funtional or aesthetic. Indeed, when asked why they decorate their pots, potters from different ethnic and linguistic groups in Kenya state they do so for aesthetic reasons (Omollo 1988; Wandibba 1995; Pers. observ.). In traditional African households pottery serves numerous purposes: ritual, carriage, storage, cooking, serving food, and tubs for washing babies (Barbour and Wandibba 1989; Wandibba 1995; Pers. observ.). It is, therefore, imperative that we distinguish between categorisations that are made to facilitate archaeological investigations from those that may have prevailed in the past.

It is the contention of the present author that the interpretive approach suggested here will enable us to recognise the importance of the roles played by the individuals and the corporate society in the economic and cultural activities of the populations who occupied Holocene East Africa. Although both individuals and the corporate society participate in decision making, the fact that the capacities of decision making are varied amongst individuals, raises important questions about the relationship between a given community and material culture. Thus, can the community be defined in terms of the material which is produced by individuals and not by the corporate society per se? Or, can a definition of material culture be appropriate when it is offered solely in terms of the corporate society at the expense of the individual?

The starting point should perhaps be to make attempts to answer the following questions: are the wares merely tools of analysis for the archaeologist or are they an expression of some form of identity? If the former is the case, then it follows that it is gratuitous to associate the wares with specific linguistic or ethnic groups, to use them as chronological markers, or to make inferences about past subsistence strategies from them. If the wares are vehicles of expressing identity, we must ask whose identity do they express? Is it that of the individual potters, their family, clan, or ethnic and/or linguistic group? Or, is it the identity of the intended consumers? Did all the inhabitants of Holocene East Africa make earthen vessels? If some of them did not, how could they have expressed their identity through pottery? If the pottery they used was specifically made for them by individuals from other groups, whose identity was expressed in it? Were the groups who made pottery self-

Ethnographic observations made in Kenya clearly demonstrate that the craft of making pots is not necessarily a monopoloy of a given ethnic or linguistic group. Among some communities it is confined to specific clans or households (Lindbloom 1920; Peristiany 1939; Wagner 1970) and in others to specific individuals. In the latter case, a ceramic tradition, which is associated with a given community but made by specific individuals, tends to become extinct with 134

Karega-Mũnene

the death of the potter, as has been the case at Kokwa Island in Lake Baringo (Wandibba 1995). Among the Luo of western Kenya, a community that is famous for making earthen vessels, potters form a mere 1% of the population; all the potters are women and tend to live close to sources of clay. While some of them learn the craft before marriage, the majority come from families without potting background. Largely, however, a priori knowledge about the craft is irrelevant since the potters have to conform to the tastes of the household or clan they marry into:

...the individual decorative innovations of one [Luo] potter came to be widely adopted by others in her community largely because of... [her] personal popularity and... willingness to help others improve their potting. Conversely, another potter expended considerable effort to maintain a personal decorative technique different from that of her co-wife..., but this has not been adopted by other potters and has remained an individual variant because she is considered a ‘complainer’ and is not a popular figure in the community, and because... [her style] requires significant extra work. In one instance, a community of potters who specialize in making one kind of pot split into two groups following a dispute; ... whereas... [the two groups] used to make identical pots, the[ir] production... has become distinguishable by a small but consistent difference in decoration. (Herbich 1987: 202)

...Much as non-potter women will learn to use local names for pots and ‘unlearn’ the names they used to employ, potters will learn to reproduce the local forms and apply decorations from a limited repertoire used in the community. They learn the craft under the supervision of a mother-in-law or senior cowives. They gather clay and temper from the same sources and learn to utilize the same techniques in constructing a range of acceptable forms and decorations.... Significantly, however, even those whose mothers were potters almost always conform to the local pattern. This is both because they did not learn the craft in earnest from their mothers [because they considered it to be ‘dirty’] and because they are profoundly influenced by the women working around them in their husbands home and in neighbouring homesteads in the community. (Herbich 1987: 200-201)

Thus, the fact that the Luo are a single ethnic group seems to be largely incosequential to the potters, whose vessels are characterised by personal variation. The pottery made by a given group of potters in Luoland is not only consumed by their community, but also by other Luo communities and by various ethnic groups. In a few instances, however, some potters might make distinct pottery for trade with a specific ethnic group (Herbich 1987). While the latter type of pottery may not be found in Luoland but in a neighbouring community, pottery that belongs to various Luo ‘traditions’ occurs both in Luoland and in other parts of Kenya settled by different ethnic and linguistic groups.

Thus, women who originally came from different parts of Luoland may end up perpetuating a ceramic ‘tradition’ that was, until their marriage, alien to them. The ‘tradition’ itself bears the ‘imprint’ of the ‘community’ concerned − where the community is the village or a group of households and not necessarily the entire Luo ethnic group − as well as that of the individual potters. It may also bear ‘foreign imprint’, resulting from interaction with other Luo potters and/or non-Luo potters. Although such influence “always occur[s] within the context of the tradition into which the potter has been indoctrinated” (Herbich 1987: 201), when combined with individual creativity, it often leads to the adaptation of new vessel forms and decoration.

Inevitably this condition raises several questions which are pertinent to interpretation of ceramic patterning. For instance, do we assign the pottery made by Luo potters specifically for trade with a given ethnic group to the Luo or to the ‘foreign’ consumers? How do we explain the presence of a ceramic ‘tradition’ that is created by a given Luo community in most of Luoland and even beyond − through the migration of the potting community or through trade? Can we really draw ethnic or linguistic boundaries with regard to the spatial distribution of the pottery, let alone attribute its presence, say, in central Kenya to Luo migrants?

Some potters are, however, able to express their individual identity more strongly than others, thus influencing the styles adopted by other potters and/or the tastes of their community:

In short, diversity may have been the norm rather than the exception in past societies in East Africa, as both the above example and the ceramic and faunal data explicated in this study 135

Holocene Foragers, Fishers and Herders of Western Kenya

suggest. Thus, it is not inconceivable that one or more of the wares represented at Gogo Falls could have been produced by individuals who lived there while others could have been obtained from elsewhere through trade or exchange. Indeed, it is quite intriguing that Akira ware and obsidian raw material are viewed as having been the only exchange or trade items to have reached the site without large scale migrations. If the occupants of the site could have acquired these items from as far as the Serengeti plains in Tanzania and the Lake Naivasha basin in the central Rift, why could they not acquire other wares and/or domestic animals in a similar manner? Further, if some of the pottery found at the site was produced there, why could it not have ‘spread’ elsewhere through trade or exchange?

communities. The Ilchamus, for instance, make their own pots and sell some of it to the Tugen (Hivernel 1978). The Luo also make their own pottery and sell large quantities of it to communities like the Tugen, Gusii and Gĩkũyũ who live in different parts of Kenya (Barbour and Wandibba 1989; Herbich 1987; Wandibba 1990, 1995; Pers. observ.). If such communities existed in Holocene East Africa, and there is hardly any reason to assume otherwise, then it is arguable that groups who never made pottery could have accumulated some of the wares in the region. It is also likely that people who made pots but who also exchanged some of their vessels with individuals from other potting families, clans, groups, or communities could have accumulated some of the pottery. Such transactions could have taken place within short and/or long distances from the occupied sites, as evidenced by movement of obsidian in the region (e.g., Merrick and Brown 1984a, 1984b; Merrick et al. 1990, 1994).

Assuming that some of the wares could have been manufactured at the site or in its vicinity, it is quite likely that the form and decoration on some, if not all the vessels, may have been influenced by contacts between individual potters who may have belonged to the same community or to different communities living contiguous to each other (Wandibba 1995: 167). But it is also likely that the requirements or preferences of the intended consumers could also have played a significant role in this. It is not inconceivable that these factors could have caused variation in the pottery in the scale of what we consider to be wares today.

In sum, the presence of certain artefacts at a given site does not necessarily indicate the people who created them occupied the site. Similarly, the occurrence of such artefacts at sites that are located far apart does not necessarily mean the linguistic or ethnic group who are presumed to have created the artefacts occupied both sites or that they migrated from one site to the other. Thus, we must recognise the fact that the task of deciphering the ethnic or linguistic identity of the occupants of a given site from material culture is a precarious undertaking.

It is also not inconceivable that the occupants of the site may not have manufactured any of the pottery represented there, but obtained it from elsewhere through exchange. As a matter of fact, such cases are not unfamiliar in present-day East Africa. The Tugen of Baringo District in Kenya, for example, have not made pots in living memory; but they used and still use pots. In the past they obtained their pottery from the Keiyo and Marakwet through exchange, while at present they obtain it from the Keiyo and Ilchamus who are their neighbours and from the Luo and Abaluyia of western Kenya. The Pokot also never made pots; rather, they bought them from the Marakwet (Hivernel 1978). Similarly, the Maasai have rarely produced their own pots in historic times, but acquire them from the Ogiek (A. H. Jacobs, personal communication to Gramly [1972: 175]). Other extant communities in the region like the Ilchamus, Luyia, Luo and Gĩkũyũ make pots for their own use and also for the market. They also buy pots from other

The faunal data from Gogo Falls must be viewed in a similar manner for, live or dead animals as well as animal products like milk, skins and hide are also traded both within and between communities. It must also be borne in mind that economic activities like fishing, hunting, gathering, herding and probably fowling are practised by individuals and not necessarily by communities en masse. But, as we have argued above, the exploitation of the food resources in question must have been formalised by the community. The occurrence of many skeletal remains from the same animal, for instance, could be indicative of the acceptance of the animal as a food resource, while underrepresentation or absence of remains of another animal could be indicative of a food taboo. Like the ceramic data, the faunal data are not in themselves indicative of the large-scale 136

Karega-Mũnene

population movements which have hitherto been evoked to explain the Neolithic phenomenon. This observation is underscored by the fact that there was no exclusive association, for instance, between herding and Elmenteitan ware, which is presumed to have been exclusively associated with pastoralists. Evidence for the foraging activities, which are thought to have been practised by the so-called Kansyore populations, is also lacking. As a matter of fact, the economic transformation that would have characterised the faunal data had the presumed migration of pastoralists to the site occurred is absent. What is evident is an economy in which individuals exploited basically all the available ecological zones for purposes of their subsistence, a situation that could have been given rise to by their adaptive behaviour. No doubt, individual members of the community in question must have played a significant role in the adoption of the subsistence strategies in question. After all, the very first step to hunt a specific animal or to keep livestock was taken by individuals and not by the community en masse.

Schepartz 1988; Rightmire 1975, 1984; cf. Cavalli-Sforza et al. 1994). That food production appears to have been gradual at Gogo Falls and Enkapune ya Muto, suggests the adoption of animal husbandry as a subsistence strategy was a process of substituting food production for food collecting in an attempt to expand the menu.

Similarly, if domestic animals were initially obtained through exchange or trade or even stolen, the gathering of the information leading to that end must have been acquired and processed by individuals themselves. The process of acquisition could have been facilitated by the emergence of social hierachies among the groups who occupied Holocene East Africa and controlled or had direct access to obsidian raw material and, therefore, played a central role in obsidian exchange/trade networks. If correct, such groups could have acquired domestic animals as prestige or ritual goods or as a means of storing surpluses. Other groups who did not control or have direct access to obsidian quarries may have acquired domestic animals through the exchange/trade networks. In time, the keeping of domestic animals may have become an acceptable way of storing surpluses, hence the increase in the size of domestic animal herds in more recent faunal assemblages, as is the case at Gogo Falls and Enkapune ya Muto, for example (Marean 1992).

Contrary to the belief that soil conditions in East Africa are not conducive to the preservation of plant remains on archaeological sites, this study has, together with Robertshaw’s (1991) work at Gogo Falls, confirmed that such remains do exist and that they can be recovered. More importantly, although at the time of writing the present author is not in a position to evaluate the floral evidence from his 1989 excavations (since these are yet to be communicated by the palaeobotanist), it is now evident that plants were actually exploited in the region during the Holocene.

Summary and Conclusions This study has demonstrated that the subsistence strategies of the occupants of Gogo Falls involved hunting, fishing, herding, exploitation of plants and probably fowling. In terms of their relative importance, fishing, hunting and herding seem to have been of more or less the same importance throughout the occupation. It was not possible, however, to assess the importance of fowling, for reasons that have been discussed above. It will, therefore, suffice to emphasise here that the inference about fowling may be contentious because of the nature of the evidence at hand. As a matter of fact, it can only be confirmed or refuted through further field research.

In light of the evidence presented in this study, it is imperative that the ‘Pastoral Neolithic’ be renamed the Neolithic. The former term not only over-emphasises the importance of herding in Neolithic subsistence economies, but also precludes the exploitation of plants (whether domestic or wild) as well as hunting, fishing and fowling. Yet, as this study has demonstrated, animal husbandry was only one of several subsistence strategies that were adopted at Gogo Falls, which also appears to be the case at other East African Neolithic sites (e.g., Barthelme 1985; Gifford et al. 1980; Gifford-Gonzalez and Kimengich 1984; Marean 1992; Marshall and Stewart 1984; Mehlman 1989; Nelson and

Needless to add, this hypothesis presupposes continued occupation of East Africa by preNeolithic populations, a proposition that is borne out by artefactual and faunal evidence (e.g., Ambrose 1984; Barut 1994; Marean 1992) and by physical anthropological evidence (e.g.,

137

Holocene Foragers, Fishers and Herders of Western Kenya

Kimengich 1984; Onyango-Abuje Robertshaw et al. 1983).

1977b;

enunciated above. The model, it will be remembered, recognises the very crucial fact that individual humans have varied intellectual and technological capacities and that these differences are mirrored in the material remains of the societies to which the individuals belong. Additionally, it recognises that the communities who occupied Holocene East Africa were not isolated from each other. Rather, they interacted with each other, thus exchanging ideas as well as goods and services. There is little doubt that these factors are crucial to any attempts that may be aimed at improving knowledge about culture change and subsistence strategies during the Holocene. That is because, unlike large-scale migrations, they offer a better explanation to the archaeological data at our disposal. In fact, these factors do not only explain quite well why variations that have been observed in the artefactual and biological remains from Gogo Falls, for instance, existed, but also how they may have come about.

Examination of the chronological representation of the fauna exploited at Gogo Falls reveals there was no subsistence change in the Neolithic/Iron Age transition. This is not to say that such a change did not occur during the transition from the pre-Neolithic (food collection) to the Neolithic (food production). Although the evidence for this transition is not available at the site, it is most likely that the transition was marked by a significant subsistence change. The change itself was probably adaptive rather than transformational, just as was the case in the Neolithic/Iron Age transition. Thus, food production in Holocene East Africa probably arose from the endeavour of the concerned peoples to cope with their ever-changing cultural, biological and physical environment, rather than as a result of large scale population movements. Although the earliest known presence of domestic animals at Gogo Falls is now securely dated, the chronometric evidence at hand does not tell us anything about the direction from which the animals may have reached the site. Rather, they indicate that the animals were present in the Lake Victoria basin by mid fourth millennium BP. Drawing arrows on a map linking the Gogo Falls dates with dates from other Neolithic sites which, as we have noted above, are concentrated in a narrow corridor defined by the Rift Valley, is not likely to reveal the direction(s) from which the animals came.

While investigating the likely coexistence of the wares we should avoid describing the pottery that we may recover from a given site as being predominantly of one ware or the other. This practice should be replaced by documentation of the chronological and spatial representation of all the identified wares. In addition, the wares should be treated as tools of analysis or as ‘styles’, not as hallmarks for specific ethnic or linguistic groups and subsistence strategies. Afterall, it is new approaches and ideas – not old data, approaches and untestable hypotheses – that propel us forward in our search for knowledge. As Daniels and Hammer (1992: 145) have advised, “we must... keep testing ideas [and approaches] and be willing to change as new data warrants [sic] change.”

Nevertheless, there is need to investigate the nature of the transitions from the foraging economies of the pre-Neolithic period to the food-producing economies of the Neolithic and Iron Age periods. This exercise could be aided by the uncovering of horizons that are exclusively pre-Neolithic, Neolithic and Iron Age at Gogo Falls and/or at any other sites in the region. Further field research should, therefore, be encouraged in this direction. Recovery of faunal and floral remains should be paramount in such research. Attempts should also be made to recover large faunal samples since these are likely to be more representative than small ones.

As this study has argued, the view that past ethnic or linguistic groups can be defined through material culture is not only potentially misleading, but also inadequate. If anything, it can only apply to insular communities and, there is no evidence to suggest that the occupants of Holocene East Africa belonged to that category. Thus, it must be appreciated that similarities and differences in material culture, like the ones that have been used to define the wares, are reflections of the dynamic relationships that existed between the people responsible for its production and consumption. Indeed, like any other artefact in any given society, pottery is “an item [that] operat[es] in an economic and social

The apparent coexistence of Kansyore, Elmenteitan, Akira and Urewe wares at Gogo Falls and, indeed, in other parts of East Africa should be investigated further in order to evaluate the validity of the interpretive model 138

Karega-Mũnene

environment” (Barnett 1995: 79). It is, therefore, pertinent that due attention be paid to its production and distribution for better understanding of the effects that these contextual variables may have on ceramic patterning in the region. There is hardly any reason to assume that production and consumption of a given ware may not have taken place among different ethnic or linguistic groups living contiguous to each other or even far apart. Couching explanations of archaeological data in terms of physical migrations, as has hitherto been the case, does not only minimise the contribution of contact and exchange/trade to culture change and adaptation (in terms of subsistence strategies), but also denies the groups concerned the dynamism which appears to have characterised their relationships with each other and with their environment.

This study has demonstrated that knowledge about past subsistence activities can be greatly improved by studying biological and cultural remains together. This approach does not only enable us to appreciate the complexity of the relevant data, but also the need to avoid simplistic interpretations of those data. It is hoped that the approach will be adopted in future research in East Africa in order to obtain as much information as possible about the cultural and economic activities of the people concerned. Lastly, it must be emphasised here that although employment of historical linguistics in East African archaeology gives archaeological explanation a narrative, from an archaeological point of view over-reliance on linguistic hypotheses obfuscates investigations of culture change and subsistence economies in the region.

139

APPENDIX I

FREQUENCY OF THE WARES BY SQUARES AND SPITS

Square 1, Spit:

Akira

Elmenteitan

Kansyore

Urewe

Total

0-10 10-20 20-30 30-40 40-50

0 1 0 0 0 1

9 12 8 1 3 33

6 8 2 4 1 21

12 5 5 0 1 23

27 26 15 5 5 78

0-10 10-20 20-30 30-40 40-50

0 0 0 0 0 0

20 17 6 3 1 47

10 5 2 2 2 21

13 8 3 3 1 28

43 30 11 8 4 96

0-10 10-20 20-30 30-40 40-50 50-60 60-70

1 1 1 0 0 0 0 3

3 14 8 3 3 3 1 35

7 6 4 4 3 13 1 38

8 6 4 2 0 3 0 23

19 27 17 9 6 19 2 99

0-10 10-20 20-30 30-40 40-50 50-60

0 0 0 1 0 0 1

3 3 2 0 1 3 12

1 1 10 6 3 1 22

5 2 4 0 1 1 13

9 6 16 7 5 5 48

0-10 10-20 20-30 30-40 40-50 50-60

0 0 0 0 0 0 0

25 14 8 39 1 14 101

16 15 12 13 5 2 63

19 12 7 22 5 5 70

60 41 27 74 11 21 234

0-10 10-20 20-30 30-40 40-50 50-60 60-70

0 0 0 0 0 0 0 0

22 12 7 15 4 3 0 63

6 10 7 6 13 6 2 50

9 13 5 4 3 3 0 37

37 35 19 25 20 12 2 150

0-10 10-20 20-30 30-40 40-50 50-60

1 0 0 0 0 0 1

18 13 5 11 6 1 54

9 7 3 7 13 1 40

7 11 4 7 3 1 33

35 31 12 25 22 3 128

0-10 10-20 20-30 30-40 40-50 50-60

0 0 0 1 0 0 1

20 15 5 9 3 7 59

9 8 4 15 5 11 52

15 5 5 5 5 5 40

44 28 14 30 13 23 152

0-10 10-20 20-30 30-40 40-50 50-60

0 0 0 0 0 0 0

29 13 8 4 3 0 57

17 9 12 11 8 7 64

17 5 2 9 1 2 36

63 27 22 24 12 9 157

Total Square 2, Spit:

Total Square 3, Spit:

Total Square 4, Spit:

Total Square 5, Spit:

Total Square 6, Spit:

Total Square 7, Spit:

Total Square 8, Spit:

Total Square 9, Spit:

Total

140

Karega-Mũnene

Square 10, Spit:

Akira

Elmenteitan

Kansyore

Urewe

Total

0-10 10-20 20-30 30-40 40-50 50-60 60-70

2 0 0 1 0 0 0 3

9 14 4 15 33 2 0 77

13 8 7 6 18 4 2 58

17 10 2 17 25 5 3 79

41 32 13 39 76 11 5 217

0-10 10-20 20-30 30-40 40-50 50-60 60-70

0 0 1 0 0 0 1 2

28 10 3 0 5 0 1 47

3 2 8 8 0 9 0 30

8 2 18 6 5 4 2 45

39 14 30 14 10 13 4 124

0-10 10-20 20-30 30-40 40-50 50-60 60-70

0 0 1 0 0 0 0 1

8 12 6 2 7 0 0 35

16 9 9 11 10 1 4 60

32 10 5 4 5 1 1 58

56 31 21 17 22 2 5 154

0-10 10-20 20-30 30-40 40-50 50-60 60-70

0 0 0 1 0 0 0 1

39 21 6 3 4 3 0 76

13 5 4 8 5 8 4 47

25 17 10 8 6 1 2 69

77 43 20 20 15 12 6 193

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

0 0 0 0 0 0 0 0 0 0

25 25 9 10 3 2 1 0 1 76

16 7 6 12 3 6 5 3 0 58

12 21 8 13 6 1 3 2 0 66

53 53 23 35 12 9 9 5 1 200

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80

0 0 0 0 0 0 0 0 0

3 0 11 10 3 1 2 2 32

9 1 11 14 10 7 10 4 66

12 0 5 5 7 3 1 0 33

24 1 27 29 20 11 13 6 131

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

0 0 0 0 0 0 0 0 0 0

15 27 9 11 6 2 4 8 4 86

14 5 6 5 6 1 7 5 1 50

8 8 3 11 7 1 5 4 1 48

37 40 18 27 19 4 16 17 6 184

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

0 0 1 0 2 1 0 0 0 4

42 16 11 10 16 5 13 4 2 119

18 2 5 4 9 13 18 3 1 73

15 12 5 12 5 4 11 3 2 69

75 30 22 26 32 23 42 10 5 265

Total Square 11, Spit:

Total Square 12, Spit:

Total Square 13, Spit:

Total Square 14, Spit:

Total Square 15, Spit:

Total Square 16, Spit:

Total Square 17, Spit:

Total

141

Holocene Foragers, Fishers and Herders of Western Kenya

Square 18, Spit:

Akira

Elmenteitan

Kansyore

Urewe

Total

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

0 1 0 1 1 0 0 0 0 3

31 30 23 19 10 23 8 7 5 156

23 18 13 6 8 18 12 7 6 111

13 16 15 14 15 18 6 5 4 106

67 65 51 40 34 59 26 19 15 376

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100

0 0 0 0 0 0 0 0 0 0 0

30 20 13 2 10 0 6 4 2 1 88

9 7 1 4 7 0 9 5 8 0 50

15 18 7 7 5 0 4 2 3 2 63

54 45 21 13 22 0 19 11 13 3 201

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100

0 0 0 1 0 0 0 0 0 0 1

30 13 12 29 12 8 6 22 3 1 136

13 4 8 11 6 11 5 15 7 2 82

22 9 9 20 19 6 3 17 1 2 108

65 26 29 61 37 25 14 54 11 5 327

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

1 0 1 0 2 0 0 0 0 4

24 6 13 15 5 4 6 2 0 75

3 2 1 1 0 5 9 2 0 23

10 1 10 11 6 10 0 1 1 50

38 9 25 27 13 19 15 5 1 152

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

1 0 0 0 0 0 1 0 0 2

45 14 18 12 4 0 6 9 0 108

10 2 5 5 3 1 4 7 0 37

12 12 4 7 4 0 8 13 0 60

68 28 27 24 11 1 19 29 0 207

0-10 10-20 20-30 30-40 40-50 50-60 60-70

0 1 0 0 0 0 0 1

31 32 23 9 5 1 7 108

7 6 6 1 4 3 1 28

21 18 12 6 1 0 2 60

59 57 41 16 10 4 10 197

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80

0 0 0 0 0 0 0 0 0

35 17 22 17 9 1 0 4 105

7 4 3 4 6 8 4 1 37

16 10 5 10 2 1 1 2 47

58 31 30 31 17 10 5 7 189

Total Square 19, Spit:

Total Square 20, Spit:

Total Square 21, Spit:

Total Square 22, Spit:

Total Square 23, Spit:

Total Square 24, Spit:

Total

142

Karega-Mũnene

Square 25, Spit:

Akira

Elmenteitan

Kansyore

Urewe

Total

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200

0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1

29 11 7 13 13 10 17 0 0 4 20 20 10 3 0 1 1 0 0 0 159

1 3 0 0 2 1 2 0 0 2 2 4 12 18 10 11 14 11 5 1 99

9 7 3 5 5 1 4 0 0 0 2 2 4 2 1 0 1 1 1 2 50

39 21 10 18 20 12 23 0 0 6 24 26 27 23 11 12 16 12 6 3 309

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210

1 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 3

35 20 13 9 7 2 18 14 9 10 8 5 3 5 5 0 0 0 0 0 0 163

8 0 1 2 0 2 3 2 8 1 4 3 2 5 8 7 11 6 8 1 1 83

14 9 9 4 3 2 4 2 3 1 2 2 2 3 0 0 0 0 0 0 0 60

58 29 24 15 10 6 25 19 20 12 14 10 7 13 13 7 11 6 8 1 1 309

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210 210-220 220-230

0 1 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3

12 23 7 15 9 21 19 12 18 11 0 0 12 11 5 3 5 0 0 0 0 0 0 183

3 2 0 1 0 1 0 3 0 2 0 0 3 6 4 5 12 18 12 6 2 0 4 84

10 17 3 4 2 2 2 2 0 1 0 0 2 1 3 1 3 2 2 0 0 0 0 57

25 43 10 20 11 24 23 17 18 14 0 0 17 18 12 9 20 20 14 6 2 0 4 327

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

6 5 24 9 4 11 11 11 6 11 11 4 6 5 3 1 128

6 2 3 9 3 3 2 2 0 4 6 3 6 6 6 1 62

4 4 9 4 11 13 5 1 0 1 2 1 0 1 0 0 56

16 11 36 22 18 27 18 14 6 16 19 8 12 12 9 2 246

Total Square 26, Spit:

Total Square 27, Spit:

Total Square 28, Spit:

Total

143

Holocene Foragers, Fishers and Herders of Western Kenya

Square 29, Spit:

Akira

Elmenteitan

Kansyore

Urewe

Total

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150

0 0 0 1 1 0 0 0 1 1 0 0 0 0 0 4

3 13 11 13 12 11 12 10 17 5 9 0 0 0 0 116

2 3 4 1 4 1 1 2 7 8 4 16 6 10 4 73

4 3 5 6 6 8 7 5 0 0 4 0 0 0 0 48

9 19 20 21 23 20 20 17 25 14 17 16 6 10 4 241

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150

1 2 0 0 1 1 0 0 0 0 0 0 0 0 0 5

6 19 12 2 12 9 7 15 7 10 5 1 0 0 0 105

4 3 4 3 3 4 0 4 3 7 8 11 26 9 5 94

4 16 8 7 14 13 3 3 2 0 3 2 0 0 0 75

15 40 24 12 30 27 10 22 12 17 16 14 26 9 5 279

45

2639

1676

1610

5970

Total Square 30, Spit:

Total Grand Total

144

APPENDIX II

COMPOSITION OF THE PREVIOUSLY STUDIED FAUNAL SAMPLE*

Trench II III III Total

*

Identifiable

Non-identifiable

Total

Analyst/Reference

1501 612 2273 4386

0 0 33397 33397

1501 612 35670 37783

Stewart (1991) Marshall (1986, 1991) Karega-Mũnene (1986, 1987)

The sample from Trench II consisted of fish elements only and was studied by Stewart’s (1991). The sample from Trench III studied by Marshall (1986, 1991) consisted of mammalian teeth only. The sample from Trench III studied by Karega-Mũnene (1986, 1987) consisted of cranial and postcranial mammalian and fish elements.

145

APPENDIX III

FREQUENCY OF THE 1989 FAUNAL COLLECTION BY SQUARES AND SPITS

Square 1, Spit:

Identifiable

Non-identifiable

Total

0-10 10-20 20-30 30-40 40-50

0 19 13 1 0 33

86 80 36 29 0 231

86 99 49 30 0 264

0-10 10-20 20-30 30-40 40-50

2 29 2 0 0 33

152 262 94 42 10 560

154 291 96 42 10 593

0-10 10-20 20-30 30-40 40-50 50-60 60-70

0 0 4 1 1 1 0 7

114 68 206 40 24 0 0 452

114 68 210 41 25 1 0 459

0-10 10-20 20-30 30-40 40-50 50-60

129 6 12 0 0 0 147

222 64 113 28 12 0 439

351 70 125 28 12 0 586

0-10 10-20 20-30 30-40 40-50 50-60

9 0 131 24 5 0 169

253 218 0 655 24 15 1165

262 218 131 679 29 15 1334

0-10 10-20 20-30 30-40 40-50 50-60 60-70

25 33 17 11 1 10 0 97

92 59 127 139 45 36 0 498

117 92 144 150 46 46 0 595

0-10 10-20 20-30 30-40 40-50 50-60

9 29 13 51 2 0 104

18 102 88 201 114 0 523

27 131 101 252 116 0 627

0-10 10-20 20-30 30-40 40-50 50-60

35 9 32 2 0 1 79

186 103 38 62 0 37 426

221 112 70 64 0 38 505

0-10 10-20 20-30 30-40 40-50 50-60

0 8 204 21 0 0 233

150 152 2 106 43 38 491

150 160 206 127 43 38 724

Total Square 2, Spit:

Total Square 3, Spit:

Total Square 4, Spit:

Total Square 5, Spit:

Total Square 6, Spit:

Total Square 7, Spit:

Total Square 8, Spit:

Total Square 9, Spit:

Total

146

Karega-Mũnene

Square 10, Spit:

Identifiable

Non-identifiable

Total

0-10 10-20 20-30 30-40 40-50 50-60 60-70

1 46 7 61 14 0 2 131

207 146 137 181 213 117 16 1017

208 192 144 242 227 117 18 1148

0-10 10-20 20-30 30-40 40-50 50-60 60-70

0 76 40 4 1 0 1 122

0 306 197 222 0 26 21 772

0 382 237 226 1 26 22 894

0-10 10-20 20-30 30-40 40-50 50-60 60-70

4 1 2 23 19 0 0 49

0 182 124 323 148 8 14 799

4 183 126 346 167 8 14 848

0-10 10-20 20-30 30-40 40-50 50-60 60-70

18 20 0 165 18 17 4 242

212 0 130 0 189 101 167 799

230 20 130 165 207 118 171 1041

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

44 41 37 3 56 37 3 0 0 221

94 250 256 273 295 128 14 20 113 1443

138 291 293 276 351 165 17 20 113 1664

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80

24 0 13 4 28 0 3 0 72

301 10 81 135 141 84 44 21 817

325 10 94 139 169 84 47 21 889

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

0 13 13 4 25 8 0 4 1 68

687 156 105 280 189 111 0 89 23 1640

687 169 118 284 214 119 0 93 24 1708

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

118 18 14 14 52 64 21 5 2 308

906 318 305 259 500 239 814 305 120 3766

1024 336 319 273 552 303 835 310 122 4074

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

78 50 20 57 2 27 4 28 4 270

287 254 211 334 166 186 64 92 22 1616

365 304 231 391 168 213 68 120 26 1886

Total Square 11, Spit:

Total Square 12, Spit:

Total Square 13, Spit:

Total Square 14, Spit:

Total Square 15, Spit:

Total Square 16, Spit:

Total Square 17, Spit:

Total Square 18, Spit:

Total

147

Holocene Foragers, Fishers and Herders of Western Kenya

Square 19, Spit:

Identifiable

Non-identifiable

Total

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100

61 33 17 22 28 20 28 5 1 2 217

458 645 253 208 390 216 97 71 22 34 2394

519 678 270 230 418 236 125 76 23 36 2611

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100

116 42 40 34 45 9 0 31 2 0 319

414 363 403 390 332 218 8 319 239 98 2784

530 405 443 424 377 227 8 350 241 98 3103

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

47 0 6 11 82 13 4 7 2 172

466 20 170 192 301 201 50 25 18 1443

513 20 176 203 383 214 54 32 20 1615

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

21 15 67 15 16 0 4 4 0 142

257 225 296 333 54 0 176 116 13 1470

278 240 363 348 70 0 180 120 13 1612

0-10 10-20 20-30 30-40 40-50 50-60 60-70

54 59 85 6 7 5 4 220

320 185 269 100 57 59 36 1026

374 244 354 106 64 64 40 1246

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80

33 20 87 21 28 59 17 6 271

231 188 388 190 252 76 75 0 1400

264 208 475 211 280 135 92 6 1671

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200

212 135 165 1141 1081 679 179 0 0 136 206 132 281 14 6 4 9 8 1 44 4433

38 34 70 222 410 497 261 0 0 538 635 624 13 118 95 120 127 138 10 93 4043

250 169 235 1363 1491 1176 440 0 0 674 841 756 294 132 101 124 136 146 11 137 8476

Total Square 20, Spit:

Total Square 21, Spit:

Total Square 22, Spit:

Total Square 23, Spit:

Total Square 24, Spit:

Total Square 25, Spit:

Total

148

Karega-Mũnene

Square 26, Spit:

Identifiable

Non-identifiable

Total

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210

51 115 50 113 101 55 171 34 68 50 29 47 46 40 20 12 1 19 8 19 25 1074

199 211 262 407 386 315 727 160 136 106 110 117 77 79 66 47 0 54 73 64 76 3672

250 326 312 520 487 370 898 194 204 156 139 164 123 119 86 59 1 73 81 83 101 4746

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210 210-220 220-230

75 17 71 135 78 79 89 117 115 52 0 0 133 63 127 0 3 1 118 32 54 15 75 1449

183 336 0 496 172 327 304 646 291 212 0 0 978 544 458 0 0 0 0 222 611 238 154 6172

258 353 71 631 250 406 393 663 406 264 0 0 1111 607 585 0 3 1 118 254 665 253 229 7621

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160

3 23 0 8 10 25 21 29 60 19 33 48 20 68 10 1 378

51 59 114 67 82 0 320 214 199 227 307 132 130 300 105 50 2357

54 82 114 75 92 25 341 243 259 246 340 180 150 368 115 51 2735

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150

0 4 0 25 127 15 21 14 45 36 218 41 65 1 4 616

43 84 240 117 6 203 134 97 192 187 0 69 147 82 63 1664

43 88 240 142 133 218 155 111 237 223 218 110 212 83 67 2280

Total Square 27, Spit:

Total Square 28, Spit:

Total Square 29, Spit:

Total

149

Holocene Foragers, Fishers and Herders of Western Kenya

Square 30, Spit:

Total Grand Total

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150

Identifiable

Non-identifiable

Total

0 7 8 2 22 311 36 641 136 32 33 40 2 1 3 1274

32 153 96 57 163 29 119 0 416 156 127 24 21 6 17 1416

32 160 104 59 185 340 155 641 552 188 160 64 23 7 20 2690

12950

47295

60245

150

APPENDIX IV

FREQUENCY OF THE 1983 FAUNAL COLLECTION BY TRENCHES AND SPITS

Trench I, Spit:

Identifiable

Non-identifiable

Total

OB M1 M2 M3 M4 M5 M6 M7 M8 M9 M10

220 154 198 179 111 85 91 126 151 174 168 1657

8 8 25 7 16 3 7 8 4 12 16 114

228 162 223 186 127 88 98 134 155 186 184 1771

OB A1 A2 A3 A4 A5 FA1 FA2 FA3 FA4 FA5 FA6 FA7 FA8 CL1 BL1 BL2 BL3 BL4 BL5 BL6 BL7 BL8

47 538 649 605 634 174 334 498 280 311 75 82 128 73 80 179 158 128 154 122 10 66 30 5355

6 143 46 36 30 11 20 3 32 17 3 7 0 6 3 27 8 5 9 16 1 4 0 433

53 681 695 641 664 185 354 501 312 328 78 89 128 79 83 206 166 133 163 138 11 70 30 5788

OB S1 S2 S3 S4 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14

0 3 26 0 2 10 9 1 2 6 13 11 60 15 7 19 0 0 0 184

0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 2

0 3 26 0 2 10 9 1 2 6 13 11 62 15 7 19 0 0 0 186

Total Trench II, Spit:

Total Trench III, Spit:

Total

151

Holocene Foragers, Fishers and Herders of Western Kenya

Trench IV, Spit :

Identifiable

Non-identifiable

Total

15-25 25-35 35-45 45-55 55-65 65-75 75-85 85-95 95-105 105-115 115-125 125-135 135-145 145-155

79 5 13 4 12 9 15 29 26 20 57 104 40 14 427

0 0 0 0 0 0 0 0 0 0 5 0 0 0 5

79 5 13 4 12 9 15 29 26 20 62 104 40 14 432

TS R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R12

47 353 425 434 478 732 876 857 797 460 317 167 45 5988

0 47 28 46 30 42 53 58 61 48 48 35 0 496

47 400 453 480 508 774 929 915 858 508 365 202 45 6484

13611

1050

14661

Total Trench V, Spit:

Total Grand Total

152

APPENDIX V

POTTERY AND TAXONOMIC REPRESENTATION BY SQUARES AND SPITS Pottery

Fauna

Akira

Elmenteitan

Kansyore

Urewe

Domestic

Wild

Fish

Square 1, Spit:

0-10 10-20 20-30 30-40 40-50

x -

x x x x x

x x x x x

x x x x

x -

-

-

Square 2, Spit:

0-10 10-20 20-30 30-40 40-50

-

x x x x x

x x x x x

x x x x x

-

-

-

Square 3, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70

x x x -

x x x x x x x

x x x x x x x

x x x x x -

-

-

-

Square 4, Spit:

0-10 10-20 20-30 30-40 40-50 50-60

x -

x x x x x

x x x x x x

x x x x x

-

-

x -

Square 5, Spit:

0-10 10-20 20-30 30-40 40-50 50-60

-

x x x x x x

x x x x x x

x x x x x x

-

x -

-

Square 6, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70

-

x x x x x x -

x x x x x x x

x x x x x x -

-

-

-

Square 7, Spit:

0-10 10-20 20-30 30-40 40-50 50-60

x -

x x x x x x

x x x x x x

x x x x x x

-

x -

x -

Square 8, Spit:

0-10 10-20 20-30 30-40 40-50 50-60

x -

x x x x x x

x x x x x x

x x x x x x

x -

-

x -

Square 9, Spit:

0-10 10-20 20-30 30-40 40-50 50-60

-

x x x x x -

x x x x x x

x x x x x x

-

x -

-

Square 10, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70

x x -

x x x x x x -

x x x x x x x

x x x x x x x

-

x -

-

153

Holocene Foragers, Fishers and Herders of Western Kenya

Pottery

Fauna

Akira

Elmenteitan

Kansyore

Urewe

Domestic

Wild

Fish

Square 11, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70

x x

x x x x x

x x x x x -

x x x x x x x

-

-

x -

Square 12, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70

x -

x x x x x -

x x x x x x x

x x x x x x x

x -

x -

-

Square 13, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70

x -

x x x x x x -

x x x x x x x

x x x x x x x

x -

-

-

Square 14, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

-

x x x x x x x x

x x x x x x x x -

x x x x x x x x -

x x -

x x -

-

Square 15, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80

-

x x x x x x x

x x x x x x x x

x x x x x x x -

-

x x -

-

Square 16, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

-

x x x x x x x x x

x x x x x x x x x

x x x x x x x x x

x x x -

-

x -

Square 17, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

x x x -

x x x x x x x x x

x x x x x x x x x

x x x x x x x x x

x x -

-

x x -

Square 18, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

x x x -

x x x x x x x x x

x x x x x x x x x

x x x x x x x x x

x -

x x x -

-

Square 19, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100

-

x x x x x x x x x x

x x x x x x x x x -

x x x x x x x x x x

x -

x -

-

154

Karega-Mũnene

Pottery

Fauna

Akira

Elmenteitan

Kansyore

Urewe

Domestic

Wild

Fish

Square 20, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100

x -

x x x x x x x x x x

x x x x x x x x x x

x x x x x x x x x x

x -

x x -

x -

Square 21, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90

x x x -

x x x x x x x x -

x x x x x x x -

x x x x x x x x

x -

x x -

x -

Square 22, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80

x x -

x x x x x x x

x x x x x x x x

x x x x x x x

x -

x x x

x -

Square 23, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70

x -

x x x x x x x

x x x x x x x

x x x x x x

-

x -

x -

Square 24, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80

-

x x x x x x x

x x x x x x x x

x x x x x x x x

x -

x -

x x -

Square 25, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200

x -

x x x x x x x x x x x x x x -

x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x

x x x x x x x x x -

x x x x x x x x x x -

x x x x x x x -

Square 26, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210

x x -

x x x x x x x x x x x x x x x -

x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x -

x x x x x x x x x x x -

x x x x x x x x x x x x x x x -

x x x x x x -

155

Holocene Foragers, Fishers and Herders of Western Kenya

Pottery

Fauna

Akira

Elmenteitan

Kansyore

Urewe

Domestic

Wild

Fish

Square 26, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210

x x -

x x x x x x x x x x x x x x x -

x x x x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x -

x x x x x x x x x x x -

x x x x x x x x x x x x x x x -

x x x x x x -

Square 27, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160 160-170 170-180 180-190 190-200 200-210 210-220 220-230

x x -

X x x x x x x x x x x x x x x -

x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x x -

x x x x x x x x x x x x

x x x x x x x x x x x x x x x x x x

x x x x x x x x x

Square 28, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150 150-160

-

x x x x x x x x x x x x x x x x

x x x x x x x x x x x x x x x

x x x x x x x x x x x x -

x x x x x x x x x -

x x x x x x x x x x x

x x x x -

Square 29, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150

x x x x -

x x x x x x x x x x x -

x x x x x x x x x x x x x x x

x x x x x x x x x -

x x x x x x x

x x x x x x -

x x -

156

Karega-Mũnene

Pottery

Square 30, Spit:

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150

Fauna

Akira

Elmenteitan

Kansyore

Urewe

Domestic

Wild

Fish

x x x x -

x x x x x x x x x x x x -

x x x x x x x x x x x x x x

x x x x x x x x x x x -

x x x x x x x x x -

x x x x x x x -

x x x x -

x = Present, - = absent.

157

REFERENCES CITED

Barut, S. 1994 Middle and Later Stone Age technology and land use in East African savannas. African Archaeological Review 12: 43-72.

Alexander, J. 1970 The Directing of Archaeological Excavations. London: John Baker. Ambrose, S.H. 1982 Archaeological and linguistic reconstructions of history in East Africa. In C. Ehret and M. Posnansky (eds.), The Archaeological and Linguistic Reconstruction of African History, Berkeley: University of California Press, pp. 104157.

Beadle, L.C. 1974 The Inland Waters of Tropical Africa: an introduction to tropical limnology. London: Longman. Behrensmeyer, A.K. 1978 Taphonomic and ecologic information from bone weathering. Paleobiology 4: 150-162.

Ambrose, S.H. 1984 The introduction of pastoral adaptations to the highlands of East Africa. In J.D. Clark and S.A. Brandt (eds.), From Hunters to Farmers: the causes and consequences of food production in Africa, Berkeley: University of California Press, pp. 212239.

Beuning, K.R.M. 1999 A re-evaluation of the Late Glacial and early-Holocene vegetation history of the Lake Victoria region, East Africa. Palaeoecology of Africa 26: 115-136. Binford, L.R. 1978 Nunamiut Ethnoarchaeology. New York: Academic Press.

Ambrose, S.H. 1985 Excavations at Masai Gorge Rockshelter, Naivasha. Azania 20: 29-67.

Binford, L.R. 1981 Bones: ancient men and modern myths. New York: Academic Press.

Barbour, J. and S. Wandibba (eds.) 1989 Kenya Pots and Potters. Nairobi: Oxford University Press.

Binford, L.R. 1984 Faunal Remains from Klasies River Mouth. Orlando: Academic Press.

Barendsen, G.W., E.S. Deevy and L.J. Gralensk 1957 Yale natural radiocarbon measurements III. Science 126: 908-919.

Bower, J.R.F. 1973a Seronera: excavations at a Stone Bowl Site in the Serengeti National Park, Tanzania. Azania 8: 71-104.

Barnett, W.K. 1995 Putting the pot before the horse: earliest ceramics and the Neolithic transition in the western Mediterranean. In W.K. Barnett and J.W. Hoopes (eds.), The Emergence of pottery: technology and innovation in ancient societies, Washington and London: Smithsonian Institution Press, pp. 79-88.

Bower, J.R.F. 1973b Early pottery and other finds from Kisii District, western Kenya. Azania 8: 131140. Bower, J.R.F. 1984 Subsistence-settlement systems of the Pastoral Neolithic in East Africa. In L. Krzyzaniak and M. Kobusiewicz (eds.), Origin and Early Development of FoodProducing Cultures in North-Eastern Africa, Poznan: Polish Academy of Sciences, pp. 473-480.

Barthelme, J.W. 1984 Early evidence for animal domestication in eastern Africa. In J.D. Clark and S.A. Brandt (eds.), From Hunters to Farmers: the causes and consequences of food production in Africa, Berkeley: University of California Press, pp. 200-205.

Bower, J. R. F. 1991 The Pastoral Neolithic of East Africa. Journal of World Prehistory 5: 49-82.

Barthelme, J.W. 1985 Fisher-Hunters and Neolithic Pastoralists in East Turkana, Kenya. Oxford: BAR International Series 254. 158

Karega-Mũnene Catt, J.A. 1988 Quaternary Geology for Scientists and Engineers. Chichester: Ellis Horwood.

Bower, J. R. F. and C. M. Nelson 1978 Early pottery and pastoral cultures of the central Rift Valley, Kenya. Man n.s. 13: 554-566.

Cavalli-Sforza, L.L., P. Menozzi and A. Piazza 1994 The History and Geography of Human Genes. Princeton: Princeton University Press.

Bower, J.R.F., C.M. Nelson, A.F. Waibel and S. Wandibba 1977 The University of Massachusetts’ Later Stone Age/Pastoral ‘Neolithic’ comparative study in central Kenya: an overview. Azania 12: 119-146.

Chaplin, R.E. 1971 The Study of Animal Bones from Archaeological Sites. London and New York: Seminar Press.

Brachi, R.M. 1960 Excavations of a rock-shelter at Hippo Bay, Entebbe. Uganda Journal 24: 62-70.

Chapman, S. 1967 Kantsyore Island. Azania 2: 165-191.

Brain, C.K. 1981 The Hunters or the Hunted?: an introduction to African cave taphonomy. Chicago: University of Chicago Press.

Childe, V.G. 1934 New Light on the Most Ancient East: the oriental prelude to European prehistory. London: Kegan Paul, Trench, Trubner and Co.

Briggs, D. and P. Smithson 1985 Fundamentals of Physical Geography. London: Hutchinson.

Childe, V.G. 1951 Social Evolution. London: Watts.

Brown, J. 1966 The excavations of a group of burial mounds in Ilkek near Gilgil, Kenya. Azania 1: 59-77.

Clark, J.D. 1962 The spread of food production in subSaharan Africa. Journal of African History 3: 211-228.

Brown, J. 1969 Some polished axes from East Africa. Azania 4: 160-166.

Clark, J.D. 1964 The prehistoric origins of African culture. Journal of African History 5: 161-183.

Bunting, B.T. 1965 The Geography of Soil. London: Hutchinson.

Clark, J.D. 1967a A record of early agriculture and metallurgy in Africa from archaeological sources. In C. Gabel and N. R. Bennett (eds.), Reconstructing African Culture, Boston: Boston University Press, pp. 1-24.

Burkitt, C.M. 1925 Prehistory: a study of early cultures in Europe and the Mediterranean basin, second edition. Cambridge: Cambridge University Press.

Clark, J.D. 1967b The problem of Neolithic culture in Subsaharan Africa. In W. W. Bishop and J. D. Clark (eds.), Background to Evolution in Africa, Chicago: University of Chicago Press, pp. 601-627.

Butterman, J.M. n.d. Kanyamkago Economic History Texts. Unpublished Text, Kenya National Archives, Nairobi. Butzer, K.W., G.Ll. Isaac, J. L. Richardson and C. Washbourn-Kamau 1972

Clark, J.D. 1976 Prehistoric populations and pressures favoring plant domestication in Africa. In J.R. Harlan, J.M.J. de Wet and A.B.L. Stemler (eds.), Origins of African Plant Domestication, The Hague: Mouton, pp. 67-105.

Radiocarbon dating of East African lake levels. Science 175: 1069-1076.

Casteel, R.W. 1977 Characterization of faunal assemblages and the minimum number of individuals determined from paired elements: continuing problems in archaeology. Journal of Archaeological Science 4: 125-133.

Clark, J.D. 1980 Human populations and cultural adaptations in the Sahara and Nile during prehistoric times. In M.A.J. Williams and H. Faure (eds.), The Sahara and the Nile: Quaternary environments and prehistoric occupation in northern Africa, Rotterdam: Balkema, pp. 527-582.

Casteel, R.W. and D.K. Grayson 1977 Terminological problems in quantitative faunal analysis. World Archaeology 9: 235-242.

159

Holocene Foragers, Fishers and Herders of Western Kenya Clist, B. 1987

Daniels, R.B. and R.D. Hammer 1992 Soil Geomorphology. New York: John Wiley & Sons.

A critical reappraisal of the chronological framework of the Early Urewe Iron Age Industry. Muntu 6: 35-62.

De Barros, P. 1990 Changing paradigms, goals and methods in the archaeology of Francophone West Africa. In P. Robertshaw (ed.), A History of African Archaeology, London: James Currey, pp. 155-172.

Clutton-Brock, J. 1989 Cattle in ancient North Africa. In J. Clutton-Brock (ed.), The Walking Larder: patterns of domestication, pastoralism, and predation, London: Unwin Hyman. Pp. 200-206.

Dobbs, C.M. 1914 A stone bowl and ring discovered in Sotik. Journal of the East African and Uganda Natural History Society 4: 145-146.Dobbs, C.M.

Coetzee, J.A. 1964 Evidence for a considerable depression of the vegetation belts during the Upper Pleistocene on the East African mountains. Nature 204: 564-566.

1918

Coetzee, J.A. 1967 Pollen analytical studies in East and southern Africa. Palaeoecology of Africa 3: 1-146.

Dunnell, R.C. 1980 Evolutionary theory and archaeology. Advances in Archaeological Method and Theory 3: 35-99

Cohen, M. 1970 A reassessment of the stone bowl cultures of the Rift Valley, Kenya. Azania 5: 27-38.

Dyson-Hudson, N. and R. Dyson-Hudson 1982 The structure of East African herds and the future of East African herders. Development and Change 13: 213-238.

Cohen, M. 1972 Deloraine Farm: a new type of pottery. Azania 7: 161-167. Cole, S. 1963

A stone bowl in Sotik. Journal of the East African and Uganda Natural History Society 6: 265-267.

Ehret, C. 1967 Cattle-keeping and milking in eastern and southern African history: the linguistic evidence. Journal of African History 8: 117.

The Prehistory of East Africa, second edition. London: Weidenfield and Nicolson.

Collett, D.P. 1982 Models of the spread of the Early Iron Age. In C. Ehret and M. Posnansky (eds.), The Archaeological and Linguistic Reconstruction of African History, Berkeley: University of California Press, pp. 182 -198.

Ehret, C. 1968 Sheep and central Sudanic peoples in southern Africa. Journal of African History 9: 213-221. Ehret, C. 1972 Bantu origins and history: critique and interpretation. Transafrican Journal of History 2: 1-9.

Collett, D.P. and P.T. Robertshaw 1980 Early Iron Age and Kansyore pottery: finds from Gogo Falls, South Nyanza. Azania 15: 133-145.

Ehret, C. 1974 Ethiopians and East Africans: the problem of contacts. Nairobi: East African Publishing House.

Collett, D.P. and P.T. Robertshaw 1983a Pottery traditions of early pastoral communities in Kenya. Azania 18: 107-125.

Ehret, C. 1979 On the antiquity of agriculture in Ethiopia. Journal of African History 20: 161-177.

Collett, D.P. and P.T. Robertshaw 1983b Problems in the interpretation of radiocarbon dates: the Pastoral Neolithic of East Africa. African Archaeological Review 1: 57-74.

Ehret, C. 1982a The first spread of food production to southern Africa. In C. Ehret and M. Posnansky (eds.), The Archaeological and Linguistic Reconstruction of African History, Berkeley: University of California Press, pp. 158-181.

Courty, M.A., P. Goldberg and R. Macphail 1989 Soils and Micromorphology in Archaeology. Cambridge: Cambridge University Press. Crane, H.R. and J.B. Griffin 1962 University of Michigan radiocarbon dates VII. Radiocarbon 4: 183-203. 160

Karega-Mũnene Gasse, F. and F.A. Street 1978 Late Quaternary lake-level fluctuations and environments of the northern Rift Valley and Afar region Ethiopia and Djibouti. Palaeogeography, Palaeoclimatology, Palaeoecology 24: 279-325.

Ehret, C. 1982b Linguistic inferences about early Bantu history. In C. Ehret and M. Posnansky (eds.), The Archaeological and Linguistic Reconstruction of African History, Berkeley: University of California Press, pp. 57-65.

Gautier, A. 1984 Quaternary mammals and archaeozoology in Egypt and the Sudan: a survey. In L. Krzyzaniak and M. Kobusiewicz (eds.), Origin and Early Development of FoodProducing Cultures in North-Eastern Africa, Poznan: Polish Academy of Sciences, pp. 43-56.

Ehret, C. 1984 Historical/linguistic evidence for early African food production. In J. D. Clark and S. A. Brandt (eds.), From Hunters to Farmers: the causes and consequences of food production in Africa, Berkeley: University of California Press, pp. 26-35.

Gautier, A. 1987 Prehistoric men and cattle in North Africa: a dearth of data and a surfeit of models. In A.E. Close (ed.), Prehistory of Arid North Africa: essays in honor of Fred Wendorf, Dallas: Southern Methodist University Press, pp. 163-187.

Estes, R.D. 1991. The Behavior Guide to African Mammals: including hoofed mammals, carnivores, primates. Berkeley: University of California Press. Fadiman, J.A. 1976 The Meru people. In B.A. Ogot (ed.), Kenya Before 1900, Nairobi: East African Publishing House, pp. 139-173.

Gebauer, A.B. 1995 Pottery production and the introduction of agriculture in southern Scandinavia. In W.K. Barnett and J.W. Hoopes (eds.), The Emergence of Pottery: technology and innovation in ancient societies, Washington and London: Smithsonian Institution, pp. 99-112.

Fagan, B.M. and L. Lofgren 1966a Archaeological reconnaissance on the Sese Islands. Uganda Journal 30: 81-86. Fagan, B.M. and L. Lofgren 1966b Archaeological sites on the Nile-Chobi confluence. Uganda Journal 30: 201-206.

Gerrard, A.J. 1981 Soils and Landforms: an integration of geomorphology and pedology. London: George Allen.

Faugust, P.M. and J.E.G. Sutton 1966 The Egerton Cave on the Njoro River. Azania 1: 149-153.

Gifford, D.P. 1978 Ethnoarchaeological observations of natural processes affecting cultural materials. In R.A. Gould (ed.), Explorations in Ethnoarchaeology, Albuquerque: University of New Mexico Press, pp. 77-101.

Flannery, K.V. (ed.) 1986 Guila Naquitz: archaic foraging and early agriculture in Oaxaca, Mexico. Orlando: Academic Press. Fleming, S. 1976 Dating Techniques in Archaeology: a guide to scientific techniques. New York: St. Martin’s Press.

Gifford, D.P. 1980 Ethnoarchaeological contributions to the taphonomy of human sites. In A.K. Behrensmeyer and A.P. Hill (eds.), Fossils in the Making: vertebrate taphonomy and paleoecology, Chicago: University of Chicago Press, pp. 93-106.

Gabel, C. 1969 Six rock shelters on the northern Kavirondo shore of Lake Victoria. African Historical Studies 2: 205-254. Gabel, C. 1974 Terminal food-collectors and agricultural initiative in East and Southern Africa. The International Journal of African Historical Studies 7: 56-68.

Gifford, D.P., G.Ll. Isaac and C.M. Nelson 1980 Evidence for predation and pastoralism at Prolonged Drift: a Pastoral Neolithic site in Kenya. Azania 15: 57-108. Gifford-Gonzalez, D.P. 1985 Faunal assemblages from Masai Gorge Rockshelter and Marula Rockshelter. Azania 20: 69-88.

Gasse, F. 1980 Late Quaternary changes in lake-levels and diatom assemblages on the south-eastern margin of the Sahara. Palaeoecology of Africa 12: 333-350. 161

Holocene Foragers, Fishers and Herders of Western Kenya Griffiths, J.F. 1962 The climate of East Africa. In E.W. Russell (ed.), The Natural Resources of East Africa, Nairobi: East African Literature Bureau, pp. 77-87.

Gifford-Gonzalez, D.P. and J. Kimengich 1984 Faunal evidence for early stock-keeping in the central Rift of Kenya: preliminary findings. In L. Krzyzaniak and M. Kobusiewicz (eds.), Origin and Early Development of Food-Producing Cultures in North-Eastern Africa, Poznan: Polish Academy of Sciences, pp. 457-471.

Guthrie, M. 1962 Some developments in the prehistory of the Bantu languages. Journal of African History 3: 273-282.

Gowlett, J.A.J., R.E.M. Hedges, I.A. Law and C. Perry 1987 Radiocarbon dates from the Oxford AMS system: archaeometry datelist 5. Archaeometry 29: 125-155.

Hamilton, A.C. 1972 The interpretation of pollen diagrams from highland Uganda. Palaeoecology of Africa 7: 45-149.

Gramly, R.M. 1972 Report on the teeth from Narosura. Azania 7: 87-91.

Hamilton, A.C. 1974 The history of the [East African] vegetation. In E. M. Lind and M. E. S. Morrison, East African Vegetation, London: Longman, pp. 188-209.

Gramly, R.M. 1975 ‘Pastoralists and Hunters: recent prehistory in southern Kenya and northern Tanzania’. Ph.D. Thesis, Harvard University. Gramly, R.M. 1976 Upper Pleistocene archaeological occurrences at site GvJm/22, Lukenya Hill, Kenya. Man n.s. 11: 319-344.

Hamilton, A.C. 1976 The significance of patterns of distribution shown by forest plants and animals in tropical Africa for the reconstruction of Upper Pleistocene palaeoenvironments: a review. Palaeoecology of Africa 9: 63-97.

Grayson, D.K. 1978 Minimum numbers and sample size in vertebrate faunal analysis. American Antiquity 43: 53-65.

Hamilton, A.C. 1982 Environmental History of East Africa: a study of the Quaternary. London: Academic Press.

Grayson, D.K. 1979 On the quantification of vertebrate archaeofaunas. Advances in Archaeological Method and Theory 2: 199-237.

Harlan, J. R. 1977 The origins of cereal agriculture in the Old World. In C.A. Reed (ed.), Origins of Agriculture, The Hague: Mouton, pp. 357383.

Grayson, D.K. 1984 Quantitative Zooarchaeology: topics in the analysis of archaeological faunas. New York: Academic Press.

Harlan, J. R., J.M.J. de Wet and A.B.L. Stemler 1976 Plant domestication and indigenous African agriculture. In J.R.Harlan, J.M.J. de Wet and A.B.L. Stemler (eds.), Origins of African Plant Domestication, The Hague: Mouton, pp. 3-19.

Greenberg, J.H. 1966 The Languages of Africa. The Hague: Mouton.

Herbich, I. 1987 Learning patterns, potter interaction and ceramic style among the Luo of Kenya. African Archaeological Review 5: 193204.

Greenberg, J.H. 1972 Linguistic evidence regarding Bantu origins. Journal of African History 13: 189-216 Gregory, J.W. 1896 The Great Rift Valley 1968 impression. London: Frank Cass.

Hillson, S. 1986 Teeth. Cambridge: Cambridge University Press.

Gregory, J.W. 1921 The Rift Valleys and Geology of East Africa. London: Seeley, Service and Co.

Hivernel, F.M.M. 1978 An Ethnoarchaeological Study of Environmental Use in the Kenya Highlands. Ph.D. Thesis, University of London.

Gribbin, J. & J. Cherfas 1982 The Monkey Puzzle: are apes descended from man? Aylesbury: Triad Paladin.

Hobley, C.W. 1913 Stone axe. Journal of the East Africa and Uganda Natural History Society 3: 60-61.

162

Karega-Mũnene Lange, G. 1991 A seed of wild finger millet from Gogo Falls. Azania 26: 191-192.

Huffman, T.N. 1970 The Early Iron Age and the spread of the Bantu. South African Archaeological Bulletin 25: 3-21.

Leakey, L.S.B. 1931 The Stone Age Cultures of Kenya Colony. Cambridge: Cambridge University Press.

Huffman, T.N. 1989 Ceramics, settlement and Late Iron Age migrations. African Archaeological Review 7: 155-182.

Leakey, L.S.B. 1935 The Stone Age Races of Kenya. London: Oxford University Press.

Jochim, M.A. 1979 Breaking down the system: recent ecological approaches in archaeology. Advances in Archaeological Method and Theory 2: 77-117.

Leakey, L.S.B. 1936 Stone Age Africa: an outline of prehistory in Africa. London: Oxford University Press.

Karega-Mũnene 1986 A Contribution to the Study of Subsistence Patterns of Elmenteitan Populations with reference to Animal Bones from Gogo Falls in South Nyanza, Kenya. B.A. Hons. Dissertation, University of Nairobi.

Leakey, L.S.B. 1942 The Naivasha fossil skull and skeleton. Journal of the East Africa Natural History Society 16: 169-177. Leakey, L.S.B. and Archdeacon W.E. Owen 1945 A contribution to the study of the Tumbian culture in East Africa. Coryndon Memorial Museum Occasional Papers, No. 1, Nairobi.

Karega-Mũnene 1987 Preliminary report on the subsistence strategies of Elmenteitan populations at Gogo Falls in South Nyanza, Kenya. Nyame Akuma 28: 19-20.

Leakey, M.D. 1943 Notes on the ground and polished stone axes of East Africa. Journal of the East Africa Natural History Society 17: 182195.

Karega-Mũnene 1991 Subsistence and social change in prehistoric Iran. In P.J. Crabtree and K. Ryan (eds.), Animal Use and Culture Change, Philadelphia: University of Pennsylvania Museum, pp. 8-18.

Leakey, M.D. 1945 Report on the excavations at Hyrax Hill, Nakuru, Kenya Colony, 1937-1938. Transactions of the Royal Society of South Africa 30: 271-409.

Karega-Mũnene 1993 Neolithic/Iron Age Subsistence Economies in the Lake Victoria Basin of East Africa. Ph.D. Dissertation, University of Cambridge.

Leakey, M.D. 1966 Excavations of burial mounds in Ngorongoro Crater. Tanzania Notes and Records 66: 123-135.

Karega-Mũnene 1996 The East African Neolithic: an alternative view. African Archaeological Review 13: 247-254.

Leakey, M.D., Archdeacon W.E. Owen and L.S.B. Leakey 1948 Dimple-based pottery from central Kavirondo, Kenya Colony. Coryndon Memorial Museum Occasional Papers, No. 2, Nairobi.

Karega-Mũnene and S.M. Kahĩnju 1992 Recent excavations at Gogo Falls in western Kenya. Nyame Akuma 37: 8-9. Kendall, R.L. 1969 An ecological history of the Lake Victoria basin. Ecological Monographs 39: 121176.

Leakey, M.D. and L.S.B. Leakey 1950 Excavations at the Njoro River Cave: Stone Age cremated burials in Kenya Colony. Oxford: Clarendon Press.

Klein, R.G. and K. Cruz-Uribe 1984 The Analysis of Animal Bones from Archaeological Sites. Chicago and London: University of Chicago Press.

Levine, M.A. 1982 The use of crown height measurements and eruption wear sequences to age horse teeth. In B. Wilson, C. Grigson and S. Payne (eds.), Ageing and Sexing Animal Bones from Archaeological Sites, Oxford: BAR, British Series 109, pp. 223-250.

Langdale-Brown, I., H.A. Osmaston and J.G. Wilson 1964 The Vegetation of Uganda and its Bearing on Land-Use. Entebbe: Government Printer.

163

Holocene Foragers, Fishers and Herders of Western Kenya Institute in Eastern Africa, Memoir 11, pp. 205-260.

Limbrey, S. 1975 Soil Science and Archaeology. London: Academic Press. Lind, E.M. and M.E.S. Morrison 1974 East African Vegetation. Longman.

Marshall, F.B. 1991 Mammalian fauna from Gogo Azania 26: 175-179.

London:

Falls.

Marshall, F.B., K. Stewart and J.W. Barthelme 1984 Early domestic stock at Dongodien in northern Kenya. Azania 19: 120-127.

Lindblom, G. 1920 The Akamba in East Africa. Uppsala: Archives d’Etudes Orientales.

Marshall, F.B. and K. Stewart 1994 Hunting, fishing and herding pastoralists of western Kenya: the fauna from Gogo Falls. Archaeozoologia 7: 7-27.

Livingstone, D.A. 1967 Postglacial vegetation of the Ruwenzori Mountains in equatorial Africa. Ecological Monographs 37: 25-52. Lofgren, L. 1967 Stone structures of South Nyanza, Kenya. Azania 2: 75-88.

Marshall, F.B. and T. Pilgram 1993 NISP vs. MNI quantification of body-part representation. American Antiquity 58: 261-269.

Lubbock, J. 1872 Prehistoric Times as Illustrated by Ancient Remains and the Manners and Customs of Modern Savages, third edition. London: William and Northgate.

Mbae, N.B. 1986 Aspects of Maasai Ethnoarchaeology: implications for archaeological interpretation. M.A. Thesis, University of Nairobi.

Lwanga-Lunyiigo, S. 1976 The Bantu problem reconsidered. Current Anthropology 17: 282-286.

McBrearty, S.A. 1986 The Archaeology of the Muguruk Site, Western Kenya. Ph.D. Thesis, University of Ilinois at Urbana-Champaign.

Lynch, B.M. and L.H. Robbins 1979 Cushitic and Nilotic prehistory: new archaeological evidence from north-west Kenya. Journal of African History 20: 319-328.

Mehlman, M.J. 1977 Excavations at Nasera Rock, Tanzania. Azania 12: 111-118. Mehlman, M.J. 1979 Mumba-Hohle revisited: the relevance of a forgotten excavation to some current issues in East African prehistory. World Archaeology 11: 80-94.

Marean, C.W. 1992 Hunter to herder: large mammal remains from the hunter- gatherer occupation at Enkapune ya Muto rockshelter, Central Rift, Kenya. African Archaeological Review 10: 65-127.

Mehlman, M.J. 1989 Later Quaternary Archaeological Sequences in Northern Tanzania. Ph.D. Dissertation, University of Illinois at Urbana-Champaign.

Marean, C.W. 1997 Hunter-gatherer foraging strategies in tropical grasslands: model building and testing in the East African Middle and Later Stone Age. Journal of Anthropological Archaeology 16: 189-225.

Merrick, H. 1973 Aspects of the size and shape variation of the East African stone bowls. Azania 8: 115-130.

Marean, C.W. and D. Gifford-Gonzalez 1991 Late Quaternary extinct ungulates of East Africa and palaeoenvironmental implications. Nature 350: 418-420.

Merrick, H.V. and F.H. Brown 1984a Obsidian sources and patterns of source utilization in Kenya and northern Tanzania: some initial findings. African Archaeological Review 2: 129-152.

Marshall, F.B. 1986 Aspects of the Advent of Pastoral Economies in East Africa. Ph.D. Dissertation, University of California at Berkeley.

Merrick, H. and F.H. Brown 1984b Rapid chemical characterization of obsidian artefacts by electron microprobe analysis. Archaeometry 26: 230-236.

Marshall, F.B. 1990 Cattle herders and caprine flocks. In P. Robertshaw (ed.), Early Pastoralists of South-Western Kenya, Nairobi: British 164

Karega-Mũnene Morrison, M.E.S. 1968 Vegetation and climate in the uplands of south-western Uganda during the later Pleistocene period: I. Muchoya swamp, Kigezi District. Journal of Ecology 56: 363-384.

Merrick, H., F.H. Brown and M. Connelly 1990 Sources of the obsidian at Ngamuriak and other south-western Kenyan sites. In P. Robertshaw (ed.), Early Pastoralists of South-Western Kenya, Nairobi: British Institute in Eastern Africa, Memoir 11, pp. 173-182.

Morrison, M.E.S. and A.C. Hamilton 1974 Vegetation and climate in the uplands of south-western Uganda during the later Pleistocene period: II. forest clearance and other vegetational changes in the Rukiga Highlands during the last 8000 years. Journal of Ecology 62: 1-32.

Merrick, H. V. and M.C. Monaghan 1984 The date of the cremated burials in Njoro River Cave. Azania 19: 7-12. Merrick, H.V., F.H. Brown, and W.P. Nash 1994 Use and movement of obsidian in the Early and Middle Stone Ages of Kenya and northern Tanzania. In S.T. Childs (ed.), Society, Culture, and Technology in Africa, Philadelphia: MASCA Research Papers in Science and Technology, Supplement to Vol. 11, pp. 29-44.

Mosley, P.N. 1992 Ugunja: a new Kansyore Oltome site. Azania 27: 129-134. Mturi, A.A. 1986 The Pastoral Neolithic of west Kilimanjaro. Azania 21: 53-63.

Mgomezulu, G.G.Y. 1981 Recent archaeological research and radiocarbon dates from eastern Africa. Journal of African History 22: 435-456.

Munro-Hay, S.C. 1989 Excavations at Aksum: an account of research at the ancient Ethiopian capital directed in 1972-4 by the late Dr. Neville Chittick. London: British Institute in Eastern Africa, Memoir 10.

Michels, J.W. 1986 The Axumite kingdom: a settlement archaeology perspective. Paper presented to the Ninth International Conference on Ethiopian Studies, Moscow, U.S.S.R., 2629 August 1986.

Munro-Hay, S.C. 1991 Aksum: an African Civilisation of late antiquity. Edinburgh: Edinburgh University Press.

Michels, J.W. 1990 Review article: excavations at Aksum. African Archaeological Review 8: 177187.

Murdock, G.P. 1959 Africa: its peoples and their cultural history. New York: McGraw-Hill.

Mithen, S.J. 1989 Evolutionary theory and post-processual archaeology. Antiquity 63: 483-494.

Muriuki, G. 1974 A History of the Kikuyu, 1500-1900. Nairobi: Oxford University Press.

Mithen, S.J. 1990 Thoughtful Foragers: a study of prehistoric decision making. Cambridge: Cambridge University Press.

Mwaniki, H.S.K. 1973 A Political History of the Embu C.A.D. 1500-1906, M.A. Thesis, University of Nairobi.

Moore, A.M.T. 1995 The inception of potting in Western Asia and its impact on economy and society. In W.K. Barnett and J.W. Hoopes (eds.), The Emergence of Pottery: technology and innovation in ancient societies, Washington and London: Smithsonian Institution, pp. 39-53.

Mworia-Maitima, J. 1991 Vegetation response to climate in central Rift Valley, Kenya. Quaternary Research 35: 234-245. Mworia-Maitima, J. 1997 Prehistoric anthropogenic fires and land cover changes in western Kenya: evidence from polle, charcoal, grass phytoliths and grass cuticle analysis from lake sediments. Holocene 7: 409-417.

Morgan, W.T.W. 1973 East Africa. London: Longman. Morris, K.R.S. 1960 Studies on the epidemiology of sleepingsickness in East Africa. Transactions of the Royal Society of Tropical Medicine and Hygiene 54: 71-86.

Mworia-Maitima, J. 1999 Late Quaternary pollen record from Lake Elmenteita, Kenya. Palaeoecology of Africa 26: 103-114.

165

Holocene Foragers, Fishers and Herders of Western Kenya Ogot, B.A. 1967 History of the Southern Luo: I. migration and settlement, 1500-1900. Nairobi: East African Publishing House.

Nelson, C.M. 1980 The Elmenteitan lithic industry. In R. E. Leakey and B. A. Ogot (eds.), Proceedings of the Eighth Panafrican Congress of Prehistory and Quaternary studies, Nairobi: TILLMIAP, pp. 275-278.

Omollo, J.O. 1988 An Ethnoachaeological Study of the Ceramic Industry of the Luo of Nyakach Division, Kisumu District, B.A. Hons. Dissertation, University of Nairobi.

Nelson, C.M. 1995 The work of the Koobi Fora Field School at the Jarigole pillar site. Kenya Past and Present 27: 49-63.

Onyango-Abuje, J.C. 1976 Reflections on culture change and distribution during the Neolithic period in East Africa. In B.A. Ogot (ed.), Hadith 6: history and social change in East Africa, Nairobi: East African Literature Bureau, pp. 14-30.

Nelson, C.M. and M. Posnansky 1970 The stone tools from the re-excavation of Nsongezi rockshelter, Uganda. Azania 5: 119-172. Nelson, C.M. and J. Kimengich 1984 Early phases of pastoral adaptation in the central highlands of Kenya. In L. Krzyzaniak and M. Kobusiewicz (eds.), Origin and Early Development of FoodProducing Cultures in North-Eastern Africa, Poznan: Polish Academy of Sciences, pp. 481-487.

Onyango-Abuje, J. C. 1977a Crescent Island: preliminary report on excavations at an East African Neolithic site. Azania 12: 147-159. Onyango-Abuje, J. C. 1977b A Contribution to the Study of the Neolithic in East Africa with Particular Reference to Nakuru-Naivasha Basins. Ph.D. Dissertation, University of California at Berkeley.

Nenquin, J. 1959 Dimple-based pots from Kasai, Belgian Congo. Man 59: 153-154. Nurse, D. 1982 Bantu expansion into East Africa: linguistic evidence. In C. Ehret and M. Posnansky (eds.), The Archaeological and Linguistic Reconstruction of African History, Berkeley: University of California Press, pp. 199-222.

Onyango-Abuje, J.C. 1980 Temporal and spatial distribution of Neolithic cultures in East Africa. In R.E. Leakey and B.A. Ogot (eds.), Proceedings of the Eighth Panafrican Congress of Prehistory and Quaternary studies, Nairobi: TILLMIAP, pp. 288-292.

O’Brien, T.P. 1939 The Prehistory of Uganda Protectorate. Cambridge: Cambridge University Press.

Owen, Archdeacon W.E. 1941 The early Smithfield culture of Kavirondo Kenya and South Africa. Man 41: 115.

Odegi-Awuondo, C. 1990 Life in the Balance: ecological sociology of Turkana nomads. Nairobi: African Centre for Technology Studies Press.

Parkes, P.A. 1986 Current Scientific Techniques in Archaeology. London and Sydney: Croom Helm. Payne, S. 1972 On the interpretation of bone samples from archaeological sites. In E.S. Higgs (ed.), Papers in Economic Prehistory, Cambridge: Cambridge University Press, pp. 65-81.

Odner, K. 1971a Usangi Hospital and other archaeological sites in the north Pare mountains, northeastern Tanzania. Azania 6: 89-130. Odner, K. 1971b A preliminary report on an archaeological survey on the slopes of Kilimanjaro. Azania 6: 131-149.

Payne, S. 1973 Kill-off patterns in sheep and goats: the mandibles from Asvan Kale. Anatolian Studies 23: 281-303.

Odner, K. 1971c An archaeological survey of Iramba, Tanzania. Azania 6: 151-198.

Pearce, S. and M. Posnansky 1963 The re-excavation of Nsongezi rockshelter, Ankole. Uganda Journal 27: 8594.

Odner, K. 1972 Excavations at Narosura, a stone bowl site in the southern Kenya highlands. Azania 7: 25-92.

166

Karega-Mũnene Posnansky, M. 1967b The Iron Age in East Africa. In W. W. Bishop and J. D. Clark (eds.), Background to Evolution in Africa, Chicago: University of Chicago Press, pp. 629-649.

Peristiany, J.G. 1939 The Social Institutions of the Kipsigis. London: Routledge and Kegan. Perrot R.A. and F.A. Street-Perrot 1982 New evidence for a late Pleistocene wet phase in northern intertropical Africa. Palaeoecology of Africa 14: 57-75.

Posnansky, M. 1968 Bantu genesis - archaeological reflexions. Journal of African History 9: 1-11.

Phillipson, D.W. 1977 The Later Prehistory of Eastern and Southern Africa. London: Heinemann.

Posnansky, M. and J.W. Sikibengo 1959 Ground stone axes and bored stones in Uganda. Uganda Journal 23: 179-181.

Phillipson, D.W. 1984a Early food-production in central and southern Africa. In J.D. Clark and S.A. Brandt (eds.), From Hunters to Farmers: the causes and consequences of food production in Africa, Berkeley: University of California Press, pp. 272-280.

Purseglove, J.W. 1972a Tropical Crops: monocotyledons, Volume 1. London: Longman. Purseglove, J.W. 1972b Tropical Crops: monocotyledons, Volume 2. London: Longman.

Phillipson, D.W. 1984b Aspects of early food production in northern Kenya. In L. Krzyzaniak and M. Kobusiewicz (eds.), Origin and Early Development of Food-Producing Cultures in North-Eastern Africa, Poznan: Polish Academy of Sciences, pp. 489-495.

Purseglove, J.W. 1976 The origins and migrations of crops in tropical Africa. In J.R. Harlan, J.M.J. de Wet and A.B.L. Stemler (eds.), Origins of African Plant Domestication, The Hague: Mouton, pp. 291-309.

Phillipson, D.W. 1993 African Archaeology, 2nd edition. Cambridge: Cambridge: University Press.

Renfrew, C. 1972 The Emergence of Civilisation: the Cyclades and the Aegean in the third millennium B.C. London: Methuen & Co.

Phillips, P. 1980 The Prehistory of Europe. London: Allen Lane.

Republic of Kenya 1984 The Antiquities and Monuments Act, Chapter 215 of the Laws of Kenya, revised edition. Nairobi: Government Printer.

Pianka, E.R. 1974 Evolutionary Ecology. New York: Harper and Row.

Republic of Kenya n.d. South Nyanza District Development Plan, 1989-1993. Nairobi: Ministry of Planning and National Development.

Posnansky, M. 1961a Bantu genesis. Uganda Journal 25: 86-93.

Richardson, J.L. 1966 Changes in the level of Lake Naivasha, Kenya, during postglacial times. Nature 209: 290-291.

Posnansky, M. 1961b Dimple-based pottery from Uganda. Man 61: 141-142. Posnansky, M. 1961c Pottery types from archaeological sites in East Africa. Journal of African History 2: 177-198.

Richardson, J.L. 1972 Palaeolimnological records from Rift lakes in central Kenya. Palaeoecology of Africa 6: 131-136.

Posnansky, M. 1962 The Neolithic cultures of East Africa. In G. Mortelmans and J. Nenquin (eds.), Actes du IV Congres Panafricain de Prehistoire et de l’Etude du Quaternaire, Tervuren: Musee Royal de l’Afrique Centrale, pp. 273-281.

Richardson, J.L. and R.A. Dussinger 1986 Paleolimnology of mid-elevation lakes in the Kenya Rift Valley. Hydrobiologia 143: 167-174. Richardson, J.L. and A.E. Richardson 1972 History of an African Rift lake and its climatic implications. Ecological Monographs 42: 499-534.

Posnansky, M. 1967a Excavations at Lanet, Kenya, 1957. Azania 2: 89-114.

167

Holocene Foragers, Fishers and Herders of Western Kenya Pastoralists of South-Western Kenya, Nairobi: British Institute in Eastern Africa, Memoir 11, pp. 303-304.

Ridings, R. 1996 Where in the world does obsidian hydration dating work? American Antiquity 61: 136-148.

Robertshaw, P. 1991 Gogo Falls: excavations at a complex archaeological site east of Lake Victoria. Azania 26: 63-195.

Rightmire, R.G. 1975a New studies of post-Pleistocene human skeletal remains from the Rift Valley, Kenya. American Journal of Physical Anthropology 42: 351-369.

Robertshaw, P. 1993 The beginnings of food production in southwestern Kenya. In T. Shaw, P. Sinclair, B. Andah and A. Okpoko (eds.), The Archaeology of Africa: food, metals and towns, London and New York: Routledge, pp. 358-371.

Rightmire, R.G. 1975b Problems in the study of later Pleistocene man in Africa. American Anthropologist 77: 28-52. Rightmire, R.G. 1984 Human skeletal remains from eastern Africa. In J.D. Clark and S.A. Brandt (eds.), From Hunters to Farmers: the causes and consequences of food production in Africa, Berkeley: University of California Press, pp. 191-199.

Robertshaw, P. and C. Cable 1990 Stone artefact assemblages from the Lemek-Mara region. In P. Robertshaw (ed.), Early Pastoralists of South-Western Kenya, Nairobi: British Institute in Eastern Africa, Memoir 11, pp. 80-172 Robertshaw, P., D. Collett, D. Gifford and N.B. Mbae 1983 Shell middens on the shores of Lake Victoria. Azania 18: 1-43.

Rindos, D. 1984 The Origins of Agriculture: an evolutionary perspective. London: Academic Press.

Robertshaw, P. and W. Wetterstrom 1989 Plant remains from Gogo Falls. Nyame Akuma 31: 25-27.

Robbins, L.H. 1972 Archaeology in the Turkana District, Kenya. Science 176: 359-366.

Roberts, N. 1998 The Holocene: an environmental history. Oxford: Blackwell Publishers.

Robbins, L.H. 1984 Late prehistoric aquatic and pastoral adaptations west of Lake Turkana, Kenya. In J.D. Clark and S.A. Brandt (eds.), From Hunters to Farmers: the causes and consequences of food production in Africa, Berkeley: University of California Press, pp. 206-211.

Ryan, K., Karega-Mũnene, S.M. Kahĩnju and P.N. Kunoni 1991 Cattle naming: the persistence of a traditional practice in modern Maasailand. MASCA Research Papers in Science and Archaeology 8: 90-96.

Robertshaw, P. 1985 Preliminary report on excavations at Gogo Falls, South Nyanza. Nyame Akuma 26: 25-26.

Sassoon, H. 1967 New views on Engaruka, northern Tanzania: excavations carried out for the Tanzania Government in 1964 and 1966. Journal of African History 8: 201-217.

Robertshaw, P. 1989 The development of pastoralism in East Africa. In J. Clutton-Brock (ed.), The Walking Larder: patterns of domestication, pastoralism, and predation, London: Unwin Hyman, pp. 207-214.

Sassoon, H. 1968 Excavations of a burial mound in Ngorongoro Crater, Dar-es-Salam. Tanzania Notes and Records 69: 15-32. Sauer, C.O. 1969 Agricultural Origins and Dispersals: the domestication of animals and foodstuffs, second edition. Cambridge, Massachusetts: Massachusetts Institute of Technology Press.

Robertshaw, P. 1990a Pottery assemblages from the Lemek-Mara region. In P. Robertshaw (ed.), Early Pastoralists of South-Western Kenya, Nairobi: British Institute in Eastern Africa, Memoir 11, pp. 183-204.

Schepartz, L.A. 1988 Who were the later Pleistocene east Africans? African Archaeological Review 6: 57-72.

Robertshaw, P. 1990b Appendix I: radiocarbon and obsidianhydration dates from sites in the LemekMara region. In P. Robertshaw (ed.), Early 168

Karega-Mũnene Soper, R.C. 1967b Iron Age sites in north-eastern Tanzania. Azania 2: 19-36.

Schmidt, P.R. 1975 A new look at interpretations of the Early Iron Age in East Africa. History in Africa 2: 127-136.

Soper, R.C. 1969 Radiocarbon dating of ‘Dimple-based Ware’ in western Kenya. Azania 4: 148153.

Schmidt, P.R. 1978 Historical Archaeology: a structural approach in an African culture. Westport: Greenwood Press.

Soper, R.C. 1971a A general review of the Early Iron Age of the southern half of Africa. Azania 6: 5-37.

Schoenbrun, D.L. 1990 Early History in East Africa’s Great Lakes Region: linguistic, ecological and archaeological approaches, ca. 500 B.C. to ca. A.D. 1,000. Ph.D. Thesis, University of California at Los Angeles.

Soper, R.C. 1971b Early Iron Age pottery types from East Africa: comparative analysis. Azania 6: 39-52.

Schoenbrun, D.L. 1993 We are what we eat: ancient agriculture between the Great Lakes. Journal of African History 34: 1-31.

Soper, R.C. 1971c Iron Age archaeological sites in the Chobi sector of Murchison Falls National Park, Uganda. Azania 6: 53-87.

Schroeder, R.A. and J.L. Bada 1973 Glacial-postglacial temperature difference deduced from aspartic acid racemization in fossil bones. Science 182: 479-482.

Soper, R.C. 1982 Bantu expansion into eastern Africa: archaeological evidence. In C. Ehret and M. Posnansky (eds.), The Archaeological and Linguistic Reconstruction of African History, Berkeley: University of California Press, pp. 223-238.

Sene, K.J. and D.T. Plinston 1994 A review and update of the hydrology of Lake Victoria in East Africa. Hydrologic Sciences 39: 47-63.

Soper, R. and B. Golden 1969 An archaeological survey of Mwanza region, Tanzania. Azania 4: 15-79.

Shaw, T. 1976 Early crops in Africa: a review of the evidence. In J.R. Harlan, J.M.J. de Wet and A.B.L. Stemler (eds.), Origins of African Plant Domestication, The Hague: Mouton, pp. 107-153.

Ssemmanda, I. and A. Vincens 1999 Preliminary pollen record from the deep waters of Lake Victoria East Africa. Palaeoecology of Africa 26: 137-145.

Shaw, T. 1977 Hunters, gatherers and first farmers in West Africa. In Hunters, Gatherers and First Farmers Beyond Europe: an archaeological survey, (ed.) J.V.S. Megaw, Leicester: Leicester University Press, pp. 69-125.

Stahl, A.B. 1984 A history and critique of investigations into early African agriculture. In J.D. Clark and S.A. Brandt (eds.), From Hunters to Farmers: the causes and consequences of food production in Africa, Berkeley: University of California Press, pp. 9-21.

Siiriainen, A. 1971 The Iron Age at Gatung’ang’a, central Kenya: contributions to the Gumba problem. Azania 6: 199-232. Simoons, F.J. 1965 Some questions on the economic prehistory of Ethiopia. Journal of African History 6: 1-13.

Stemler, A.B.L. 1980 Origin of plant domestication in the Sahara and the Nile Valley. In M.A.J. Williams and H. Faure (eds.), The Sahara and the Nile: Quaternary environments and prehistoric occupation in northern Africa, Rotterdam: Balkema, pp. 503-526.

Soff, H.G. 1969 Sleeping sickness in the Lake Victoria region of British East Africa, 1900-1915. African Historical Studies 2: 255-268.

Stemler, A.B.L., J.R. Harlan and J.M. De Wet 1975 Caudatum sorghums and speakers of Chari-Nile languages in Africa. Journal of African History 6: 161-183.

Soper, R.C. 1967a Kwale: an Early Iron Age site in southeastern Kenya. Azania 2: 1-17.

Stewart, K. 1991 Fish remains from Gogo Falls. Azania 26: 179-180. 169

Holocene Foragers, Fishers and Herders of Western Kenya Street, F.A. and A.T. Grove 1976 Environmental and climatic implications of late Quaternary lake-level fluctuations in Africa. Nature 261: 385-390.

Van Zinderen Bakker, E.M. 1962 A late-glacial and post-glacial climatic correlation between East Africa and Europe. Nature 194: 201-203.

Survey of Kenya 1970 National Atlas of Kenya, third edition. Nairobi: Survey of Kenya.

Van Zinderen Bakker, E.M. 1964 A pollen diagram from equatorial Africa, Cherangani, Kenya. Geologie en Mijnbouw 43: 123-128.

Sutton, J.E.G. 1964 A review of pottery from the Kenya highlands. South African Archaeological Bulletin 19: 27-35.

Vavilov, N.I. 1926 Studies on the origins of cultivated plants. Bulletin of Applied Botany and Plant Breeding 16: 1-245.

Sutton, J.E.G. 1966 The archaeology and early peoples of the highlands of Kenya and northern Tanzania. Azania 1: 37-57.

Vavilov, N.I. 1951 The origin, variation, immunity and breeding of cultivated plants. Chronica Botanica 13: 1-364.

Sutton, J.E.G. 1974 The aquatic civilization of middle Africa. Journal of African History 15: 527-546.

Waane, S.A.C. 1979 The Distribution of Iron Age Pottery in East Africa: an ethnoarchaeological approach. Ph.D. Dissertation, University of Illionis at Urbana-Champaign.

Sutton, J.E.G. 1977 The African aqualithic. Antiquity 51: 2534.

Wagner, G. 1970 The Bantu of Western Kenya, Vol. 2, London: Frank Cass.

Sutton, J.E.G. 1980 Aquatic reflections. In R.E. Leakey and B.A. Ogot (eds.), Proceedings of the Eighth Panafrican Congress of Prehistory and Quaternary studies, Nairobi: TILLMIAP, pp. 321-322.

Wandibba, S. 1977 An Attribute Analysis of the Ceramics of the Early Pastoralist Period from the Southern Rift Valley, Kenya. M.A. Thesis, University of Nairobi.

Temple, P. H. 1967 Causes of intermittent decline of the level of Lake Victoria during the late Pleistocene and Holocene. In R. W. Steel and R. Lawton (eds.), Liverpool Essays in Geography: a jubilee collection, London: Longman, pp. 43-63.

Wandibba, S. 1980 The application of attribute analysis to the study of Later Stone Age/Neolithic pottery ceramics in Kenya summary. In R.E. Leakey and B.A. Ogot (eds.), Proceedings of the Eighth Panafrican Congress of Prehistory and Quaternary studies, Nairobi: TILLMIAP, pp. 283-285.

Terasmae, J. 1974 An evaluation of methods used for reconstruction of Quaternary environments. In W.C. Mahaney (ed.), Quaternary Environments: proceedings of a symposium, York University, Ontario: Geographical Monographs No. 5, pp. 5-32.

Wandibba, S. 1990 Ancient and modern ceramic traditions in the Lake Victoria basin of Kenya. Azania 25: 69-78. Wandibba, S. 1995 Seeking the past in the present: archaeological implications of ethnographic pottery studies in Kenya. In A. Lindahl and O. Stilborg (eds.), The Aim of Laboratory Analyses of Ceramics in Archaeology, Konferenser 34, Stockholm: Kungl. Vitterhets Historie och Antikvitets Akademien, pp. 161-169.

Trapnell, C.G. and I. Langdale-Brown 1962 The natural vegetation of East Africa. In E.W. Russell (ed.), The Natural Resources of East Africa, Nairobi: East African Literature Bureau, pp. 92-102. Van Noten, F. 1979 The Early Iron Age in the interlacustrine region: the diffusion of iron technology. Azania 14: 61-80.

Wandibba, S. 1999 The past and the present in the present: ceramic ethnoarchaeology in Kenya. Inaugural lecture, University of Nairobi, Kenya.

Vansina, J. 1985 Oral Tradition as History. London: James Currey.

170

Karega-Mũnene Wetterstrom, W. 1991 Plant remains from Gogo Falls. Appendix IV in Robertshaw, P., Gogo Falls: excavations at a complex archaeological site east of Lake Victoria. Azania 26: 180191.

Washbourn-Kamau, C. 1975 Lake Quaternary shorelines of Lake Naivasha, Kenya. Azania 10: 72-92. Waters, M.R. 1992 Principles of Geoarchaeology: a North American perspective. Tucson and London: University of Arizona Press.

Woodburn, J. 1968 An introduction to Hadza ecology. In R. Lee and I. DeVore (eds.), Man the Hunter, Chicago: Aldine, pp. 49-55.

Wayland, E.J. 1924 The Stone Age in Uganda. Man 24: 124. Wayland, E.J. 1934 Rifts, rivers, rains and early man in Uganda. Journal of the Royal Anthropological Institute 64: 333-352.

Wrigley, C. 1960 Speculations on the economic prehistory of Africa. Journal of African History 1: 189-203.

Wendorf, F. and F.A. Hassan 1980 Holocene ecology and prehistory in the Egyptian Sahara. In M.A.J. Williams and H. Faure (eds.), The Sahara and the Nile: Quaternary environments and prehistoric occupation in northern Africa, Rotterdam: Balkema, pp. 407-419.

Zohary, D. and M. Hopf 1988 Domestication of Plants in the Old World: the origin and spread of cultivated plants in West Asia, Europe, and the Nile Valley. Oxford: Clarendon Press.

Were, G.S. 1967 A History of the Abaluyia of Western Kenya. Nairobi: East African Publishing House.

171

CAMBRIDGE MONOGRAPHS IN AFRICAN ARCHAEOLOGY No 1 No 2 No 3 No 4 No 5 No 6 No 7 No 8 No 9 No 10 No 11 No 12 No 13 No 14 No 15 No 16 No 17 No 18 No 19 No 20 No 21 No 22 No 23 No 24 No 25 No 26 No 27 No 28 No 29 No 30

BAR S75, 1980 The Niger Delta Aspects of its Prehistoric Economy and Culture by Nwanna Nzewunwa. ISBN 0 86054 083 9 BAR S89, 1980 Prehistoric Investigations in the Region of Jenne, Mali A Study in the Development of Urbanism in the Sahel by Susan Keech McIntosh and Roderick J. McIntosh ISBN 0 86054 103 7 BAR S97, 1981 Off-Site Archaeology and Human Adaptation in Eastern Africa An Analysis of Regional Artefact Density in the Amboseli, Southern Kenya by Robert Foley. ISBN 0 86054 114 2 BAR S114, 1981 Later Pleistocene Cultural Adaptations in Sudanese Nubia by Yousif Mukhtar el Amin. ISBN 0 86054 134 7 BAR S119, 1981 Settlement Patterns in the Iron Age of Zululand An Ecological Interpretation by Martin Hall. ISBN 0 86054 143 6 BAR S139, 1982 The Neolithic Period in the Sudan, c. 6000-2500 B.C. by Abbas S. Mohammed-Ali. ISBN 0 86054 170 3 BAR S195, 1984 History and Ethnoarchaeology in Eastern Nigeria A Study of Igbo-Igala relations with special reference to the Anambra Valley by Philip Adigwe Oguagha and Alex Ikechukwu Okpoko. ISBN 0 86054 249 1 BAR S197, 1984 Meroitic Settlement in the Central Sudan An Analysis of Sites in the Nile Valley and the Western Butana by Khidir Abdelkarim Ahmed. ISBN 0 86054 252 1 BAR S201, 1984 Economy and Technology in the Late Stone Age of Southern Natal by Charles Cable. ISBN 0 86054 258 0 BAR S207, 1984 Frontiers Southern African Archaeology Today edited by M. Hall, G. Avery, D.M. Avery, M.L. Wilson and A.J.B. Humphreys. ISBN 0 86054 268 8. £23.00. BAR S215, 1984 Archaeology and History in Southern Nigeria The ancient linear earthworks of Benin and Ishan by P.J. Darling. ISBN 0 86054 275 0 BAR S213, 1984 The Later Stone Age of Southernmost Africa by Janette Deacon. ISBN 0 86054 276 9 BAR S254, 1985 Fisher-Hunters and Neolithic Pastoralists in East Turkana, Kenya by John Webster Barthelme. ISBN 0 86054 325 0 BAR S285, 1986 The Archaeology of Central Darfur (Sudan) in the 1st Millennium A.D. by Ibrahim Musa Mohammed. ISBN 0 86054 367 6. BAR S293, 1986 Stable Carbon Isotopes and Prehistoric Diets in the South-Western Cape Province, South Africa by Judith Sealy. ISBN 0 86054 376 5. BAR S318, 1986 L'art rupestre préhistorique des massifs centraux sahariens by Alfred Muzzolini.. ISBN 0 86054 406 0 BAR S321, 1987 Spheriods and Battered Stones in the African Early and Middle Stone Age by Pamela R. Willoughby. ISBN 0 86054 410 9 BAR S338, 1987 The Royal Crowns of Kush A study in Middle Nile Valley regalia and iconography in the 1st millennia B.C. and A.D. by Lázló Török.. ISBN 0 86054 432 X BAR S339, 1987 The Later Stone Age of the Drakensberg Range and its Foothills by H. Opperman. ISBN 0 86054 437 0 BAR S350, 1987 Socio-Economic Differentiation in the Neolithic Sudan by Randi Haaland. ISBN 0 86054 453 2 BAR S351, 1987 Later Stone Age Settlement Patterns in the Sandveld of the South-Western Cape Province, South Africa by Anthony Manhire. ISBN 0 86054 454 0 BAR S365, 1987 L'art rupestre du Fezzan septentrional (Libye) Widyan Zreda et Tarut (Wadi eshShati) by Jean-Loïc Le Quellec. ISBN 0 86054 473 7 BAR S368, 1987 Archaeology and Environment in the Libyan Sahara The excavations in the Tadrart Acacus, 1978-1983 edited by Barbara E. Barich. ISBN 0 86054 474 5 BAR S378, 1987 The Early Farmers of Transkei, Southern Africa Before A.D. 1870 by J.M. Feely. ISBN 0 86054 486 9 BAR S380, 1987 Later Stone Age Hunters and Gatherers of the Southern Transvaal Social and ecological interpretation by Lyn Wadley. ISBN 0 86054 492 3 BAR S405, 1988 Prehistoric Cultures and Environments in the Late Quaternary of Africa edited by John Bower and David Lubell. ISBN 0 86054 520 2 BAR S418, 1988 Zooarchaeology in the Middle Nile Valley A Study of four Neolithic Sites near Khartoum by Ali Tigani El Mahi. ISBN 0 86054 539 3 BAR S422, 1988 L'Ancienne Métallurgie du Fer à Madagascar by Chantal Radimilahy. ISBN 0 86054 544 X BAR S424, 1988 El Geili The History of a Middle Nile Environment, 7000 B.C.-A.D. 1500 edited by I. Caneva. ISBN 0 86054 548 2 BAR S445, 1988 The Ethnoarchaeology of the Zaghawa of Darfur (Sudan) Settlement and Transcience by Natalie Tobert. ISBN 0 86054 574 1

No 31

BAR S455, 1988 Shellfish in Prehistoric Diet Elands Bay, S.W. Cape Coast, South Africa by W.F. Buchanan. ISBN 0 86054 584 9 No 32 BAR S456, 1988 Houlouf I Archéologie des sociétés protohistoriques du Nord-Cameroun by Augustin Holl. ISBN 0 86054 586 5 No 33 BAR S469, 1989 The Predynastic Lithic Industries of Upper Egypt by Liane L. Holmes. ISBN 0 86054 601 2 (two volumes) No 34 BAR S521, 1989 Fishing Sites of North and East Africa in the Late Pleistocene and Holocene Environmental Change and Human Adaptation by Kathlyn Moore Stewart. ISBN 0 86054 662 4 No 35 BAR S523, 1989 Plant Domestication in the Middle Nile Basin An Archaeoethnobotanical Case Study by Anwar Abdel-Magid. ISBN 0 86054 664 0 No 36 BAR S537, 1989 Archaeology and Settlement in Upper Nubia in the 1st Millennium A.D. by David N. Edwards. ISBN 0 86054 682 9 No 37 BAR S541, 1989 Prehistoric Settlement and Subsistence in the Kaduna Valley, Nigeria by Kolawole David Aiyedun and Thurstan Shaw. ISBN 0 86054 684 5 No 38 BAR S640, 1996 The Archaeology of the Meroitic State New perspectives on its social and political organisation by David N. Edwards. ISBN 0 86054 825 2 No 39 BAR S647, 1996 Islam, Archaeology and History Gao Region (Mali) ca. AD 900 - 1250 by Timothy Insoll. ISBN 0 86054 832 5 No 40 BAR S651, 1996 State Formation in Egypt: Chronology and society by Toby A.H. Wilkinson. ISBN 0 86054 838 4 No 41 BAR S680, 1997 Recherches archéologiques sur la capitale de l’empire de Ghana Etude d’un secteur d’habitat à Koumbi Saleh, Mauritanie. Campagnes II-III-IV-V (1975-1976)-(1980-1981) by S. Berthier. ISBN 0 86054 868 6 No 42 BAR S689, 1998 The Lower Palaeolithic of the Maghreb Excavations and analyses at Ain Hanech, Algeria by Mohamed Sahnouni. ISBN0 86954 875 9 No 43 BAR S715, 1998 The Waterberg Plateau in the Northern Province, Republic of South Africa, in the Later Stone Age by Maria M. Van der Ryst. ISBN 0 86054 893 7 No 44 BAR S734, 1998 Cultural Succession and Continuity in S.E. Nigeria Excavations in Afikpo by V. Emenike Chikwendu. ISBN 0 86054 921 6 No 45 BAR S763, 1999 The Emergence of Food Production in Ethiopia by Tertia Barnett. ISBN 0 86054 971 2 No 46 BAR S768, 1999 Sociétés préhistoriques et Mégalithes dans le Nord-Ouest de la République Centrafricaine by Étienne Zangato. ISBN 0 86054 980 1 No 47 BAR S775, 1999 Ethnohistoric Archaeology of the Mukogodo in North-Central Kenya Hunter-gatherer subsistence and the transition to pastoralism in secondary settings by Kennedy K. Mutundu. ISBN 0 86054 990 9 No 48 BAR S782, 1999 Échanges et contacts le long du Nil et de la Mer Rouge dans l'époque protohistorique (IIIe et IIe millénaires avant J.-C.) Une synthèse préliminaire by Andrea Manzo. ISBN 1 84171 002 4. £28.00. No 49 BAR S838, 2000 Ethno-Archaeology in Jenné, Mali Craft and status among smiths, potters and masons by Adria LaViolette. ISBN 1 84171 043 1 No 50 BAR S860, 2000 Hunter-Gatherers and Farmers An enduring Frontier in the Caledon Valley, South Africa by Carolyn R. Thorp. ISBN 1 84171 061 X. £25.00. No 51 BAR S906, 2000 The Kintampo Complex The Late Holocene on the Gambaga Escarpment, Northern Ghana by Joanna Casey. ISBN 1 84171 202 7. £30.00. No 52 BAR S964, 2000 The Middle and Later Stone Ages in the Mukogodo Hills of Central Kenya A Comparative Analysis of Lithic Artefacts from Shurmai (GnJm1) and Kakwa Lelash (GnJm2) Rockshelters by G-Young Gang. ISBN 1 84171 251 5. £25.00.

No 53 BAR S1006, 2001 Darfur (Sudan) In the Age of Stone Architecture c. 1000 - 1750 AD Problems in historical reconstruction by Andrew James McGregor. ISBN 1 84171 285 X. £34.00.