African pastoralists’ preoccupation with the sexual, social, and productive attributes of livestock is mirrored metaphorically in the same spheres of human life (Galaty 1982; Llewelyn-Davies 1979; Meeker 1989; Saitoti 1986; Saitoti and Beckwith 1980; Spencer 1965, 1988). The self-identity of young man and name-ox or, as Deng (1972) has put it, “personality ox,” among many Nilotic-speakers (Evans-Pritchard 1940; Gulliver 1955; Klima 1970; Thomas 1965) and some Cushitic speakers (Almagor 1978) is the best-known version of such blurring of the line between animals and human actors, or, more properly, the merging of the social and productive identities of each. This is not a universal trait of pastoral societies; other herding groups in Eurasia and the Americas stress neither age-grading nor identification of person and animal to such a degree. Yet this is apparently an old trait of African pastoral systems; Saharan rock art probably testifies to the gender of both humans and domestic animals. Holl (1995) has pointed to the likely testimony of some paintings to age-based transitions that male and female actors made in negotiating their lives in that milieu. The sexual attributes of livestock (and wild animals) are clearly depicted in well-preserved paintings, and the sex of human actors can usually be readily inferred from their depictions. In cases where actual physical attributes of sex are not emphasized, it appears that cultural means of denoting gender (and age?), such as hairdo, body ornamentation, and personal gear are employed (Holl 1995). Thus, in their design choices and execution, early testimonials to African pastoralist existence bear witness to the importance of representing sex and gender systems of both humans and animals. The intertwined aspects of gender, person, and society among African pastoral peoples are pervasive, but what can we realistically hope to recover of the social or ideological contexts in which prehistoric pastoralists lived? And how? This chapter explores these questions using East African data.
Two points are generally agreed on by researchers currently studying the development of food production in East Africa. First, paleontological and archaeological evidence points to the introduction of all major domestic species into East African ecosystems, be they domestic food animals (cattle, sheep, goats, and, later, dromedaries) or domestic plants (millets, sorghum, varieties of legume). Second, though a diversity of opinion exists over the nature of early food-producing economies in Kenya and Tanzania, investigators agree that the evidence points to a relatively greater commitment to pastoral livestock than to farming in the location and layout of encampments, and in the biological remains and associated artifacts. It is therefore appropriate to focus on how to investigate archaeologically the social relations, including the role of gender, of people with strong commitments to pastoral livestock.
The term gender as used here does not mean “women.” It refers to socially defined, differentiated roles, based usually but not always on the discernible sex of individuals. The study of gender in eastern Africa should address roles of both male and female actors in prehistoric, stock-owning societies. We who have spent time with modern agropastoralists and pastoralists in the region have had it forcefully impressed on us by members of those societies that age and gender are fundamental organizing principles that structure individuals’ experience of life and their roles as producers and consumers.1
This chapter lays groundwork for such research by resituating some known archaeological evidence within a more socially and gender-conscious context. It begins with a theoretical passage now well traveled in other geographic contexts but novel in Africa (see also Kent, Chapter 1; Wadley, Chapter 4; Stahl and Cruz, Chapter 11; Segobye, Chapter 12; Hall, Chapter 13): whether and how archaeological study of gender is possible. The extant literature on past and present East African pastoralists is shown—as has done the literature in other venues—to already feature gender, in the form of unconscious, androcentric perspectives on human action. Such methodological androcentrism impedes intelligent exegesis of archaeological sites, because it hinders collection of data on the full range of human activities that create them. I argue that gender cannot be investigated systematically without a wider reorientation of research toward a socially focused archaeology and without specifying those approaches best suited to the prehistoric East African archaeological cases dealt with here.
Framing a more balanced approach to gender in studying prehistoric East African pastoralists is especially challenging, because ethnographically documented pastoralists and agropastoralists in the region themselves often stress the centrality of male actors in kinship, public discourse, ritual, and aesthetics. Moreover, this material presents another kind of challenge, that of the prudent use of ethnographic analogy. In contrast to cases discussed by many other authors in this volume, the East African “Pastoral Neolithic” is beyond the reach of ethnographically based “direct historic approach.” It is therefore useful to consider how ethnographic and other actualistic information can be used in such cases (see also Kent, Chapter 3).
The second part of the chapter turns to an example of a more socially focused approach to interpreting archaeological materials, specifically the widespread phenomenon of Nderit ceramics in what is now Kenya and Tanzania. My approach demonstrates the mobilization of “uniformitarian” assumptions within a socially directed inquiry advocated in the first half of the paper, as well as the virtual inevitability of raising questions about gender within such a framework.
This section takes up issues that exist in the background of any more socially inclined interpretation of archaeological sites, including those incorporating gender. It begins with a review of the ways that extant archaeological and ethnographic accounts of East African pastoralists are implicitly gendered. It then develops an argument for making the goals of archaeological inferences and reconstruction more social, specifying the types of “social archaeology” that seem most appropriate to the societies under study. Finally, it offers a brief review of the appropriate use of ethnographic and other actualistically derived analogies in defining archaeological problems and construction narratives based on archaeological data.
Gender has been a topic in anglophone archaeology for about a decade and a half, rather later than the topic emerged in other social sciences and humanities. Wylie (1991) has discussed the reasons for this late emergence in some detail, as did Conkey and Spector (1984) and Gero (1985). A central point of these analyses is that, though it has often been asserted that “you can’t dig up gender,” traditional archaeological approaches both to first-order behavioral interpretations of artifacts and sites and to development of higher-order historical narratives are themselves implicitly or explicitly gendered. Male agency is assumed for nearly any significant technological breakthrough or directional change in social organization (Watson and Kennedy 1991). Berns (1993) has made the case that European analyses of West African “high art” ceramic artifacts implicitly assume that these were products of men, despite overwhelming historic and ethnographic evidence of long and widespread ceramic manufacture by women of the region. Wright (1991) has made similar observations regarding ceramic production in southwest Asia, as has Handsman (1991) regarding exegesis of prehistoric figurines from Lepenski Vir. Gero (1985, 1991) and Kehoe (1990) have noted that fabrication and use of stone tools, especially more labor-intensive types, is normally ascribed to male hunters or food producers rather than to women. Jarvenpa and Brumbach (1995) note that, when mentioned at all, women’s toolkits are assumed to be less technologically complex than men’s. They cite recent theorization by Hayden (1992b) as a case in point. One may add Binford’s earlier (1987b) argument that Acheulian handaxes are products of wider-foraging, butchering males, whereas females account for contemporaneous scatters of nondescript flake tools. Jarvenpa and Brumbach (1995) go on to describe as counterexamples the labor-intensive, high-curation tools in historic and modern Chipewyan women’s hide-working kits, also documented by Spector (1982, 1991) for Plains Indian women.
As in other regions of the world, archaeological research in East Africa has often set analytic categories and findings into implicitly gendered research designs and narratives. In the East African case, hunters, pastoral stock owners, or ironworkers are assumed to be the active agents creating the form and extent of the archaeological record, and these occupational roles are assumed to be filled by males. Few archaeological analyses of the Pastoral Neolithic or preceding phases of East African prehistory have focused on such social interactions as exchange and political relations, but those few that do interpret these too as male domains. None consider the role of women in creating the media with which and in which these interactions might have taken place, despite some obvious—if theoretically arguable—linkages between archaeological evidence linked to production, exchange, and social display (for example, ostrich eggshell beads, gourds, milling stones, ceramics) and spheres of female activity in East African ethnographic cases. The same cannot be said of elision of modern gender roles with specific classes of archaeological data imputed to men—wild animal bones signifying decision making by male hunters, domestic animal bones likewise standing for male herders and their actions. I stress that I am not advocating such simplistic uses of “ethnographic analogy” as a way to incorporate “women’s roles” into archaeological analyses—quite the contrary (see below). The example is simply intended to indicate that selective mobilization of analogy is part of the implicitly gendered way archaeological interpretation is done.2
The only archaeological analysis of early pastoral evidence that has offered a more explicitly gendered (and more social) reading of the early pastoral archaeological record in the region is Robertshaw’s (1989, 1990) scenario for “big man” lineages and monopolistic trade networks among makers of Elmenteitan lithics and ceramics in the Mara-Loita region of Kenya. I would interpret Robertshaw’s project less as a deliberate exercise in gendering the archaeological evidence than as one that proceeds from assuming male agency in any economically important activity.
It is small wonder, however, that most archaeologists’ models of pastoral society accept women as noninstrumental, and women’s work as marginally important to production, trade, and social life. Most older and many more recent anthropological accounts of pastoralism are largely androcentric in both focus and tone—what is reported is what men do, and what men do is what is important. That men wield overt economic and political power and that they dominate public discourse in most African pastoral societies has apparently blinded (male?) ethnographers to women’s roles, areas of autonomy, and subtler exercises of power. It may indeed be difficult for male ethnographers to work closely with female informants among peoples practicing strong social partitioning of men and women. But in East African pastoral encampments women are everywhere, milking, preparing milk products, tending adult livestock, nurturing their children and young livestock, butchering and cooking animals, preparing other kinds of foodstuffs, carrying water and firewood, making and cleaning houses, singing, arguing, importuning, and otherwise participating in a rich web of socially contextualized work and experience. Lacunae in traditional ethnographies on roles of women—and of children—thus reflect a literal turning away from the daily business of the pastoral encampment.
For archaeologists, the prevailing androcentrism of most ethnographic literature on pastoralists is not just a peripheral problem of gender politics, to be taken or left according to one’s own personal inclinations. It especially ill serves those of us who study the remains of everyday life, whether in residential sites or in special-purpose locales. If we lack actualistic evidence on details of house construction (how much time and labor per structure type), food preparation (energetic trade-offs of different methods of cooking and processing), water and wood gathering (how much is required by specific technologies of cooking), cleaning and refuse disposal of a settlement (efficiencies versus symbolic action)—all because these tasks are in the hands of women and children and thus have been discounted—what can serve as our referential sets for working with archaeological materials? To the extent to which our own cultural mind-sets keep us from documenting important categories of behavior that produce archaeological traces, we are deprived of important classes of data about archaeological materials (Gifford-Gonzalez 1993). This is not a matter of “paying attention to women,” because we are well-advised not to associate certain tasks and roles a priori with either gender, or to adults over children. Rather, we must not refrain from investigating important areas of resource use and site formation simply because they are in the hands of persons whom our culture—or even the culture under study—depreciates.
Exceptions to androcentric norms of description are found in the work of human ecologists and development economists. These include McCabes and Dyson-Hudson’s (1982) and Wienpahl s (1984) research on livestock management and household production in Turkana homesteads, Arhem’s (1985) close study of Maasai divisions of labor and output among households in the Serengeti Conservation area, and Homewood and Rogers’ (1991) ecologically and regionally contextualized study of Maasai subsistence strategies in the same area. Like archaeologists, these researchers must study the overall situation of humans within a regional context, linking details of household social relations and economy to socioeconomic systems and regional ecology. Such an approach leads to ample documentation of pastoral and agropastoral women’s actions and manipulation—if not outright ownership—of major resources. As well, economic historians and anthropologists such as Kjekshus (1977) and Hakansson (1994) also have documented women’s roles in local and regional agropastoral and pastoral economic systems (see below). Recent academic literature and cinema from more feminist perspectives have also enhanced understandings of gender relations in East African pastoral systems (Hodder 1986; Llewelyn-Davies 1979; Moore 1982; Williams 1987).
What may we glean from such research that is relevant to archaeological cases? Historically, East African pastoral women actively participated in production and distribution systems, as well as in actual exchange negotiations. Confined neither to house nor to homestead, they sometimes traveled impressive distances solely in the company of other women, carrying goods and conducting trade. Kjekshus (1977), citing early Arab and European travelers’ accounts in what is now Tanzania, notes that Maasai women regularly traveled scores of kilometers to preordained, nonurban market locales for trade, even during times of intertribal warfare or raiding. Similar patterns of interethnic trade by women are summarized by Hakansson (1994) for Gusii and Luo in the 19th century. Late 20th-century pastoral women’s minibus expeditions to trade in East African metropoles thus have precolonial antecedents.
Detailed economic analyses indicate that females commonly held rights both to allocate harvested and stored plant foods and to the products of flocks and herds under their care. Despite male ownership of productive agrarian land and livestock, women’s on-the-ground control was the basis for their manipulation of household economics and negotiation of social success within patrilineal, patrilocal societies. For example, Hakansson (1994) has noted that Gusii men wishing to mobilize cattle for bridewealth and other status-enhancing activities, or grains with which to brew beer and stage feasts for visitors, had to negotiate with wives for grain and labor, and even were forced to bargain for the “loan” of cattle they had previously committed to a specific wife’s sons. Thus, far from being a simplistic patriarchal world in which men controlled land, livestock, and women, historic Gusii society comprised a complex set of competing entitlements and obligations, in which persons of different ages and genders attempted to negotiate their most advantageous possible life courses.
The foregoing discussion of the gendered nature of mainstream archaeological interpretation—and some counterexamples that emphasize women—might appear to be an argument for filling in the present androcentric “half-truth” of archaeological interpretation with a gynecocentric exegesis of the missing half. It is not. In Conkey and Spector’s (1984) terms, such a “just add women and stir” remedy for redressing androcentrism merely follows and reifies the culturally specific gender categories implicit in the Western archaeological tradition. Insistence on a priori, “boys-do-this, girls-do-that” perspectives blind us to the potential social variability with which we may be confronted archaeologically.
The archaeological study of gender will be productive only if a broader shift is made toward more socially focused analytic approaches. Gender is one of a number of socially constructed categories that place individuals and groups within larger sets of economic and political relationships and guide the allocation of rights and obligations within a society (see Conkey and Spector 1984). A socially focused approach requires that we abandon anglophone archaeologists’ traditional inclination to view archaeological evidence as products either of individual effort or of poorly delimited, yet unitary, “groups.” Traditional archaeological perspectives may testify more to our own cultural construal of work as individualistic and autonomous within a largely undifferentiated societal unit, than to any quality inherent in the archaeological evidence itself. What kind of “social archaeology” is useful for exegeses of sites created by unranked, unstratified, food-producing or foraging groups? The “chiefdoms–civilizations” social archaeology (Earle 1987; Renfrew 1984; Renfrew and Cherry 1986; Sanders and Price 1968) does not, from my point of view, readily integrate with the empirical evidence of smaller-scale social groupings nor with the more fluid social arrangements such as presumably characterized pastoral and forager societies in prehistoric East Africa.
East African archaeology has shared with other regional traditions the tendency to view hunter-gatherers as fundamentally ecologically driven. Moreover, ethnographers and historians have until recently assumed that East African farmers and pastoralists were subsistence-oriented producers who, if involved in exchange at all, passively responded to the stimuli of trade initiated from the Indian Ocean Coast (Kjekshus 1977; Koponen 1988). Both assumptions have come into question more recently, and evidence is mounting that at least some East African food producers regularly planned to produce surpluses for intergroup trade as part of their normal social and economic regimes, with the aim of acquiring prestige goods (Hakansson 1994).
Recent work on the archaeology of “complex hunter-gatherers” elsewhere in the world (Price and Brown 1985; Price and Feinman 1995) indicates that socially directed theory and method can be developed and applied to such groups. I believe approaches that start with a fundamentally economic analysis of social relations among individuals and small groups in the procurement, production, and distribution of resources and commodities (Bender 1985; Hayden 1992a; Marquardt 1985) are most useful. With a focus on resource acquisition, production, and distribution in the context of social relations, one can accommodate consideration—if not immediate discernment—of gender (see also Stahl and Cruz, Chapter 11; Segobye, Chapter 12; Hall, Chapter 13). Such an approach does not necessarily negate or dismiss ecological models but rather includes them in a more complex model of human social and economic behaviors (Hayden 1992a; Jones 1996). With their varied emphases on the “structures of everyday life,” these approaches allow us to link information on present-day groups and the material remnants of the quotidian activities of ancient peoples.
Although I have done ethnoarchaeological research among modern pastoralists and agropastoralists in East Africa, I do not believe it prudent to equate pastoralists then with pastoralists now in a facile way. The most obvious difference between the prehistoric period and recent times is the colonial—and now, postcolonial—contexts in which historic and ethnographic documentation has taken place. Most of us are aware of how and why apparently “essential features” of modern pastoralist socioeconomic status, ethnicity, and territoriality may stem from developments in their very recent histories (Chang and Koster 1994; Chang 1967; Denbow 1984; Denbow and Wilmsen 1986; Galaty 1991; Galaty 1977; O’Brien 1977; Schrire 1984b; Yellen 1990b). Of even greater concern to archaeologists is the possibility that the situation of the earliest pastoral colonists of East Africa may well have been very different from that of pastoralists in the precolonial 2nd millennium of the Christian era. The latter point will be developed in the last section of this paper, but it raises the question of how to use analogies with modern pastoral groups in doing archaeological analysis. One must choose one’s parallel cases based on an understanding of the nature of analogical reasoning, avoiding, for example, the notion that all pastoralists live (and lived) in segmentary patrilineages, that a modal pastoralist personality type exists, that all of them conceptualize and handle cattle, camels, small stock, and wild animals the same way, or that all mobilize labor and reduce economic risk similarly.
In an earlier paper (Gifford-Gonzalez 1991), I discussed problems and potentials of analogical reasoning in historical sciences such as archaeology in some detail. Here I note two points relevant to using contemporary studies in archaeological research. First, most archaeological reasoning, including that based on “uniformitarian” assumptions and on the “direct historic” approach, is analogical (Wylie 1982, 1985). Archaeological epistemology, hypothesis formation and selection, as well as methods of inference, are normally and unavoidably analogical (Binford 1981; Gifford 1981; Gifford-Gonzalez 1991; Gould and Watson 1982; Smith 1977; Wylie 1982, 1985). Even the logical sorting of alternatives before explicit articulation of hypotheses or models occurs within a matrix of analogical comparisons and judgments that Wylie (1989) has called “suppressed analogy” (see also Salmon 1982).
Second, criticisms of misused ethnographic analogy (Ascher 1961; Binford 1967, 1987a; Freeman 1968; Gould 1978, 1980; Gould and Watson 1982) in fact object to the failings of formal analogies, in contrast to relational ones. Formal analogies extend to the less-known case-associated traits of the known case, based solely on points of resemblance in form between the two. This is the very problem critics of the abuse of ethnographic analogy identify as the error of unjustifiably imputing aspects of a living society to an ancient one. In the case of gender studies, the grossest such errors might include assuming men were never gatherers, women were always potters, and so forth. Of more concern, perhaps, are the potential pitfalls of the “direct historic” approach, which deploys an elaborate system of formal analogies (for example, assuming that protohistoric Sotho gender arrangements were like those of ethnographic Sotho in all respects). Stahl and Cruz (Chapter 11) and Hall (Chapter 13) work through this issue discerningly, as have Lewis-Williams and Dowson (1988) in their discussion of the relevance of San ritual and rock painting to European paleolithic art. More strongly warranted analogical inferences depend on relational analogies, where properties are ascribed to the less-understood partner in an analogical pair based on knowledge of the functional or causal relationships of those qualities in the better-known member of the pair (Copi 1982; Hesse 1966). For example, paleontologists routinely impute an ontogenetic history to a fossil bone, based on its resemblances to modern bones for which such histories are universal causal and functional conditions.
How does the foregoing relate to using research findings on modern pastoralists (or others) to investigate gender or other social aspects of ancient pastoralist society? If we recognize that analogies are more warranted if based on systematic or causal relations, we must employ ethnographic evidence to move us beyond facile gender and cultural stereotyping, to locate those enduring and universal facts of pastoral life to which all groups and households engaged in the keeping of herds and flocks must respond. From the multiplicity of modern cases, one can discern underlying factors with which any group relying on cattle, sheep, goats, donkeys, or camels must cope. These include the climatic patterns characterizing the areas inhabited, basic biological requirements and behaviors of the species managed, and other synecological interactions particular to the region, all of which require strategic, socially coordinated responses from pastoralists. By arguing from first principles established by these factors, we may develop a framework for thinking about the parameters and challenges of early pastoral life. In turn, this can lead to better definition of research problems and relevant types of archaeological evidence. The second part of this chapter considers the specifics of the East African case, focusing on gender in modern groups and three planes of analysis that follow logically from the foregoing and contribute to an understanding of gender in the archaeological record: (1) risks and risk reduction in pastoral systems, (2) regional ecology and epizootiology, and (3) functional artifact analysis.
Details of the sequence of food-producing economies in eastern Africa are not yet widely published, and this section attempts to summarize the current state of knowledge.
Unlike many other regions where food production either developed or was introduced, research in East Africa has been sporadic, largely done on the individual initiative of foreign researchers from several different national traditions of archaeology, and uncoordinated through central planning or voluntary, consensual research design. Indigenous scientists have been hampered by the paucity of resources devoted to archaeological field research and by their administrative obligations to museums, universities, and antiquity services. These factors account for some, but not all, of the rather skimpy nature of archaeological data points, advanced materials analyses, and contingent nature of interpretations.
Analysis of pollen cores and limnological analyses of lakes of the Kenyan Central Rift and adjacent highland lakes of Kenya, together with more limited studies of the Lake Turkana basin, have produced enough information to outline climatic and vegetational history of the region. Table 7.1 summarizes a dynamic pattern of climate and vegetation change in the terminal Pleistocene and Holocene, to which humans would have had to adjust. Lake Turkana, the area with the earliest evidence of domestic animals in the region, underwent dynamic variations in level from terminal Pleistocene through Holocene (Butzer 1971), and during the early Holocene moist phase, it was an outlet lake joined to the Nile drainage system. As in more northern parts of Africa, this moist phase seems to have seen development of a lacustrine-focused economy that included introduced domestic animals and ceramics in its later phases (Table 7.1). The mid-Holocene arid phase may have prompted emigration from the Lake Turkana basin toward pastures farther south. As they moved south, immigrants would have encountered hunter-gatherer populations also adjusting not only to differences in the locations and relative proportions of biotic communities but also to a shift in rainfall from a unimodal to a bimodal pattern in south-central Kenya (Marshall 1990).
Ambrose (1984a,b) has modeled a dynamic vegetational sequence for the steep-sided Central Rift of Kenya (Figure 7.1 and Table 7.1), asserting that these both called for reorganization in indigenous foragers’ land use and influenced the timing of domestic bovines’ and caprines’ introduction into south-central Kenya. Only after 3000 BP would optimal habitats have opened up for savanna-adapted ungulate species, including domestic cattle, sheep and goats. Less is known of vegetational history of the more southern Athi-Kapiti, Loita, and Serengeti plains. We may expect that wet phases would have led to expansion of woodlands from adjacent uplands into savanna habitats, while during times of increased aridity, savannas would have expanded into wooded zones and arid steppe into former grasslands, with concomitant reorganizations in animal communities. Marean (1990, 1992b; see also Marean and Gifford-Gonzalez 1991) has documented presence of arid-adapted large animal taxa in terminal Pleistocene deposits at Lukenya Hill, on the Athi Plains south of Nairobi, reflecting the late Pleistocene dry phase.
It seems likely that pastoral colonization of what is now the savanna region of Kenya and northern Tanzania was inhibited by climatic and ecological factors until sometime around the late 4th and early 3rd millennium BP. At that point, one can discern at least three (possibly more) coeval flaked-stone industries and three ceramic traditions associated with remains of early domestic animals in Kenya and Tanzania. All ceramic traditions are associated with ground stone palettes, rubbing stones, and so-called stone bowls. The last were interpreted by Leakey and Leakey (1950) as mortars and some possibly as indirect heat cooking vessels, and again as grinding equipment by Robertshaw and Collett (1983a).
The Elmenteitan and the Eburran (L. S. B. Leakey’s [1931, 1935] “Kenya Capsian”) lithic industries are typically Later Stone Age, blade-based and incorporating geometrics. The Eburran is local to the Central Rift of Kenya, enduring from around 12,000 to about 3000 BP (Ambrose et al. 1980). The third lithic grouping is in fact a more heterogeneous assortment of microlithic Later Stone Age assemblages united under the heading “Savanna Pastoral Neolithic” (Ambrose 1984b; Bower 1991; Bower and Nelson 1978; Bower et al. 1977). This term may mean different things to different researchers. I suggest that the heterogeneity within this grouping, its general similarities with East African Later Stone Age industries, and, in well-studied cases, its specific continuities with earlier, local core-reduction and tool-making practices (see Mehlman 1989), suggest that it may not be a new imported technology, but rather an array of diverse and in some cases autochthonous technologies.
Table 7.1. Archaeological and climatic sequence in East Africa, from around 20,000 BP, with references appended. All dates approximate.
Table references appear at end of chapter.
Figure 7.1. Map of East Africa, showing sites and regions mentioned in the text. Key to numbered sites: 1 = FWJJ5; 2 = GAJ14; 3 = Jaragole; 4 = Njoro River Cave; 5 = Hyrax Hill; 6 = Prolonged Drift; 7 = Enkapune ya Muto; 8 = Crescent Island Causeway; 9 = Narosura; 10 = Lukenya Hill; 11 = Olduvai; 12 = Nasera; 13 = Mumba Cave.
Reliable radiocarbon dates for Elmenteitan and Savanna Pastoral Neolithic sites are few, but the earliest coincide relatively closely with the onset of modern climatic conditions (Collett and Robertshaw 1983b; Merrick and Monaghan 1984). In south-central Kenya, the three lithic industries overlap in time, and two have adjacent, often mutually exclusive geographic ranges. Thus, there is reasonable archaeological evidence for the coexistence of groups with distinctive artifact-making practices in the 3rd millennium BP.
In contrast, different ceramic traditions cross boundaries between some lithic industries, suggesting a complex social situation. Of the ceramic traditions, the Nderit predates the others and appears to have spread southward from the Lake Turkana basin to the Serengeti (Bower 1991; Marshall in press). Nderit pottery is normally associated with Savanna Pastoral Neolithic lithics but is encountered with what Ambrose (1984b, 1990) characterizes as Eburran 5 lithics at Enkapune ya Muto Rock Shelter and Hyrax Hill Neolithic Village sites near Lake Nakuru (Figure 7.1). Later sites containing Savanna Pastoral Neolithic lithics yield another, ceramic tradition, including Narosura/Akira pottery. According to Ambrose (1984b) these ceramics are also associated with Eburran Phase 5 lithics at the Crescent Island Causeway Site, Lake Naivasha (Figure 7.1). Elmenteitan (formerly Remnant) ceramics are usually associated with Elmenteitan lithics but have sometimes been recovered from Eburran Phase 5 deposits (Ambrose 1984b) and were among the ceramics diagnosed at Prolonged Drift, considered a Savanna Pastoral Neolithic site on the basis of lithics (Robertshaw and Collett 1983a).
Subsistence diversity is reflected in these assemblages, but again not always neatly along “cultural” lines. Savanna Pastoral Neolithic and larger Elmenteitan sites contain predominantly domestic animal remains and lie in locales favored by modern pastoralists. Bone isotope research (Ambrose and DeNiro 1986) suggests that Savanna Pastoral Neolithic makers relied more on animal products than did makers of Elmenteitan artifacts. Robertshaw and Collett (1983a) note that numerous Savanna Pastoral Neolithic and Elmenteitan sites could have supported rainfall-dependent farming and that groundstone axes or adzes and stone bowls may have been agricultural implements. Sites with late Eburran lithics contain domestic animals. At Enkapune ya Muto Rock Shelter, the Eburran 5 level has domestic caprines (and Nderit/Ileret ceramics) in a level dating 4900–3000 14C years BP (Ambrose 1990; Marean 1992a), the earliest uncontested dates for domestic stock in the Central Rift.
At the risk of glossing over problematic aspects of as yet sketchy evidence, I offer a summary of what these data suggest. As the climate and biotic communities of southern-central Kenya approached conditions similar to those of today, successive immigrations of food-producing groups took place, the earliest of which involved makers of Nderit ceramics. Indigenous foragers would have had to respond to these incursions in some way, and evidence is accumulating of incorporation of domestic animals and of various ceramic wares into sites with local lithic traditions.
By mid- to late 3rd millennium BP, the upper Mara River drainage of southwestern Kenya saw replacement of Savanna Pastoral Neolithic lithics and associated ceramics by a third tradition, the Elmenteitan, which overlays Narosura ceramics in some occurrences (Robertshaw 1990). The two groups then occupied adjacent, mutually exclusive ranges for several hundred years, both using domestic cattle, sheep, and goats, but maintaining distinctive lithic and ceramic manufacturing techniques over time. The makers of Elmenteitan artifacts, perhaps more committed to cultivation, occupied woodland-grassland ecotones and open grasslands of the Mau Escarpment and Loita–Mara plains, while makers of Narosura/Akira ceramics occupied the grasslands of the Central Rift and plains south into Tanzania and were heavily committed to livestock rearing (Ambrose 1982a).
The widespread phenomenon of Nderit ceramics presents an opportunity to explore a social approach to East African archaeological materials. Nderit pottery, Louis Leakey’s (1931,1935) “Gumban A,” plus so-called Ileret Ware from east Lake Turkana, includes vessel forms ranging from very large bowls to smaller vessels, all of which are typified by internal scoring of the bowls, application of multiple, broad panels of incised or comb-stamped impressions over most of the outer surface of the vessels, and incised or milled rims. The earliest examples of this tradition are in the northeastern Lake Turkana basin, with dates between 4500 and 4000 14C years BP (Barthelme 1985), associated with remains of domestic livestock, fish, and other aquatic vertebrates in apparently residential encampments. At least five nonresidential funerary sites in the Lake Turkana basin, as yet neither well dated nor well published, have yielded Nderit-style ceramics (Koch 1994). Based on commonalities of artifacts, architecture, and handling of the dead, Koch (1994) defines the Jaragole Mortuary Tradition and the Jaragole Ossuary Tradition by Nelson (1993). The sites comprise implanted stone pillars up to 2 meters long, funerary mounds containing apparently deliberately fragmented human remains (interpreted by Koch and Nelson as secondary burials), fragments of many large and often exquisitely wrought Nderit-style ceramics, zoomorphic ceramic figurines (of both wild and domestic animals, the only such known from East Africa), marine shells and lithics from distant sources (Nelson 1993). Nelson (1993) reports that fragments of at least 650 large ceramic bowls were recovered from the Jaragole Pillar Site. Bones of aquatic and terrestrial vertebrates have been excavated, but no taxonomic identifications have yet been published.
Nderit-style ceramics in south-central Kenya and northern Tanzania are not associated with such elaborate funerary practices, nor have ceramic figurines been recovered from any sites. At GvJm44, one of many sites clustered around Lukenya Hill, south of Nairobi on the Athi Plains, Nderit ceramics are associated with stone bowls and domestic animals (Figure 7.1). Nelson (1993; 1992: personal communication) has noted that decoration on Nderit sherds around Lukenya seems more cursorily applied than it is on sherds in the Central Rift area, much less the Lake Turkana examples. Nderit sherds are associated with a predominantly nondomestic animal food base in some sites, as at Ambrose’s Enkapune ya Muto rockshelter in the Mau Escarpment (Marean 1992a). On the Serengeti, Mehlman (1989:458) reports rare Nderit-style sherds, confirmed by Bower and Wandibba, in association with a local and long-lived lithic industrial tradition at Nasera Rock, north of Olduvai Gorge. Nderit ceramics are also found associated with an exclusively wild fauna at the Gol Kopjes site, also on the Serengeti (Bower 1988, 1991; Bower and Chadderdon 1986). Overlying levels at the same site continue in the association of wild migratory and resident fauna with the later, Akira ceramic tradition.
How best to interpret sites with predominantly or exclusively wild fauna and Nderit ceramics? Two alternative explanations have been advanced in the literature. The first interprets these as camps of hunter-gatherers who obtained the pots through exchange with food- (and ceramic-) producing neighbors, a case strengthened by independent lines of evidence suggesting continuities with earlier, preceramic local traditions (e.g., Eburran Phase 5, Mehlman’s Nasera data).3 Others believe the sites are best understood as evidence of immigrating groups practicing a flexible mix of herding, foraging, and farming (Bower 1986, 1991).4 In the absence of other information, either scenario would seem tenable, and both have been previously proposed for some combination of ceramics and wild fauna in the East African Neolithic.
I propose a third alternative here, that Nderit pottery makers were present at the camps of local foraging peoples, reflecting networks of alliance and exchange during the early pastoral colonization of Kenya and northern Tanzania, in which gender may have figured prominently. This suggestion is necessarily speculative, inasmuch as I have not engaged in field research from this point of view, and relevant excavated collections are half a world away. There is no a priori reason why one of these explanations should be privileged over any other, and it will ultimately require assessing the relevant evidence on a site-by-site basis. However, exploring why the last option may be viable for the period when Nderit ceramics spread into East Africa is an example of a more socially focused approach. The next section develops an argument from first principles drawn from contemporary pastoralist studies and epizootiology.
Researches on modern pastoralists emphasize the constant and pressing need for herding labor within such systems (Ingold 1980) and the relation of this to pastoral social systems. Because successful stock owners are by definition perpetually short of labor, pastoralists are open to recruiting new members by any and all means, and “definitions of identity tend to be inclusive rather than exclusive” (Waller 1985:349). Historically documented pastoralists expanding into East African savannas did not act as hermetically sealed, “tribal” units, emphasizing their differences with newly encountered groups, but rather engaged in flexible and intense, if often ambivalent, relations with neighbors (Galaty 1991; Sobania 1991; Turton 1991). Alliances were common, mediated by intermarriage, in some cases by wholesale transfers of larger social groups from one linguistic and “ethnic” entity to another, and by rituals.
Demands of prospering herds and flocks for herding personnel may be a universal aspect of pastoral systems, but other social relations, particularly those between food producers and peoples who, if not exclusively foragers, gather and hunt many wild foods, may be much more dependent on historic context. Among the most influential recent writings on such relations have been those of Bailey (1988; Bailey and DeVore 1989) on the asymmetrical social and economic relations of Efe foragers with their Lese food-producing neighbors in Zaire. Bailey argues that male Efe hunters are at great disadvantage in finding spouses, because Efe families favor marrying daughters into Lese households as a form of risk reduction and “upward mobility.” This reduces the number of potential spouses for Efe men and heightens competition among them for mates. Cronk (1991) documented a similar process of “hypergyny” among the Mukogodo of northern Kenya, who recently shifted from a foraging-dominated life to pastoral subsistence very similar to that of their neighbors the Samburu. Samburu men “marry down” to Mukogodo women, providing their parents with livestock, now preferred to traditional hunters’ bride service, and thus forcing Mukogodo men to accumulate livestock to stay competitive for wives.5 Blackburn (1996) documents a similar pattern of asymmetrical out-marriage among Okiek of the Mau Escarpment in Kenya. Social and economic asymmetries between contemporary pastoralists and hunter-gatherers today are so pervasive that they may gradually drive one social system, if not the people themselves, to extinction.
It is tempting to extend this model to prehistoric cases in which communication between food producers and foragers is evident. Speth (1990) cited hypergyny as a force driving social and subsistence change among hunter-gatherers on the American plains who entered into close exchange relations with Pueblo food producers in protohistoric times. Taking a somewhat different tack, Robertshaw (1989) uses ethnographic data and Ingold’s (1980) theories on the ideological correlates of pastoral production to develop a scenario for status and power asymmetries between hunting peoples and pastoralists, assuming again that presently observable power relations pertained in prehistory as well.
However, archaeologists should use such contemporary models with caution. Ethnographically documented interactions have taken place in the colonial and postcolonial context, when food producers emphasized their bounded, “tribal” identity rather than inclusiveness, to vindicate land claims (see Waller 1985; Sobania 1991). Today’s observed asymmetries may have obtained in precolonial times, but they should not be assumed a priori, without arguments about why this should be the case. The earliest food producers with domestic stock faced a different social and ecological milieu than do recent herding peoples, and this may well have placed them in a different relation to indigenous foragers than is typical of historically documented cases.
Contemporary pastoralists recognize that climatic fluctuations typical of the grasslands they inhabit, as well as unpredictable but recurrent epizootics, put them at constant risk of herd decimation (Dahl and Hjort 1976; Dyson-Hudson and Dyson-Hudson 1970, 1980; Homewood and Rogers 1991; Ingold 1980). To buffer the impacts of largely unavoidable disasters, herders employ various risk-reduction strategies. These include: dispersing stock holdings widely through loans to other herders; parceling out animals of one male stock owner to multiple wives and their offspring in far-flung locales; developing exchange relations with farming peoples, often supported or mediated by marriage; farming and, last and ideologically least, gathering and hunting.
For the earliest pastoral peoples moving into what is now Kenya and Tanzania, some such strategies were less likely options. Stock loans, bond friendships based on livestock exchange, and far-flung marriages require a certain density of pastoralists in the landscape to be reliable insurance in times of herd disasters. At least for the span of time over which we see Nderit pottery as a widespread phenomenon, it is not evident that such densities existed. Instead, prudent pastoral stock owners may well have depended both on a highly flexible subsistence strategy of their own, as suggested by Robertshaw and Collett (1983b; see also Galaty 1991) and on establishing close and relatively peerlike relationships with indigenous hunter-gatherers.
For pastoralists entering the Central Rift and regions to the south, new disease challenges to their livestock would have made the chancy world of pastoral production even more dangerous. Possibly novel animal diseases include trypanosomiasis (sleeping sickness), spread by tsetse fly in brushy habitats, and the tick-borne East Coast fever (ECF) and ungulate-borne malignant catarrhal fever (MCF). Present-day African pastoralists avoid sleeping sickness by keeping their cattle away from brushy zones. However, herders have proactively modified tsetse distributions by burning underbrush and moving goats and cattle into riparian habitats during drier months, the time of lowest infection risk (Lamprey and Waller 1990; Stenning 1958).
The third novel disease threat to livestock may have been both less manageable and avoidable, even once recognized: wildebeest-derived malignant catarrhal fever (WD-MCF), a herpes virus close to 100 percent fatal in infected cattle, is transmitted by wildebeest calves a few days to less than three months old (Mushi and Rurangirwa 1981). Wildebeest have not been documented in or to the north of the Lake Turkana basin (Dorst and Dandelot 1970; Gautier 1981; Van Neer and Uerpmann 1989). They did, however, definitely inhabit the Central Rift as far north as Lake Baringo in the 3rd millennium BP; their bones are sometimes found in sites bearing domestic stock (Gifford et al. 1980; Gifford-Gonzalez and Kimengich 1984; Hivernel 1983). Thus, their presence in newly entered areas would have presented potentially disastrous circumstances to herders.
Heightened disease vulnerabilities, combined with lower stock densities for herd-replacement through pastoral alliances, may have made ties to local hunting and gathering peoples a considerably more important risk-reduction strategy for pastoralists than has been recently observed during the first entry of stock-owning groups into East Africa. Local foraging groups could offer labor during good years for livestock and local knowledge of exploitable wild foods during bad. Hunter-gatherers reciprocally would have gained pastoral products and perhaps cultivated cereals, valuable inputs to seasonally fat- and carbohydrate-poor diets (Speth and Spielmann 1983). The Kenyan forest environment has been shown to be poor in edible plant resources, limiting foragers’ ability to obtain carbohydrates in that zone, though some other East African savanna-woodlands are more productive of dry season foods (Vincent 1985a,b). Introduced domestic African grains, typically harvested in summer after planting with the spring rains, would have been available at precisely the point in hunter-gatherers’ yearly cycles at which nutritional stress reached its highest levels.
Ceramic technology itself, as I believe is evident from the “Nderit” sites, would have been a major medium of interaction. Recent zooarchaeological discussions have converged in concluding that substantial nutritional gains, especially in extraction of fat, can be realized by boiling in comparison to roasting or baking (Gifford 1993; Lupo 1995; O’Connell et al. 1988). Archaeological sequences in East Africa testify to the appearance of ceramic vessels with or without domestic stock. Where some domestic animals appear at the same time as ceramics, as at Enkapune ya Muto, ceramics need not reflect a wholesale shift to a food-producing way of life, but rather a technological enhancement of the existing foraging pattern (Marean 1992a; see also Parkington, Chapter 2, and Wadley, Chapter 4). Historically documented groups that rely heavily on wild foods, such as the Okiek and the Hadza, traditionally made their own ceramic vessels; others obtained their pots through exchange—but all use them.
With much of the spectrum of insurance against herd decimation seen in contemporary situations unavailable to early pastoralists, more equitable, reciprocal exchange relationships may have obtained between them and indigenous foragers than are observable today. Spouses may have moved in both directions. The concept of hypergyny is thoroughly gendered, and we may prudently refrain for the moment from assigning either relative status or gender to the “spouses” in the prehistoric Nderit case. To do so requires that we advance logical arguments about why we might suppose, for example, women were potters, or men were stone workers. If the genders of people who may have moved between groups cannot currently be discerned, we can at least conceive of circumstances in which pastoralists had much less of an upper hand in marriage transactions than they might today. A similar point has been made by Habicht-Mauche (1991) regarding the Plains–Pueblo Indian interactions in the protohistoric period of which Speth wrote.
Nonetheless, one wonders, did certain aspects of food producers’ ways of life—not their emplacement in an essentialized category of “pastoralist,” but concrete, socially manipulable features of society and technology—accord new colonists advantages when dealing with indigenous groups? If so, might these practices, and the social statuses they produced, induce foragers to shift toward intensive production of wild or domestic commodities for trade with such groups? For example, modern Okiek of the Mau Forest of Kenya “mine” the forest for honey and wild animal pelts, not only for themselves but also as their most valuable item of exchange with food producers (Blackburn 1982, 1996; Kratz 1986). Okiek honey is often made into a mead wine essential to Maa-speakers’ rituals, which also require wild animal skins. Intensified grain production, often the option of poorer agropastoralists, might provide foragers with social leverage via an important item of exchange, a luxury food for pastoralist feasts and rituals, especially when brewed into beer (Hayden 1992a). Nderit-making pastoralists entering a riskier, less known, and less productively stable environment than the Lake Turkana basin in its wet phase heyday may have paradoxically been more motivated to emphasize feasting and give-aways during times of relative herd affluence to solidify ties with local foraging groups against the bad times.
What archaeological evidence exists to assess these possibilities? Space does not permit a comprehensive enumeration, but a few lines of inquiry can be suggested. First, do sites with local continuities in lithic traditions incorporate new technologies in ways that might reflect specialized production? If Ambrose has correctly identified Eburran 5 lithics at the Crescent Island Causeway Site at Lake Naivasha, it may be such a case. The site lies next to a constant source of nonalkaline water, in excellent soil for cultivating sorghum (Robertshaw and Collett 1983a), and has yielded very high concentrations of “stone bowls,” interpreted by most as seed processing equipment (C. M. Nelson 1992: personal communication; John Bower 1993: personal communication).
Ceramic analyses should be especially informative in these terms. If the use of Nderit ceramics signified membership in exchange networks, rather than in a specific cultural-economic “ethnic” group, paste composition of vessels could indicate whether pots were moving between groups, or whether potters were doing so. Habicht-Mauche’s (1987, 1991) demonstration that persons in “bison hunters” camps on the western plains of Texas made cookpots virtually indistinguishable from those of the Rio Grande Pueblos from local Plains clay sources is an example of this type of research. Kiriama’s (1984) petrographic analysis of Nderit ceramics and that of Langdon and Robertshaw (1985) of Elmenteitan ceramics suggest localized pottery production. Systematic analysis of other East African Neolithic ceramics will be relevant to assessing spatially extensive patterns of social alliance and exchange. We already know obsidian was traveling over great distances during this period (Merrick and Brown 1984; Robertshaw 1988).
Closer technological and stylistic analysis of ceramics can also elucidate the place of pottery production in social display. For example, one explanation of the rather “sloppy” decoration on the Lukenya Hill Nderit pottery noted earlier is that these were made by foragers attempting to signify their membership in the social network in which such pots circulated. Yet another explanation for the less intensive and carefully applied decoration of Lukenya Nderit pots is that their makers may have been food producers who were simply not participating in display of “valued-added” wares as intensely as others making and using the ceramics, perhaps because of their location or resource base.
Future analysis of dietary status and life history in light of determinable sex within the Neolithic skeletal sample, and closer examination of funerary handling of individuals, should shed further light on early East African food producers’ society and aspects of gender. This project is complicated by lack of cemeteries and a rather small skeletal sample (Schepartz 1988), but some interesting possibilities exist. For example, stone vessels apparently had great significance in the social and ritual lives of Nderit and other early food producers, because these appear in burials associated with Nderit, Narosura, and Elmenteitan lithics or ceramics. The Njoro River Cave crematory presents an interesting if unique case. The cave yielded at least 47 males, 20 females, plus 11 individuals too poorly preserved to make a sex determination (Leakey and Leakey 1950). Lavish funerary furniture includes decorated gourds, necklaces of semiprecious stones and sedge beads, worn in some cases by putatively female individuals, and an intricately carved wooden vessel about 15 cm high with proportions of a drinking vessel. Equally interesting is the virtually identical number of three separate types of grinding-related equipment—stone bowls (78), lower grindstones (77), and pestle-rubbing stones (78)—and their equivalence to the number of individuals estimated to be represented in the excavated sample (78). While bodily ornamentation seemed to vary with gender and age (Leakey and Leakey 1950:26–37), deposit of grinding equipment and persons in equal numbers reflects a different system of signification in which gender differentiation was not apparently so marked (see also Robertshaw and Collett 1983a). These examples illustrate the potential of conceptualizing artifacts in a social context.
Given its speculative nature, I do not purport the foregoing Nderit example as the only possible reading of that data; rather, it suggests how a more socially oriented approach to the East African archaeological materials might develop. I will end with two observations. First, including gender in archaeological research need not entail loss of methodological rigor. Considering socially coordinated production, distribution, and consumption (and gender as a factor in them) in fact demands that we dispense with sloppy thinking about relations of ecology, technology, and production (Bender 1979), and with implicit, ethnocentric or androcentric epistemologies of archaeological objects and sites. Far from sanctioning scientifically weak archaeological interpretations, adding a consideration of gender to archaeological exegeses can unite various strands of archaeological theory into an undoubtedly challenging but nonetheless well-grounded enterprise.
Second, as we move toward archaeological study of human social relations, we move to another, more complex level of inference, in which no single body of theory or method that has served us well at earlier levels may suffice (see Gifford-Gonzalez 1991). Well-worn paths to inference available to us at less complex levels won’t get us where we want to go, nor do they necessarily even indicate promising directions to take. We must innovate on what we know, while remaining committed to archaeology as an evidence-based and -monitored practice. Archaeologists in Europe and the Americas have gone some way in this project already, and we may learn about how to conduct such research from their work. Interesting examples juxtapose multiple lines of evidence and multiple bodies of relevant theory and method, resituating plausible lower-level inferences within more complex arguments, generating new expectations, and testing them with archaeological evidence. This work requires a discerning use of actualistic knowledge, plus a critical assessment of our own working concepts of “the social.” It should not abandon the concreteness of empirical archaeological evidence, nor the rigor of a systematic, self-testing methodology. Jones’ ecologically grounded (1996) study of prehistoric food procurement systems along the central coast of California is such a rigorous approach to assessing the role of gender, as are the chapters in this volume by Parkington, Stahl and Cruz, and Hall. This kind of work demands a more sustained and intense collaboration than has hitherto typified research on the later span of East African prehistory.
Africa seems to have always presented humans with great challenges. We who study her past can choose whether we will engage with the challenge of knowing more of the distinctive lives of African men, women, and children who lived before us.
Writings of and informal conversations with John Bower, Fiona Marshall, Bernard Mbae, Michael Mehlman, and Charles Nelson, many of whom have shared as-yet unpublished information with me, have helped me greatly. As well, I have profited from reading the excellent research of Stanley Ambrose, Curtis Marean, and Peter Robertshaw. My analyses of East African faunas were undertaken with the permission of the Kenyan and Tanzanian governments and the support of the National Museums of Kenya. Susan Kent, Arek Marciniak, and Michael Mehlman offered very helpful comments on earlier drafts. I alone am responsible for the errors of fact and judgment herein.
1. For example, after my initial interviews with the local head of Kenyan Police and the state-sponsored Administrative Chief about my planned ethnoarchaeological fieldwork, I first physically entered the large Dassanetch settlement at Ileret in far northern Kenya. An assertive female adolescent immediately strode up to me and demanded in Swahili, “Are you a girl or a woman?” It was obvious that which option I specified would determine my status among the scores of onlooking women, whose interest in me was quite different from that of the Administrative Chief, who had treated me not only as a foreign white under the protection of a powerful (white Kenyan male) player in the informal politics of the area, but also as an honorary male.
2. Let me stress that this structure of research is neither the product of conscious “bias against women” nor restricted to male researchers. I believe most of my own ethnoarchaeological and archaeological research in East African prehistory has been as androcentric as anyone else’s, and that we all have participated in theoretical systems that downplay the social context of human action.
3. Mehlman (1989) suggests a very similar scenario for incorporation of Kansyore tradition ceramics, without domestic animals, into local forager groups circulating around both Nasera Rock and Mumba Cave.
4. Collett and Robertshaw (1983a) have proposed that sites containing Elmenteitan ceramics and wild fauna, such as Prolonged Drift (Gifford et al. 1980), are the products of “pastoralists in recovery” from disastrous losses of their livestock.
5. As well, Wilmsen (1989; Wilmsen and Denbow 1991) has argued that San speakers of the Kalahari have been in close and asymmetrical relations with neighboring food-producing groups, and with imperialist European powers beyond, for at least a century, which relations have determined their participation in or dropping out of food production as a subsistence base.
1. Hills 1978; Marshall 1990.
2. Ambrose 1980; Bower 1991; Langdon and Robertshaw 1985; Leakey 1931; Leakey and Leakey 1950; Marshall 1986; Merrick and Monaghan 1984; Nelson 1980; Robertshaw 1988, 1989, 1990; Wandibba 1980.
3. Ibid.
4. Leakey and Leakey 1950; Merrick and Monaghan 1984.
5. Ambrose 1984a,b for all Eburran citations.
6. Marean 1992.
7. Richardson 1972; Richardson and Richardson 1972.
8. Ambrose 1984a,b; Ambrose et al. 1980.
9. Robertshaw et al. 1983; Robertshaw 1991.
10. Mehlman 1989.
11. Butzer 1971; Butzer et al. 1972; Hamilton 1982.
12. Butzer et al. 1972; Coetzee 1987; Hastenrath and Kutzbach 1983.
13. Butzer et al. 1972; Hamilton 1982; Isaac et al. 1972.
14. Butzer 1971; Butzer et al. 1972; Hamilton 1982
15. Butzer et al. 1972; Hamilton 1982; Isaac et al. 1972; Johnson et al. 1996; Livingstone 1975, 1980; Richardson 1972; Richardson and Richardson 1972.
16. Butzer et al. 1972; Hamilton 1982; Isaac et al. 1972; Livingstone 1975, 1980; Richardson and Richardson 1972.
