Demography of the Hadza an increasing and high density population of savanna foragers.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 89:159-181 (1992) Demography of the Hadza, an Increasing and High Density Population of Savanna Foragers NICHOLAS G. BLURTON JONES, LARS C. SMITH, JAMES F. OCONNELL, KRISTEN HAWKES. AND C.L. KAMUZORA Departments of Education, Anthropology, and Psychiatry, University of California Los Angeles, Los Angeles, California 90024-1521 (N.G.B.J.); 11 Rue Waasserklapp, L-5681, Dalheim, Luxembourg (L.C.S.); Archaeological Center, Department of Anthropology, Uniuersity of Utah, Snlt L u h ~City. TJtnh 841 12 fJ.F.O.’C.): Ikpnrtmmt ofAnthro.polog~, Universitn, of Utah, Salt Lake City, Utah 84112 (K.H.1; Demography Unit, IJniuersity of Dar es Salaam, P.O. Box 35047, Dar es Salaam, Taiizania (C.L.K.) KEY WORDS Hunters and gatherers, Costs of children, Demography, Sub-Saharan Africa ABSTRACT This is a report on the demography of the Hadza, a population of East African hunter-gatherers. In it, we describe the results of a census, and our estimation of age structure, survivorship, mean age of women at childbearing, number of live children, total population size and density, and rate of change since 1967. We show that relevant measures fit closely the stable population model North 6 chosen by Dyson to represent Hadza demography in the 1960s. We compare aspects of Hadza demography with surrounding non-Hadza and with the !Kung. Among other things, we find that the Hadza have a higher population density, higher fertility, and a faster population growth rate than do the !Kung. These demographic differences are consistent with our expectations, which were based on differences in the costs and benefits of foraging in the two regions. We also show that Hadza demographic parameters display remarkable consistency over the past 20 years. Since neighboring populations have been encroaching on the area used by the Hadza, and Hadza foragers have been subject to interludes of externally imposed settlement, this consistency is surprising. We discuss some of the implications. o 1992 Wiley-Liss, Inc. In this paper our first aim is to present data on the demography of the Hadza, a hunting and gathering population of eastern Africa. Our second and more general goal is to contribute to inquiry into hunter gatherer reproductive strategies and the ways in which these may be shaped by features of local ecology. To this end we test some simple predictions, drawn from biological principles, about differences between the Hadza and the !Kung. The demography of people who obtain their food by hunting and gathering, and especially the demography of the !Kung, has strongly influenced ideas about human pop0 1992 WILEY LISS, INC. ulation regulation, behavior, and evolution. The !Kung deserve attention because they have been so well studied (e.g., Howell, 1979; Lee, 1979). But they also have been considered important because, as foragers in the sub-Saharan savanna, they have been held to present the best available model of Pleistocene forager adaptations (e.g., Lee and DeVore, 1968). Given the shortage of material on sub-Saharan savanna huntergatherers, it seems important to present data on a second population. Received February 11,1991;accepted March 19,1992. 160 N G BLURTON JONES ET AL. Contemporary discussion of hunter-gatherer population regulation took its lead from Birdsell (1953, 1968). Investigating the ecologically reasonable expectation that population would be linked to resources, Birdsell showed that aboriginal Australian population densities were greater in areas of greater rainfall. He assumed rainfall to be a good proxy measure of resources, an assumption supported by the observations of Coe et al. (1976). Martin and Read (1981) obtained the same result for a small sample of Atrican hunters and gatherers iiduJing the Hadza. While biological anthropologists and some demographers pursued the physiological, proximate causes of fertility (Wood et al., 1985; Wood and Weinstein, 1988; Ellison, 1990), others continued to pursue ecological, ultimate causes. Many anthropologists followed the zeitgeist of the 1960s and 1970s towards the group function, “prudent breeder” side of the argument among Birdsell’s biological contemporaries, and concluded that hunter-gatherers were able to achieve zero population growth safety below “carrying capacity,” and to receive with this a good deal of leisure time. In biology, individual-centered adaptationist thinking prevailed, and gave rise to evolutionary or behavioral ecology. Recently this has reentered anthropology (Winterhalder and Smith, 1981). Subsequently, three views of modern hunter-gatherer society and population became widely influential. One portrays them as the “original affluent society,” unencumbered by the “infinite wants” of commercial societies. Free of market temptations, they want little, and so spend little time at work, share what they have, and prudently restrain their reproduction t o remain safely below a Malthusian equilibrium with their resources (e.g., Sahlins, 1972; Harris, 1988: 250-251; Haviland, 1989: 363-371). The second view portrays modern hunter-gatherers as specialized components of regional or larger economies, whose behavior is determined by their dependence on farmers or herders or trading involvements with the world political economy (Headland and Reid, 1989). They may be seen as the lowest underclass. Dispossessed of lands and resources, and subject to fluctuating opportu- nities for trade and employment, they are at the bottom of a deeply stratified international economy, their behavior determined by the exploitation of those better placed t o gain from the world system (Schrire, 1980; Wilmsen, 1989). Each of these two views may serve a useful purpose, countering, on the one hand, a notion that foraging is a poor and unsatisfying way to make a living and, on the other, that foragers are timeless isolates who have no neighbors and no history. But the first implie; r; suspensior, of M d t h t l s i 3 n processes that deserves careful scrutiny. And both these influential portraits suggest a sameness in habits among hunter-gatherers which available ethnography shows to be false. Each implies that variation in patterns of living among foragers in different places is due to differences in the temporal depth and extent of colonial domination (Service, 1962; Steward, 1955; Solway and Lee, 1990). The third view, which has provoked the greatest body of recent quantitative fieldwork, attends to the particular features of local plant and animal resources and the way these alter the costs and benefits for varying solutions to the problems of living which people confront day by day. This view is often characterized by predictions drawn from models and thinking in evolutionary ecology (Smith and Winterhalder, in press; Hill and Hurtado 1989). and holds that people vary opportunistically in behavior ar?d reproductive patterns, their largely culturally transmitted behavior conforming to normal biological concepts such as adaptation. An adaptationist approach draws its predictions by more or less rigorous argument, from the proposition that organisms maximize fitness. For example, the combinations of fertility, offspring survivorship, and offspring reproductive success that maximizes lifetime reproductive success can be examined (e.g., Smith and Fretwell, 1974; Pennington and Harpending, 1988). In many circumstances expenditure on increased fertility will pay more than increased expenditure on offspring survivorship. Conversely, lowering fertility could pay off in circumstances in which its beneficial effect on offspring survivorship was suf- DEMOGRAPHY (3F THE HADW ficiently large. We assume here that on the whole, where it is easier to feed offspring and thus easier t o keep them alive, lifetime reproductive success is more likely to be maximized by increasing fertility than where it is difficult to feed offspring. How much of the variation in behavior within and between populations of hunter gatherers can be explained by reference to differences in these trade-offs of costs and benefits is an empirical question. Some of our expectations about Hadza behavior and demography, and their differences from the !Kung, stem from models used in this third view. Using Howell’s reproductive histories, Blurton Jones (1986, 1987)has shown that in spite of the average four-year interval between births, !Kung women may have been maximizing the rate at which they produced children who survived to reproductive age (Pennington and Harpending [ 19881, who present no data on inter-birth intervals, imply that another component of reproductive success, the total number of births, which is influenced by the length of the reproductive career, must be at least as important as inter birth interval). Blurton Jones’s analyses showed that many features of the birth spacing patterns made biological sense given quite specific features of local wild resources and their spatial distribution. A critical determinant seems to have been the marked dependence of !Kung children on their mothers for food, and the great distance between niongongo nut groves and dry season water (Blurton Jones and Sibly, 1978). The finding that Hadza children gather much of their food for themselves (Blurton Jones et al., 1989), suggests that a Hadza woman who withholds provisioning effort from her children in order to support another pregnancy risks their survival less than would a !Kung woman, whose children are entirely dependent on her and other adults (see also Blurton Jones et al., in press). Expecting women to maximize their reproductive success, we then expect Hadza women to divert effort to their next pregnancy earlier in the life of each child, and thus to produce more children. In short we expect the Hadza to have shorter inter birth intervals and higher fertility than the 161 !Kung. If these expectations are met they will add further weight to the mounting evidence that quite specific features of local ecology have important effects on the reproductive costs and benefits of foraging and childcare alternatives and so on the behavioral and reproductive strategies that people display. Furthermore, if we expect forager populations to perform according t o biological common sense, the apparently richer habitat of the Hadza might be expected to lead to a more dense population. If we expect forager populations to be subject to density dependent control like many other biological populations, we should also not be surprised to find some forager populations which are increasing, and in contrast to find others like the apparently constant !Kung population. Many alternative explanations for differences between the demographic parameters of populations can be entertained. Post hoc explanations greatly outnumber predictions. For example, Harpending and Draper (19901, and Harpending (pers. comm.) suggest that the low fertility of the !Kung may result from sexually transmitted diseases (STD), and that the region of African low fertility should be extended to enclose Ngamiland. It would be easy to attribute differences between !Kung and Hadza fertility to differences in STD or treatment. But in the absence of good data on incidence of STD and availability of antibiotics it would be hard to raise these ideas to the level at which predictions could be derived. Predictions can however be drawn from ecology. THEEASTERNHADZA The Eastern Hadza occupy a 2,500 km2 area in the Eastern Rift Valley, southeast of Lake Eyasi (Woodburn, l964,1968a,b). This region has a warm, dry climate, with a marked 6-7-month rainy season (mean annual rainfall 300-600 mm [Schultz, 19711). Much of the country is rock-strewn and hilly. Vegetation is primarily mixed savanna woodland (Schultz, 1971); medium and large mammals are locally abundant (Smith, 1980). The Hadza divide this region into several loosely bounded units, including Mangola in the north, Siponga in the east, Tli’ika in the 162 N G BLURTON JONES ET AL southwest (Woodburn, 1968b). These terms are used often in the analyses reported below. At the time of European contact, around the beginning of this century, the Hadza had this country largely to themselves, according to Woodburn (1964, and references therein). They apparently lived entirely by hunting and gathering, except at the southwestern extreme of their range. Local incursions by non-Hadza pastoral and agricultural groups are recorded in historic times as early ds the bagirmiilg d Lhe 20th century (Obst, 1912), and the 1920s (Woodburn, 1964, 1988; McDowell, 1981). Archaeological evidence indicates the periodic presence of farmers and pastoralists over several millennia (Mehlman, 1988). Non-Hadza settlement is now heaviest in Mangola and Siponga. During the past 50 years, various segments of the Hadza population have been subjected to a series of government- and mission-sponsored settlement schemes designed to encourage them to abandon foraging in favor of full time farming (McDowell, 1981; Ndagala, 1988). None of these schemes has been successful; and in every case, most of the Hadza involved have returned to the bush, usually within a few months. In each instance, some Hadza have managed to avoid settlement altogether, and continued to live as full-time huntergatherers. One settlement scheme, at Yaeda (south Siponga), was more substantial than most. Scores of Hadza were moved there in 1964, and many children attended the local school. One hundred forty-two of the 380 eastern Hadza censused by Barnicot’s team in 1967 (see next section) were at Yaeda. The settlement broke up in the mid-1970s; in Smith’s census in late 1977 only 35 Hadza were living a t Yaeda, and by 1985, only about a dozen Hadza were still living there. Throughout this period and before, Hadza families have also farmed at the village of Munguli (south Tli’ika [Woodburn, 1964, 1968a; see also Cooper, 19491).A school and clinic were built there in the 1970s by missionaries, who have since left. The village remains a thriving community. South of Munguli, a few Hadza live in the Isanzu village of Mwangeza. PREVIOUS DEMOGRAPHIC WORK AMONG THE HADZA The Hadza were subjects of demographic, genetic, and biometric investigation in 1966-67 as part of the International Biological Program (IBP), Human Adaptability Project. The project was instigated and led by Nigel Barnicot, funded by The Royal Society. As the ethnographer of the Hadza, with several years of fieldwork behind him, the role of James Woodburn in this fieldwork was clearly vital to its SIICCPC‘: These inyestigators conducted a census, measured heights and weights, estimated ages, and collected blood, stool, dermatoglyphic and urine samples. These data were for a time kept as an archive a t University College London, Anthropology Department, available to interested researchers. Publications resulting from this work have covered topics in anthropometry, demography, dermatoglyphics, and epidemiology (e.g., Barnicot et al., 1972; Bennett et al., 1970, 1973; Hiernaux and Hartono, 1980). The 1967 census data were studied by W. Brass and Tim Dyson, whose analysis provides a basis for much of the work reported here. In particular, Dyson used data from women’s reproductive histories to estimate infant mortality, and an age structure using age estimates made in the field by Woodburn and Bennett. From these he selected a stable population model (North, level 6, Coale and Demeny. 1966) with similar characteristics, and then used the model to estimate fertility and mortality rates for the population as a whole. Lars Smith conducted a second census in the dry season of 1977. In it, he recorded the height, weight, approximate age, parentage, and household affiliation of all Hadza encountered. Smith’s data on the ages of infants have proven especially useful in the analyses described below. METHODS Procedures for studying demography of small populations of people who keep no written records have been most clearly set out by Howell (1979). Both Howell and Dyson used stable population models in their analyses. The use of such models has its own DEMOGRAPHY OF THE W Z A inherent robustness. The model populations are those that inevitably arise from long term persistence of given schedules of fertility and mortality. Many measures of populations are quantitatively dependent on each other, to extents displayed in the model life tables. For instance, a certain fertility rate, coupled with a certain mortality rate, inevitably implies a certain rate of population increase or decrease. A certain combination of mortality rate and fertility rate is inevitably tied to a certain age structure. Given two or more such parameters it is possible to select a model and read from it several other parameters which inevitably follow from those measured. If observed values do not approximate the expected levels, either some measure is wrong, or an assumption of the model is seriously violated. Conversely, if several independent measures support the same model, we may have increased faith in their accuracy and in the likely accuracy of other predictions of the model. The models assume limited migration in or out of the population, and unchanging levels and patterns of age specific fertility and mortality. In reality these must vary to greater or lesser extents but if they do not vary too much, nor vary in a consistent direction we may expect the models to perform adequately. Commenting on departures between observed and modeled values, Howell remarks that “the vital rates in the stable population model should be expected t,o characterize not a particular year, but rather a much longer period, perhaps as long as a century. The stable population model, then, should describe the population on the average over many points of time, but not at every point” (1976:144). Like Howell (1979) we approach the problem of determining the demographic characteristics of a hunter gatherer population not as an empirical, inductive, problem, but by using what data we can acquire to test the null hypothesis that the population conforms to a stable population model. Furthermore, instead of attempting to select a model de nouo, we test two null hypotheses: that our data fit the model chosen by Dyson t o represent the Hadza censused in 1967, or that our data fit the model chosen by Howell for the Kung. 163 We report whatever data we could obtain that can be compared to Dyson’s and Howell’s data or models. We are fortunate that several measures could be obtained, and that they were independent of assumptions about which model was appropriate for the Hadza. We present five sets of figures to compare the Hadza in 1985 with the Dyson and Howell models: 1. The age structure of the population in 1985, derived from an age ranking, and the place in this ranking of 35 people itf hiLu*ii age. 2 . Survivorship to 1985 of people who featured in the 1967 census. 3. Mean age a t childbearing, for mothers of small babies in 1985, and in previous censuses (we collected no reproductive histories). 4. Number of live children per woman against age of woman. 5. Calculation of the total population in 1967 using a “capture-recapture” argument, and then using the difference between the 1967 and 1985 populations to estimate rate of increase. To help place the Hadza in a broader perspective we use data from the Tanzanian national census to compare the Hadza age structure with the age structure of three electoral wards that surround (and include) Hadza country. 1985 census procedures The 1985 census was conducted by Smith and Blurton Jones, with the help of two Hadza field assistants. The goal was to identify all Hadza currently living on or near traditional Eastern Hadza country. Starting at the northernmost camp in Mangola region a t the beginning of June, the census team worked south, at each camp asking where nearby camps were to be found, and recruiting help in locating them. The tour ended at the villages of Munguli and Mwangeza in the last days of the month. In each camp they visited, team members recorded the names, parentage, and household affiliations of all residents. Data were collected from a total of 36 camps plus the two villages, including about five camps in Si- 164 N G BLURTON JONES ET AL ponga not visited by the team, but reported by residents visiting camps where team members were working. Data from these reports were cross checked by Smith and the two field assistants. Periodically, at two or three camps in each local area, team members gathered groups of four or five older people for interviews on the survivorship of individuals recorded in the 1967 census. (This exercise was facilitated by the availability of anthropometric data obtained from the University College IBP archive wilh iliie pcliiii&un Uf'khe then head of the Anthropology department.) Similar groups were assembled to discuss the age rank of individuals listed in the 1977 and 1985 censuses. The two Hadza field assistants played a crucial role in this exercise. Both are young married men who are literate, have attended secondary school, and had worked with Smith on the 1977 census. Each has an encyclopedic knowledge of Hadza identities and genealogies. Their role was to collect the basic census data, i.e., to enumerate camp membership by observation, and to develop information on parentage and household affiliation through discussion with camp residents. This approach overcame the extreme resistance, sometimes overt hostility, shown by many Hadza to questions about their names. Such questions are especially important because the Hadza often change their names, or are known by more than one name. Clarity about names is vita! ic: questions about age ranking, survivorship of the 1967 census population, and the identities of individuals recorded as infants in the 1967 and 1977 censuses. We have two independent checks on the accuracy of the family compositions and genealogies collected in this census. One is provided by Smith himself, who lived with the Hadza for four years in the rnid-l970s, and in the process came t o know many individual Hadza well. More recently, Blurton Jones, Hawkes, and OConnell have conducted extended periods of behavioral ecology fieldwork in Tli'ika (Blurton Jones et al., 1989; Hawkes et al., 1989; O'Connell et al., 1988a,b, 1990),and as a result, know most of the 200-250 people who normally live there. Data gathered in the 1985 census match well with information developed during these longer periods of fieldwork. Data from both sources also match those recorded in the 1967 census. As a result of all this, we have substantial faith in the validity of our data. ANALYSIS AND RESULTS Hadza numbers and density The 1985 census recorded 719 Eastern Hadza. This is a slight underestimate of the actual Hadza population. Fic!dwork cmditions, coupled with the Hadza habit of changing names, led to various omissions and overinclusions: 719 people were recorded in the census but 13 of these did not receive an age ranking. Our box of cards used for age ranking contained 775 cards and all these were placed in the ranking. Subsequently we discovered that 22 people were entered twice in the age ranking (but only once in the census). Forty-seven people had been age ranked but were not recorded in the census. They were known to us from previous censuses, and known to our assistants and informants. Some were living alone or with non-Hadza. Thus the actual number of eastern Hadza in June 1985 was probably around 750. The census includes 125 people living at Munguli and Mwangeza, most involved in both farming and foraging; 233 in Tli'ika, all essentially full time hunter-gatherers (see also Bunn et al., 1988); and about 180 each in Sipunga and Mangola, foraging, working for farmers, or farming themselves (McDowell, 1981; Vincent, 1985). The mean population of 36 bush camps is 16.5; the range is 2-48. Camp size varies with season. To calculate population density we measured the area occupied by the Hadza from the Lake Manyara 1/250,000 scale map (series Y503, sheet SA-36-16, edition 1-TSD), published by the Surveys and Mapping Division, Ministry of Lands, Housing, and Urban Development, Tanzania, 1980. On this map, we identified twenty-six 10 x 10 km squares in which Hadza were living at the time of the census. Since several of these squares include areas where Hadza seldom go, we conservatively subtracted one square from the total. Still included is the Yaeda D E M O G W H Y O F THE HADZA Plain, which the Hadza regularly cross, but rarely exploit. The resulting figure is somewhat lower than Woodburn’s (1968a: 49) estimate of “well over” 1,000 mi2. Given our estimate that Hadza country includes some 2,500 km2, the overall regional population density for the Hadza is about 0.30/km2 (0.74/mi2). Actual density varies locally and with the seasons, probably reaching a peak in Tli’ika during the late dry season, when large game animals are concentrated around a small number of water sources (Smith, 1980; O’Cofinel! et a]., 1988a,b). Excluding villagers, the regional density is about 0.24/km2(0.61/mi2). The density of the Hadza population is nearly an order of magnitude greater than the figure reported for the central Kalahari San (0.03/km2; Tanaka, 1976) and three to six times higher than those noted for the Dobe !Kung (0.042-0.080/km2). Yellen and Lee (1976: 28) give “the Dobe-/Du/da region” as 11,O.OO km2, while Lee (1979: 41) makes the Dobe area about 140 x 65 km, or “roughly 9,000 km2” (actually 9,100 km2). Lee (1979: 42-43) gives the “Dobe area” population as 379-584 (the number varies from year to year), which in an area of 9,100 km2, yields densities in the range 0.042-0.064 people/km2. Some doubt remains about including the /Du/da people. If we count these 145 souls, and use Lee’s 9,100 km2 area figure, the population density is 0.08Okm’. If we use Yellen and Lee’s 11,000 km2 figure, we get 0.066/km2. The use of stable population models supposes limited migration. We believe that Hadza migration is quite limited, both as emigration, and as movement between eastern and western Hadza. There are apparently some 200-300 Western Hadza. Smith has visited them and noted that they have for some time led a much more village centered life than most Eastern Hadza. We concur with Woodburn (1968a): “Little contact occurs between the eastern and western Hadza: only very occasionally have individuals moved temporarily or, more rarely, permanently from one of these areas to another.” Our efforts to locate survivors from the 1967 census give some information about emigration. Of people aged less than 20 in 1967 (the group most likely to have 165 emigrated by 19851, 9 out of 63 males and 6 out of 73 females were living outside Eastern or Western Hadza country. Immigration would prima facie seem impossible. But the ancestries of a number of Hadza include Isanzu parents. (The Bantu speaking Waisanzu farmers live to the south of Hadza country.) Barnicott et al. (1972:634) report that ‘‘for 437 subjects the proportions of the ancestral components are Hadza 79.8%, Isanzu 17.3%, Sukuma 1.7% and Iramba l.2%.” (These figures apparently include gra1idpdieiitb.j Of the 275 children in the 1985 census, nine had non-Hadza fathers. Age structure Developing an age structure for the Hadza is fairly difficult. As is common for modern hunter-gatherers, there is no comprehensive set of written birth records to which an investigator can refer, and most Hadza know neither their own age nor the year in which they were born. Moreover, relative age appears not to be extensively distinguished in Hadza kinship nomenclature, nor is it generally given attention. Age ranking While they lack the !Kung concern with relative ages, the Hadza showed some interest in comparing people on the basis of who was born first. Informants worked long hours on this task with Smith. Major problems involved the relative ages of people from different parts of Hadza country. who are not well known to all informants, and those of children, who may have changed names often, or who may be confused with siblings. The final ranking includes 775 individuals, ranked from oldest (1)to youngest (775). Of these, 706 are listed in the 1985 census, 22 turned out to be duplicates (for instance due to changing their name between 1977 and 19861, another 47 were known from previous censuses but not encountered during 1985. Thirteen people recorded in the 1985 census were inadvertently excluded from the age ranking. Known ages Forty-four people in the ranking are of known ages, ranging from 1-29 years. These data come from several sources. Some of 166 N.G. BLURTON JONES ET AL. these individuals were small babies in the 1967 census, and we assume their ages were accurately estimated to the nearest year by IBP investigators Woodburn and Bennett, both of whom had lengthy experience working with people in East Africa. These people were thus about eighteen years of age in 1985. Others were small babies at the time of the 1977 census, and therefore about eight years old in 1985. Still others had their birth years or dates recorded, either in the IBP data or Smiths mid-1970s journal. Some r , c x ~ b c r ~were s dse noted dixing Blurton Jones and Smiths visits in 1982 and 1984. Ages of four older people are estimated less directly, and unfortunately the estimates for three of them are not in every respect independent of all the measures one would wish to predict. Nonetheless, these estimates are not based on visual estimates of their ages as old people, since many are not prepared to believe older folks are as old as they look, and because margins of error are undoubtedly greater for seniors. Woodburn (1964) provides lists of adults, camp compositions, and genealogies recorded during his 1959-60 fieldwork. Some of these people are identifiable in the 1985 census and age ranking. We assume, on the basis of our own later field experiences, that the youngest man in our ranking listed by Woodburn as married in 1960 would then have been 20-25 years of age, the youngest woman then married, 15-20. We estimated their ages in 1985 accordingly. Cooper (1949) provides an account of his visit t o the Hadza in 1945. In it, he gives the name of his guide and information about the guide’s parents. These data match precisely with a man listed in both the 1967 and 1977 censuses, but who died before 1985. Judging from a photograph taken by Cooper, the man was in his late 20s at the time of Cooper’s visit. This fits well with Woodburn and Bennett’s estimate of age 50 twenty-two years later, and gives him an age of 68 had he survived to 1985. We allotted the age 68 to the rank of his closest 1967 contemporary still living in 1985. Least direct is our estimate of the age of an old woman who died early in 1986. At the time of the 1985 census, she had four surviv- ing children, all placed in the age ranking. The older two also appear in the 1967 census. We made visual estimates of the ages of the oldest and youngest, the former a man of about 60, the latter a woman in her low thirties, well known to us from our ecological fieldwork, who also appeared in the 1977 census as a young woman with two small children. The old woman herself appeared to be in her late 70s in 1985. Since it seems unlikely that she had her first child when she was younger than sixteen, or her oldest when She was more than 46 we estimate her age in 1985 at 76. Fitting ages to age ranks Our procedure for calculating ages for age ranks differs from that used on the !Kung by Howell (1979: 23-46), who fitted her rank order to a stable population model. Our results are independent of any assumption about the shape of Hadza age structure. We simply fitted a (three-term) polynomial regression (Dixon, 1981) of individuals of known age (n = 48) against the distribution of all individuals by age rank, and used the regression equation to estimate ages for each rank. Estimated ages are obtained from this program by including age ranks as if they were data, but with age as missing data, a useful trick not described in the BMDP manual. We chose polynomial regression because the technique will fit any curve if enough terms are added. We wanted t o minimize the assumptions built into our procedure for deriving an age structure. The only assumption inherent in this technique is that the age structure is fairly smooth, i.e., individual ages are neither massively under- nor over-represented. Checks on the age calculations Howell (1979: 31-41) checked for systematic distortions in the calculation of ages from age ranks by examining differences in age between mothers and their children, expecting these differences to be constant across all age ranks. The particular measure used was age of mother at birth of first child. Lacking detailed reproductive histories for a large number of women in the census, this 167 DEMOGRAPHY OF THE HAUZA TABLE 1. Age structure of‘the eastern Hazda population in 1985 (706 individuals eensused and age-ranked) and in 1967 C380 eastern 93 western Hadza ceiisused) 1985 _______________ N in Age ( y r L - - 5 10 15 20 25 30 35 40 45 50 55 60 65 70 701 Total interval ( 0 4 _, _etc ) _ -__ 112 84 79 67 61 48 47 39 32 0“ Y‘ 27 21 24 20 18 706 Q below age ~ 15 9 27 7 39 1 48 6 57 2 64 0 70 7 76 2 80 7 34 1 88 2 91 2 94 6 97 4 100 0 Dyson model (% below age) -___ 16 4 29 1 40 3 50 2 59 0 66 7 73 4 79 1 84 1 ss 2 91 7 94 6 96 8 98 3 100 0 - ~~ 1967 _______________ N in __-___~ ~ Dyson interval observed ( 0 4 , __etc )________ (‘?c below age) - ___ -75 15 9 82 33 2 42 42 1 26 47 6 27 53 3 37 61 1 39 69 3 46 79 1 27 84 4 ~ 5 04 3i 14 9 15’ 94 9 96 8 __ __ 473 15 people over 60 years old A second way to check our age estimates is measure is unavailable to us. The oldest surviving child of a woman in our sample might to compare them with those made visually be her second or third born. Instead, we cal- by Woodburn and Bennett for people listed culated age differences for all possible in both the 1967 and 1986 censuses. Visual mother-child pairs in the 1985 census. Mean estimation has often been criticized as misdifference in ages of mother and oldest sur- leading (e.g., Howell 1979: 23-24), yet we viving child is essentially the same for moth- found strong correlations between the reers aged 40-50 as for mothers aged over 50 sults of our method and those of the Wood(30 years for the latter, 29 for the former, burn-Bennett visual estimates: r = 0.9717, some of whom might yet raise their score by and slope 0.9376 for men (n = 122, giving birth again). Mean ages of mothers of P < 0.001), r = 0.9546 with the slightly less children born in census years (discussed be- impressive slope of 0.8945 for women low) are also similar across all three census in = 121. P c; 0.001). There are almost certainly some errors in cohorts. Both results indicate no systematic both the age rankings and in the estimates distortions in age estimates. Age estimates for older people are very of ages of particular individuals. The errors sensitive to variation in the ages given for are potentially most important in analyses the four oldest adults of “known”age. They involving subsamples of older individuals. are also sensitive to the number of terms Nevertheless, we are confident that our age used in the polynomial. If the regression calculations provide an accurate overall picprogram is allowed to run until it reduces ture of Hadza age structure in 1985. the residuals for the four oldest “known”age adults t o very low levels, then a strong dis- Resulting age structure crepancy emerges in the age gap between The age structure is a count of people remother and child for mothers over 60 versus those aged 40-60. More consistent results corded in the 1985 census. People listed in are produced by the three-term equation, previous censuses who were alive, but not which in effect smooths out the estimates of counted in the 1985 census are excluded, the four “known” age adults, rather than by even if they had been ranked on relative age. insisting that we regard our estimates as Table 1and Figure 1show the age structure given by our aging system. Table 1 also precisely correct. N.G. BLURTON JONES ET AL 168 Hadza Age Structure 70- 1 65-69 - fi0-64 . 55-59 . 50-54 . 45-49 - 40-44 * 30-34 20-24 ,. ,“ ,- ‘ 0 10-14 I 1 0-4 60 40 20 No of males 0 20 40 60 No of females Fig. 1. Age pyramid for Eastern Hadza in 1985 shows Dyson’s observed and modeled age structure for the 1967 census, based on Woodburn and Bennett’s visual estimates. A Kolmogoroff-Smirnoff test indicates that our 1985 age structure is not significantly different from either the 1967 age structure or Dyson’s model. Comparison of the 1985 Hadza data with Howell’s model of the !Kung indicates that the Hadza population is significantly younger (Table 1)(Kolmogoroff-Smirnov test. P < 0.01, Siegel, 19561. We return to these results below. However, Hadza age structure varies with location. Specifically, the Tli’ika population (foragers in unencroached habitat who obtain more than 95% of their food by hunting and gathering) is markedly younger than the Hadza population as a whole, and the Siponga and Mangola populations (with more mixed subsistence in the most degraded habitat) correspondingly older. Table 2 shows the age structure in three Hadza regions and the village of Munguli. (The small number of people found in the north of the region called Han!abe are included in Mangola, where they often spend time. Those found in the southern border of Han!abe and Tli’ika are included in Tli’ika). We can compare the figures in various ways, for example by examining the proportion of peo- ple aged less than 15. This proportion ranges from .4469 for Tli’ika to .3351 for Siponga. Changing the cut-off point makes little difference (Table 2). Tli’ika and Munguli have substantially younger populations than Siponga and Mangola. The latter regions, most heavily settled by other tribes in recent decades, are significantly older than Tli’ika (testing proportion under 15, chi square = 5.98, P < .025 > .01). The regional differences among the Hadza, while they confuse the overall demographic pictiirr-.,arP likely te FW;C instructive about the costs and benefits of various resource acquisition, parenting and mating strategies. The data suggest that those who live in the bush, or who farm in a predominantly Hadza village, have the most living children, and differ most strongly from the !Kung, while those who have least surviving children are those who live in areas where people of other tribes farm, who occasionally employ Hadza as laborers. Survivorship over the past 18 years: An estimate of age specific mortality In 1985, Smith and Blurton Jones asked Hadza about the identities of all individuals listed in the 1967 census, whether they were alive or dead, and if alive, where they were living. The process was repeated many times, and multiple reports of a death were required before it was considered confirmed, Efforts were made to locate all survivors. Of the 380 Hadza listed in that census, 102 (27%) were dead, 258 (68%) alive, and 20 (5%)unaccounted for (Table 3 ) . Eight of the 20 unaccounted for were males aged 23-28 (a group with predictably great mobility) in 1985. Since 1985 we have encountered 5 of the 20. We conclude that it is incorrect to assume that those unaccounted for were dead, and we have omitted all 20 from our calculations. Table 3 and Figures 2 and 3 show the 18 year survivorship of the 1967 sample, compared with the survivorship calculated for a series of Coale and Demeny (1966)model life tables chosen to vary about the models chosen by Howell to represent the !Kung (West 5 ) ,and Dyson to represent the Hadza (North 6). Thc scrics comprised West levels 1, 5, 169 DEMOGRAPHY OF THE HADW TABLE 2. Percentage below each age point for each, Hadza region, compared to Dyson and LKung models 5 10 15 20 25 30 35 40 45 50 55 60 65 70 100 16.4 29.1 40.3 50.2 59.0 66.7 73.4 79.1 84.1 88.2 91.7 94.6 96.8 98.3 100 17.7 28.3 44.7 53.1 61.1 67.2 72.6 79.2 82.3 84.1 88.9 93.2 35.6 97.8 100 11.4 24.7 35.5 43.9 54.2 61.4 68.7 72.3 77.7 84.3 88.5 89.7 93.9 96.9 100 18.7 29.31 41.5 51.2 60.1 64.2 69.1 73.9 82.1 86.2 89.4 90.2 33.; 96.7 100 15.7 27.2 33.5 45.0 52.9 61.8 70.7 76.9 80.1 83.8 86.4 90.8 94.6 97.9 100 11.8 21.9 31.5 40.6 49.2 57.2 64.5 71.2 77.3 82.8 87 6 91.6 Y4.Y 97.3 I on TABLE 3. Deaths and survivorship to 1985 by people in 1967 census and traced i n 1985' Women ___ Fate unknown Age (yr) Dead Alive 0-4 11 3 2 4 6 6 1 27 24 9 7 6 6 16 16 5 6 4 2 3 1 0 5-9 10-14 15-19 20-24 25-29 30-34 35-39 4044 4549 50-54 55-59 60 + 1 1 0 1 0 1 0 1 0 0 1 Dead-~ 4 11 3 2 1 4 3 2 2 5 3 4 4 . Alive ____ 10 26 20 8 8 11 12 7 9 10 6 2 0 Men ._ Fate Unknown 1 6 2 0 0 1 0 1 0 0 0 0 0 Observed survivorship _____.-____. ,7115 ,7813 ,8750 ,7931 ,7368 ,6957 ,7143 ,3333 .0833 'Column on right uses people dead or located alive, ignores people of unknown fate and 9; and North levels 1 , 6 , and 9; with life expectancies at birth of 20, 30, 32, and 40 years. Expected survivorships for each 5 year cohort (5-9, 10-14, 15-19, etc.) over the following 18 years was calculated as follows. P(x + 2 t o x + 20) = I(x + 20)/1(x + 21, where P = probability of survival for next 18 years; x = age a t start of age interval in which the individual belonged in 1967; 1 = number of survivors in a model cohort of 100,000 births, at time x (from Coale and Demeny tables). The quantity l(x + 2) is calculated by linear interpolation as L(lx + 2) = LOX) - ([I(x) - l(x + 5)] X .4). Expected survivorship for the 0-4-yearold cohort was calculated by interpolating l(x + 2) using x = 1 instead of x = 0. The assumption here was that the 1967 sample of under 5s would include few under l-year- '7 09 0.8 0.7 p 0.6 v) :0.5 f 0.4 .- 0.3 0.2 0.1 p 3 $ oq L 4 o ? mr n, % ,x Js %,S o , %, g , O l E % F i R 8 5 L ? 8 age cohorts Fig. 2 Observed survivorship over 18 year interval between censuses, plotted with survivorship predicted from models West 1,5,9. olds, and that linear interpolation from 0 to 5 would be misleading. Since mortality declines rapidly from birth to 5 this method , 8 170 N.G. BLURTON $JONESET AL TABLE 5. Models allowed calculation of total expected dead after 18 years (Table presents goodness of fit of observations to the models) 0.9 0.8 0.7 ~ . _ _ Dead _ _ ~ _ Alive _ Chi square' ~ ~ CL r 0.6 West Y) .s5 0.5 1 5 9 North 1 6 9 Observed 0.4 0.3 north 6 0.2 north 9 147 113 87 213 247 273 23.3 1.55 3.41 146 105 85 102 214 255 275 258 22.3 0.11 4.45 Data differ significantly from model prediction when chisauared 23.84 df = 1. Fig. 3. Observed survivorship over 18 year interval between censuses plotted with survivorship predicted from models North 1, 6,9. The models allow us to predict the total number of deaths expected in the 1967 census population. The match of the observaTABLE 4. Comparison of the fit of the model predictions to tions to the predictions is summarized in observed survivorship ofeach, 5 year cohort over 18 years Table 5 . Again, the closest match is to North Significance 6, and next closest to West 5 . There are Goodness of fit of difference from model Model chi-squared significant differences from West 1 and . _ _ _ _ ~ . . _ _ _ _ _ North 1. West 1 45.7 P < ,001" 5 20.5 P > .05 < .10 While it is impressive that this parameter 9 25.6 P > .01 < ,025" so closely to model North 6 we must note fits North 1 41.6 P i ,001" that the differences from models predicting 6 15.6 P > .10 9 22.0 P > ,025 < .05" substantially lower mortality levels are barely significant. When only 100 deaths * Data significantly differ from model, P < 0.05. are observed, the standard errors of predicted survivorship is around 5% (Coale and Demeny in Howell 1979, Table 5.2, p. 111). could have predicted too high a survivorship This sets the 95% confidence limits for for the 0 4 cohort. As a check, the calcula- model West 5 a t around the levels of model tions were repeated, examining survivor- West 1 and model West 9. We must accept ship of under 3 -year-olds separately from that our analysis of mortality cannot be very the 14-year-olds. This increased all the chi- sensitive. Although on inspection of Figure 2 squareds slightly (but also the d.f.1, suggest- one is drawn to the high mortality of several ing a closer fit to West 5 ( P > .10) than young age groups, and the low mortality of shown in Table 4 but leaving other signifi- the older age groups, this observation cancance levels, and the overall result un- not be taken as more than a stimulus to further research. changed. We are unable to demonstrate a difference Inspection of Figures 2 and 3 suggests that the data vary widely about the model from either null hypothesis. Our mortality West 5 and North 6 predictions. However data are most closely compatible with Dysome of the visually impressive data points son's estimate for the Hadza, but are indisare based on very small samples. The fit tinguishable from Howell's estimate for the between the observations and the models Kung. Hadza mortality is within the range was tested with chi-squared. The closest fit expected for people living a hard life, with was with North 6, closely followed by West limited access to medical facilities. 5 . Mortality was significantly lower than Mean age at childbearing predicted from models West 1 and North 1, We have data on age of mother at birth for and somewhat higher than predicted from three sets of live, censused babies, those remodels West 9 and North 9 (Table 4). ~ 171 DEMOGRAPHY OF THE HADZA TABLE 6. Age of mothers at birth of babies recorded in each census (See comments in text about oldest ages) 1985 1977 1967 Combined total n Mean SD Range 22 9 27 58 31.07 30.67 30.85 30.90 11.7 11.6 6.1 9.44 19-53 18-45 2 2 4 0 [91 19-53 TABLE 7. Numbers o f births to mothers older and younger than 35, comparing Hadza data from 198.5, 1977. and 1967 with !Kung data from Howell 1979 (page 141, Table 7.2, and page 124, Table 6.1 Mother‘s age at birth of child >35 __ s35 . . ~ Hadza !King (1963-1973) 41 153 17 26 __58 179 ‘The value of this comparison i s reduced by rounding errors for ages in 1967 census, and by comparing cross-sectionas with longitudinal data. corded in the 1967,1977, and 1985 censuses, respectively. Ages of mothers of 1985 babies are calculated from the 1985 age rankings, with the exception uf ulie 01 t w u iliutlirrj who were babies themselves in 1967 and so are of known ages. Ages of 1977 mothers are calculated from their estimated 1985 ages, and 1967 mothers from the Woodburn and Bennett estimates (because older bearers of children in 1967 had died by 1985, and so received no age ranking). The resulting age estimates for each cohort are very similar (Table 6). Examination of the raw data suggests that it is common for a Hadza woman t o bear a child when aged 40-45. The extreme value of 53 seen in Table 6 could arise from an error in age ranking. These figures, with overall mean 30.9 years, are higher than Howell’s estimated average age for !Kung mothers at birth: 27.93 t 6.58 years. A t-test indicates that the combined Hadza mean is significantly different from the !Kung mean (t = 2.23, P < 0.05). This difference has two dimensions: l i the difference in age structures, 2! differences in age specific fertility. The latter may in turn have two components. First, Hadza women may marry at a later age than do !Kung women. Later marriage would on average be expected to lead to later first births. Second, Hadza women also appear to continue giving birth at later ages than do the !Kung. Table 7 shows the number of Hadza women in the combined sample aged older and younger than 35 at birth of child in the sample years, and the number of !Kung women similarly divided for births over a 10 year period, 1963-73. The difference is significant (chi-squared = 10.04, P < 0.011. A greater proportion of births are to women aged 35 or more among the Hadza than among the !Kung. Given that !Kung have a slightly older age structure than Hadza, this suggests that Hadza indeed have higher feriility ill th& 30s than do !Kung. Phtllips et al. (1991) suggest that Hadza age specific fertility remains high after 35 years of age. Number of live children Based on the 1967 reproductive histories, Dyson (1977) calculated the number of children ever born and the number that had died for women in various age groups. The number alive can be tallied from his summary tables. Our recent census allows us to compute the number of children alive and living in Hadza territory in 1985. Because this figure is not based on reproductive histories, it is likely t o be an underestimate. Some children may have left the area, or been missed in the census. As Table 8 indicates, our numbers are similar to Dyson’s. One-sample t-tests show no significant differences from Dyson’s figures. Determining rate of increase by estimating the size of the 1967 Hadza population The 1967 census was partial; no serious effort was made t o include every Hadza. Nevertheless, data were recorded on a total of 473 individuals, including 380 Eastern and 93 Western Hadza. Here we use a kind of “capture-recapture” argument to estimate the number of Eastern Hadza in 1967. From this number, we can then calculate an annual rate of change in population size. As indicated above, Smith, Blurton Jones, and their Hadza assistants counted and age ranked 706 Eastern Hadza in the course of the 1985 census. Of these, 384 were 18 or older, and thus had been born at the time of the 1967 census. Two hundred fifty-eight of these had been recorded in the 1967 census. N.G. BLURTON JONES Er AL 172 TABLE 8. Numbers of live children for wom,en of uarious ages (Comparison of 1985 datu with Dyson’s report o f 1967 data) 1985 Age of women (yr) ..-.._-._______ 20-29 30-39 4049 .50 + ___ No. of __ children alive No. of Women __ 62 117 85 146 No. of 52 40 27 61 ___ No. of Dyson __ alive ever born children alive No. of women Mean children alive 1.19 2.93 3.15 2.39 66 164 71 41 103 42 31 32 12 1.32 3.22 3.5 - - __ T -1.29 --0.35 --0.37 ‘Mothers more than 50 years old have fewer living children than young:er mothers. This cannot he taken to indicate lower fertility in bygone years. It may instead arise because their children have died a t a n ad\ranced age, having been subjected to the reduced survivorship of the over-40s. This leaves 126 of the 1985 adults who must have been alive, but missed in the 1967 census. These 126 must also be survivors of a larger group alive in 1967, some of whom have since died or disappeared. Let us assume that their number is proportionally similar to those who died or disappeared from the group seen in 1967. Of the 380 Eastern Hadza counted in 1967, 258 were seen in 1985, another 102 were reported dead, and another 20 were unaccounted for in 1985. Thus from 122 missing/258 seen, we get the proportion 0.473. We calculate that .473 of the 126 seen in 1985 but missed in 1967 must have been alive in 1967 but died or disappeared by 1985. This number is 60 (0.473 x 126). Thus the total missed by the 1967 census includes those who survived to 1985 (126), plus those who didn’t (60),or 186 in all. The total 1967 population must have comprised these uncensused 186, plus the censused 380, or 566 in all. The difference between the estimated 1967 (566) and censused 1985 (706) populations permits us to calculate the average rate of increase (r)as follows: (formula presented in the U.S. Bureau of the Census publication: ISP Supplemental Course Series No 2. Demography Lectures p. 9.) The resulting rate of increase is 13.0 per thousand per year. Dyson (1977) estimated the rate at 13.90 per thousand (1.39%).These figures are much closer than necessary for a reasonable match. Rate of increase is inexorably determined by fertility and mortality. Since our independent es- timate of this rate matches t.hat implied by the model indicated by age structure and mortality data, we are justified in having greater confidence in the appropriateness of the chosen model for describing Hadza demography. Comparison with neighboring populations The Hadza can be placed in local demographic perspective by comparison with the results of the Tanzania National Census for 1978. Table 9 gives the population by age group in each of three electoral wards which collectively encompass Hadza territory: Mangola on the north, Yaeda Chini on the east, and Matongo (including the villages of Munguli and Mwangeza) on the south. The census does not distinguish individuals by language or tribal affiliation, but most of people in these wards are farmers and herders. Their numbers far exceed those of the Hadza. A Kolmogoroff-Smirnov test indicates that the age structure in each ward is significantly younger than that derived above for the Hadza (two-sample, two-tailed tests, P < 0.01). This suggests a much higher fertility for at least some non-Hadza, and an even higher rate of population increase, Mangola is an area of noticeable immigration; the other wards may have low levels of emigration. Mortality is also lower among non-Hadza (Sembwaje, 1984). Although we have emphasized the younger age structure and apparent greater fertility of the Hadza relative to the !Kung, they are, as one might have expected, older and less fertile than 173 DEMOGRAPHY OF THE HADZA TABLE 9. Conzparison of Hadza age structure with age structure in neighboring villages enumerated in national census . Age (w) 5 10 15 25 35 45 55 65 65 + Total .- ...-- Hadza ___________ No. % 112 84 79 128 95 71 54 45 38 706 15.9 27.7 39.1 57.2 70.7 80.7 88.2 94.6 ~ No. of people in age group and o/c people below age Matongo Mangola ____ No. 56 No. IC 1,979 1.888 1,462 1,757 1,166 815 633 512 18.3 35.8 49.3 65.5 76.3 83.8 89.7 94.4 600 10,812 their non-Hadza farming and herding neighbors. 1,339 1,158 756 1,150 863 653 422 269 236 6,846 19.6 36.5 47.5 64.3 76.9 86.5 92.6 96.5 -.__ ~- Yaeda chini ~ _ _ No. 570 1,941 1,661 1,189 1,469 1,057 757 494 331 394 9,293 20.9 38.7 51.5 67.4 78.7 86.9 92.2 95.7 demographic characteristics of the Hadza in 1985, as recorded and analyzed by us, are very similar to the Hadza in 1967, as reDISCUSSION corded by Barnicott et al. and analyzed by We have presented data on age structure, Dyson. The data and analyses in these studmortality, mean age at childbearing, and ies are substantially independent of each numbers of live children, for the eastern other, which makes the agreement between Hadza censused in 1985, and an estimation the two all the more striking. The Hadza of the change in population since 1967. To probably have had, over the long term, a life interpret these data we have compared expectancy at birth of 31-32 years, a populathem with the analysis by Dyson of the 1967 tion growth rate of 1.3-1.4% per annum, and Hadza data, the analysis by Howell of the a total fertility rate of 6.15. Their vital rates Kung data, and data from the Tanzanian must be close to the figures given by Dyson national census for the surrounding vil- in 1977: birth rate 46.7, death rate 32.8, natlages. The comparisons raise several impor- ural increase 13.9 per thousand. Thus this tant issues, some pertinent primarily to the analysis suggests, surprisingly, that there Hadza case, others more general. have been no meat changes in Hadza demography since the IBP data were collected Comparison with Dyson’s analysis of the in 1966-1967. Hadza in 1967 Comparison between the Hadza Dyson based his analysis on age strucand the !Kung ture, and on reproductive histories collected We suggested that comparing the Hadza by Woodburn. He drew his estimate of the level of mortality from the infant mortality with the Kung would be instructive. On the reported in the reproductive histories. He one hand, the Hadza might have displayed indicated the need for information on adult very low fertility, low population density, mortality. We have provided such data, and and near zero population growth as do the they best fit the model chosen by Dyson. Kung, fitting the pattern often treated as This is particularly encouraging since the typical of hunter gatherers. Such a finding source ofour data on mortality is so different could have strengthened either the view from the source of Dyson’s estimates of mor- that human populations depart from Maltality. Our calculation of rate of increase (a thusian expectations, or the view that marcapture-recapture method), which was inde- ginality to the world system shapes the bependent of our choice of model, results in havior of modern hunter gatherers to much complete agreement with the figure Dyson of a sameness. But the data show the Hadza to differ from the Kung in the directions exderives from his model. We are inclined to interpret our findings pected from the details of local ecology and as strengthening Dyson’s conclusions. The biological common sense: higher foraging re- 174 N G BLURTON JONES ET AT,. turns are observed for Hadza adults (Blurton Jones et al. 1990), and children (Blurton Jones et al., 1989), Hadza show higher density, higher fertility, and a higher rate of increase than the !Kung. Elsewhere, we reported data showing that Hadza children are economically “cheap”to raise (Blurton Jones et al., 1989). They can provide about half their own food from the age of five, and perform many useful services for adults. Hadza women are thus spared many of the constraints that apparelilly piuCluLe a sharp increase ir, offspring mortality as interbirth intervals shorten among the !Kung (Lee, 1972; Blurton Jones and Sibly, 1978; Blurton Jones, 1986,1987). For these reasons, we expected Hadza mothers to give birth to a greater number of children at shorter intervals (Blurton Jones et al., 1989). The available demographic data, ours and those of Barnicot et al., as analyzed by Dyson, imply that the Hadza are indeed substantially more fertile than the !Kung. The Hadza have a significantly younger age structure, and there was no evidence that their mortality was substantially different. The difference in age structure can then be comfortably attributed to differences in fertility. We have done no reproductive history interviews, but our data are compatible with Dyson’s conclusion, that Hadza total fertility rate averages 6.15 (compared to the !Kung rate of 4.7 for women over 45 in 1968). As Dyson points out, the Hadza rate is similar to that of many traditional agricultural populations. Indeed, it coincides with the mean of Campbell and Wood‘s (1989) entire sample of natural fertility populations. The Hadza cannot be said to duplicate the Kung restraint on fertility. The Hadza age structure is, however, not so young as that of surrounding village populations. Both the Hadza and the !Kung populations under comparison here include individuals who have not always lived as hunter-gatherers. We must ask whether differences between the two populations could arise merely from differences in the fraction of each associated with agricultural or pastoral settlements. !Kung who lived primarily as foragers had longer inter-birth intervals than !Kung who lived largely on resources at cattle posts (Lee, 1972; Blurton Jones, 1987). The cattle post subset might therefore be expected to have a younger age structure than the bush living subset. Our data (Table 2) show that the Hadza pattern is just the reverse: bush dependent Hadza have a younger age structure than Hadza who have more contact with settlements. Thus, bush living Hadza are more different from the !Kung as a whole, and so presumably even more different from the bush living !Kung, than are the Hadza as a whole. Differences between the two popiilatinne. cannot be due merely to differences in the proportions living more or less settled lives. Although the Hadza are more fertile than the !Kung, we have yet to demonstrate anything about their inter-birth intervals. Indeed, the greater number of births to Hadza women may be partly due to the practice of continuing their reproductive careers to a greater age than do !Kung women. However, they may also begin their reproductive careers slightly later than do the !Kung. Both of these patterns would require explanation. Nevertheless, the data and analyses presented here give us no reason to reject our prediction based on the abundance of resources and independence of children: Hadza IBI’s should be shorter than those documented for the !Kung. The very much higher density of Hadza population, and the high rate of increase of the Hadza, are incompatible with the notion that hunter gatherers prudently maintain their population a t a low density and almost zero population growth. That the Hadza population is increasing shows that it is neither constant nor at carrying capacity. Hadza population increase may be a recent phenomenon, due to successful exploitation of modern opportunities leading to a new carrying capacity. However, if the population models are interpreted to summarize long standing conditions (constant mortality during the preceding 25-30 years, and constant fertility for some two generations (United Nations 19671, desiderata that our match across an 18 year interval begins to approach) the increase may represent long standing conditions, such as recovery from a population crash some decades ago, or removal of some previous constraint such as DEMOGRAPHY f3F THE HADZA hostile neighbors (discussed by Obst 1912). Remarks by earlier visitors, including Woodburn (in the field 1958-60), Cooper (1949, in the field 19451, Kohl-Larsen (1958, fieldwork in 1931-19361, and Obst (19121, are compatible with a steady increase in the numbers of Hadza from the beginning of the century a t the rate of increase reported for 1967 and 1985. 175 effects. The method was certainly not designed to examine individual, short-term, and sub-population differences in reproductive strategies and it should come as no surprise if it fails to do this. It remains an excellent descriptive tool, with the attraction that it describes major patterns and long term trends. Nonetheless we should discuss the effects that we expect encroachment, and time at settlements to have upon Hadza ferOn the use of stable population models tility, mortality and migration. In addition, Our data and analysis could be taken to given this doubt that we have cast on the illustrate the strength and robustness ofthe value ofthe close fit between our predictions stable population method. Several indepen- and findings, we must consider some of the dent measures each fit the same model. Our alternative explanations for differences bedata set, and Dyson’s collected largely inde- tween Hadza and !Kung demographic papendent of each other and 18 years apart, rameters. both support the same model. But is this robustness gained a t a cost t o sensitivity? Why do we see no demographic effect of settlement schemes? Our apparently tidy results must be disIt has been widely accepted that sedentacussed in relation t o recent changes in the rization produces many changes. Changes Hadza environment. Recent decades have witnessed an accel- in fertility andlor mortality are among the erated influx of non-Hadza into the region most commonly claimed (Harpending and and a series of government and mission set- Wansnider, 1982; Handwerker, 1983; Lee, tlement schemes. We have shown that there 1972, Roth, 1985). Thus one expectation are differences in age structure between might be that Hadza who spend time a t setHadza regions, with the most encroached- tlements, or who have more contact with upon areas having the oldest age structures, farmers and villagers might show higher suggesting lowered fertility. Yet this has not fertility than those who live primarily upon changed the overall picture since 1967, nor bush foods. Our regional comparisons sugis there any trace in our age structure of the gest the opposite. The most fertile Hadza excess 11-20-year-olds who would result appear to be those who live in the bush. This from a baby boom during the life of the could of course be an effect of migration, inYaeda settlement. Individuals of’known age cluding emigration of young adults to seek bracket the main period of the Yaeda settle- the opportunities of village life. This in turn ment. The proportions of people under 20 in implies the decision on the part of parents to 1985 and over 10 do not differ from the num- stay in the bush. As Cooper (1949) rebers expected from Dyson’s model (North 6 marked, the numbers of children in Hadza with increase at 1.0% or 1.5%; chi- bush camps are truly impressive. When peosquared = 1.33 and 0.86, respectively). ple moved to a recent forcefully promoted Can the stable population method be ex- settlement scheme, some of the last people pected to be sensitive to effects of such in the bush were children, left there with changes in ecology? If the method is insensi- their grandparents for food and safety. We tive to these presumably large ecological in- see no reason to assume that settlement influences can we still claim that such ecologi- creases fertility or survivorship of Hadza. cal features as foraging return rates of The records of epidemics, and the Hadza adults and children are responsible for the view that the arrival of people of other tribes observed differences between Kung and in settlements presages disease, suggest Hadza demographic parameters? that mortality may even increase in settleMost demographers would argue that we ments. If people pursue their reproductive are expecting too much of the stable popula- interests, we would expect settlements that tion method when we expect it to show such people have to be pressured to join to have 176 N G RLURTON JONES ET AL different demographic effects from spontaneous settlement. That considerable pressure to settle has been exerted on the Hadza, while many !Kung settle spontaneously, is another interesting difference between these populations. The encroachment upon Hadza country has included noticeable destruction of vegetation, including trees, valuable berry bushes, and wildlife habitat. One might expect this loss of resources to be damaging to Hadza fertility and survivorship. But Hadza who lire near farmers are a h a~puseciLO a greater variety of economic opportunity. These include helping harvest sweet potatoes (not helping to plant them!), and guarding maize fields as harvest time approaches. In exchange for protecting the crop from wild animals (an excellent hunting opportunity) Hadza are allowed an apparently unmonitored share of the crop. Some exchange of meat for maize may occur at any time of the year where there are villagers. Thus even the loss of habitat cannot be assumed a priori to have detrimental effects upon Hadza reproduction. However, we were able to show a difference in age structure between encroached and non-encroached regions. We cannot at this stage dismiss local migration as a reason for this. But the differences strengthen our attribution of the differences between the !Kung and the Hadza to the differences in ease of acquiring food from the bush. Alternative explanations for differences between !Kung and Hadza What are the other plausible competing explanations for the difference between Hadza and !Kung density, fertility, and population change? One might be that the differences are indeed due to ecology but to very general features such as overall productivity of the habitat, further illustrating the principle demonstrated by Birdsell, and are not due to the microscopic and particular features (distance between food and water in the dry season) to which we have attended. Thus the association shown by Martin and Read (1981) between rainfall and population density may be all there is to be said. But it is worth noting that the Hadza populatiuri is more dense than predicted by their regression of density on rainfall. We are tempted to attribute this to the local proximity of food and water, and the success of children’s foraging. Another explanation might be to claim that the Hadza have lost the traditional restraints upon reproduction shown by the !Kung. Since (even after re-reading the discussion ofthe topic in Lee and DeVore, 1968) we know little about such restraint and how it is held to arise, it is hard to say what kind of event might remove it. One could, however, argue that the recent history of the Hadza has been more disrupted than that of the !Kung as portrayed by Lee (1972,19791, though less disrupted than the !Kung as portrayed by Wilmsen (1989,1990). We have not investigated the proximal causes of higher Hadza fertility but age a t weaning is surely the first place to look. Hadza evidently wean about a year earlier than !Kung (Blurton Jones et al., 1989). The physiological effect of weaning on fecundability is now well known (Konner and Worthman, 1980; Wood et al., 1985; Ellison, 1990) and the importance of this effect for demographic parameters has been demonstrated. But there is little research on what determines the time of weaning. We seek answers to this question among ecological variables such as the “costs” of rearing a child. Age a t weaning seems unlikely to be explained by availability of “weaning foods.” In both societies children are eating substantial quantities of adult food from about a year old. A recent paper by Lee et al. (1991) shows that mammals tend to wean when the offspring reaches 4 x birthweight. We might seek relationships between human infant growth rates and age at weaning. Both mother and baby might be expected to try to adjust weaning to the survival prospects of the infant, though not to agree on its exact timing (Trivers, 1974, and others). Weight might be a good predictor of these prospects. Research on the effects of mothers diet and activity upon fertility is also underway (Ellison, 1990, and others). We may note that Hiernaux and Hartono (1980) comment on the weight and relative fatness of Hadza women. Hadza women are the same height as !Kung women but some 8 kg heavier. While Hadza women forage daily and their DEMOGRAPHY OF THE HADZA digging must be energy expensive, their journeys are shorter (Hawkes et al., 19891, in a less demanding climate (Blurton Jones et al., 1989). Thus we are not short of potential proximal causes of higher Hadza fertility. But the thrust of our research has been to see how these proximal mechanisms may serve or fail t o serve the selective pressures that brought them into being. Suggestions by Harpending based on recent observations on the Herero (Harpending and Draper, 1990; Harpending pers c0mrn.l. ahout the possible influence of sexually transmitted diseases (STD) in the history of southern African populations provide a compellingly simple alternative hypothesis about low fertility among the !Kung and about some of the differences between the !Kung and the Hadza. One way to identify a substantial effect of STD is to compute the number of older women who have borne no children. Belsey (1976) contrasts populations in which this figure is 3 4 %with others where it ranges between 25 and 40%. Since we have no reproductive histories we are unable to give this figure for the Hadza. Howell (1979, p. 127) reports a figure of 10% for the !Kung. To properly test the implications of Harpending’s suggestions, that Hadza have been less exposed to sexually transmitted diseases than !Kung, or have had better access to antibiotics, we need detailed documentation and dating of the history of contact with populations infected with syphilis, gonorrhea, chlamydial, and other infections of the reproductive tract, and similar detailed history of access to antibiotics. Bennett et al. (1973) reported a surprisingly low incidence of antibodies to treponemal infection among the Hadza in 1967 but informants listed STD as one reason for leaving in search of modern medical help. Incidence of STD would be a good candidate for explaining the regional variation observed among the Hadza. But the apparent recent increase in contact with outsiders does not coincide with a discernible overall decline in fertility. Among many populations, changes in access to STDs and to antibiotics must be accompanied with so many other changes that it would be hard to disentangle their effects. Howell (1979) appeared to deal ade- 177 quately with the issue of infertility due to STDs, and it is hard to see how the presence of STD would account for the numerous features of !Kung reproduction successfully predicted by Blurton Jones and Sibly’s ecological model (Blurton Jones, 1986, 19871, such as the close relationship between interbirth interval and child mortality. Those who hold the view that contemporary hunter gatherers are best understood as a rural proletariat could explain the demographic differences, if they claimed that the !Kung have been and still are more exploited than the Hadza. There might be some truth to this, despite the massive difference in proximity to much higher population densities. Hadza have for long lived in the vortex of a whirlpool of major African language families. They have been close neighbours on all sides of quite densely populated and successful agricultural and herding societies-Sukuma, Iramba, Isanzu, Iraqw, Datoga, and Masai. Woodburn (1988) discusses in depth the relations between Hadza and their neighbors. He argues that the Hadza “immediate returns” ethos enhances their independence from their neighbors. We support much of Woodburn’s masterly social analysis, and from our more materialist perspective would emphasize only two additional factors. The demands put upon the Hadza by their neighbors may differ from the demands put on the !Kung. For example, the apparent disinterest among the Datoga in employing Hadza as herdsmen contrasts with the Herero and Tswana, and as far as we know the Datoga do not offer the inducements to laborers that Herero and Tswana offer. The relative richness of Hadza and !Kung environments may play its part. The Hadza bush environment around Lake Eyasi may provide greater opportunities, which enables Hadza to resist exploitation more effectively than the !Kung. Note that Hadza take employment protecting and harvesting a crop, not planting it! Another aspect of Hadza history may be even more germane. We argued that Hadza population may have been increasing throughout this century. If we extrapolate back far enough the Hadza vanish sometime in the nineteenth century! It seems likely 178 N G BLURTON JONES ET AL that the increase in Hadza population represents a recovery from a decline, or the removal of some previous external restraint. Three possibilities are presented by the literature. Obst (19121, Bleek (19311, and Kohl-Larsen (1958) were told of wars with the Masai. Oral history among contemporary Hadza and Datoga informants matches these accounts. The effects of this seem not to have lasted long, except for one Datoga sect who lost their livestock and took to the hills and lived for a time as foragers alongside thc Hsdza. The Masai had apparently left this area by the time of Obst’s visit. Obst’s informants reported another external influence-raids by Isanzu and Sukuma who abducted women and children. Obst gives no indication that these raids were to capture people for trade as slaves but slave trade from far in the interior was extensive until the 1870s (Sheriff, 1979). Obst’s Hadza informants claimed that these raids had declined as the abundance of elephants (the main target of the visits) declined. These raids may have held Hadza population at a level well below the carrying capacity of the environment, either through the direct effect of removal of women, or by rendering access to some resources too risky. The decline and eventual cessation of the raids as they became forbidden by central authority may have permitted an increase of Hadza population that has yet to end. The rate of the increase may have been determined by the natural resources and the technology used to acquire them. A historical factor that may underlie both the above, and which seems t o be being neglected in the discussions of Kalahari and !Kung history, is discussed by Kjekshus (1977). The rinderpest epidemic late in the 19th century apparently had radical effects on African economies and populations. Herders were subject to famine, epidemics of smallpox broke out, along with other diseases (Ferguson, 1980) and tsetse flies enlarged their range, spreading sleeping sickness with them. The Hadza could have been subjected t o these hazards, or have gained from the weakening of their neighbours. Thus in our comparison of the Hadza and the !Kung we may not be comparing two populations a t equilibrium. We have al- ready established that at least one of them, the Hadza, was not at carrying capacity. We may be comparing rates of recovery from late 19th-century disaster. These rates, given that both populations whether by choice or by the removal of other opportunities, were largely subsisting on non-domesticated foods, may well be shaped by the richness of the environments and the ecological constraints upon foraging success. It seems to us, in conclusion, that attempting to account for forager demography without ecolow. would be a t least as difficult and potentially misleading as attempting the account without examining the history of relationships with neighbours. ACKNOWLEDGMENTS We are grateful to the following for help and advice in the field, David Bygott and Jeannette Hanby; Mrs. Flora Matemu, Nick and Bridget Evans; Michael Leach; Margaret Gibb and many of her staff; our field assistants Gudo Mahiya and Sokolo Mpanda and the small teams of informants they gathered in each locality, without whom we would have collected no accurate data in the census; and to all those Hadza who have tolerated our intrusions. During the lengthy process of trying to learn demographic methods some of us received much help and tutelage from Nancy Howell. Any mistakes that remain are ours. Helpful comments on earlier drafts were also received from Tim Dyson, Henry Harpending, Robert Bailey, JGm Hill, and two anonymous referees. N.B.J. wishes to thank David Coleman, Fred Brett and Geoffrey Harrisson (convenor of the IBP Human Adaptability program) for leading him to the IBP archive, and R.D. Martin for permission to use it. We are grateful to Lynne Fairbanks, Henry Harpending, Linda Muthen, and Rich Cincotta for advice on statistics and computing, to Gabriella Kopahl for translating Obst, and Darlene Smucny for translating Kohl-Larsen, and to Darlene Smucny and Diane Crumley for entering data on the computer. We are grateful to The Tanzanian Commission on Science and Technology for permission to do research in Tanzania, and t o Mrs. A. Lyaruu for help, guidance and en- DEMOGRAPHY OF THE HADZA couragement. The fieldwork was conducted with support from The National Science Foundation, The Swan Fund, and UCLA International Studies Program. LITERATURE CITED Barnicot N, Bennett F, Woodburn J , Pilkington T, Antonis A (1972) Blood pressure and serum cholesterol in the Hadza of Tanzania. Hum. Biol. 44t87-116. Belsey, M.A. (1976) The epidemiology of infertility: a review with particular reference to sub-Saharan Africa. Bull. WHO 54t319-341. Bennett F, Kagan I, Barnicot N, Woodburn J (1970) Helminth and protozoal parasites of the Hadza of Tanzania. Transactions of the R. SOC. Trop. Med. Hyg. 64t857-880. Bennett F, Kagan I, Barnicot N, Woodburn J (1973) Studies on viral, bacterial, rickettsial, and treponemal diseases of the Hadza, and a note on injuries. Hum. Biol. 45t61-68. Birdsell J B (1953)Some environmental and cultural factors influencing the structure of Australian aboriginal populations. Am. Nat. 871834);171-207. Birdsell J B (1968) Some predictions for the Pleistocene based on equilibrium systems among recent huntergatherers. In RB Lee and I DeVore (eds.): Man the Hunter. Chicago: Aldine, pp. 229-240. Bleek DF (1931) The Hadzapi or Watindega ofTanganyika territory. Africa 4r273-286. Blurton Jones N (1986) Bushman birth spacing: A test for optimal interbirth intervals. Ethol. Sociobiol. 7t91-105. Blurton Jones N (1987) Bushman birth spacing: a direct test of some simple predictions. Ethol. Sociobiol. 8;183-203. Blurton Jones N, Hawkes K, OConnell J (1989) Studying the costs of children in two foraging societies: Implications for schedules of reproduction. In V Standen and R Foley (eds.!: Comparative Socioecology: The Behavioral Ecology of Humans and Other Animals. Spccia1 Publication no. 8 of the British Ecological Society. Oxford: Blackwell, pp. 367-390. Blurton Jones N, Hawkes K, Draper P (1990) Differences between Hadza and !Kung childrens work: original affluence or practical reason. Paper presented at 6th International Conference on hunting and gathering peoples. University of Alaska Fairbanks, May 1990. Blurton Jones N, Hawkes K, Draper P (in press) Differences between Hadza and !Kung children's work: Original affluence or practical reason? In T Burch (ed.): Issues in Hunter-gatherer Research. Blurton Jones N, Sibly R (1978)Testing adaptiveness of culturally determined behavior: Do Bushman women maximize their reproductive success by spacing births widely and foraging seldom? In V Reynolds and Blurton Jones N (eds.): Human Behavior and Adaptation. Symposia for the Study of Human Biology, Volume 18. London: Taylor and Francis, pp. 135-157. Bunn H, Bartram L, Kroll E (1988) Variability in bone assemblage formation from Hadza hunting, scaveng- 179 ing, and carcass processing. J . Anthropol. Arch. 7t412-457. Campbell KL, Wood JW (1989) Fertility in traditional societies. In P Diggory and S Teper (eds.): Natural Human Fertility: Social & Biological Mechanisms. Macmillan, London. Code A, Demeny P (1966) Regional Model Life Tables and Stable Populations. Princeton: Princeton University Press. Coe MJ, Cummings DH, and Phillipson J (1976) Biomass and production of large African herbivores in relation to rainfall and primary production. Oecologia 22t341-354. Cooper B 11949) The Kindiga. Tanzania Notes and Records 2 7s-15. Dixon W (ed.) (1981)BMDP Statistical Software. Berkeley: University of California Press. Dyson T (1977)The demography of the Hadza in historical perspective. Proceedings of a seminar on historical demography. Edinburgh: Centre for African Studies, University of Edinburgh, pp. 139-154. Ellison FT ' (1990) Human ovarian function and reproductive ecology: new hypotheses. Am. Anthropol. 92:933-952. Ferguson DE (1980). The political economy of health and medicine in colonial Tanganyika. In MHY Kaniki (ed.): Tanzania under Colonial Rule. London: Longman, pp. 307-343. Handwerker WP (1983) The first demographic transition: an analysis of subsistence choices and reproductive consequences. Am. Anthropol. 85t5-27. Harpending HC, Draper P (1990) Curr. Anthropol. 31:127-129. Harpending HC, Wansnider L (1982) Population structure of the Ghanzi and Ngamiland !Kung. In M Crawford and J Mielke (eds.): Current Developments in Anthropological Genetics. New York: Plenum, pp. 2950. Harris M (1988) Culture, People, and Nature: An Introduction to General Anthropology (5th edition). New York: Harper and Row. Haviland W (1989) Anthropology (5th edition). New York: Holt, Rinehart and Winston. Hawkes K, OConnell J , Blurton Jones N (1989) Hardworking Hadza grandmothers. In V Standen and R Foley (eds.): Comparative Socioecology: The Behavioral Ecology of Humans and Other Animals. Special Publication No. 8 of the British Ecological Society. Oxford: Blackwell, pp. 341-366. Headland T, Reid L (1989) Hunter-gatherers and their neighbors from prehistory to the present. Curr. Anthropol. 30:43-66. Hiernaux J, Hartono D (1980) Physical measurements of the adult Hadza. Ann. Hum. Biol. 7t339-346. Hill K, Hurtado M (1989) Hunter-gatherers of the New World. Am. Scientist 77:437443. Howell N (1976) The population of the Dobe area !Kung. In RB Lee and I DeVore (eds.): Kalahari HunterGatherers. Cambridge: Harvard University Press, pp. 137-151. Howell N (1979) Demography of the Dobe !Kung. New York Academic Press. 180 N.G. BLURTON JONES ET AL Kjekshus K (1977)Ecology Control and Economic Development in East African History: The Case of Tanganyika 1850-1950. London: Heineman. Kohl-Larsen L (1958)Wildbeuter in Ost-Afrika: die Tindiga, ein Jager- und Sammlervolk. Berlin: Dietrich Reimer. Konner MJ, and Worthman C (1980)Nursing frequency, gonadal function, and birth spacing among !Kung hunter-gatherers. Science 207,788-791. Lee PC, Majluf P, Gordon I J (1991) Growth, weaning and maternal investment from a comparative perspective. J . Zool. 225t99-114. Lee RB (1972)Population growth and the beginnings of sedentary life among the !Kung bushmen. In Spooner B (ed.i: Population Growth: Anthropological Implications. Cambridge, Mass: MIT Press, pp. 329-342. Lee RB (1979) The !Kung San: Men, Women, and Work in a Foraging Society. Cambridge: Cambridge University Press. Lee RB, DeVore I (1968) (eds.): Man the Hunter. Chicago: Aldine. Martin CR, Read D (19811 The relation of mean annual rainfall to population density for some African hunters and gatherers. Anthropol. UCLA 7,151-160. McDowell W 119811 A brief history of the Mangola Hadza. Unpublished manuscript prepared for the Rift Valley Project, Ministry of Information and Culture, Division of Research, Dar es Salaam, Tanzania. Mehlman M (1988) Later Quaternary archaeological sequences in northern Tanzania. Unpublished PhD dissertation, Department of Anthropology, University of Illinois, Champaign-Urbana. Ndagala D (1988) Free or doomed? Images of the Hadzabe hunters and gatherers of Tanzania. In T Ingold, D Riches, and J Woodburn (eds.): Hunters and Gatherers 1: History, Evolution and Social Change. Oxford: Berg Publishers Ltd., pp. 65-72. Obst E (1912) Von Mkalama ins Land der Watindiga. Mitteilungen der Geographiscen Gesellschaft in Hamburg, 26,2-27. OConnell J. Hawkes K, Blurton Jones N (l988ai Hadza scavenging: Implications for Plio-Pleistocene hominid subsistence. Curr. Anthropol. 29,356363. OConnell J, Hawkes K, Blurton Jones N (1988b)Hadza hunting, butchering, and bone transport and their archaeological implications. J . Anthropol. Res. 44,113162. OConnell J, Hawkes K, Blurton Jones N (1990)Reanalysis of large mammal body part transport among the Hadza. Journal of Archaeol. Sci. 17,301-316. Pennington R, Harpending HC (1988)Fitness and fertility among Kalahari !Kung. Am. J . Phy. Anthropol. 77:303-319. Phillips J , Worthman CM, Stallings JF, Blurton Jones N, Sellen D (1991) New field techniques for detection of female reproductive status. Presented at American Association of Physical Anthropologists Annual Meeting, Milwaukee, April 1991. Roth E (1985) A note on the demographic concomitants of sedentism. Am. Anthropol. 87:380-382. Sahlins M (1972) Stone Age Economics. Chicago, Aldine. Schrire C (1980)Aninquiry into the evolutionary status and apparent identity of San hunter gatherers. Hum. Ecol. 8t9-32. Schultz J (19711 Agrarlandschaftliche Veranderungen in Tanzania (MbuldHanang District.s). Hamburg: Weltform Verlag. Sembajwe I(1984)The status of knowledge on mortality in Tanzania. Utafiti 6:179-191. Service E (1962) Primitive Social Organization: An Evolutionary Perspective. New York: Random House. Sheriff AMH (1979) Tanzanian societies at the time of the partition. In MAY Kaniki (ed.): Tanzania under Colonial Rule. London: Longman, pp. 11-50. Siege1 S (1956) Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill. Smith CC, Fretwell SD (1974) The optimal balance between size and number of offspring. Am. Nat. 1O8:499-506. Smith EA, Winterhalder B (eds.): (in press) Ecology, Evolution, and Human Behavior. Chicago: Aldine de Gruyter. Smith L (1980) The distribution of Hadza hunter-gatherers in relation to their resources. Unpublished manuscript on file, Department of Anthropology, University of Utah, Salt Lake City. Solway J, Lee RB (1990) Foragers genuine or spurious? Situating the Kalahari San in history. Curr. Anthropol. 31t109-146. Steward J (1955)Theory of culture change: The methodology of multilinear evolution. Urbana: University of Illinois Press. Tanaka J (1976) Subsistence ecology of central Kalahari San. In Lee RB and DeVore I (ed.): Kalahari HunterGatherers. Cambridge: Harvard University Press, pp. 98-119. Trivers RL (1974) Parent-offspring conflict. American Zoologist 11,249-264. United Nations (1967) Methods of Estimating Basic Demographic Measures from Incomplete Data. Popul. Stud. 42. Vincent A (1985)Wild tubers as a harvestable resource in the East African savannas: Ecological and ethnographic studies. Unpublished PhD dissertation, Department of Anthropology, University of California, Berkeley. Wilmsen EN (1989) Land filled with Flies. A political economy of the Kalahari. Chicago: University of Chicago Press. Wilmsen EN, Denbow J R (1990) Paradigmatic History of San-speaking Peoples and Current attempts at Revision. Curr. Anthropol. 31,489-524. Winterhalder B, Smith EA (1981)Hunter-Gatherer Foraging Strategies: Ethnographic and Archaeological Analyses. Chicago: University of Chicago Press. Wood JM, Johnson PL, Campbell KL (1985) Demographic and endocrinological aspects of low natural fertility in highland New Guinea. J. Biosoc. Sci. 17:57-79. Wood JW, Wemstein M (1988) A Model of Age-specific Fecundability. Population Studies 42:85-113. Woodburn J (1964)The social organization of the Hadza DEMOGRAPHY OF THE HADZA of north Tanganyika. Unpublished PhD dissertation, Department of hinihropoiopy , Vriiversiiy of Cambridge. Woodburn J (1968a) An introduction to Hadza ecology. In R Lee and I DeVore (eds.): Man the Hunter. Chicago: Aldine, pp. 49-55. Woodburn J (1968b) Stability and flexibility in Hadza residential groupings. In R Lee and I DeVore (eds.): Man the Hunter. Chicago: Aldine, pp. 103-110. 181 Woodburn J (1988) African hunter-gatherer social orgailiaation: is it best understood as a pi.od.uct ofei~,capaulation? In T Ingold, D Riches, J Woodburn (eds.): Hunters and Gatherers 1:History, Evolution and social change. Oxford: Berg Publishers Ltd., pp. 31-64. Yellen J, Lee RB (1976) The Dobe-/Du/da environment: background to a hunting and gathering way of life. In R Lee and I DeVore (eds.): Kalahari Hunter-Gatherers. Cambridge: Harvard University Press, pp. 2 7 4 6 .