This paper was originally presented at the annual
meeting of the American Anthropological Association. November 23, 1996,
San Francisco.
Comments welcome. Do not cite without permission.
Copyright Raymond Hames, 1996.
Introduction
Simple foraging and horticultural societies, commonly known as bands and tribes (Service, 1962), are commonly characterized as being economically egalitarian. How economic egalitarianism is defined varies. In relation to food sharing, the focus of this chapter, economic egalitarianism may mean "all individuals of the same age and sex cohort have equal access to life sustaining resources" (Speth, 1990: 3). This formulation suggests that food consumption within a group will be equalized according to age and sex status and differential productive capacities and needs as it varies between households or individuals will be compensated through the exchange system. Related to egalitarian exchange is Sahlins’ well-known concept of generalized reciprocity which "refers to transactions that are putatively altruistic, transactions on the line of assistance give, and, if possible, and necessary, assistance returned" (Sahlins, 1990:193-194). These transactions tend to occur among close kin or among those who live in close residential proximity. The flow of food in such cases is "in favor of the have not" (Sahlins 1990:194). Again, we find relative need and productive ability determine who give and gets. Woodburn in his widely cited distinction between immediate and delayed return egalitarian societies notes that sharing, especially of large game, serves as a leveling mechanism to enforce an equality of consumption. In fact, he views egalitarian sharing as a kind of graduated tax system where the most successful are required to chronically give more than those who are less successful (Woodburn, 1982:441-442). Aspelin (1979; and see below), in an admirably detailed analysis of food exchange among the Mamaindê, shows that the amount of food given to a household is directly proportional to household need as indexed by household size. Egalitarian exchange therefore means that food is distributed among households in direct proportion to a household’s need and a household’s ability to give (Dowling, 1968:503) which may be simply indexed by the number of producers, consumers and, especially the consumer to producer ratio of the household.
A theoretical alternative to egalitarian exchange is the theory of reciprocal altruism (Trivers, 1971) which suggests that food exchanges are strategic in nature and are designed to enhance the fitness of the individual engaged in exchange. In contrast to egalitarian models, resource needs or requirements of potential recipients of food resources are not relevant to exchange considerations. Instead recipients are chosen based on their likelihood and ability to reciprocate in the future. Two predictions are derivable from reciprocal altruism theory that may be tested with Yanomamö data. The first is that the scope of exchange should be inversely correlated with village size. As village size increases there should be an increasing number of families who do not share with one another. Theoretical justification for this prediction stems from the work of Boyd and Richerson (1988). Secondly, there should be a positive correlation between giving and receiving. By this I mean that a family that gives to another family should, in return, receive food from that family. In effect, exchange should be balanced through some degree of reciprocation.
Studies of food exchanges inspired by evolutionary theory show that food exchanges are not balanced. In hierarchical societies (Betzig, 1988; Bird and Bird, 1997) clearly show that high status households receive disproportionately more and give disproportionately less in food exchanges. In an egalitarian context, Kaplan and Hill’s (1984) research on the foraging Aché show that while inequalities in giving and receiving in persist at the economic level, they are compensated or balanced at the level of sexual access and reproduction.
With few exceptions (e.g. Aspelin, 1979; Betzig, 1988; Bird and Bird, 1997; Hames, 1990, 1996) most of the detailed empirical or quantitative accounts of food sharing focus on egalitarian foragers (Cashdan, 1985; Henry, 1951; Kaplan and Hill, 1985; Marshall 1961). These studies have shown that food sharing within small bands is intense, widespread, and frequently appears not to immediately enhance the nutritional status of the person who acquired the resource or his or her immediate household. Other studies, (Burch, 1992; Damas, 1975; Endicott 1992; Gould, 1982), indicate that egalitarian sharing may be highly restricted. For example, Burch notes "This raises the possibility that many, if not most, accounts of generalized sharing among hunter-gatherers have been based on studies of the internal dynamics of single local-household villages. To the extent that this is so, the accounts are not wrong, they simply tell only part of the story." (1992: 109). This observation suggests that our characterizations of sharing are critically dependent on the size of the group or whether one is documenting intra-familial sharing within large extended household units.
The goal of this chapter is to contrast predictions made by what I will call the generalized sharing model of food exchange compared to a model based on reciprocal altruism. In doing so I will examine the roles village size, exchange balance, and household demographic attributes as determinants of exchange patterns among the Yanomamö, an egalitarian foraging horticultural people of the Venezuelan Amazon. I will endeavor to demonstrate that reciprocal altruism provides a perspective superior to that of generalized sharing to account for food exchange patterns among the Yanomamö.
Methods
The method used to gather data on Yanomamö food exchanges basically follow a modified time allocation technique first employed by Kaplan and Hill (1984) in their studies of Aché exchange. In the course of a time allocation study, any time someone was observed eating I noted what was being eaten and I asked who produced the food and who gave the food. Later variables were added indicating household membership of food consumer and food giver. These observations led to the creation of a series of matrices which quantifies the flow of resources between households in a village. A facsimile of one of these matrices is presented in Figure 1. It should be read in the following way: household B gave food six times to household A; household B received food four times from household A, and household B gave food 44 times to itself. This technique differs from measurements of food exchange where weighed inflows and outflows of foods are sampled (see for example Hames, 1990). Four matrices of three villages (one village was studied in two different years) are the basis for most of the analyses that follow. Total food exchange observations for each village are as follows: Bisaasi-teri (1986) 217; Bisaasi-teri (1987) 741; Rakoiwä 254; and Krihisiwä 232. General ethnographic characteristics of these villages and more detailed information on methods have been described elsewhere (Hames, 1996).
Entities in Exchange
General evolutionary models of reciprocal exchanges focus on transactions between individuals. Nevertheless, attempts by anthropologists to evaluate these models in the context of food exchange among humans have tended to look at exchanges between individuals who acquire resources and the families to which they gave food (Kaplan and Hill, 1985; Hames, 1990) or exchanges between families (Hames, 1996). This perspective is taken because humans live in households which, in part, are designed to pool resources acquired by members. As such, food flows not so much between individuals as between households. In a behavioral sense it is difficult for one individual to give another something for his or her sole consumption. For example, if an individual gives a raceme of plantains (approximately 20-30 kilograms of food) to another, the recipient carries it back to his or her place of residence, where all household members have access to the resource. Unlike many animals, the receiver does not consume it on the spot or run to a hiding place to consume it out of sight of family members. Characteristics of this sort of exchange are easily visible when one observers the distribution of medium to large game animals in a Yanomamö village. The distributor calls out to individual households to fetch butchered and/or cooked meat, whereupon a boy or girl runs across the plaza, takes the meat from the distributor, and returns home with it. As I have noted elsewhere in the context of garden labor exchange (Hames, 1987) this is a targeting problem: givers, in most cases, cannot easily give to individuals.
It should also be understood that the giver of a resource cannot easily or unilaterally decide the household or households to which he or she is going to give. This is because other household members have the ability to influence which households receive food from their household. These constraints on individual choice in giving derive from a number of sources. Garden food, for example, is the result of joint labor by husband and wife. Therefore, they both have the ability to influence how garden products are distributed. In many other cases it is the consequence of competing interests in ensuring that family needs are met or that kin, affines, or social allies receive due consideration in distributions of resources collected by family members. As a consequence exchange in this paper is analyzed in terms of food flows between families.
Households in this analysis are defined
as collections of individuals who share a garden or set of gardens, a common
hearth or set of hearths, collaborate extensively in food production, and
are residentially distinct from other households. Households typically
correspond to a variety of family forms such as extended (stem form) and
nuclear polygynous and monogamous families. In some cases households are
in transition which typically involves the dissolution of an extended family
as for example as a son-in-law is terminating bride service.
The Function of Exchange
Before one can make predictions about patterns in food exchange (i.e., who gets what and how much) one needs to understand the function of exchange or what adaptive problem or problems it is designed to overcome. For the Yanomamö, what exchange is designed to accomplish is not completely clear. Based on previous research (Hames, 1990), much exchange appears to be a strategy for risk reduction to insure low variance in food intake. On a day to day basis, the amount of food a household produces is not a simple function of the amount or effort or time its members put into work. Unpredictable game encounters or localized thunderstorms that blow-down heavily laden plantains can leave a household destitute for varying lengths of time. Sharing prevents such unhealthy periods of low food availability by directing flows of food from households who have an temporary excess of food to households who have a temporary deficit. Food storage, an alternative adaptation to the problem of risk, is extremely costly in the tropics. So too is a focus on less risky resources (Hames, 1990) which would lower food getting efficiency. Risk pervades foraging and gardening pursuits in different ways.
In terms of risk, hunting is the most variable of all subsistence pursuits. Compared to other food getting activities, the possibility of complete failure is high and success can range from a one-kilogram monkey to a 150-kilogram tapir. Widespread distributions of game appear to be means to reduce variance in consumption of high quality protein that is poorly available in garden crops which compose more than 70% of the Yanomamö diet. There is a tendency to assume that gardening is a reliable food getting activity. Indeed, to insure a continuos and reliable supply of plantains the Yanomamö stagger plant. However, plantains are sensitive to a variety of environmental perturbations that make the production of a steady supply difficult to achieve. Since plantains cannot be stored for long periods of time, households are confronted with chronic problems of surplus and scarcity.
Intensity, Scope, and Balance of Exchange
The vocabulary we use to describe patterns of food exchange in anthropology is poorly developed largely because quantitative accounts of the actual flow of food resources in a community are still in their infancy. In this section I will introduce several concepts to describe food exchange: intensity, scope, and balance. Intensity is a measure of what fraction of a household’s food comes from other households. It directly measures the degree to which a household’s consumption is a consequence of food resources from other households. In a sense, it is a measure of subsidy. The scope of exchange is a measure of the number of households from which a household receives food resources in a village. This concept is illustrated in Figure 2 which shows linkages (arrows) in exchange between households (solid circles) in a village. Under a narrow scope of exchange there are three groups of households who intensively exchange food but only among themselves. In contrast, a wide scope of exchange shown in the right-hand side of the figure indicates that each household has exchange relationships with nearly every other household in the village. In this example, there are no restricted exchange circles and sharing is nearly coterminous with the entire village.
There is no obvious relationship between the intensity and scope of exchange among the Yanomamö. Although exchange is intense among the Yanomamö, it tends to be limited to a small fraction of households in the village. Figure 3 shows the intensity of exchange in four Yanomamö villages for the four basic food resources. The Y axis shows, for example, that nearly 39% of all the garden food consumed by a household was produced by other households in the village. Overall 40% of all food resources a household consumes was given to them by other households. Despite this relatively high degree of dependency a household has on other households for its food supply, the scope of exchanged, measured by the number of households that actually engage in exchange, is variable. This pattern is revealed in Figure 4 which shows the number of household exchange partners in the four villages which range from two to six households. For example, on average, households in the village of Krihisiwa received or gave food to about four other households. Of course, whether this is a narrow or broad scope depends on the number of households in a village. Accordingly, they range from 13% of available households to 80% (Table 1).
Table 1. Percent dyads of exchanging
households.
|
|
|
| Bisaasi 1986 |
13%
|
| Bisaasi 1987 |
35%
|
| Krihisiwa a |
34%
|
| Rakoiwa |
80%
|
A final way in which to view exchange is through the concept of balance. There are two ways in which it can be calculated: general and specific. General balance is the amount received in relation to the amount given to and from a household in relation to all other households. It allows one to determine which households are net givers or receivers of food resources such that net receivers receive more than they give and net givers give more than they receive. Specific balance refers to exchanges between individual households and it allows one to determine whether there is a relationship between how much a household gave and received from specific households in the village. Calculation of balance is done through a correlational analysis employing a matrix identical to the one portrayed in Figure 1. General is calculated by summing the row values (less giving to one’ self ) which represent giving and correlating that with the sum of the column values (less receiving from one’s self) which represent receiving. Specific balance is calculated by comparing how much A gave to B and B gave to A, A gave to C and C gave to A, etc. for all household dyads in the village.
Exchange and Household Demography
As described above, there is a generalized belief that sharing in tribal societies is egalitarian. Although what amounts to an egalitarian pattern of sharing is not sharply defined, the general consensus is that all households tend to receive shares of food when distributions are made and the amount received is proportional to the number of consumers in the household. For example, Aspelin’s quantitative account (1979) provides clear evidence that among the Brazilian Mamaindê the amount of the food a household received during distributions was proportional to its size (as measured by the number of normalized consumers). Conversely, large households should give more food than smaller households given their greater productive capacity. Reciprocal altruism makes no prediction regarding the relationship between household size and amount of food given or received.
In a recent publication (Hames, 1996) I provided quantitative evidence that household type was a determinant of an exchange pattern: polygynous households received proportionally more food from other village households than monogamous households. By this I mean that of the total food consumed by a polygynous household a greater proportion came from sources outside the polygynous household. Now I would like to consider the role of household demographic factors to determine if a generalized egalitarian pattern holds for the Yanomamo. If an egalitarian principle were at work, one would predict that household size and certain attributes of household size such as number of producers, number of consumers, and consumer to producer ratios should correlate with the amount of food given and received. Specifically, the number of consumers in a household should positively correlate with demand or the amount received, the number of producers should determine the amount given (ability to give), and consumer to producer ratio should positively correlate with the amount of food received and negatively with the amount given. The reasoning is as follows. Households with a larger number of producers should be able to produce more and therefore have more to give to others; and households with a large number of consumers should have greater needs and therefore receive more than households with few consumers. Finally, the ratio of consumers to producers provides the best measure of relative ability to give and need to receive in a single measure. As the number of consumers relative to producers rises, individuals must work relatively harder to meet household demand. If there were an egalitarian principle designed to equalize work or insure adequate household consumption then there should be a positive correlation between the amount of food received and consumer to producer ratio. At the same time, consumer to producer ratios should negatively correlate with the amount of food given since households with low consumer to producer ratios and therefore can more easily give. These predictions are a reflection of the old Marxian adage: from each according to his ability to work and to each according to his need.
Table 2. Receiving and giving as a function of household size
in four Yanomamo
villages.
| Village | Receiving Pearson’s r | Significance 1-tailed | Giving Pearson’s r | Significance 1-tailed |
| Rakoiwa | 0.73 | 0.05 | 0.35 | 0.25 |
| Krihisiwa | 0.57 | 0.03 | 0.17 | 0.30 |
| Bisaasi 1987 | 0.58 | 0.02 | 0.28 | 0.16 |
| Bisaasi 1986 | 0.44 | 0.06 | 0.30 | 0.12 |
Table 2 can be interpreted to indicate that household size is correlated with the absolute amount of food received by a household. However, it does not mean that large households receive relatively more food resources than small households. To gain a relative measure of the relationship between household size and food received, one can divide the amount of food received by the number of times a household was observed eating. This normalized measure of consumption of food received from other households represents the percent of time individuals in a household were observed to eat food that was not produced by a household member. This measure is identical to household dependency or subsidy I made in a previous study linking food exchange and marriage form (Hames, 1997). In none of the four villages is there a correlation between household size and percent of total food intake supplied by other households. These two results suggest that large households receive absolutely more food than small households but the amount they receive is not relatively greater.
Calculations of consumers and producers in a household is common in economic anthropology as a consequence of long-term interest in Chayanovian analyses of households (e.g., Durrenberger, 1984). Normally, estimates of full to fractional producers are not based on time allocation studies of adults or non-adults. Here the calculation of producers was based on Yanomamö labor time allocation statistics and is published elsewhere (Hames, 1994). For Yanomamö men and women, labor time peaks between thirty and forty years of age and declines slightly thereafter. These individuals were counted as 1 producer while all others were calculated less than one depending on how closely they came to the maximum adult. To calculate consumers in a household, I used a similar procedure but relied on Kaplan’s data on food consumption data for all age and sex groups for the Machiguenga (Kaplan, in press), a people living in the Peruvian Amazon who possess an economic system very similar to the Yanomamö.
The results of these analyses presented in Table 3 show no consistent relationship between amount of food received or given and consumer to producer ration. In fact, one of the two statistically significant correlations (Bisaasi 1986 "Gave") is in the opposite direction predicted. If some sort of egalitarian principle were at work we would expect to see a positive correlation between consumer to producer ratios and amount of food received and a negative correlation between consumer to producer ratio and amount of food given.
Table 3. Correlation (Pearson) between
consumer to producer ratios and amount of
food received and given in four Yanomamö
villages.
| Village | C/P
Received |
C/P
Gave |
| Bisaasi 1986 | 0.107 | 0.637* |
| Bisaasi 1987 | 0.587* | 0.065 |
| Krihisiwa | 0.236 | -0.498 |
| Rakoiwa | -0.563 | 0.204 |
| **Correlation is
significant at the 0.01 level (1-tailed)
*Correlation is significant at the 0.05 level (1-tailed) |
||
Village Size and Exchange Scope
If the function of exchange were to reduce variance then one would not need to have more exchange partners than necessary to reduce variance in consumption to an acceptable level. In this regard, Winterhalder (1986; see also Winterhalder et al. this volume for a more detailed analysis) has shown the exchange networks for hunters require no more than five to six households (assuming one hunter per household) to reduce hunting variance (the most risky of subsistence activities) to an acceptable level. This means that while additions to the sharing pool would further reduce variance in consumption the impact is minimal beyond five or six hunters. Although Winterhalder’s model was designed to predict the size of groups whose rationale for grouping is to reduce variation in consumption, it is reasonable to hypothesize that the scope of exchange is partially a negative function of village size.
Reciprocal altruism theory predicts that exchange should chronically occur between individuals who have reason to believe that exchanges will occur indefinitely into the future and who have a means to insure that partners will not defect. Boyd and Richerson have shown that as sharing cliques or coalitions become larger it becomes more difficult to deal with defectors. As a consequence, sharing among the Yanomamö should be restricted among a small circle of reliable reciprocators. Models of egalitarian exchange do not predict that exchangers should habitually limit their giving to a restricted circle of individuals.
If widespread sharing is more likely in small villages then we may be able to test this proposition for the Yanomamö by correlating the number of households in a village and the percent of household dyads who did not exchange food during the sampling period. For the four Yanomamö villages in this study, the number of households in a village ranged from 6 to 16 (or 24 to 102 people) and they were present approximately 100% of the time.
To test this proposition a matrix of village households as givers and receivers of resources to and from other households for each of the four villages was produced (shown in Figure 1 above). Row households were receivers of resources from column households, column households were givers of resources to row households, and upper left to lower right diagonal cells represented what each household gave and received to and from itself. The number of empty cells was counted for receivers and givers as a percentage of one half of the matrix less the diagonal cells. These two measures as they relate to number of households in a village are represented in Figure 5. The percent no exchange measures on the Y-axis is frequency at which household dyads were either observed not to give or receive food resources from other. For example, in Krihi approximately 60% of all possible household dyads neither gave or received food resources to or from each other. The results of number of households regressed on percent of dyads who neither gave or received to one another show that the relationship is strongly positive in both cases (about r2=0.71) in the predicted direction but it falls short of statistical significance (about p=0.08, one-tailed, see figure for exact significance levels).
There is additional Yanomamö evidence that scope of sharing may be a function of village size. In an earlier publication John Saffirio and I (1983) described variation in sharing in five different Yanomamö villages in Brazil. My reanalysis of that data (presented in Figure 6) indicates that the scope of sharing, as measured by percent of village households who received gathered, hunted, or fished resources each time they were brought in to the village declines with village size. As Table 4 shows, the correlations coefficients for gathered foods and fish are highly significant but not for game. Finally, Good (1987: 136-137, Tables 18 and 19) shows after a Yanomamö village fissioned, reducing its size from 114 to 41 residents, the probability of a household not receiving a portion of a major animal kill went from 5% for tapir, 29% for white-lipped peccary, and 25% for collared peccary to 0% for all three species in the smaller village. Good uses this data to suggest that village fissioning is a consequence of members being unhappy with the distribution of game animals and fissioning leads to the formation of villages in which game is more widely shared. The analysis presented here suggests an alternative interpretation: food exchangers carefully limit their giving to a small set of partners whom they deem reliable and the number they select is no more than necessary to reduce variance in food intake.
| Table 4. Sharing scope and village size in 5 Brazilian Yanomamö villages | ||
| Resource | Pearson’s r | Significance, 1-tailed |
| Fish | -.873 | 0.027 |
| Game | -.808 | 0.08 |
| Gathered | -.90 | 0.019 |
In addition, there is some comparative evidence that exchange scope is related to village size. Among the Mamaindê (or Nambiquara) of Brazil, 80% of the time five or fewer families were present in the village when food was distributed (Aspelin, 1979:318). Of the 116 village-wide distributions recorded by Aspelin, only in five cases were present families excluded from the distribution. Significantly, perhaps, in four of these five cases of exclusion there were more than five families in the village.
Although my data fail to achieve statistical significance at the 0.05 level, I am reluctant to accept the null hypothesis of no relationship between village size and scope of sharing for a number of reasons. The clear trend in the three independent research projects (Saffirio, Aspelin, and Good) is that village size and widespread sharing are negatively related and my results are consistent with this pattern.
It should be kept in mind that the three studies described above measure exchange differently. My behavioral scan method measures the frequency of eating any food that was produced by one’s household or that of another; Good measured the distribution of large game kills (those who did and did not receive portions of a kill); Saffirio measured the distribution of all wild resources, and Aspelin measures the percent of the amount received of the total distributed in relation to number of households present and household size.
Balance in Exchange
As mentioned, a bivariate correlation of the amount given and received between individual households and each household compared all households in the village will allow an assessment of balance in exchange. In specific balance reciprocal altruism would predict a positive correlation between giving and receiving between specific households. It is not critical that the amounts given and received balance precisely over a period as short as sampling period of these studies (about two months in each village). Egalitarian models of exchange hold divergent views on balance. In one case, Sahlins (1965) predicts that balance in exchange is strongly influenced by closeness of kinship and residential proximity such that the closer the degree of kinship and residential proximity between families the more likely that exchanges will be unbalanced in a positive way such that those who are able to give, give to those who are in need. As residential and kinship distances increase, exchanges become more balanced and as they further increase, exchanges become negatively balanced such that households attempt to take advantage of one another (Sahlins, 1965). It is clear that this model in some respects closely parallels simple kin selection models (Hames, 1989) which predict that close genetic kin are more likely to be tolerant of imbalances in exchange than distant kin. However, Sahlins specifically disavows such an interpretation (Sahlins, 1976) when he claims that closeness of kinship is defined culturally and not genetically. A different view held by Aspelin (1979) who reasons that balance is likely to be an artifact of individual family ability to produce and need to consume. To some extent this proposition has already been evaluated in analyzing the relationship between amount received and given in relation to family demographic attributes.
In all four villages the correlation between giving and receiving between household dyads was positive and significant but the correlations were weak to moderate (Table 5). The straightforward interpretation of this result is that households tend to limit their exchange relations to particular households. The strength of these associations is powerfully influenced by the fact that a significant number of households in all the villages were observed to not exchange at all. For example, in Rakoiwa where we have the strongest correlation (r=0.50), 12 of the 36 pairs did not exchange at all. If these individuals are deleted from the sample, then the correlation diminishes to r=0.25 (p=0.035, 1-tailed).
Table 5. Specific balance between household
dyads in four Yanomamö villages.
| Village | Pearson’s r | Significance (1-tailed) |
| Rakoiwa | 0.50 | 0.001 |
| Krihisiwa | 0.29 | 0.019 |
| Bisaasi 1987 | 0.21 | 0.026 |
| Bisaasi 1986 | 0.16 | 0.026 |
Discussion
The data and analysis on Yanomamö food exchanges presented here suggests that reciprocal altruism may be an important factor in Yanomamö food exchanges between families. Yanomamö tend to restrict their food exchange to a limited number of families as seen in the analysis of village size and scope of exchange and exchanges tend to be balanced. Although some support is seen for the notion of egalitarian exchange by the fact that large families received more food than small families, large families do not give more food compared to small families and they do not receive proportionately more food than small families.
Hawkes (1993) has suggested that lack of balance in giving and receiving is evidence against reciprocal altruism. Yanomamö data reveal a positive correlation between giving and receiving between families which leads me to conclude that exchange among the Yanomamö is balanced. This positive relationship is strongly influenced by the fact that many families were not observed to engage in exchanges. Although I have not done a detailed analysis of the balances, it is clear some families are in near perfect balance while other show moderate disparity. I would argue that there are at least three different reasons why lack of perfect balance may occur. First, the short time period of the data collection may reveal imbalance that would be more closely balanced over a long period of time. Second, as noted by Winterhalder (1996) and Gurven (MS) marginal valuations as determined by resource type (meat versus crops) or costs of giving or benefits of receiving may differ significantly from exchange to exchange owing to a variety of circumstances. For example, the benefit of receiving or cost of giving increases with the length of time one has gone without for both the giver and the receiver. Finally, it has long been noted by ethnographers studying people as diverse as foragers (Mauss, 1967) and Irish small holders (Arensberg, 1959) that attempts to balance exchanges are tantamount to ending exchange relationships. For these reasons I would conclude that a positive correlation between giving and receiving provide reasonable evidence that exchange is based on some form of reciprocal altruism. I believe the critical point demonstrated here is that exchange among the Yanomamö is strongly contingent such that families bias their giving to those who give to them. However, it is clear that a closer analysis of the nature of balances is required especially in terms of the kinship and other social bonds that characterize families in exchange relationships.
Patterns of exchange found among the Yanomamö differ significantly from those found among foragers. Sharing among the Aché has a much wider scope such that nearly all families receive portions of food brought into the camp, a pattern similar to what is found among the foraging-horticultural Mamaindê. To some extent these contrasts may be a simple function of group size since Yanomamö villages in this study ranged from about 50 to 100 residents compared to 30 to 40 found among the Mamaindê and Aché. As noted by Burch (1992) patterns of widespread sharing we have come to expect among foragers may be an consequence of observing sharing in small foraging groups. As groups become larger, the utility of widespread sharing may diminish which may lead to sharing focused on subsets of families within a settlement.
These findings also suggest an alternative interpretation of Good’s explanation of Yanomamö village fissioning. As mentioned, he notes that as a village becomes larger game animals are not as widely shared as they are in a small village. My data and that of Saffirio shows this effect for all food resources. Good argues that in large villages, rancorous accusations of stinginess frequently accompany less than complete game distributions and this ultimately leads to a village fissioning into two smaller villages where game is shared more widely. He argues that not being included in food distributions is an insult to one’s status and such insults precipitate a village fission leading to the creation of a smaller village where such insults have a lower probability of occurring. His model does not specify the function of game sharing other than to preemptively reduce accusations of stinginess over a critically scarce resource. This perspective seems to invoke motivations inherent in models of tolerated theft Blurton Jones (1984) or what Peterson refers to as "demand sharing" (1993). Yanomamö can be extremely aggressive in their demands for all resources, as Chagnon has documented on numerous occasions (Chagnon, 1974, 1992). Although my data also suggest that sharing may be more widespread (i.e. has greater scope) in small compared to large villages, I am unable to demonstrate a significant correlation between village size or number of families in a village and the intensity of sharing any (game, fish, garden, or gathered foods) food resources. What this means is that the amount of food a family is likely to receive from all other families is independent of village size and therefore fissioning to form smaller villages will not increase one’s absolute intake of resources from other families. This is because sharing in large or smaller villages seems to be restricted to sets of families who form reciprocal sharing partnerships.
The fact that there appears to be a relationship
between the scope of exchange and village size and that Yanomamö households
appear to restrict their giving and receiving to specific households has
interesting implications for village structure. This pattern if further
implicated by patterns of specific balance between households. These trends
suggest that medium and large villages may be economically subdivided into
groups of households based on habitual sharing relationships. It would
be interesting to determine whether these patterns parallel or replicate
predictable lines of village fissioning which seem to be based on patterns
of marital alliance and kinship, documented by Chagnon (Chagnon and Bugos
1979; Chagnon,1981). It suggests that villages, in the absence of intense
warfare, should be no larger than the necessary number of households needed
to reduce subsistence risk. These small village probably operate at an
economic advantage because there is less economic competition for local
resources. As warfare intensifies villages must become larger for defensive
and offensive purposes (Chagnon, 1974) and villages begin to internally
subdivide into groups of households who maintain intense exchange relationships.
Summary
The Yanomamö are an egalitarian people, yet village-wide patterns of food sharing are not designed to preferentially channel food to households with difficulties in meeting consumer demand by taking it from households who meet consumer demand more easily. In a number of analyses I demonstrated that there was no relationship between consumer, producers, or consumer to producer ratios and quantities of food flows between households. The patterns of sharing that did emerge appear to be based on a strategy of reciprocal altruism. The following point to this strategy:
