‘There is only one method in social anthropology, the comparative method – and that is impossible.’ So, it is often reported, said E.E. Evans-Pritchard – though the aphorism appears nowhere in his published writings.
In fact, the comparative method is far from impossible, although in social and cultural anthropology it is constrained by severe limitations. The first to point this out was Sir Francis Galton in a discussion of a paper delivered by E. B. Tylor at the Anthropological Institute in 1888. Tylor argued from a sample of 350 societies that the evolution of cultural complexity leads from matrilineal to patrilineal institutions. Galton disagreed, noting that correlations may result either from evolution or from common origin:
It was extremely desirable for the sake of those who may wish to study the evidence for Dr. Tylor’s conclusions, that full information should be given as to the degree in which the customs of the tribes and races which are compared together are independent. It might be, that some of the tribes had derived them from a common source, so that they were duplicate copies of the same original.
‘Galton’s Problem’, as it became known, has plagued statistical studies not only in anthropology but in other social sciences too, ever since.
In his book The Methodology of Anthropological Comparison, Gopala Sarana (1975) distinguishes three kinds of comparative method: global-sample comparison, controlled comparison, and illustrative comparison. Global-sample comparison, or global comparison, was the kind of comparison to which Galton took exception. From the 1940s to the 1970s, it was the mainstay of the school of George Peter Murdock and his followers. It is also reminiscent of the earlier method of Sir James Frazer and, arguably, of Claude Levi-Strauss. Levi-Strauss’s position, though, is ambiguous, since the degree to which his choice of examples is regional and the degree to which it is intended to represent universal principles is not always clear. In the works of the Murdockians (e.g. Murdock 1949), a sample of the world’s societies is chosen. Then the sample is analysed with respect of the distribution of selected cultural features. Conclusions are drawn on cause and effect, and thus the sample is believed to yield explanation of relations between cultural features broadly applicable worldwide. For example, if agriculture is in the hands of women, this might yield a tendency for uxorilocal residence. That, in turn, might lead to the recognition of matrilineal descent groups and ultimately to ‘Crow’ kinship terminologies (in which an entire descent group would be classified as the ‘fathers’ or ‘fathers’ sisters’ of one’s own).
At the other extreme, there is illustrative comparison. This is used especially, and legitimately, for pedagogical purposes and more broadly for highlighting social or cultural phenomena that may be different in diverse contexts. For instance, a first-year anthropology course may include an in-depth study of the Nuer, as an example of a society based on patrilineal descent, and an in-depth study of the Trobrianders, as an example of a society based on matrilineal descent. Students might be encouraged to compare the Nuer and the Tro-brianders with respect to the role of the father or the mother’s brother in relation to bringing up children or in relation to his place in the social structure. Many important points could be raised. Among Trobrianders, the mother’s brother, rather than the father, is the ‘father figure’ for children and is treated in a more formal manner. Among Nuer, this is not true.
Trobrianders have localized matrilineal groups through a norm of avunculocal residence. Men grow up in ‘foreign’ villages, namely those of their respective fathers, then move at puberty to the villages of their designated matrilineal groups. Women move to their husbands’ villages upon marriage. This rule keeps the men of the groups together, but not the women through whom they are related. They never live in their own villages but grow up in their fathers’ and then move to their husbands’. Nuer descent groups, in contrast, are in theory localized around patrilineal descent groups, although at least at the time of key ethnographic studies the rules of residence were not strictly adhered to. The Trobriand Islands represent a chiefly society, with chiefships inherited matrilineally; Nuer society is not hierarchically organized, with social control essentially in the hands of acephalous lineages and ‘leopard-skin chiefs’ who are in fact adjudicators in disputes between lineages.
The problem with illustrative comparison, though, comes when a student, or an anthropologist, makes unjustified generalizations on the assumption that the Nuer or the Trobianders are necessarily typical of patrilineal or matrili-neal societies. One cannot say on the basis of just one example of each that, say, matrilineal societies are chiefly and patrilineal ones are not. This is avoided when the contrasts are clearly pedagogical, as in some studies within the ‘Culture and Personality’ school (e.g. Benedict 1934), where examples represented either extreme types or simply societies with which the comparative ethnographers had some familiarity.
In between global-sample comparison and illustrative comparison lies controlled comparison, of which the most informative type is usually regional comparison. This is the approach advocated by Fred Eggan (1954) and which characterized his comparative work in Native North America. In a more structuralist form it was found too in the Dutch studies of the East Indies as a ‘field of ethnological study’ (Josselin de Jong 1977). It is also found in later work, for example, on Southern Bantu social and symbolic structures (Kuper 1982) and on Khoisan Southern Africa.The idea is that by narrowing the range of variables through working on similar societies, especially but not necessarily ones within an ethnographic region or culture area, more meaningful comparisons can be made. For example, taking three cases where close-kin marriage is common, Kuper showed that the Southern Bantu system in each case enables powerful individuals to perpetuate their power through bridewealth transactions. Among the Tswana, men tend to marry women of lower social status than themselves, and bride-wealth is relatively low. Among the Southern Sotho, men tend to marry women of higher status, and bridewealth is relatively high. Among the Swazi, men may marry either way. Those who marry ‘down’ (in the Tswana way) tend to pay less bridewealth, and those who marry ‘up’ (in the Southern Sotho way) tend to pay more bridewealth.
The distinctions between the different types of comparison is not strictly definable, and some comparative studies are open to interpretation as to exactly what the theory of comparison is. One excellent and classic study shows this well: S.F. Nadel’s (1952) essay on witchcraft in four African societies. Two of his examples are Nigerian and two are Sudanese. Each regional pair has both similarities and differences, but comparisons across the two regions also reveal features of witchcraft that seem to be associated with variations in gender relations, age structures, kinship, and so on. Nadel sacrifices an element of control in order to illustrate structural constraints which might otherwise not be obvious. His study is structural but only partly regional, and it is illustrative as well as, in part, controlled.
Comparison is perhaps less fashionable as a ‘method’ than it was in the past, but it is always with us as part of social anthropology’s essence. Yet it remains an elusive ideal, in part, for the very reasons Galton gave for his rejection of Tylor’s attempt. It is often difficult to articulate what comparison is for, as it is troublesome to identify exactly what is being compared.
