B. Tylor, "On a Method of Investigating the Development of Institutions, applied to Laws of Marriage and Descent," Jour. Anthroj. Inst. XVIII., 1889, 245-272. Rivers, Kinship and Social Organization.
Lowie, "Exogamy and the Classificatory System of Relationships," Am. Anth., 1915, we are not usually in a position to determine a simple cause-effect relationship. There are often alternative explanations : the mother's brother may be identified with the father's sister's hus band because of cross-cousin marriage or because they both belong to the same exogamous moiety or because two households arrange to marry off their respective daughters to the boys in the other household (Rivers). The safe rule is to reject any social de terminant of a nomenclatorial trait unless it is reported as extant in the tribe or at least in the general area in question.
Other phases of the problem must be considered. An institu tion may be overshadowed by coexisting institutions or it may not yet have had time to assert itself terminologically. The Miwok, e.g., practise cross-cousin marriage but their terminology shows a far deeper impress of rival forms of marriage. It seems a fair conclusion that the latter are of greater antiquity.' Diffusion presents a further complication. While harmony be tween the nomenclature and the social structure is frequent, there are discrepancies that can be most readily interpreted by bor rowing. To exemplify, there are tribes with a bifurcate merging terminology that lack clans, but their location suggests that they have borrowed a neighbor's nomenclature. It would be dangerous
to infer that they had at one time been organized into clans, unless extraneous data so suggest.
Finally, there are the linguistic factors stressed by Kroeber. When nearly all Siouan languages have a separate term for pa ternal aunt, the lack of such a word (in address only) among the Crow must be interpreted as a late development ; and the use of the "mother" word appears as a characteristic sample of linguistic extension,—the same phenomenon as when English uses "wife" as a special term of affinity while the cognate German Weib applies to woman generically. The reality of such phenomena militates against any attempt to reduce the whole of kinship terminology to social causes. There will always be residual phenomena resisting interpretation on any but linguistic lines. This means that they are in a sense unique facts that can be understood after they have been observed but that could not be deduced from general These cautions limit but do not destroy the sociological signifi cance of relationship terminologies. When all allowances are made for disturbing factors, a host of correlations remain between kin ship features and sociological factors. When more intensive studies of large linguistic families shall be available, it will be pos sible to balance with greater nicety the relative importance of sociological and other factors as determinants. (R. H. Lo.)