This pioneer attempt requires both revision and amplification. The last mentioned feature, e.g., is of purely local significance in western North America and of limited application where found.
The differentiation between elder and younger sibling (brother or sister) means psychologically simply the intercalation of two additional generations. On the other hand, the reciprocity prin ciple, linking by a common term distinct generations, merits inde pendent status. Cognizance must also be taken of the frequent duplication of terms in address and mere reference, the mode of classification itself sometimes differing. Extra-American termi nologies also suggest that the categories laid down by Kroeber might be materially increased. Even as it stands, however, his sketch furnishes a valuable instrument for the precise definition and comparison of distinct systems.' In the present state of our knowledge it is impracticable to do more than concentrate upon an unequivocally significant feature as a basis for classification in a world-wide survey, and to refine areally by supplementary characteristics. The trait that ob trudes itself on our notice is the treatment of collateral as com pared with lineal kin in the first ascending and descending gen erations. It is implicitly the phenomenon that impressed Mor gan; it has a bearing on Tylor's and Rivers' theories (see below) ; and constitutes one of Kroeber's categories. Practically, the data usually suffice to characterize systems from this point of view.
The logical possibilities are as follows : (a) Collateral lines are wholly merged in the lineal within a particular generation (generation system) ; (b) each generation is bisected so that only half the collateral kin are merged in the lineal (bifurcate merging system) ; (c) the immediate collateral lines are dis tinguished from the lineal and from each other (bifurcate col lateral system) ; (d) collateral lines are distinguished from the lineal, but not from each other (lineal).
In standard samples of these four systems logical coherence prevails: if the father's brother is called father, he addresses his brother's son as son ; if there is a separate word for mother's brother, there is likely to be one for nephew. However, discrep ancies occur, and a system must sometimes be classed by the pre ponderance of its affiliations. Taking the classification of uncles for illustration, the scheme may be illustrated by the table on page 85.
The second system, which Morgan and Rivers genetically con nect with the first (though in a reverse sense), is not one whit farther from the third; and the bifurcate collateral system sets off the immediate family as sharply from the rest of the universe as Organization, 55 sq.
W. Gifford, California Kinship Terminologies (Univ. Cal. Pub., vol. 18, 1922).
our English lineal system. All four systems actually occur in prim itive communities, notwithstanding the widespread notion that they have only the first two, corresponding to Morgan's classi ficatory types.
With this provisional scheme it is possible to characterize the main areas of the globe. Since regional summaries are available for Australia and North America, it will be best to begin with these continents.
Australian kinship terminologies follow the bifurcate merging plan.
Certain other traits are characteristic. Among these is the use of separate terms for maternal and paternal grandparents. Thus, the Kakadu in Northern Australia call the father's father kaga, but the mother's father peipi. With many tribes this trait is coupled with the reciprocity principle : in Arunta, arunga desig nates simultaneously the father's father and the son's child. Again, the world-wide distinction between the speaker's elder and younger siblings often displays refinement, so that in Arunta the father's elder brother's sons are "elder brothers," while the father's younger brother's sons are "younger brothers." More rarely, as among the Mungarai, the uncles and aunts themselves are distinguished with reference to their age as compared with the speaker's parents.
While these features have parallels in remote areas, others seem distinctive. According to Radcliffe-Brown, all nomenclatures of the region conform to two types, correlated with distinct forms of marriage law. In type I., as found among the West Australian Kariera and the Urabunna of the Lake Eyre region, the father's father is classed with the mother's mother's brother; by the cor related matrimonial rule a man is prohibited from marrying any woman unless she stands to him in the relation of a mother's brother's daughter. In type II., the mother's mother's brother is distinguished from the father's father, being often classed with the mother's mother; here the prescribed marriage is between the children of two female cross-cousins, a man marrying his moth er's mother's brother's daughter's daughter. This second type is of wider distribution than the first, having been discovered among the Dieri, the Central tribes generally, from the Arunta north ward to the Anula and Mara, and among such West Australians as the Mardudhunera.' Probably most of the tribes substitute relationship terms for personal names in ordinary intercourse, but the central Australians only tabu sacred appellations, while secular names are used in address, interchangeably with the kinship terms.' A feature worth noting is the absence of terms of affinity. This is evidently to be correlated with the fact that in Australia mar riage is prescribed with definite blood relatives. Accordingly there is no necessity for coining new words for connections by marriage. This same trait may be expected to occur in greater or lesser de gree in other parts of the world where corresponding forms of marriage with blood relatives are either obligatory or at least preferred.