Reviews. 473
of the Tjingilli, and consequently differs from that of Messrs. Spencer and Gillen. The probability is (as we shall see directly) that the tribes in question are in a state of transition, and that Mathews and Howitt's statement may be (or may have been at the time) accurate as regards one portion of a tribe which is spread over a considerable extent of country, while Spencer and Gillen's statements may be equally accurate for another portion of the tribe, or (seeing that their report is the most recent) may be now accurate for the entire tribe.
A Warramunga tradition, reported by the latter, concerning the change of phratry of certain mythical beings lends countenance to this suggestion. Such a change of phratry seems actually to have happened in the Arunta tribe, where the Panunga class has been transferred from phratry B to phratry A^ and conversely the Purula class from phratry A to phratry B. Nothing else will account for the Urabunna rearrangement of the classes to suit their system of female descent, when intermarriage takes place between the Urabunna and Arunta. According to Mathews' account a similar change has been effected by the Tjingilli. In this tribe the totems descend on the male side, while the phratries descend on the female side. The result is that while in tracing descent on the male side the totem is retained through- out the generations, the phratry is changed at every step in the genealogy ; and conversely, on the female side the phratry is retained, but the totem changes. A woman as the first step of phratry A (call her A\) gives birth to children who are A2. The women of A2 give birth to A^, those of A3 to ^4, those of Aa, X.O A\. The women of Ax marry the men of B\ ; but the women of A2 cannot marry the men of B2, because, seeing that descent by totems is in the male line, these men would (or might) belong to their totem, in fact would (or might) be by tribal reckoning their brothers. They therefore marry the men of B4. But this they could not do either, if B4 were the descendants of Bi, because they would belong to their fathers' totem. The people of B4 must therefore belong either to a gens of B, severed by what I have called vertical cleavage from B, or to a gens severed by the same process from A. Let us suppose the former. If we now compare the arrangement of
