It remains however to be established that this segregation of totems is actually found in the tribes in question. For the Warramunga Spencer and Gillen distinctly state[136] that the arrangement is dichotomous, in which case the alleged result would not be brought about. The Anula and Mara are exceptional tribes with direct male descent; it is hardly likely that the eight-class system spread from them. The Mayoo have not yet been reported on by an expert. Finally some of the tribes have not even the dichotomous arrangement of totems but distribute them in both phratries. The basis of the hypothesis, therefore, is hardly established.
Singularly enough, Dr Durkheim[137] expresses his adherence to a previous theory of his own as to the method of effecting the change from female to male descent in four-class tribes. This he supposes to have been done by transferring one of the two classes from each phratry to the opposite one; and in the former discussion (Année Soc. V, 82 sq.) he showed that this procedure would result in scattering the totems through both phratries, as we find them to be in the case of the Arunta. It is therefore singular to find that he adheres to this theory when his new hypothesis demands that the totems, so far from being more widely distributed, should be actually confined to the members of one couple. Beyond the Urabunna custom in intertribal marriages, however, which is hardly decisive evidence, there does not appear to be any proof that the transference from one phratry to the other ever took place.
The further support claimed by Dr Durkheim for his hypothesis from the alleged male descent of the totem in tribes where female descent of the class names prevails, rests on too uncertain a basis to make it necessary to deal with it at length; some criticism of the evidence will be found elsewhere.
We have seen above that the Dieri rule is precisely parallel to that of the eight-class tribes in practice; it is however expressed, not by a class system, but by enacting that people standing in a certain degree of kinship or consanguinity shall marry. If Dr Durkheim's theory of the origin of the eight-class system is correct, it should also apply to the Dieri. Now the rule that a man must marry his maternal great-uncle's daughter clearly prevents intermarriage with one of the mother's totem; but this cannot be the object of the rule, for it is prevented already by the phratry system. Dr Durkheim's theory therefore finds no support in the Dieri rule.
On the other hand, unless the totems have been scattered through the phratries since the southern Arunta divided their classes, Dr Durkheim will have difficulty in explaining why a tribe where the totem does not concern marriage at all has found it necessary to split the classes; and that though the child does not take its totem from mother or father.
Herr Cunow has advanced the view that the classes correspond to distinctions of age; but he took as his basis, not the differentia of elder and younger, but the distinction made by the initiation customs, which divide the community, in his view, into three strata—young, adult and old. Into the difficulties created by this theory we need not here enter. Suffice it to say that the theory depends on the supposition that an age-grade had to marry within itself. Now the age-grade is not a fixed body, but is continually changing its personnel; not only so, but it is difficult to see how marriage could take place, given the initiation ceremonies, in any other way; unions of "old men" with adult women apart, which are not, in fact, prohibited, so far as is known, the only marriages possible are those within the adult grade. Although father and son can rarely belong to the adult grade simultaneously, mother and daughter can readily do so. If not, these grades are clearly generation classes, and what Herr Cunow really takes as the basis of his theory is the generation in each family. This can readily be shown by a consideration of the kinship terms.
[130] Roth, Eth. Stud. p. 182; Spencer and Gillen, Nor. Tr. p. 616; Howitt, p. 262; J. R. S. N. S. W. XXXI, 166.
[131] J. A. I. VII, 249, cf. J. R. S. N. S. W. XXXI, 172.
[132] Howitt, p. 110.
[133] Nor. Tr. p. 167; Proc. R. G. S. Qu. XVI, 70; J. R. S. N. S. W. XXX, 111, 112.