What is the effect of dividing a phratry into two classes? Firstly and most obviously, to reduce by one half the number of women from whom a man may take his spouse. Secondarily, to put in the forbidden class both his mother's generation and his daughters' generation. It must however not be overlooked that it is the whole class of individuals that are thus put beyond his reach and not those only who stand to him in the relation of daughters in the European sense. Now it is certain that the savage of the present day distinguishes blood relationship from tribal relationship; of this there are plenty of examples in Australia itself[130]. In fact the hypothesis that the introduction of class regulations was due to a desire to prevent the intermarriage of parents and children, more especially of fathers and daughters, the mothers being of course of the same phratries as their sons in the normal tribe, depends for its existence on the assumption that consanguinity was recognised. But it is clearly a clumsy expedient to limit a man's right of choice to the extent we have indicated solely in order to prevent him from marrying his daughter, when the simple prohibition to marry her would, so far as we can see, have been equally effective.

Dr Durkheim has suggested that phratries and classes originated together.

If we start with two exogamous local groups in which the determinant spouse removes, the result is two groups in which both phratries are found, as is evident from the following graphic representation. The two sides represent the local grouping, the letters A and B the phratry names, and m or f male or female; the = denotes marriage, the vertical lines show the children, the brackets show that the person whose symbol is bracketed removes, and the italics that the symbol in question is that of a spouse introduced from without.

mA=fB mB=fA
[fB]mB=fA fB=mA[mB]
[fA]mA=fB fA=mB[fB]
[fB]mB=fA fB=mA[fA]
etc. etc.

We see from this that the alternate generations are in each group A and B, whose spouses are in the same alternation B and A, the male remaining in the group, the female removing in each case, if we assume that the matrilineal kinship is the rule. The permanent members of each group therefore, and in like manner the imported members, are by alternate generations A and B, though of course there is no difference of age actually corresponding to the difference of generation.

By the simple phratry law that A can only marry B, and may marry any B, local group mates are marriageable. The law however which forbids the marriage of phratry mates is on Mr Lang's original theory founded on the prohibition to marry group mates. If we suppose that the primal law or the memory of it continued to work, we have at once a sufficient explanation of the origin of the four-class system. The tribes or nations in which the instinct against intra-group marriage was strong enough to persist as an active principle after the law against intra-phratry marriage had become recognised, may have proceeded to create four classes at a very early stage, while those in whom the feeling for the primal law was less strong adhered to the simple phratry system.

But it is an insuperable objection to this theory that it makes the four-class system originate simultaneously with, or at any rate shortly after, the rise of the phratries. For we cannot suppose that the feeling for the primal law remained dormant for long ages and then suddenly revived. On the other hand we have seen that if the difference in the distribution of the phratry and class names is any guide, a considerable interval must have separated the rise of the one from the rise of the other. Unless therefore it can be shown that some other explanation accounts for the non-coincidence of phratry and class areas, we can hardly accept any explanation of the origin of classes which makes them originate at a period not far removed from the introduction of the phratries.

The fact that a certain number of class names are in character totemic, that is, bear animal names, suggests that the class system may be a development of the totem kins, which in certain cases are grouped within the phratries or otherwise subject to special regulations. In the Urabunna the choice of a man of one totem is said to be limited to women of the right status in a single totem of the opposite phratry. Among the similarly organised Yandairunga the limitation is to certain totems, and Dr Howitt gives other examples of the same order. In the Kongulu tribe these totemic classes seem to have been known by special names. In the Wotjoballuk tribe there are sub-totems, grouped with certain totems, which again seem to be collected into aggregates intermediate between the phratry and the simple totem kin. But it is difficult to see why, if the classes have arisen out of such organisations, there should be found over the great part of Australia four, and only four, classes from which the eight have obviously developed. In any case we have no parallel in these modifications to the alternate generations of the class system.

These find an analogue, according to an old report, not subsequently confirmed, in the Wailwun tribe, where, however, it is supplementary to the classes. We are told that there are four totems in this tribe, though this does not agree with other reports, and that they are found in both phratries indiscriminately. A woman's children do not take her totem, nor, apparently, the totem of her brother, who belongs to a different kin, but are of the remaining two totems according to their sex[131]. From this it follows that the totems alternate, precisely as do the classes; the difference in the arrangement consists in the distinction of totem falling to males and females, which has no analogue in the class system. But such arrangements, even if we may take them as established facts, are clearly of secondary origin, and can hardly give a clue to the origin of the classes.

There is an important difference between the four-class and eight-class organisations in respect of the totem kins. In the former systems the kins are almost invariably divided between the phratries; but within them they do not belong to either of the classes, though certain classes claim them[132]; but on the contrary, of necessity are divided between them. In the eight-class tribes this seems to be the case in some tribes also; in others, like the Arunta, abnormalities of development cause the totems to fall in both phratries. But in the Mara, the Mayoo, and the Warramunga[133] they fall, or are stated to fall, in the first case into groups according to the four classes, in the other cases according to the "couples," i.e. the two classes which stand in the relation of parent and child (the son of Panunga is Appungerta, his son is again Panunga, and so for the other pairs). This suggests that totemism has something to do with the division of the four classes into eight, as was pointed out by Dr Durkheim in 1905[134]. His argument is that as long as descent was in the female line, the rule was that a man could not marry a woman of his mother's totem. When the change to male descent took place, the mother's totem, as we see by actual examples[135], did not lose the respect which it formerly enjoyed; there is in more than one tribe a tabu of the mother's as well as of the father's totem. That being so, it is natural to suppose that the new marriage organisation according to male descent might be modified to take account of this fact. By dividing the classes and arranging that one member of a couple should be debarred not only from intermarrying with the class of his mother, for which the four-class system also provides, but also from intermarrying with the second member of the same couple too, this result was attained, in the view of Dr Durkheim.