Sometimes, but no less exceptionally, it is an ancestor or a group of ancestors which serves as totem directly. In this case, the clan takes its name, not from a thing or a species of real things, but from a purely mythical being. Spencer and Gillen had already mentioned two or three totems of this sort. Among the Warramunga and among the Tjingilli there are clans which bear the name of an ancestor named Thaballa who seems to be gaiety incarnate.[234] Another Warramunga clan bears the name of a huge fabulous serpent named Wollunqua, from which the clan considers itself descended.[235] We owe other similar facts to Strehlow.[236] In any case, it is easy enough to see what probably took place. Under the influence of diverse causes and by the very development of mythological thought, the collective and impersonal totem became effaced before certain mythical personages who advanced to the first rank and became totems themselves.
Howsoever interesting these different irregularities may be, they contain nothing which forces us to modify our definition of a totem. They are not, as has sometimes been believed,[237] different varieties of totems which are more or less irreducible into each other or into the normal totem, such as we have defined it. They are merely secondary and sometimes even aberrant forms of a single notion which is much more general, and there is every ground for believing it the more primitive.
The manner in which the name is acquired is more important for the organization and recruiting of the clan than for religion; it belongs to the sociology of the family rather than to religious sociology.[238] So we shall confine ourselves to indicating summarily the most essential principles which regulate the matter.
In the different tribes, three different systems are in use.
In a great number, or it might even be said, in the greater number of the societies, the child takes the totem of its mother, by right of birth: this is what happens among the Dieri and the Urabunna of the centre of Southern Australia; the Wotjobaluk and the Gournditch-Mara of Victoria; the Kamilaroi, the Wiradjuri, the Wonghibon and the Euahlayi of New South Wales; and the Wakelbura, the Pitta-Pitta and the Kurnandaburi of Queensland, to mention only the most important names. In this case, owing to a law of exogamy, the mother is necessarily of a different totem from her husband, and on the other hand, as she lives in his community, the members of a single totem are necessarily dispersed in different localities according to the chances of their marriages. As a result, the totemic group lacks a territorial base.
Elsewhere the totem is transmitted in the paternal line. In this case, if the child remains with his father, the local group is largely made up of people belonging to a single totem; only the married women there represent foreign totems. In other words, each locality has its particular totem. Up until recent times, this scheme of organization was found in Australia only among the tribes where totemism was in decadence, such as the Narrinyeri, where the totem has almost no religious character at all any more.[239] It was therefore possible to believe that there was a close connection between the totemic system and descent in the uterine line. But Spencer and Gillen have observed, in the northern part of central Australia, a whole group of tribes where the totemic religion is still practised but where the transmission of the totem is in the paternal line: these are the Warramunga, the Quanji, the Umbia, the Binbinga, the Mara and the Anula.[240]
Finally, a third combination is the one observed among the Arunta and Loritja. Here the totem of the child is not necessarily either that of the mother or that of the father; it is that of a mythical ancestor who came, by processes which the observers recount in different ways,[241] and mysteriously fecundated the mother at the moment of conception. A special process makes it possible to learn which ancestor it was and to which totemic group he belonged.[242] But since it was only chance which determined that this ancestor happened to be near the mother, rather than another, the totem of the child is thus found to depend finally upon fortuitous circumstances.[243]
Outside of and above the totems of clans there are totems of phratries which, though not differing from the former in nature, must none the less be distinguished from them.
A phratry is a group of clans which are united to each other by particular bonds of fraternity. Ordinarily the Australian tribe is divided into two phratries between which the different clans are distributed. Of course there are some tribes where this organization has disappeared, but everything leads us to believe that it was once general. In any case, there are no tribes in Australia where the number of phratries is greater than two.
Now in nearly all the cases where the phratries have a name whose meaning has been established, this name is that of an animal; it would therefore seem that it is a totem. This has been well demonstrated in a recent work by A. Lang.[244] Thus, among the Gournditch (Victoria), the phratries are called Krokitch and Kaputch; the former of the words designates the white cockatoo and the latter the black cockatoo.[245] The same expressions are found again among the Buandik and the Wotjobaluk.[246] Among the Wurunjerri, the names employed are Bunjil and Waang, which designate the eagle-hawk and the crow.[247] The words Mukwara and Kilpara are used for the same purpose in a large number of tribes of New South Wales;[248] they designate the same birds.[249] It is also the eagle-hawk and the crow which have given their names to the two phratries of the Ngarigo and the Wolgal.[250] Among the Kuinmurbura, it is the white cockatoo and the crow.[251] Many other examples might be cited. Thus we are led to regard the phratry as an ancient clan which has been dismembered; the actual clans are the product of this dismemberment, and the solidarity which unites them is a souvenir of their primitive unity.[252] It is true that in certain tribes, the phratries no longer have special names, as it seems; in others where these names exist, their meaning is no longer known, even to the members. But there is nothing surprising in this. The phratries are certainly a primitive institution, for they are everywhere in a state of regression; their descendants the clans have passed to the first rank. So it is but natural that the names which they bore should have been effaced from memory little by little, when they were no longer understood; for they must belong to a very archaic language no longer in use. This is proved by the fact that in many cases where we know the animal whose name the phratry bears, the word designating this animal in the current language is very different from the one employed here.[253]
Between the totem of the phratry and the totems of the clans there exists a sort of relation of subordination. In fact, in principle each clan belongs to one and only one phratry; it is very exceptional that it has representatives in the other phratry. This is not met with at all except among certain central tribes, notably the Arunta;[254] also even where, owing to disturbing influences, overlappings of this sort have taken place, the great part of the clan is included entirely within one or the other of the two groups of the tribe; only a small minority is to be found in the other one.[255] As a rule then, the two phratries do not overlap each other; consequently, the list of totems which an individual may have is predetermined by the phratry to which he belongs. In other words, the phratry is like a species of which the clans are varieties. We shall presently see that this comparison is not purely metaphorical.
In addition to the phratries and clans, another secondary group is frequently met with in Australian societies, which is not without a certain individuality: these are the matrimonial classes.
By this name they designate certain subdivisions of the phratry, whose number varies with the tribe: there are sometimes two and sometimes four per phratry.[256] Their recruiting and operation are regulated by the two following principles. In the first place, each generation in a phratry belongs to different clans from the immediately preceding one. Thus, when there are only two classes per phratry, they necessarily alternate with each other every generation. The children make up the class of which their parents