root (which, as we have seen, is probably welu, the terminations _-uku_, _-itchi_, and _-uru_ (=_-aree_) being formative suffixes), we have here too a single phratry name on the one side and three sister names on the other.  While it is clear that the names cannot be in any sense of the term recent, from the fact that linguistic differentiation had already gone some distance in what we may call, for want of a better term, groups speaking a stock language (in proof of which we have only to look at the formative suffixes), it seems equally clear that the present phratry names must be considerably later than the final settlement of the country.  At the same time it must not be forgotten that the existence of numerous small phratries, the number of which may yet be largely increased by more exact research, is prima facie a proof that the groups which adopted them had not reached the stage at which anything like that tribal (still less national) organisation was known, which is at the present day characteristic of the Arunta, and, perhaps, we may say, of all groups organised on a class system with class names known and used over an area far beyond that over which the (in a restricted sense) tribal language extends.

The recurrence of crow in the phratry name of the far west lends further support to the view that the phratry names were selected in some way, and were not due to some accident of savage wit.  The view has been taken that the phratry animals were originally totems, or animals that became totems at a later stage.  In view of the large number of totems found in many tribes, or even restricting their number to six or eight in each phratry, it is not difficult to estimate the probability that cockatoo and crow would recur in different areas, and that an opposition of characters should be found in other cases.  The hypothesis needs at any rate to be combined with a theory, firstly, of borrowing of phratry names, a process which must indeed have played a large part in the development of the present system, but which does not necessarily involve the supposition that the borrowed names replaced previously existing home-made names; and, secondly, of selection of such names as were not borrowed.

It has been mentioned that the principle of tribal property in land or, to be strictly accurate, in hunting grounds, is, at the present day, a fundamental one in native Australian jurisprudence.  But, as is shown by the map, in some cases the phratries are split into two or more segments[106], more or less remote from one another, geographically speaking.  Now this apparent segmentation must be due to migration; it can hardly arise from the chance adoption of identical names; for the groups in which the names occur are, though separated by a considerable distance, not so remote as, on the theory of chance selection, we should expect them to be, in other words the probability is in favour of the segmentation of an original group