bound up with the question concerning the temporal relation of the two important phenomena last mentioned. An unambiguous answer to the latter question, however, may be gathered precisely by a study of Australian conditions, at least so far as the development in these regions is concerned. If we recall our previous schema (p. 141), representing the tribal organization of the Kamilaroi, and here, as there, designate the totemic groups (emu, kangaroo, opossum, etc.) comprised within the clan by _m n o p...,_ it is apparent that the totems must be at least as old as the division into the two tribal halves. Unless this were the case, we could not explain the fact that, with very minor exceptions, precisely the same totems exist in the two tribal divisions. The condition might be represented thus:-- I II _mnopq opmsn_ It is also evident, however, that the totems could not have influenced this first division, otherwise their members would not have separated and passed over into the two tribal divisions, as they did in almost every case. Remembering that the totemic groups are also cult associations, we might express the matter thus: At the time of the first tribal division, the cult groups were not yet strong enough to offer resistance to the separation of the tribal divisions, or to determine the mode of division; therefore, members of totem _m_, for example, went here or there according as other external conditions determined. Conditions were quite different at the time of the second division, when the tribal half I separated into clans A and B, and II into C and D, according to the schema:-- I II / \ / \ / \ / \ A B C D mnop qrst mpqs nort