It is mathematically impossible that this exclusiveness should be the result of accident. On a first consideration, therefore, I took it to be the result of deliberate legislative design, at the moment when on my hypothesis two _local_ totem groups, containing members of several _totems of descent_, united in _connubium_. The totem names, I at first conceived, with reluctance, must have been consciously and deliberately meted out between the two local totem groups, now become phratries. This idea did not involve so stringent and useless a measure as that of segmenting the two phratries into minor totem groups: however the idea was still too much akin to that of Harpocration as regards the arbitrary drafting of the Attic population into γένη. But, on further reflection, I conceived that my first theory was superfluous. Given the existence of local groups, as such totemic, and of totem kins of descent within the original local totem groups, the actual results, I thought, arise automatically, as soon as two local totem groups agree to intermarry. Men and women must many out of their local totem group (now ‘phratry’) and must marry out of their totem of descent. Consequently, no one totem could possibly exist in both phratries. This I now, on third thoughts, ‘which are a wiser first,’ deem erroneous. The automatic arraying of one set of totems into one, or another set into the other, phratry, would not occur. The totems have been divided between the two phratries.[3] This condition of affairs is universal in Australia, except where, as among the Arunta and similar tribes, the same totem comes to exist in both phratries, so that men and women of the same totem, but of opposite phratries, may intermarry. That breach of old rule, we shall try to show, arises from the peculiar animistic philosophy of the Arunta, by virtue of which totems are no longer totems of descent, but are otherwise obtained. The Kamilaroi practice of interphratry marriage arises out of respect for totem and neglect of phratry law. My conjecture takes for granted, let me repeat, that, before the ‘bisection,’ or the amalgamation, which produced the two exogamous ‘classes,’ the totem kindreds were already exogamous. My reasons for this opinion have already been given, in the discussion of Mr. Crawley’s theory of the origin of exogamy (_supra_), to which the reader may refer. My suggestion makes the growth of exogamy non-moral, gradual, and almost unconscious, till it is clinched and stereotyped by the totem tabu.[4] The opposite theory--namely, the deliberate bisection into exogamous ‘classes,’ of totem groups, or of an ‘undivided commune’ not previously exogamous, appeals too much, I repeat, to conscious and--as far as we can see--motiveless legislation, at an early stage. The bisection must have had a purpose, and has no visible purpose except the establishment o f exogamy, and why did the ‘undivided commune’ establish that? THE IDEAS OF MR. FRAZER HIS EARLIER THEORY