Each of these ‘phratries’ has five totems, not found in the other class, and how this occurred, if not by actual deliberate arrangement, I do not know. One thing is clear: totem and phratry are prior to ‘class’ divisions. They occur where ‘class’ divisions do not. But my theory does not involve the deliberate introduction of exogamy, by an exogamous bisection of groups not hitherto exogamous, or by making two pre-existing totem groups exogamous. I take the groups to have been exogamous already, before the blending in _connubium_ of two local totem groups (now ‘phratries’), each including numbers of already exogamous totem kindreds. They were exogamous before the ‘phratries’ existed, and after their falling into the two phratries, exogamous they remained. DISTRIBUTION OF TOTEMS IN THE ‘PHRATRIES’ Mr. McLennan, ere he had the information now before us, wrote, in 1865, ‘Most probably contiguous groups would be composed of exactly the same stocks’ (we can now, for ‘stocks,’ read ‘totem kins’)--‘would contain gentes of precisely the same names.’[2] This is obvious, for Emu, Kangaroo, Wild Duck, Opossum, Snake, and Lizard, living in the same region, would raid each other (by the hypothesis) for wives, and each foreign wife would bring her own totem name into each group. Yet we find that the two ‘primary classes’ (phratries) of the Urabunna (which, on my theory, represent two primitive totem local groups, say Emu and Kangaroo, each with its representatives of all other totem groups within raiding distance) _never_ contain the same totems.