objects of art, can be reckoned primitive. Yet, where such stones do
not exist, the usage of acquiring totems by locality does not exist;
even where the belief in reincarnation and in local centres haunted by
totemic spirits is found in North Australia.[For an hypothesis of the
origin of the CHURINGA NANJA belief, see my SECRET OF THE TOTEM,
chapter iv.]

On these grounds it appears that the hereditary totem is the earlier,
and that the Arunta usage is the result of the special and inseparable
superstition about the sacred stones. It may be a relatively recent
complication of and addition to the theory of reincarnation. Meanwhile,
the belief and usage produce an unique effect. The Arunta and Kaitish,
we saw, are so advanced socially that they possess not two, or four,
but eight matrimonial classes. The tribe is divided into two sets of
four classes each, and no person in A division (nameless) of four
classes may marry another person of any one of these four, but must
marry a person of a given class among the four in B division
(nameless). The succession to the class is hereditary in the mate line.
But any person among the Arunta, contrary to universal custom
elsewhere, may marry another person of his or her own totem, if that
person be in the right class of the opposite division. Nowhere else can
a person of division A and totem Grub find a Grub to marry in the
opposite division B. But this is possible among the Arunta and Kaitish,
because their totems are acquired by pure accident, are not hereditary,
and all totems exist, or may exist, in division A and also in division B.

Mr. Frazer argues that the Arunta is the earlier state of affairs. He
supposes that men acquired their totems, at first, by local accident,
before they had laid any restrictions on marriage. Later, they divided
their tribe, first into two, then into four, then into eight classes;