reatly preponderate in
each of the two exogamous intermarrying divisions of the tribe. This
must be because the present practice has not yet quite upset the
ancient usage, by which no totem ever occurred in both divisions. There
is even an Arunta myth asserting that this was so, but it is, of
course, of no historical value as evidence. Here it is proper to give
Mr. Frazer's contrary theory in his own words:--

'This [Arunta] mode of determining the totem has all the appearance of
extreme antiquity. For it ignores altogether the intercourse of the
sexes as the cause of offspring, and further, it ignores the tie of
blood on the maternal as well as the paternal side, substituting for it
a purely local bond, since the members of a totem stock are merely
those who gave the first sign of life in the womb at one or other of
certain definite spots. This form of totemism, which may be called
conceptional or local to distinguish it from hereditary totemism, may
with great probability be regarded as the most primitive known to exist
at the present day, since it seems to date from a time when blood
relationship was not yet recognised, and when even the idea of
paternity had not yet presented itself to the savage mind. Moreover, it
is hardly possible that this peculiar form of local totemism, with its
implied ignorance of such a thing as paternity at all, could be derived
from hereditary totemism, whereas it is easy to understand how
hereditary totemism, either in the paternal or in the maternal line,
could be derived from it. Indeed, among the Umbaia and Gnanji tribes we
can see at the present day how the change from local to hereditary
totemism has been effected. These tribes, like the Arunta and Kaitish,
believe that conception is caused by the entrance into a woman of a
spirit who has lived in its disembodied state, along with other spirits
of the same totem, at any one of a number of totem centres scattered
over the country; but, unlike the Arunta and Kaitish, they almost
always assign the father's totem to the child, even though the infant
may have given the first sign of life at a place haunted by spirits of
a different totem. For example, the wife of a snake man may first feel
her womb quickened at a tree haunted by spirits of goshawk people; yet
the child will not be a goshawk but a snake, like its father. The
theory by which the Umbaia and Gnanji reconcile these apparently
inconsistent beliefs is that a spirit of the husba