tive and yet reckon descent (of the class) in the _male_ line.
If, as he conceives, conceptional totemism was transformed in the
central tribes into patrilineal totemism, I fail to see why the
phratries or classes should descend in the female line.

If in the third place, it was proposed to prevent children of sisters or
of brothers from intermarrying, it is completely mysterious why children
of brothers and sisters should not only not have been prevented in the
same way, but absolutely be regarded as the proper mates for each other.
Even if a single community reformed itself on these lines, it is hardly
conceivable that many should have done so, even if we suppose that the
advantages of prohibition were preached from tribe to tribe by
missionaries of the new order of things. _Ex hypothesi_, cousin marriage
was not regarded as harmful; and it is highly improbable that any people
in the lower stages of culture should have discovered that in-and-in
breeding is harmful, for the results, especially in a people which
contained no degenerates, would not appear at once, even if they
appeared at all.

On this point therefore the probabilities are wholly on the side of
development as against reformation.

An additional reason against the reformation theory is found in the fact
that phratries, on this theory, would never exceed two in number, but in
practice there are, as shown in Chapter II, wide variations.

FOOTNOTES:

[107] _Secret of the Totem_, pp. 31, 91 sq.

[108] Mr Lang's view is that the women from the first retained their
original group names wherever they went. _Letter of July 27th_, 1906.

[109] See pp. 31, 50.

[110] _Fortn. Rev._ LXXVIII, 459.

CHAPTER VII.

CLASS NAMES.

Classes later than Phratries. Anomalous Phratry Areas. Four-class
Systems. Borrowing of Names. Eight-class System. Resemblances and
Differences of Names. Place of Origin. Formative Elements of the
Names: Suffixes, Prefixes. Meanings of the Class Names.

The priority of phratries over classes is commonly admitted and it is
unnecessary to argue the question at length. The main grounds for the
assumption are: (1) that it is _a priori_ probable that the fourfold
division succeeded the twofold division, exactly as the eightfold
division has succeeded, and apparently is still gaining ground, at the
expense of the four-class system. (2) Over a considerable and compact
area phratries alone are found without a trace