ecial creation. But, upon the theory of natural
evolution, we can without difficulty understand why the earlier forms
should have been the simpler forms, and also why they should have been
the most generalized. For it is out of the older forms that the newer
must have grown; and, as they multiplied, they must have become more and
more differentiated.

Again, we have seen that there is no correlation between the importance
of any structure from a classificatory point of view, and the importance
of that structure to the organism which presents it. On the contrary,
it is a general rule that "the less any part of the organization is
concerned with special habits, the more important it becomes for
classification." Now, from the point of view of special creation it is
unintelligible why unity of ideal should be most manifested by least
important structures, whereas from the point of view of evolution it is
to be expected that these life-serving structures should have been most
liable to divergent modification in divergent lines of descent, or in
adaptation to different conditions of life, while the trivial or less
important characters should have been allowed to remain unmodified. Thus
we can now understand why all primitive classifications were wrong in
principle when they went upon the assumption that divine ideals were
best exhibited by resemblances between life-serving (and therefore
adaptive) structures, with the result that whales were classed with
fishes, birds with bats, and so on. Nevertheless, these primitive
naturalists were quite logical; for, from the premises furnished by the
theory of special creation, it is much more reasonable to expect that
unity of ideal should be shown in plainly adaptive characters than in
trivial and more or less hidden anatomical characters. Moreover, long
after biological science had ceased consciously to follow any
theological theory, the apparent axiom continued to be entertained, that
structures of most importance to organisms must also be structures of
most importance to systematists. And when at last, in the present
century, this was found not to be the case, no reason could be suggested
why it was not the case. But now we are able fully to explain this
apparent anomaly.

Once more, we have seen that aggregates of characters presenting
resemblances to one another have always been found to be of special
importance as guides to classification. This, of course, is what we
should ha