tories to
which they go? When the 400 existing species of humming-birds have all
been created on the same continent for some reason supposed to be
unknown, why should this reason give way before the accident of any
means of migration being furnished to humming-birds, so that they should
be able to visit, say the continents of Africa and Asia, there gain a
footing beside the sun-birds, and henceforth determine a new centre for
the separate creation of additional species of humming-birds peculiar to
the Old World--as has happened in the case of the majority of species
which, unlike the humming-birds, have been at any time free to migrate
from their original homes?

Lastly, my third consideration is, that the supposed policy in question
does not extend to affinities which are wider than those between species
and genera--more rarely to families, scarcely ever to orders, and never
to classes. In other words, nature shows a double correlation in her
geographical distribution of organic types:--first, that which we have
already considered between geographical restriction and natural
affinity among inhabitants of the same areas; second, another of a more
detailed character between _degrees_ of geographical restriction and
_degrees_ of natural affinity. The more distant the affinity, the more
general is the extension. This, of course, is what we should expect on
the theory of descent with modification, because the more distant the
affinity, and therefore, _ex hypothesi_, the larger and the older the
original group of organisms, the greater must be the chance of
dispersal. The 400 species of humming-birds may well be unable to
migrate from their native continent; but it would indeed have been an
unaccountable fact if no other species of all the class of birds had
ever been able to have crossed the Atlantic Ocean. Thus, on the theory
of evolution, we can well understand the second correlation now before
us--namely, between remoteness of affinity and generality of
dispersal,--so that there is no considerable portion of the habitable
globe without representatives of all the classes of animals, few
portions without representatives of all the orders, but many portions
without many of the families, innumerable portions without innumerable
genera, and, of course, all portions without the great majority of
species. Now, while this general correlation thus obviously supports the
theory of natural descent with progressive modification, it