26 THE EVOLUTION OF NATURAL SPECIES. 
are usually variable, that is, polymorphic, and that these same genera 
are the ones in which the production of new species is most rapidly 
progressing; but he has nowhere drawn the conclusion that freedom 
from rigid natural selection can in any way favor the production of 
new varieties and species. On the contrary, he teaches that it is only 
through the agency of diversity in natural selection that individual 
variations can be accumulated in diverging lines that become more ° 
and more distinct. (See Origin of Species, Chapter IV.) 
3. Facts in the Distribution of Hawaiian Snatls. 
It was through the study of island fauna that I was first led to 
doubt the correctness of Darwin’s theory at this point. The terres- 
trial mollusks of the West Indies present important differences as we 
pass from island to island, but it was in the Hawaiian Islands that I 
found the greatest difference in the species inhabiting the forests in 
different parts of the same island. The remarkable features in the 
distribution and affinities of these forms will be most easily presented 
by giving a brief statement of some of the facts relating to those 
found on the island of Oahu. ‘This island, about 45 miles in length 
and 20 miles in width, is inhabited by over 200 species of land snails, 
represented by 800 or 1,000 varieties; and these are nearly all confined 
to the forests covering two ranges of mountains, the one 15 or 20 miles 
and the other 35 milesin length. But the most remarkable fact is not 
the great number of species and varieties inhabiting this small area, 
nor yet that all of them (with, perhaps, one or two exceptions) are 
peculiar to this island; but that each of these forms is confined to only 
a small section of this small area. Not only are the species on each 
of the ranges of mountains different, but those found on one range 
and inhabiting one continuous region of forest are not distributed at 
random over that region. Onthe contrary, each valley not more than 
half a mile in width and perhaps two miles in length has its own 
peculiar varieties, and in some cases its own species, which are found 
nowhere else. 
As the explorer passes from valley to valley he will find six or seven 
quite distinct groups of forms, each group being a genus or a sub- 
genus of the one family of Achatinellide. Taking any one genus and 
tracing its distribution from valley to valley, he will find that, as far 
as it extends, it is represented in each valley by one or more closely 
allied species, each species being represented by several varieties. 
One species may be confined to a single valley, or it may extend over 
several, being represented in each by varieties peculiar to that valley. 
The more widely divergent forms of one group or genus will be 
