ottle do not apply to evolution in the "open", nature "at large" or to
"wild" types. To be consistent, this same objection should be extended to
the use of the spectroscope in the study of the evolution of the stars, to
the use of the test tube and the balance by the chemist, of the
galvanometer by the physicist. All these are unnatural instruments used to
torture Nature's secrets from her. I venture to think that the real
antithesis is not between unnatural and natural treatment of Nature, but
rather between controlled or verifiable data on the one hand, and
unrestrained generalization on the other.

If a systematist were asked whether these new races of Drosophila are
comparable to wild species, he would not hesitate for a moment. He would
call them all one species. If he were asked why, he would say, I think,
"These races differ only in one or two striking points, while in a hundred
other respects they are identical even to the minutest details." He would
add, that as large a group of wild species of flies would show on the whole
the reverse relations, _viz._, they would differ in nearly every detail and
be identical in only a few points. In all this I entirely agree with the
systematist, for I do not think such a group of types differing by one
character each, is comparable to most wild groups of species because the
difference between wild species is due to a large number of such single
differences. The characters that have been accumulated in wild species are
of significance in the maintenance of the species, or at least we are led
to infer that even though the visible character that we attend to may not
itself be important, one at least of the other effects of the factors that
represent these characters is significant. It is, of course, hardly to be
expected that _any_ random change in as complex a mechanism as an insect
would improve the mechanism, and as a matter of fact it is doubtful whether
any of the mutant types so far discovered are better adapted to those
conditions to which a fly of this structure and habits is already adjusted.
But this is beside the mark, for modern genetics shows very positively that
adaptive characters are inherited in exactly the same way as are those that
are not adaptive; and I have already pointed out that we cannot study a
single mutant factor without at the same time studying one of the factors
responsible for normal characters, for the two together constitute the
Mendelian pair.