itative forms and colours in insects.
Drawn from nature (_R. Coll. Surg. Mus._).]

_Warning Colours._

In strange and glaring contrast to all these cases of protective
colouring, stand other cases of conspicuous colouring. Thus, for
example, although there are numberless species of caterpillars which
present in an astonishing degree the phenomena of protective colouring,
there are numberless other species which not only fail to present these
phenomena in any degree, but actually go to the opposite extreme of
presenting colours which appear to have been developed for the sake of
their conspicuousness. At all events, these caterpillars are usually the
most conspicuous objects in their surroundings, and therefore in the
early days of Darwinism they were regarded by Darwin himself as
presenting a formidable difficulty in the way of his theory. To Mr.
Wallace belongs the merit of having cleared up this difficulty in an
extraordinarily successful manner. He virtually reasoned thus. If the
_raison d'etre_ of protective colouring be that of concealing agreeably
flavoured caterpillars from the eye-sight of birds, may not the _raison
d'etre_ of conspicuous colouring be that of protecting disagreeably
flavoured caterpillars from any possibility of being mistaken by birds?
Should this be the case, of course the more conspicuous the colouring
the better would it be for the caterpillars presenting it. Now as soon
as this suggestion was acted upon experimentally, it was found to be
borne out by facts. Birds could not be induced to eat caterpillars of
the kinds in question; and there is now no longer any doubt that their
conspicuous colouring is correlated with their distastefulness to birds,
in the same way as the inconspicuous or imitative colouring of other
caterpillars is correlated with their tastefulness to birds. Here then
is yet another instance, added to those already given, of the
verification yielded to the theory of natural selection by its proved
competency as a guide to facts in nature; for assuredly this particular
class of facts would never have been suspected but for its suggestive
agency.

As in the case of protective imitation, so in this case of warning
conspicuousness, not only colour, but structure may be greatly modified
for the purpose of securing immunity from attack. Here, of course, the
object is to assume, as far as possible, a touch-me-not appearance; so
that, although destitute of any real means o