middle feathers, which
are themselves quite unmarked and protectively tinted like the rest of
the upper surface of the body. The figures of the expanded tails of two
species of East Asiatic snipe, whose geographical ranges overlap each
other, will serve to illustrate this difference; which is frequently
much greater and modified in an endless variety of ways (Fig. 22).

Numbers of species of pigeons, hawks, finches, warblers, ducks, and
innumerable other birds possess this class of markings; and they
correspond so exactly in general character with those of the mammalia,
already described, that we cannot doubt they serve a similar
purpose.[86]

[Illustration: FIG. 21.--Secondary quills.]

[Illustration: FIG. 22.--Scolopax megala (upper). S. stenura (lower).]

Those birds which are inhabitants of tropical forests, and which need
recognition marks that shall be at all times visible among the dense
foliage, and not solely or chiefly during flight, have usually small but
brilliant patches of colour on the head or neck, often not interfering
with the generally protective character of their plumage. Such are the
bright patches of blue, red, or yellow, by which the usually green
Eastern barbets are distinguished; and similar bright patches of colour
characterise the separate species of small green fruit-doves. To this
necessity for specialisation in colour, by which each bird may easily
recognise its kind, is probably due that marvellous variety in the
peculiar beauties of some groups of birds. The Duke of Argyll, speaking
of the humming birds, made the objection that "A crest of topaz is no
better in the struggle for existence than a crest of sapphire. A frill
ending in spangles of the emerald is no better in the battle of life
than a frill ending in spangles of the ruby. A tail is not affected for
the purposes of flight, whether its marginal or its central feathers are
decorated with white;" and he goes on to urge that mere beauty and
variety for their own sake are the only causes of these differences.
But, on the principles here suggested, the divergence itself is useful,
and must have been produced _pari passu_ with the structural differences
on which the differentiation of species depends; and thus we have
explained the curious fact that prominent differences of colour often
distinguish species otherwise very closely allied to each other.

Among insects, the principle of distinctive coloration for recognition
has prob