e which feed on or
among grasses.

[Illustration: FIG. 2.--THE CATERPILLAR OF THE EYED HAWK-MOTH
(_Smerinthus ocellatus_).]

Now, similar lines occur on a great number of caterpillars belonging
to most different groups of butterflies and moths, as you may see by
turning over the illustrations of any monograph of the group. They
exist among the Hawk-moths--as, for instance, in the Humming-bird
Hawk-moth; they occur in many butterflies, especially in those which
feed on grass; and in many moths. But you will find that the smallest
caterpillars rarely possess these white streaks. As regards the second
point, also, the streaks are generally wanting in caterpillars which
feed on large-leaved plants. The _Satyridae_, on the contrary, all
possess them, and all live on grass. In fact we may say, as a general
rule, that these longitudinal streaks only occur on caterpillars which
live on or among narrow-leaved plants. As the insect grows, these
lines often disappear on certain segments, and are replaced by
diagonal lines. These diagonal lines (Fig. 2) occur in a great many
caterpillars, belonging to the most distinct families of butterflies
and moths. They come off just at the same angle as the ribs of leaves,
and resemble them very much in general effect. They occur also
especially on species which feed on large-leaved plants; and I believe
I may say that though a great many species of caterpillars present
these lines, they rarely, if ever, occur in species which live on
grass; while, on the contrary, they are very frequent in those
species which live on large-leaved plants.

It might at first be objected to this view that there are many cases,
as in the Elephant-Hawk-moth, in which caterpillars have both. A
little consideration, however, will explain this. In small
caterpillars these oblique lines would be useless, because they must
have some relation, not only in color, but in their distance apart, to
the ribs of the leaves. Hence, while there are a great many species
which have, longitudinal lines when young, and diagonal ones when they
are older and larger, there is not, I believe, a single one which
begins with diagonal lines, and then replaces them with longitudinal
ones. The disappearance of the longitudinal lines on those segments
which have diagonal ones, is striking, where the lines are marked. It
is an advantage, because white lines crossing one another at such an
angle have no relation to anything which occurs in