these comparisons, it is to be noted that the dragon-flies and may-flies
are more highly specialised insects than stone-flies, divergent
specialisation of the adult and larva is therefore well illustrated in
these groups, which nevertheless have, like the Hemiptera and
Orthoptera, visible external wing-rudiments.

From the vast array of insects that show internal wing-growth and a true
pupal stage, a few larval types were chosen for description in Chapter
VI, and a review of these suggests again the thought of increasing
divergence between larva and imago. Reference has been made previously
to the many instances in which the former has become pre-eminently the
feeding, and the latter the breeding stage in the life-cycle. It seems
impossible to avoid the conclusion that the active, armoured
campodeiform grub differing less from its parent than an eruciform larva
differs from its parent, is as a larval type more primitive than the
caterpillar or maggot. A. Lameere has indeed, while admitting the
adaptive character of insect larvae generally, argued (1899) with much
ingenuity that the eruciform or vermiform type must have been primitive
among the Endopterygota, believing that the original environment of the
larvae of the ancestral stock of all these insects must have been the
interior of plant tissues. He is thus forced to the necessity of
suggesting that the campodeiform larvae of ground-beetles or lacewings
must be regarded as due to secondarily acquired adaptations; 'they
resemble Thysanura and the larvae of Heterometabola only as whales
resemble fishes.' There are two considerations which render these
theories untenable. The Neuroptera and Coleoptera among which
campodeiform larvae are common, are less specialised than Lepidoptera,
Hymenoptera, and Diptera, in which they are unknown. And among the
Coleoptera which as we have seen (pp. 50 _f._) display a most
interesting variety of larval structure, the legless, eruciform larva