The change that is required to transform Exopterygota into Endopterygota is merely that a cell of hypodermis should proliferate inwards instead of outwards, or that a minute hypodermal evaginated bud should be forced to the interior of the body by the pressure of a contracted cuticle.
If it should be objected that the wings so developed would be rudimentary, and that there would be nothing to encourage their development into perfect functional organs, we may remind the reader that we have already pointed out that imperfect wings of Exopterygota do, even at the present time under certain conditions, become perfect organs; and we may also add that there are, even among existing Endopterygota, species in which the wings are usually vestiges and yet sometimes become perfectly developed. In fact, almost every condition that is required for the change from exopterygotism to endopterygotism exists among the insects that surround us.
But it may perhaps be considered improbable that organs like the wings, having once been lost, should have been reacquired on the large scale suggested by the theory just put forward. If so, there is an alternative method by which the endopterygotous may have arisen from the exopterygotous condition. The sub-imago of the Ephemeroptera suggests that a moult, after the wings had become functional, was at one time general among the Hexapoda, and that the resting nymph of the Thysanoptera or the pupa of the Endopterygota represents a formerly active stage in the life-history. Further, although the wing-rudiments appear externally in an early instar of an exopterygotous insect, the earliest instars are wingless and wing-rudiments have been previously developing beneath the cuticle, growing however outwards, not inwards as in the larva of an endopterygote. The change from an exopterygote to an endopterygote development could, therefore, be brought about by the gradual postponement to a later and later instar of the appearance of the wing-rudiments outside the body, and their correlated growth inwards as imaginal disks. For in the post-embryonic development of the ancestors of the Endopterygota we may imagine two or three instars with wing-rudiments to have existed, the last represented by the sub-imago of the may-flies. As the life-conditions and feeding-habits of the larva and imago become constantly more divergent, the appearance of the wing-rudiments would be postponed to the pre-imaginal instar, and that instar would become predominantly passive.
Relationships of the Orders.—Reasons have been given for regarding the Thysanura as representing, more nearly than any other living group, the primitive stock of the Hexapoda. It is believed that insects of this group are represented among Silurian fossils. We may conclude, therefore, that they were preceded, in Cambrian times or earlier, by Arthropods possessing well developed appendages on all the trunk-segments. Of such Arthropods the living Symphyla—of which the delicate little Scutigerella is a fairly well-known example—give us some representation.
No indications beyond those furnished by comparative anatomy help us to unravel the phylogeny of the Collembola. In most respects, the shortened abdomen, for example, they are more specialized than the Thysanura, and most of the features in which they appear to be simple, such as the absence of a tracheal system and of compound eyes, can be explained as the result of degradation. In their insunken mouth and their jaws retracted within the head-capsule, the Collembola resemble the entotrophous division of the Thysanura (see Aptera), from which they are probably descended.
From the thysanuroid stock of the Apterygota, the Exopterygota took their rise. We have undoubted fossil evidence that winged insects lived in the Devonian and became numerous in the Carboniferous period. These ancient Exopterygota were synthetic in type, and included insects that may, with probability, be regarded as ancestral to most of the existing orders. It is hard to arrange the Exopterygota in a linear series, for some of the orders that are remarkably primitive in some respects are rather highly specialized in others. As regards wing-structure, the Isoptera with the two pairs closely similar are the most primitive of all winged insects; while in the paired mesodermal genital ducts, the elongate cerci and the conspicuous maxillulae of their larvae the Ephemeroptera retain notable ancestral characters. But the vestigial jaws, numerous Malpighian tubes, and specialized wings of may-flies forbid us to consider the order as on the whole primitive. So the Dermaptera, which retain distinct maxillulae and have no ectodermal genital ducts, have either specialized or aborted wings and a large number of Malpighian tubes. The Corrodentia retain vestigial maxillulae and two pairs of Malpighian tubes, but the wings are somewhat specialized in the Copeognatha and absent in the degraded and parasitic Mallophaga. The Plecoptera and Orthoptera agree in their numerous Malpighian tubes and in the development of a folding anal area in the hind-wing. As shown by the number and variety of species, the Orthoptera are the most dominant order of this group. Eminently terrestrial in habit, the differentiation of their fore-wings and hind-wings can be traced from Carboniferous, isopteroid ancestors through intermediate Mesozoic forms. The Plecoptera resemble the Ephemeroptera and Odonata in the aquatic habits of their larvae, and by the occasional presence of tufted thoracic gills in the imago exhibit an aquatic character unknown in any other winged insects. The Odonata are in many imaginal and larval characters highly specialized; yet they probably arose with the Ephemeroptera as a divergent offshoot of the same primitive isopteroid stock which developed more directly into the living Isoptera, Plecoptera, Dermaptera and Orthoptera.
All these orders agree in the possession of biting mandibles, while their second maxillae have the inner and outer lobes usually distinct. The Hemiptera, with their piercing mandibles and first maxillae and with their second maxillae fused to form a jointed beak, stand far apart from them. This order can be traced with certainty back to the early Jurassic epoch, while the Permian fossil Eugereon, and the living order—specially modified in many respects—of the Thysanoptera indicate steps by which the aberrant suctorial and piercing mouth of the Hemiptera may have been developed from the biting mouth of primitive Isopteroids, by the elongation of some parts and the suppression of others. The Anoplura may probably be regarded as a degraded offshoot of the Hemiptera.
The importance of great cardinal features of the life-history as indicative of relationship leads us to consider the Endopterygota as a natural assemblage of orders. The occurrence of weevils—among the most specialized of the Coleoptera—in Triassic rocks shows us that this great order of metabolous insects had become differentiated into its leading families at the dawn of the Mesozoic era, and that we must go far back into the Palaeozoic for the origin of the Endopterygota. In this view we are confirmed by the impossibility of deriving the Endopterygota from any living order of Exopterygota. We conclude, therefore, that the primitive stock of the former sub-class became early differentiated from that of the latter. So widely have most of the higher orders of the Hexapoda now diverged from each other, that it is exceedingly difficult in most cases to trace their relationships with any confidence. The Neuroptera, with their similar fore- and hind-wings and their campodeiform larvae, seem to stand nearest to the presumed isopteroid ancestry, but the imago and larva are often specialized. The campodeiform larvae of many Coleoptera are indeed far more primitive than the neuropteran larvae, and suggest to us that the Coleoptera—modified as their wing-structure has become—arose very early from the primitive metabolous stock. The antiquity of the Coleoptera is further shown by the great diversity of larval form and habit that has arisen in the order, and the proof afforded by the hypermetamorphic beetles that the campodeiform preceded the eruciform larva has already been emphasized.
In all the remaining orders of the Endopterygota the larva is eruciform or vermiform. The Mecaptera, with their predominantly longitudinal wing-nervuration, serve as a link between the Neuroptera and the Trichoptera, their retention of small cerci being an archaic character which stamps them as
