“In all protected pupæ,” as Chapman says, “the problem has to be faced, how is the imago to free itself from the cocoon or other envelope protecting the pupa.” In the Coleoptera and Hymenoptera the imago becomes perfected within the cocoon or cell, as the case may be, and as Chapman states, “not only throws off the pupal skin within the cocoon, but remains there till its appendages have become fully expanded and completely hardened, and then the mandibles are used to force an outlet of escape,” and he calls attention to the fact that “in many cases, even in some entire families, they are of no use whatever to the imago except in this one particular,” and he cites the Cynipidæ as perhaps the most striking instance of this circumstance.

In those Neuroptera which spin a silken cocoon, e.g. the Hemerobiidæ, the Trichoptera, and in Micropteryx (Fig. 588), the jaws used by the pupa for cutting its way out of the cocoon are even larger in proportion than in the pupa of caddis-flies (Fig. 588), being of extraordinary size.

In Myrmeleon the pupa pushes its way half out of the cocoon, and then remains, while the imago ruptures the skin and escapes (Fig. 589, a).

Thus in the Neuroptera and Trichoptera we have already established the more fundamental methods of escape from the cocoon, which we see carried out in various ways in the more generalized or primitive Lepidoptera.

The most primitive method in the Lepidoptera of escaping from the cocoon seems to be that of Micropteryx.

“In this genus,” says Chapman, “though it is nominally the pupa that escapes from the cocoon, it is in reality still the imago, the imago clothed in the effete pupal skin. To rupture the cocoon it uses not its own jaws, but those of the pupal skin, energizing them, however, in some totally different way from ordinary direct muscular action, their movements being the result of the vermicular movements of the pupa, acting probably by fluid pressure on the articular structure of the jaws, by some arrangement not altogether different perhaps from the frontal sac of the higher Diptera. In the Micropteryges the jaws of the pupa not only rupture the cocoon, but appear to be the most active agents in dragging the pupa through the opening in the cocoon and through any superincumbent earth, being merely assisted by the vermicular action of the abdominal segments, and we find in accordance with this circumstance that the pupal envelope is still very thin and delicate, and has little or no hardening or roughness by which to obtain a leverage against the walls of the channel of escape.” (Trans. Ent. Soc. London, 1896, pp. 570, 571.)

Some sort of a beak or hard process, more or less developed, according to Chapman, adapted for breaking open the cocoon exists in nearly all the Lepidoptera with incomplete pupæ (pupæ incompletæ), except the limacodid and nepticulid section. “In all these instances the pupa emerges from the cocoon precisely as in the Micropteryges, that is, the moth it really is that emerges, but does so encased in the pupal skin. To achieve this object, it seems to have been found most efficient to have three, four, or five abdominal segments capable of movement, but to have the terminal sections (segments) soldered together.”

This cocoon-breaker, as we may call it, is especially developed in Lithocolletis hamadryadella. As described by Comstock, it forms a toothed crest on the forehead which enables it to pierce or saw through the cocoon.

“Each pupa first sawed through the cocoon near its juncture with the leaf and worked its way through the gap, by means of the minute backward-directed spines upon its back, until it reached the upper cuticle of the leaf. Through this cuticle it sawed in the same way that it did through the cocoon. The hole was in each case just large enough to permit the chrysalis to work its way out, holding it firmly when partly emerged. When half-way out it stopped, and presently the skin split across the back of the neck and down in front along the antennal sheaths, and allowed the moth to emerge.”^[104]

We have observed and figured the cocoon-breaker in Bucculatrix, Talæporia (Fig. 590, a), Thyridopteryx, and Œceticus, and rough knobs or slight projection answering the purpose in Hepialidæ, Megalopyge, Zeuzera, and in Datana.^[105] See also the spine on the head of Sesia tipuliformis (Fig. 578).

The imago of the attacine moths cuts or saws through its cocoon by means of a pair of large, stout, black spines (sectores coconis), one on each side of the thorax at the base of the fore wings (Fig. 591), and provided with five or six teeth on the cutting edge (C, D).

Our attention^[106] was drawn to this subject by a rustling, cutting, and tearing noise issuing from a cocoon of Actias luna. On examination a sharp black point was seen moving to and fro, and then another, until both points had cut a rough irregular slit, through which the shoulder of the moth could be seen vigorously moving from side to side. The hole or slit was made in one or two minutes, and the moth worked its way at once out of the slit. The cocoon was perfectly dry. The cocoon-cutter occurs in all the American genera, in Samia cynthia, and is large and well marked in the European Saturnia pavonia-minor and Endromis versicolora. In Bombyx mori the spines are not well marked, and they are quite different from those in the Attaci. There are three sharp points, being acute angles of the pieces at the base of the wing, and it must be these spines which at times perform the cutting through of the threads of the cocoon described by Réaumur, and which he thought was done by the facets of the eyes. It is well known that in order to guard against the moths cutting the threads, silkraisers expose the cocoon to heat sufficient to destroy the enclosed pupa. In Platysamia the cocoon-cutters, though well developed, do not appear to be used at all, and the pupa, like that of the silkworm and other moths protected by a cocoon, moistens the silk threads by a fluid issuing from the mouth, which also moistens the hairs of the head and thorax, together with the antennæ. It remains to be seen whether these structures are only occasionally used, and whether the emission of the fluid is not the usual and normal means of egress of the moth from its cocoon. Dr. Chapman remarks that throughout the obtected moths “there are many devices for breaking through the cocoon: specially constructed weak places in the cocoon, softening fluid, applied by the moth, assisted by special appliances of diverse sorts, such as in Hybocampa^[107] and Attacus,” etc.

As to the fluid mentioned above, Trouvelot states that it is secreted during the last few days of the pupa state, and is a dissolvent for the gum so firmly uniting the fibres of the cocoon. “This liquid is composed in great part of bombycic acid.” (Amer. Naturalist, i, p. 33.)

The pupa of the dipterous genus Sciara (S. ocellaris O. S.) resembles a tineid pupa, and before transforming emerges for about two-thirds of its length from the cocoon; the pupa-skin remaining firmly attached in this position.^[108]

Certain hymenopterous pupæ are provided with temporary deciduous conical processes. Thus we have observed in the pupa of Rhopalum pedicellatum two very prominent acute tubercles between the eyes (h, Fig. 592). As the cocoon is very slight, these may be of use either in extracting itself from the silken threads or in pushing its way along before emerging from the tunnel in the stem of plants. (See also p. 611.)

c. The cremaster

Although this structure is in general confined to lepidopterous pupæ, and is not always present even in them, since it is purely adaptive in its nature, yet on account of its singular mode of development from the larval organs, and the accompanying changes in the pupal abdomen, it should be mentioned in this connection. The cremaster is the stout, triangular, flattened, terminal spine of the abdomen, which aids the pupa in working its way out of the earth when the pupa is subterranean, or in the pupa of silk-spinning caterpillars its armature of secondary hooks and curved setæ enables it to retain its hold on the threads of the interior of its cocoon after the pupa has partially emerged from the cocoon, restraining it, as Chapman well says, “at precisely that degree of emergence from the cocoon that is most desirable.” He also informs us that while in the “pupæ incompletæ the cremaster is attached to an extensible cable, which always allows some emergence of the pupa, in the pupæ obtectæ there is no doubt but that in such cases as the Ichthyuræ, Acronyctæ, and many others, it retains the pupal case in the same position within the cocoon that the living pupa occupied; this is also very usually the case in the Geometræ and in the higher tineids (my pyraloids).”

In many of the more generalized moths there is no cremaster (Micropteryx, Gracilaria, Prodoxus, Tantura, Talæporia, Psychidæ, Hepialidæ, Zeuzera, Nola, Harrisina), though in Tischeria and Talæporia (Fig. 590, but not in Solenobia) and Psychidæ, two stout terminal spines perform the office of a cremaster, or there are simply curved setæ on the rounded, unarmed end of the abdomen, as in Solenobia.

In the obtected Lepidoptera, for example in such a group as the Notodontidæ, where the cremaster is present, though variable in shape, it may from disuse, owing to the dense cocoon, be without the spines and hooks in Cerura, or the cremaster itself is entirely wanting in Gluphisia, and only partially developed in Notodonta. In the butterflies whose pupæ are suspended (Suspensi), the cremaster is especially well developed. Reference might here be made to the temporary pupal structures in certain generalized moths, which take the place of a cremaster, such as the transverse terminal row of spines in Tinea, the two stout spines in Tischeria, and the dense rough integument and thickened callosities of the pupal head and end of abdomen of Phassus, which bores in trees with very hard wood; also the numerous stout spines at the end and sides of the abdomen in Ægerians. These various projections and spines, besides acting as anchors and grappling hooks, in some cases serve to resist strains and blows, and have undoubtedly, like the armature in the larvæ and imagines of other insects, arisen in response to intermittent or occasional pressure, stresses, and impacts.

Mode of formation of the cremaster and suspension of the chrysalis in butterflies.—We are indebted to Riley^[109] for an explanation of the way the cremaster has originated, his observations having been made on species of over a dozen genera of butterflies (Suspensi).

He shows that the cremaster is the homologue of the suranal plate of the larva.^[110] The preliminary acts of the larva have been observed by various authors since the days of Vallisneri, i.e. the larva hanging by the end of the abdomen, turning up the anterior part of the body in a more or less complete curve, and the skin finally splitting from the head to the front edge of the metathoracic segment, and being worked back in a shrivelled mass toward the point of attachment. The critical feat, adds Riley, which has most puzzled naturalists, is the independent attachment of the chrysalis and the withdrawal from and riddance of the larval skin which such attachment implies. Réaumur explained this in 1734 by the clutching of the larval skin between sutures of the terminal segments of the chrysalis, and this is the case, though the sutures act in a somewhat different way.

Before pupation the larva spins a mass or heap of silk, the shape of which is like an inverted settee or a ship’s knee, and “one of the most interesting acts of the larva, preliminary to suspension, is the bending and working of the anal parts in order to fasten the back of the (suranal) plate to the inside of the back of the settee, while the crotchets of the legs are entangled in the more flattened position or seat.”

In shedding the larval skin, the following parts are also shed, and have some part to play in the act of suspension: i.e. 1st, the tracheal ligaments (Fig. 593, tl), or the shed tracheæ from the last or 9th pair of spiracles; 2d, the rectal ligament (Fig. 593, rl), or shed intestinal canal; 3d, the Osborne or retaining membrane (membrana retinens, Fig. 593, mr), which is the stretched part of the membrane around the rectum and in the anal legs, and which is intimately associated with the rectal ligament.

The structures in the chrysalis are, first, the cremaster, with its dorsal (Fig. 594, dcr) and ventral (vcr) ridges, and the cremastral hook-pad (chp), said by Riley to be “thickly studded with minute but stout hooks, which are sometimes compound or furnished with barbs, very much as are some of our fishing-hooks, and which are most admirably adapted to the purpose for which they are intended.”

Secondly, there are the other structures, viz., the sustainers (sustentors), two projections which Riley states “homologize with the soles (plantæ) of the anal prolegs, which take on various forms (3), but are always directed forward so as easily to catch hold of the retaining membrane.” These sustentors are, however, as Jackson^[111] has shown, and as we are satisfied, the vestiges of the anal legs.

Thirdly, the sustentor ridges, which, as Riley states, may be more or less obsolete in some forms, in Paphia (Fig. 596, B) and Limenitis form “quite a deep notch, which doubtless assists in catching hold of the larval skin in the efforts to attach the cremaster.”

“It is principally,” adds Riley, “by the leverage obtained by the hooking of the sustainers in the retaining membrane, which acts as a swimming fulcrum, that the chrysalis is prevented from falling after the cremaster is withdrawn from the larval skin. It is also principally by this same means that it is enabled to reach the silk with the cremastral hook-pads.”

“Dissected immediately after suspension, the last abdominal segment of the larva is found to be bathed, especially between the legs and around the rectum, in an abundance of translucent, membranous material.”

“An hour or more after suspension the end of the forming chrysalis begins to separate from the larval skin, except at the tip of the cremaster (Fig. 597, b). Gradually the skin of the legs and of the whole subjoint (10th segment) stretches, and with the stretching, the cremaster elongates, the rectal piece recedes more and more from the larval rectum, and the sustentor ridges diverge more and more from the cremaster, carrying with them, on the sustainers, a part of the soft membrane.” The rectal ligament will sustain at least 10 or 12 times the weight of the chrysalis. That of Apatura seems to rely almost entirely on the rectal ligament, assisted by the partial holding of the delicate larval skin.

FORMATION OF THE PUPA AND IMAGO IN THE HOLOMETABOLOUS INSECTS (THE DIPTERA EXCEPTED)

We have seen that in the incomplete metamorphosis, although there may be as many as five, and possibly seven moults, and in Chloëon as many as 20, and in Cicada septemdecim perhaps 25 or 30, there is but a slight change of form from one stage to another, and no period of inactivity. And this gradual outer transformation is so far as yet known paralleled by that of the internal organs, the slight successive changes of which do not differ from those observed in the growth of ametabolous insects. With the growth of the internal organs there probably goes on a series of gradual regenerative processes, and Korschelt and Heider state that we may venture to assume that each changed cell or group of cells which have become exhausted by the exercise of the functions of life are reabsorbed and become restored through the vital powers of the tissues, so that as the result there goes on a constant, gradual regeneration of the organs.

While the Hemiptera have only an incomplete metamorphosis, the males of the Coccidæ are, as shown by O. Schmidt, remarkable for passing through a complete or holometabolous development, with four stages, three of which are pupal and inactive. Hence, as Schmidt observes, there is here a hypermetamorphosis, like that of the Meloidæ, Stylopidæ, etc.

Shortly before the end of the larval stage of the male appear the imaginal buds of the eyes, legs, and wings. In the 2d or 1st pupal stage there is an atrophy of the antennæ and legs. On the other hand, at this stage the female completes its metamorphosis.

The rudiments of the wings arise on the edge of the dorsal and ventral side of the 2d thoracic segment, and this, we would remark, is significant as showing a mode of origin of the wings intermediate between that of the manometamorphic and holometamorphic insects. (See pp. 137–142.) While Schmidt could not ascertain the exact structure of the imaginal buds, he says “in general the process of formation of the extremities is exactly as Weismann has described in Corethra.” The two later pupal stages are “as in other metabolic insects.” (See p. 690, Fig. 637.)

Thus far the internal changes in the metamorphosis of the Coleoptera have not been thoroughly studied. They are less complete than in the other holometabolous insects, the differences between the larva and imago being much less marked than in the more specialized orders, and so far as known all the larval organs pass, though not without some great changes, directly into the imaginal ones, the only apparent exception being the mid-intestine, which, as stated by Kowalevsky, undergoes a complete transformation during metamorphosis. The following account, then, refers almost wholly to the Lepidoptera, Hymenoptera, and Diptera.

a. The Lepidoptera

The first observations on the complete metamorphosis of insects which were in any way exact were those of Malpighi, in 1667, and of Swammerdam, in 1733. While the observations of Swammerdam, as far as they extended, were correct, his conclusions were extraordinary. They were, however, accepted by Réaumur and by Bonnet, and generally held until the time of Herold in 1815, and lingered on for some years after. The rather famous theory of incasement (“emboîtement”) propounded by Swammerdam was that the form of the larva, pupa, and imago preëxisted in the egg, and even in the ovary; and that the insects in these stages were distinct animals, contained one inside the other, like a nest of boxes, or a series of envelopes one within the other, or, to use his own words: “Animal in animali, seu papilio intra erucam reconditus.”

This theory Swammerdam extended to the whole animal kingdom. It was based on the fact that by throwing the caterpillar, when about to pupate, in boiling water, and then stripping off the skin, the immature form of the butterfly with its appendages was disclosed. Malpighi had previously observed the same fact in the silkworm, perceiving that before pupation the antennæ are concealed in the head of the larva, where they occupy the place previously taken by the mandibular muscles; also that the legs of the moth grew in those of the larva, and that the wings developed from the sides of the worm.

Even Réaumur (1734) remarked: “Les parties du papillon cachées sous le fourreau de chenille sont d’autant plus faciles à trouver que la transformation est plus proche. Elles y sont neanmoins de tout temps.” He also believed in the simultaneous existence of two distinct beings in the insect. “Il serait très curieux de connaître toutes les communications intimes qui sont entre la chenille et le papillon.... La chenille hache, broye, digere les aliments qu’elle distribué au papillon; comme les mères préparent ceux qui sont portés aux fœtus. Notre chenille en un mot est destineé à nourrir et à defendre le papillon qu’elle renferme.” (T. i, 8^e Mémoire, p. 363.)

Lyonet (1760), even, did not expose the error of this view that the larva enveloped the pupa and imago, and, as Gonin says, it was undoubtedly because he did not use for his dissections of the caterpillar of Cossus any specimens about to pupate. Yet he detected the wing-germs and those of the legs, stating that he presumed the bodies he saw to be the rudiments of the legs of the moth (p. 450).

Herold, in his work on the development of the butterfly (1815), was the first to object to this erroneous theory, showing that the wings did not become visible until the very end of larval life; that as the larval organs disappear, they are transformed or are replaced by entirely new organs, which is not reconcilable with a simple putting off of the outer envelope. The whole secret of metamorphosis, in Herold’s opinion, consisted in this fact, that the butterfly in the larva state increases and accumulates a supply of fat until it has reached the volume of the perfect state; then it begins the chrysalis period, during which the organs are developed and take their definite form.^[112] (Abstract mostly from Gonin.) Still the old ideas prevailed, and even Lacordaire, in his Introduction à l’Entomologie published in 1834, held on to Swammerdam’s theory, declaring that “a caterpillar is not a simple animal, but compound,” and he actually goes so far as to say that “a caterpillar, at first scarcely as large as a bit of thread, contains its own teguments threefold and even eightfold in number, besides the case of a chrysalis, and a complete butterfly, all lying one inside the other.” This view, however, we find is not original with Lacordaire, but was borrowed from Kirby and Spence without acknowledgment. These authors, in their Introduction to Entomology (1828), combated Herold’s views and stoutly maintained the old opinions of Swammerdam. They based their opinions on the fact, then known, that certain parts of the imago occur in the caterpillar. On the other hand, Herold denied that the successive skins of the pupa and imago existed as germs, holding that they are formed successively from the “rete mucosum,” which we suppose to be the hypodermis of later authors. In a slight degree the Swammerdam-Kirby and Spence doctrine was correct, as the imago does arise from germs, i.e. the imaginal disks of Weismann, while this was not discovered by Herold, though they do at the outset arise from the hypodermis, his rete mucosum. Thus there was a grain of truth in the Swammerdam-Kirby and Spence doctrine, and also a mixture of truth and error in the opinions of Herold.

The real nature of the internal changes wrought during the process of metamorphosis was first revealed by Weismann in 1864. His discovery of the germs of the imago (imaginal buds) of the Diptera, and his theory of histolysis, or of the complete destruction of the larval organs by a gradual process, was the result of the application of modern methods of embryology and histology, although his observations were first made on the extremely modified type of the Muscidæ or flies, and, at first, he did not extend his view to include all the holometabolous insects. Now, thanks to his successors in this field, Ganin, Dewitz, Kowalevsky, Van Rees, Bugnion, Gonin, and others, we see that metamorphosis is, after all, only an extension of embryonic life, the moults and great changes being similar to those undergone by the embryo, and that metamorphosis and alternation of generations are but terms in a single series. Moreover, the metamorphoses of insects are of the same general nature as those of certain worms, of the echinoderms, and the frog, the different stages of larva, pupa, and imago being adaptational and secondary.

While the changes in form from the larva to the pupa are apparently sudden, the internal histogenetic steps which lead to them are gradual. In the Lepidoptera a few days (usually from one to three) before assuming the pupa stage, the caterpillar becomes restless and ceases to take food. Its excrements are now hard, dry, and, according to Gonin, are “stained carmine red by the secretions of the urinary tubes.” Under the microscope we find that they are almost exclusively composed of fragments of the intestinal epithelium. These red dejections were noticed by Réaumur, and afterwards by Herold, and they are sure indications of the approach of the transformations. It now wanders about, and, if it is a spinner, spins its cocoon, and then lies quietly at rest while the changes are going on within its body. Meanwhile, it lives on the stores of fat in the fat-body, and this supply enables it to survive the pupal period.

The amount of fat is sometimes very great. Newport removed from the larva of Cossus ligniperda 42 grains of fat, being more than one-fourth of the whole weight of the insect, he adds that the supply is soon nearly exhausted during the rapid development of the reproductive organs, “since, when these have become perfected, the quantity that remains is very inconsiderable.”

Although the larval skin of a lepidopterous insect is suddenly cast off, the pupa quickly emerging front it, yet there are several intermediate stages, all graduating into each other. If a caterpillar of a Clisiocampa, which, as we have observed, is much shortened and thickened a day or two before changing to a pupa, is hardened in alcohol and the larval skin is stripped off, the semipupa (pro-nymph, pro-pupa of different authors) is found to be in different stages of development, and the changes of the mouth-parts are interesting, though not yet sufficiently studied.

Newport attributes the great enlargement and changes in the shape of the thoracic segments of the larva of Vanessa urticæ at this time, to the contraction or shortening of the muscles of the interior of those segments, “which are repeatedly slowly extended and shortened, as if the insect were in the act of laborious respiration.” This, he adds, generally takes place at short intervals during the two hours immediately preceding the change to the pupa, and increases in frequency as that period approaches. He thus describes the mode of moulting the larval skin: “When the period has arrived, the skin bursts along the dorsal part of the 3d segment, or mesothorax, and is extended along the 2d and 4th, while the coverings of the head separate into three pieces. The insect then exerts itself to the utmost to extend the fissure along the segment of the abdomen, and, in the meantime, pressing its body through the opening, gradually withdraws its antennæ and legs, while the skin, by successive contortions of the abdomen, is slipped backwards, and forced towards the extremity of the body, just as a person would slip off his glove or his stocking. The efforts of the insect to get entirely rid of it are then very great; it twirls itself in every direction in order to burst the skin, and, when it has exerted itself in this manner for some time, twirls itself swiftly, first in one direction, then in the opposite, until at last the skin is broken through and falls to the ground, or is forced to some distance from it. The new pupa then hangs for a few seconds at rest, but its change is not yet complete. The legs and antennæ, which when withdrawn from the old skin were disposed along the under surface of the body, are yet separate, and do not adhere together as they do a short time afterwards. The wings are also separate and very small. In a few seconds the pupa makes several slow, but powerful, respiratory efforts; during which the abdominal segments become more contracted along their under surface, and the wings are much enlarged and extended along the lateral inferior surface of the body, while a very transparent fluid, which facilitated the slipping off of the skin, is now diffused among the limbs, and when the pupa becomes quiet dries, and unites the whole into one compact covering.”

The changes in the head and mouth-parts.—The changes of form from the active mandibulate caterpillar to the quiescent pupa, and then to the adult butterfly, are, as we have seen, in direct adaptation to their changed habits and surroundings, and they differ greatly in details in insects of different orders. In many Lepidoptera and certain Diptera the pupa and imago are without the mandibles of the larva, and, instead, the 1st maxillæ in the former order, and the 2d maxillæ in the latter, are highly developed and specialized. The changes in the shape of the head, with the antennæ, the latter rudimentary in the larvæ of the two orders named, are noteworthy, and will be referred to under those orders. The same may be said of the thorax with the legs and wings, and the abdomen with the ovipositor. Every part of the body undergoes a profound change, though in the Coleoptera, Trichoptera, and the more generalized and primitive Diptera, each segment and appendage of the larva are directly transformed into the corresponding parts of the pupa, and subsequently of the imago. We shall see, however, beyond, that this general statement does not apply to the Hymenoptera, in which there is a process of cephalization or transfer of parts headward, peculiar to that order.