This suggestion seems justifiable if we confine ourselves to superficial considerations. We read in every handbook of entomology that the wings only arise during the life of the caterpillar, and in a certain sense this is true, for the primary constituents or primordia of wings only develop into the fully formed wing during the larval period. But even if these primordia were only formed during the caterpillar-stage, what could they develop from? Only out of the material parts of the caterpillar, that is, from some of its living cells or cell-groups. The constitution of the wings would therefore be dependent on that of the cells of the caterpillar from which they arose, so that if these varied transmissibly through the variation of their determinants contained in the germ, the determinants of the butterfly which were just developing would vary with them; every transmissible variation of the caterpillar would necessarily cause a similar variation in the butterfly, and this does not happen. If any one hazarded the assumption that the determinants of the butterfly develop only in the caterpillar, but quite independently of its constitution, he would either be making an absurd statement, namely, that the characters of the butterfly were not transmissible at all, or he would be unconsciously admitting that the determinants of the butterfly were already contained in the parts of the caterpillar, and come direct from the germ-plasm.
That the characters of the butterfly do vary independently of those of the caterpillar I demonstrated many years ago, when we were still very far away from the idea of the germ-plasm or of determinants. I demonstrated then that the constancy of the markings of a species can be quite different in the two chief stages; that the caterpillar may be very variable, while the butterfly or the moth may be very constant in all its markings, or conversely. I called attention to the dimorphic caterpillars which are green or brown, and yet become the same moth (for instance, Deilephila elpenor and Sphinx convolvuli); I cited the case of the spurge hawk-moth (Deilephila euphorbiæ), whose dark but at the same time motley caterpillars occur in the Riviera at Nice as a local variety (Nicæa), and there wear quite a different dress—pale clay-yellow, with a double row of large conspicuous dark yellow eye-spots—while the moth does not differ from our variety in a single definite character, except in its larger size. At that time, too, I instituted experiments with the caterpillars of the smallest of our indigenous Vanessa species (Vanessa levana), of which the majority are black with black thorns, while a minority are yellowish-brown with yellow thorns; reared separately, both yielded the same butterfly, though in this case one would be inclined to suppose that there was some internal connexion between the colour of the caterpillar and that of the butterfly, since the butterfly also occurs in two colours. It was shown, however, that the colour of the butterfly had nothing to do with that of the caterpillar, for it is known to be dependent on the season, and is a seasonal dimorphism, 'while the two forms of caterpillar may occur side by side at all times of the year.'
Subsequently I made a similar experiment with the dimorphic caterpillars of the 'fire'-butterfly (Polyommatus phlæas), and it yielded the same result. The pure green caterpillars became the same butterflies as those marked with broad red longitudinal stripes, and in this case we can definitely describe both colours as protective, for the green form is adapted to the green under surface of the leaf, the red-striped to the green red-edged stalk of the lesser sorrel (Rumex acetosella).
There was really no necessity for special proofs that the caterpillar and butterfly vary transmissibly in complete independence of each other, for the facts of metamorphosis alone are enough to prove it. How would it have been possible otherwise that the jaws adapted for biting should, in the primitive insects, and in the locusts which are nearest to them, remain as a biting apparatus throughout life, while in the caterpillar they are modified during its pupal stage into the suctorial proboscis of the butterfly? The parts of insects, therefore, must be capable of transmissible variation in the stages of life independently of each other. Not only have the jaws of the leaf-eating caterpillars remained unaltered, while in the sexually mature animal they have been gradually modified into a very long and extremely complex suctorial apparatus, but when at a much later time this proboscis became superfluous in a species, because the butterfly or moth, from some cause or another, lost the habit of taking any nourishment at all, its degeneration exercised no effect on the jaws of the caterpillar, as we can observe in many hawk-moths, silk-moths and Geometridæ. How could such a degeneration become transmissible if the caterpillar's jaws, from which those of the adult are developed, remain the same? We are thus forced to assume that there is something in the latter which can vary from the germ, without the jaws themselves being altered thereby. This 'something' it is which I call 'determinants,' vital particles, which—however we may try to picture them—are indeed contained in the cells of the caterpillar's jaws, but are there inactive and do not influence the structure of these, while, on the other hand, it is their constitution which determines the form and structure of the suctorial proboscis of the butterfly down to the minutest details. It must be these alone which cause the suctorial proboscis to develop, and in some cases to degenerate again, without bringing about any change in the corresponding parts in the caterpillar.
Fig. 89. Anterior region of the larva
of a Midge (Corethra plumicornis). K, head.
Th, thorax. ui, inferior imaginal disks.
oi, superior imaginal disks. ui1, ui2, and
ui3, the primordia of the limbs. oi2
and oi3, the primordia of the wings and
'balancers.' g, brain. bg, chain of ventral
ganglia with nerves which enter the
imaginal disks. trb, tracheal vesicle.
Enlarged about 15 times.
This example seems to me to be preferable to that of the wings of insects in this respect, that there is no organ in the caterpillar with a specific function corresponding to the wing of the butterfly. Yet the two cases are exactly alike, and it would be a mistake to say that the first primordium of the wing within the caterpillar is not a part of the caterpillar at all. At first, certainly, it is only a group of cells on the skin, occurring at a particular spot on the dorsal surface of the second and third segments of the caterpillar, and doubtless arising from a single cell of the embryo, the 'primitive wing-cell,' which, however, has not as yet been demonstrated. But it is nevertheless an integral part of the caterpillar, which could neither be wanting, nor be larger or smaller, and so on; which, in short, does mean something for the caterpillar, although perhaps not more than any other of the skin-cells. For the butterfly, however, this area on the skin means the rudiment of the wing; for from it alone can there arise by multiplication the aggregate of cells which grows out into a hollow protuberance, enlarges by degrees into a disk, the imaginal disk, and eventually develops into the form of wing peculiar to the species. This imaginal disk is connected very early with nerves and with tracheæ, as may be beautifully seen especially in dipterous larvæ (Fig. 89, oi), and these become later the nerves and tracheæ of the wing, while thousands of peculiar scale-like hairs develop on the upper surface; in short, the rudiment becomes a perfect wing with its specific venation, and with the marking and colouring which is often so complicated in Lepidoptera. Almost every little spot and stripe of the latter is handed down with the most tenacious power of transmission from generation to generation, and each can at the same time be transmissibly varied; the same is true of the venation, which is so important systematically just because it is so strictly hereditary, yet it too can vary transmissibly, as can also the hooked bristles, the odoriferous apparatus, and, in short, the whole complex structure of the wing, with all its specific adaptations to the mode of flight, to the manner of life, and to the colour of the environment. How is it possible that all this can develop from a skin-cell? Is it the influence of position that effects it, and could any other cell of the caterpillar's skin do the same if it were placed in the same position? Could any neighbour-cell of the primitive wing-cell replace it if it were destroyed? It is hardly probable, and I think I can even prove that this is not so. The experiment of killing such a cell in the living animal has not yet been made; if it should succeed, we may venture to say in advance that none of the neighbouring skin-cells will be able to do its work and take its place in developing a wing; the wing in question will simply remain undeveloped. In the summer of 1897 I hatched a specimen of Vanessa antiope from the pupa, which, though otherwise normal and well-developed, lacked the left posterior wing altogether; no trace of it could be recognized. In this case, from some cause which could no longer be discovered, the first formative cell of the wing in the hypodermis, or its descendants, must have been destroyed, and no substitution of another took place, as the defect showed.
The young science of developmental mechanics attributes to the position of a cell in the midst of a group of cells a determining value as regards its further fate, and as far as the cells of the segmenting ovum are concerned this seems to be true in certain cases, but the assumption cannot be generally true except in a very subordinate sense. The formative cell of the wing does not become what it is because of its relative position in the organism. If this were so it could not happen that a wing should develop instead of a leg, as was observed in a Zygæna, nor could there be any of those deformities already referred to, to which the name 'Heterotopia' is applied, and which consist in the development of organs of definite normal structure, or at any rate of apparently normal structure in quite unusual places, e. g. an antenna on the coxa of a leg, or of a leg instead of an antenna (in Sirex), or instead of a wing. It is therefore not some influence from without that makes that particular skin-cell of the caterpillar the rudiment of the wing, but the reason lies within itself, in its own constitution. As the whole mass of determinants for the whole body and for all the stages of its development must be contained within the ovum and the sperm-cell, so the primitive cell of the butterfly's wing must contain all the determinants for the building up of this complicated part; and if the cell gets into a wrong position in the course of development because of some disturbance or other, a wing may develop from it in that position if the conditions are not too utterly divergent. These heterotopic phenomena afford a further proof of the existence of determinants, because they are quite unintelligible without the assumption of 'primary constituents' or Anlagen.
The hypothesis of determinants in the germ-plasm is so fundamental to my theory of development that I should like to adduce another case in its support and justification. The limbs of the jointed-footed animals, or Arthropods, originally arose as a pair on each segment of the body, and they were at first alike or very similar both in their function and in their form. We find illustration of this in the millipedes, and still more in the species of the interesting genus Peripatus, which resembles them externally, as well as in the swimming and creeping bristle-footed marine worms (Chætopods) belonging to the Annelid phylum. We can quite well picture to ourselves that the whole series of these appendages was represented in the germ-plasm by a single determinant or group of determinants, which only required to be multiplied in development. Without disputing whether this has really been the case in the primitive Arthropods or not, it is certain that it can no longer be the case in the germ-plasm of the Arthropods of to-day. In these each pair of appendages must be represented by a particular determinant. We must infer this from the fact that the several pairs of these appendages have varied transmissibly, independently of each other, for some are jaws, others swimming legs, or merely bearers of the gills or of the eggs; others are walking legs, digging legs, or jumping legs. In Crustaceans a forceps-like claw is often borne by the first of the otherwise similarly constructed appendages, or also by the second or the third, or there may be no forceps, and so on; in short, we see that each individual pair has adapted itself independently to the mode of life of its species. This could only have been possible if each was represented in the germ-plasm by an element, whose variations caused a variation only in that one pair of legs, and in no other.