of its parent. If this moment can be detected very early in the development he will be inclined to project the morphological differentiation back into the germ-plasm and to regard the efforts of the physiologist as relatively unimportant if not altogether futile. Now in his study of the social insects the embryologist is at a serious disadvantage, since he has hitherto been unable to distinguish any prospective worker or queen characters in the eggs or even in the young larvæ. Compelled, therefore, to confine his attention to the older larvæ, whose development as mere processes of histogenesis and metamorphosis throws little or no light on the meaning of polymorphism, he is bound to leave the physiologist in possession of the problem.

The physiologist, in seeking to determine whether there is in the environment of the developing social Hymenopteron any normal

FIG. 64. Gynandromorph of Epipheidole inquilina; male on the left, female on the right side. (Original.)

stimulus that may account for the deviation towards the worker or queen type, can hardly overlook one of the most important of all stimuli, the food of the larva. At first sight this bids fair greatly to simplify the problem of polymorphism, for the mere size of the adult insect would seem to be attributable to the quantity, its morphological deviations to the quality of the food administered to it during its larval life. Closer examination of the subject, however, cannot fail to show that larval alimentation among such highly specialized animals. as the social insects, and especially in the honey-bees and ants, where the differences between the queens and workers are most salient, is a matter of considerable complexity. In the first place, it is evident that it is not the food administered that acts as a stimulus but the portion

of it that is assimilated by the living tissues of the larva. In other words, the larva is not altogether a passive organism, compelled to utilize all the food that is forced upon it, but an active agent, at least to a considerable extent, in determining its own development. And the physiologist might have difficulty in meeting the assertion that the larva utilizes only those portions of the proffered food that are most

FIG. 65. Gynandromorph of Formica microgyna; head almost purely female, gaster male, thorax, petiole and legs male on left, female on right side. (Original.)

conducive to the specific, predetermined trend of its development. In the second place, while experiments on many organisms have shown that the quantity of assimilated food may produce great changes in size and stature, there is practically nothing to show that even very great differences in the quality of the food can bring about morphological differences of such magnitude as those which separate the queens and workers of many ants.

These more general considerations are reinforced by the following inferences from the known facts of larval feeding:

1. There seems to be no valid reason for supposing that the morphogeny of the queens, among the social Hymenoptera depends on a particular diet, since with the possible exception of the honey- and stingless bees, to be considered presently, they differ in no essential respect from the corresponding sexual phase of the solitary species. In both cases they are the normal females of the species and bear the same morphological relations to their males quite irrespective of the nature of their larval food. Hence, with the above mentioned exception of the honey- and stingless bees, the question of the morphogenic value of the larval food may be restricted to the worker forms.

2. Observations show that although the nature of the food administered to the larvæ of the various social insects is often very different, even in closely related species, the structure of the workers may be extremely uniform and exhibit only slight specific differences. Among ants, as we have seen (p. 74), the larvæ are fed with a great variety of substances. The quality of the food itself cannot, therefore, be supposed to have a morphogenic value. And even if we admit what seems to be very probable, namely, that a salivary secretion-possibly

containing an enzyme-may be administered by some ants, at least to their younger larvæ, the case against the morphogenic effects of qualitative feeding is not materially altered, as we see from the following considerations:

3. In incipient ant-colonies the queen mother takes no food often for as long a period as eight or nine months, and during all this time is compelled to feed her first brood of larvæ exclusively on the secretions of her salivary glands. This diet, which is purely qualitative, though very limited in quantity, produces only workers and these of an unusually small size (micrergates).

4. In the honey-bees, on the other hand, qualitative feeding, namely with a secretion, the so-called "royal jelly," which according to some authors (Schiemenz) is derived from the salivary glands, according to others (Planta) from the chylific stomach of the nurses, does not produce workers, but queens. In this case, however, the food is administered in considerable quantity, and is not provided by a single starving. mother, as in the case of the ants, but by a host of vigorous and wellfed nurses. Although it has been taken for granted that the fertilized honey-bee becomes a queen as the result of this peculiar diet, the matter appears in a different light when it is considered in connection with von Ihering's recent observations on the stingless bees (Meliponida) of South America (1903). He has shown that in the species of Melipona the cells in which the males, queens and workers are reared are all of the same size. These cells are provisioned with the same kind of food (honey and pollen) and an egg is laid in each of them. Thereupon they are sealed up, and although the larvæ are not fed from day to day, as in the honey-bees, but like those of the solitary bees subsist on stored provisions, this uniform treatment nevertheless results in the production of three sharply differentiated castes. On hatching the queen Melipona has very small ovaries with immature eggs, but in the allied genus Trigona, the species of which differ from the Melipona in constructing large queen cells and in storing them with a greater quantity of honey and pollen, the queen hatches. with her ovaries full of ripe eggs. These facts indicate that the large size of the queen cell and its greater store of provisions are merely adaptations for accelerating the development of the ovaries. Now on reverting to the honey-bee we may adopt a similar explanation for the feeding of the queen larva with a special secretion like the "royal jelly." As is well known, the queen honey-bee hatches in about sixteen days from the time the egg is laid, while the worker, though a smaller insect and possessing imperfect ovaries, requires four or five days more to complete her development. That the special feeding of the

queen larva is merely an adaptation for accelerating the development of the ovaries is also indicated by the fact that this insect is able to lay within ten days from the date of hatching. If this interpretation is correct, the qualitative feeding of the queen larva is not primarily a morphogenic but a growth stimulus.

5. The grossly mechanical withdrawal, by parasites like Orasema (see p. 418), of food substances already assimilated by the larva, produces changes of the same kind as those which distinguish the worker ant from the queen, i. e., microcephaly, microphthalmy, stenonoty and

FIG. 66. Ergatan dromorph of Aphanogaster picea; male on left, worker on right side. (Original.)

aptery. This case is of unusual interest because the semipupa, after the detachment of the parasite, seems to undergo a kind of regeneration and produces a small but harmonious whole out of the depleted formative substances at its disposal. What is certainly a female or soldier semipupa takes on worker characters while the worker semipupa may be said to become infraergatoid as the result of the sudden loss of the formative substances. These observations clearly indicate that the normal worker traits may be the result of starvation or withholding of food rather than the administration of a particular diet.

6. The pseudogynes of Formica admit of a similar interpretation if it be true, as I am inclined to believe (see p. 408), that they arise from starved female larvæ. Here too, the organism undergoes a kind of regeneration or regulation and assumes the worker aspect owing to a dearth of sufficient formative substances with which to complete the development as originally planned. 7. In the preceding cases the ants take on peculiar structural modifications as the result of tolerating parasites that bring about unusual perturbations in the trophic status of the colony. When ants themselves become parasitic on other ants a similar perturbation ensues, but in these cases the morphological effects are confined to the parasitic species and do not extend to their hosts. This must be attributed to the fact that the parasitic species live in affluence and are no longer required to take part in the arduous and exacting labors of the colony. Under such circumstances the inhibitory effects of nutricial castration on the development of the ovaries of the workers are removed and there is a tendency for this caste to be replaced by egglaying gynæcoid individuals or by ergatogynes, or for it to disappear

completely. These effects are clearly visible in nearly all parasitic ants. In the European Harpagoxenus sublevis, for example, the only known females in certain localities are gynæcoid workers. In the American Leptothorax emersoni, as I have shown (1903f), gynæcoid workers and ergatogynes are unusually abundant, while the true females seem to be on the verge of disappearing. Among the typical amazon ants (Polyergus rufescens) of Europe, ergatogynes are not uncommon. In Strongylognathus testaceus the worker caste seems to be dwindling, while in several permanently parasitic genera (Anergates, Wheeleriella, Epacus, Epipheidole and Sympheidole) it has completely disappeared. Only one cause can be assigned to these remarkable effects-the abundance of food with which the parasites are provided by their hosts.

8. In the Ponerinæ and certain Myrmicinæ, like Pheidole, Pogonomyrmex and Aphanogaster, the larvæ are fed on pieces of insects or seeds, the exact assimilative value of which as food can neither be determined nor controlled by the nurses. And while they may perhaps regulate the quantity of food administered, it is more probable that this must fluctuate within limits so wide and indefinite as to fail altogether to account for the uniform and precise morphological results that we witness in the personnel of the various colonies (Fig. 51). Moreover, accurate determination of the food supply by the workers must be quite impossible in cases like that of the Pachycondyla larva bearing the commensal Metopina (see p. 412).

9. The dependence of the different castes of the social insects on the seasons may also be adduced as evidence of the direct effects of the food supply in producing workers and queens. The latter are reared only when the trophic condition of the colony is most favorable, and this coincides with the summer months; in the great majority of species only workers and males are produced at other seasons. Here, too, the cause is to be sought in the deficient quantity of food rather than in its quality, which is in all probability the same throughout the year, especially in such ants as the fungus-growing Attii.

While these considerations tend to invalidate the supposition that qualitative feeding is responsible for the morphological peculiarities of the worker type, they are less equivocal in regard to the morphogenic effects of quantitative feeding. Indeed several of the observations above cited show very clearly that diminution in stature and, in pathological cases, even reversion to the worker form may be the direct effect of underfeeding. To the same cause we may confidently assign several of the atypical phases among ants, such as the micrergates, microgynes, and micranērs, just as we may regard the macrergates,