of labor. It attains its clearest expression in the social insects, in some of which, like the termites, we find both sexes equally polymorphic, while in others, like the ants, social bees, and wasps, the female alone, with rare exceptions, is differentiated into distinct castes. This restriction of polymorphism to the female in the social Hymenoptera, with which we are here especially concerned, is easily intelligible if it be traceable, as is usually supposed, to a physiological division of labor, for the colonies of ants, bees and wasps are essentially more or less permanent families of females, the male representing merely a fertilizing agency temporarily intruding itself on the activities of the community at the moment it becomes necessary to start other colonies.

FIG. 50.

We may say, therefore, that polymorphism among social Hymenoptera is a physical expression of the high degree of social plasticity and efficiency of the female sex among these insects. This is shown more, specifically in two characteristics of the female, namely the extraordinary intricacy and amplitude of her instincts, which are thoroughly representative of the species, and her ability to reproduce parthenogenetically. This, of course, means a considerable degree of autonomy even in the reproductive sphere. But parthenogenesis, while undoubtedly contributing to the social efficiency of the female, must be regarded and treated as an independent phenomenon, without closer connection with polymorphism, for the ability to develop from unfertilized eggs is an ancient characteristic of the Hymenoptera and

FIG. 51. Colony of Acanthomyops claviger, showing workers, deälated and virgin females, males, worker, male and female cocoons, X 2. (Photograph by J. G. Hubbard and Dr. O. S. Strong).

many other insects, which made its appearance among the solitary species, like the Tenthredinidæ and Cynipidæ, long before the development of social life. Moreover, polymorphism may occur in male insects which, of course, are not parthenogenetic. That parthenogenesis is intimately connected with sexual dimorphism, at least among the

FIG. 52. Pheidole instabilis. (Original.) a, Soldier; b-e, intermediate workers; f, typical worker (micrergate); g, deälated female; h, male.

social Hymenoptera, seems to be evident from the fact that the males usually if not always develop from unfertilized, the females from fertilized eggs.

While the bumble-bees and wasps show us the ancient stages in the development of polymorphism, the ants as a group, with the exception of a few parasitic genera that have secondarily lost this character, are all completely polymorphic. It is conceivable that the

FIG. 53. Cryptocerus varians. (Original.) a, Soldier; b, same in profile; c, head of same from above; d, worker; e, female; f, male.

development of different castes in the female may have arisen independently in each of the three groups of the social Hymenoptera, although it is equally probable that they may have inherited a tendency to polymorphism from a common extinct ancestry. On either hypothesis, however, we must admit that the ants have carried the development of the female castes much further than the social bees and wasps,

since they have not only produced a wingless form of the worker, in addition to the winged female, or queen, but in many cases also two distinct castes of workers known as the worker proper and the soldier.

Different authors have framed very different conceptions of the phylogenetic beginnings of social life among the Hymenoptera and consequently also of the phylogenetic origin and development of polymorphism. Thus Herbert Spencer (1893) evidently conceived the colony as having arisen from consociation of adult individuals, and although he unfortunately selected a parasitic ant, the amazon (Polyergus rufescens), on which to hang his hypothesis, there are a few facts which at first sight seem to make his view applicable to other social Hymenoptera. Fabre (1894) once found some hundreds of specimens of a solitary wasp (Ammophila hirsuta) huddled together under a stone on the summit of Mt. Ventoux in the Provence, at an altitude of about 5,500 feet, and Forel (1874) found more than fifty deälated females of Formica rufa under similar conditions on the Simplon. I have myself seen collections of a large red and yellow Amblyteles under stones on Pike's Peak at an altitude of more than 13,000 feet, and a mass of about seventy deälated females of Formica gnava apparently hibernating after the nuptial flight under a stone. near Austin, Texas. I am convinced, however, that such congregations are either entirely fortuitous, especially where the insects of one species are very abundant and there are few available stones, or, that they are, as in the case of F. rufa and gnava, merely a manifestation of highly developed social proclivities and not of such proclivities in process of development.

A very different view from that of Spencer is adopted by most authors, who regard the insect colony as having arisen, not from a chance concourse of adult individuals, but from a natural affiliation of mother and offspring. This view, which has been elaborated by Marshall (1889) among others, presents many advantages over that of Spencer, not the least of which is its agreement with what actually occurs in the founding of the existing colonies of wasps, bumble-bees and ants. These colonies pass through an ontogenetic stage which has all the appearance of repeating the conditions under which colonial life first made its appearance in the phylogenetic history of the species-the solitary mother insect rearing and affiliating her offspring under conditions which would seem to arise naturally from the breeding habits of the nonsocial Hymenoptera. The exceptional methods of colony formation seen in the swarming of the honey bee and in the temporary and permanent parasitism of certain ants, are too obviously secondary and of too recent a development to require extended comment. The