are necessary for the development of an individual: these secondary units I have spoken of as 'ids.' Starting with this assumption, there is no difficulty in conceiving that the germ-plasm of bees at the present day is composed of different kinds of ids, some of which contain the primary constituents of workers, some those of queens, and others those of males; and there is no reason why the worker-ids of ants should not be supposed to be of two kinds—those of workers in the strict sense of the word, and those of soldiers (see Note XII, p. 63). The 'male' ids—if I may be allowed the expression—become active in the absence of fertilization, the 'female' ids on the occurrence of fertilization, and the kind of food supplies the stimulus for the worker-ids or queen-ids. Slow processes of selection have gradually changed the 'female' ids in two directions, and finally led to the establishment of two perfectly distinct forms of females. That the process has not been sudden, but has been brought about step by step, is apparent; for even at the present day a number of stages in these metamorphoses are still to be found among the workers of the different species of ants. Moreover, sporadic transition·forms between workers and females also occur, and show varied combinations of characteristics, just as in the case of hermaphrodites that occur abnormally from time to time and exhibit extraordinary and often perfectly methodless mingling of sexual distinctions; in bees especially, wonderful combinations of the characters of both sexes have been observed.
The facts that are furnished by the gradual meta-
