the common offspring. Now, among the many characteristics thus inherited from both sides in the offspring, there is a sort of struggle for life and a survival of only the fittest and strongest, and thus the offspring improves by the cross. Now, such cross is most completely secured by the separation of the sexes in different individuals—i. e., by unisexuality.
(b.) Facts which bear on the next Preceding Step, viz., the Derivation of Bisexuality from Asexuality.—This is doubtless the most obscure step; yet I believe some light is visible. Here is the greatest gap in the process; yet this gap may be largely filled.
Remember, then, that there is a striking correspondence between the embryonic or ontogenic series and the evolution or phylogenic series—that the former is a rapid recapitulation, as it were by memory, of the main points of the latter. The embryo repeats by a kind of organic memory the main point of its descent from primordial protoplasm. The lesser points, and especially the earliest points, often indeed drop out of memory, but usually the main points remain. Now, in all the higher animals, ontogeny is a continuous change without break, and completed in one generation. In many lower animals, however, there are apparent pauses, and sudden great changes in this process of ontogenic development. These are called metamorphoses. In insects, for example, there are two active conditions, the larva and the perfect insect, and a sort of second passive egg-stage between—the pupa. Here we have a semblance of, but not really, two generations. Of course, only the perfect insect reproduces. But in many still lower animals we find the metamorphoses occupying two or even more distinct generations. It follows, of course, that in these animals (contrary to what occurs in all higher animals) reproduction takes place both in the larval condition and in the perfect or mature condition. Now, the mode of reproduction in these two conditions is of wholly different kinds, the former being non-sexual and the latter sexual. A single example will suffice: The common medusæ or jelly-fishes, as you know, are unisexual—i. e., male and female. The fertilized females produce eggs which grow, not into medusæ, but into polyp-like animals which are the larval form. These polyps produce buds which open into flower-like bolls, then separate and swim away as male and female medusæ, which again produce eggs that spring up as polyp-like larvæ, etc. Here ontogenesis requires two generations to complete itself. In ontogenesis when both kinds of reproduction occur, the non-sexual (gemmation) precedes the sexual (ovulation). This fact strongly suggests, in fact renders almost certain, that the same is true in phylogenesis, or at least in the phylogenesis of this class.
But again: Aphids (plant-lice) also reproduce in the larval condition, and only reach maturity after many successive generations, sometimes as many as nine or ten. In spring these insects are hatched from eggs in a larval wingless condition. From an internal organ analogous
