But how are we to understand the fact that pure parthenogenesis may last through innumerable generations, and yet no degeneration sets in? I believe very simply. In this case too, the same ids which were peculiar to the mother of the race are contained in the descendants, but they do not diminish in number, for in pure and normal parthenogenesis, such as that of _Cypris reptans_, the second maturation-division of the ovum does not take place, and this is precisely the nuclear division which effects reduction. In addition, the introduction of identical ids, which must take place in the case of inbreeding at every amphimixis, does not occur, and, what is certainly of great importance, all these cases are old species, living under natural conditions--the same conditions under which they lived as amphigonous species, and not newly formed breeds under artificial conditions, as has probably always been the case in the experiments in inbreeding. It is true that even in old species, living in a state of nature, unfavourable variations may arise in the germ-plasm, and may go on increasing during purely parthenogenetic multiplication, for the ids with unfavourably varying determinants will no longer be set aside by means of reducing division. But those individuals in which the unfavourable variational tendency increases until it has attained selection-value will be subject to selection and will be gradually eliminated; indeed, the weeding out of the inferior individuals will be more drastic here than where amphigony obtains, because in this case all the offspring of one mother are nearly alike, so that the whole progeny is exterminated if the mother varies unfavourably.