But I shall not enter more minutely into the question whether epigenesis or evolution is the right foundation of the theory of development, but shall content myself with having shown, first, that it is illusory to imagine that epigenesis admits of a simpler structure of the germ, (the precise opposite is true,) and secondly, that there are phenomena that can be understood only by an evolution-theory. Such a phenomenon is {60} the _guidance of variation by utility_, which we have considered to-day. For without primary constituents of the germ, whether they are called as I call them, determinants, or something else, _germinal selection_, or guidance of variation by personal selection, is impossible; for where all units are alike there can be no struggle, no preference of the best. And yet such a guidance of variation exists and demands its explanation, and the early assumptions of a "definitely directed variation" such as Naegeli and Askenasy made are insufficient, for the reason that they posit only _internal_ forces as the foundations thereof, and because, as I have attempted to show, the harmony of the direction of variation with the requirements of the conditions of life subsists and represents the riddle to be solved. _The degree of adaptiveness which a part possesses itself evokes the direction of variation of that part._ This proposition seems to me to round off the whole theory of selection and to give to it that degree of inner perfection and completeness which is necessary to protect it against the many doubts which have gathered around it on all sides like so many lowering thunder-clouds. The moment variation is determined substantially though not exclusively by the adaptiveness itself, all these doubts fall to the ground, with _one_ exception, that of the utility of the initial steps. But just this objection is the least weighty. Without doubt the theory requires that the initial steps of a variation should also have selective value; otherwise personal selection and hence germinal selection could not set in. Since, however, as I have before pointed out, _in no case can we pretend to a judgment regarding the selective value of a modification, or have any_ {61} _experience thereof_, therefore the assumption that in a given case where a character is transformed the original initial steps of the variation did have selective value, is not only as probable as the opposed assumption that they had none, but is _infinitely more probable_, for with this we can give an intelligible explanation of the mysterious fact of adaptation, while with that we cannot. Consequently, unless we are resolved to give up all attempts whatsoever at explanation, we are forced to the assumption that the initial steps of all actually affected adaptations possessed selective value.