The abrupt transformation of species implies sudden change in the conditions of life, since a Medusa does not live like a Polype, nor a Trematode like its “nurse.” For this reason it is impossible that natural selection can be an aiding principle of “heterogeneous generation.” If such abrupt transformation takes place it must produce the new form instantly equipped for the struggle for existence, and adapted in all its organs and systems of organs to the special conditions of its new life. But would not this be “pure magic”? It is not thereby even taken into consideration that here--as in the cases of mimicry--time and place must agree. The requirements of a pre-established harmony (“_prästabilirte Harmonie_”) further demand that an animal fitted for special conditions of life should only make its appearance at that precise period of the earth’s history when these special conditions are all fulfilled, and so forth. But he who has learnt to perceive the numerous and fine relations which, in every species of animal, bring the details of structure into harmony with function, and who keeps in view the impelling power of these conditions, cannot possibly hold to the idea of a _per saltum_ development of animal forms. If development has taken place, it must have occurred gradually and by minute steps--in such a manner indeed that each modification had time to become equilibrated to the other parts, and in this way a succession of modifications gradually brought about the total transformation of the organism, and at the same time secured complete adaptation to new conditions of life. Not only abrupt modification however, but every transformation is to be rejected when based upon the interference of a metaphysical principle of development. Those to whom the arguments already advanced against such a principle appear insufficient may once more be asked, how and where should this principle properly interfere? I am of opinion that one effect can have but one sufficient cause; if this suffices to produce it, no second cause is required. The hand of a watch necessarily turns once round in a circle in a given time as soon as the spring which sets the mechanism in movement is wound up; in an unwound watch a skilful finger can perhaps give the same movement to the hand, but it is impossible that the latter can receive both from the operator and from the spring at _the same time, the same motion_ as that which it would receive through either of these two powers _alone_. In the same manner it appears to me that the variations which lead to transformation cannot be at the same time determined by physical and by metaphysical causes, but must depend upon either one or the other.