which arise from the proliferation of the germ cells, then of course acquired variations cannot be transmitted, and evolution can proceed only on lines of inborn variations. If it is not true, if the germ cells are in part the products of the other co-existing cells, and receive from them "gemmules," and acquired variations are therefore transmissible, then besides the difficulties in the way of belief caused, as already indicated, by the incredible number of the assumed "gemmules," and the difficulties in the way of belief due to the impossibility of conceiving their arrangement in the germ cells, we are faced by the further difficulty of explaining when and by what method of evolution such a process arose. Nothing like it can possibly occur among the descendants of a pair of conjugated unicellular organisms; if it exists at all it must have arisen de novo among multicellular organisms of a high order, for among lower multicellular organisms there exists no vehicle for the conveyance of "gemmules" to distant cells.
The closer, therefore, the theory that acquired variations are transmissible is examined, the more incredible on à priori grounds does it appear. The upholders of it, however, allege that natural selection—i.e. the accumulation of inborn variations by natural selection—is quite inadequate to account for evolution in its totality. However this may be, and it is certainly disputable, the fact remains, that the more our knowledge of nature expands the better able are we to explain the whole process of evolution by the accumulation of inborn variations alone; moreover, those who maintain that many cases of evolution are explainable only on the assumption that acquired variations are transmissible, are compelled to admit that in many cases of evolution, more complex and far-reaching than any they instance to the contrary, the accumulation of acquired variations can have played no part. For instance, compare the
