The opposite view has, up to the present time, been maintained, and it has been assumed, as a matter of course, that acquired characters can be transmitted; furthermore, extremely complicated and artificial theories have been constructed in order to explain how it may be possible for changes produced by the action of external influences, in the course of a life-time, to be communicated to the germ and thus to become hereditary. But no single fact is known which really proves that acquired characters can be transmitted, for the ascertained facts which seem to point to the transmission of artificially produced diseases cannot be considered as a proof; and as long as such proof is wanting we have no right to make this supposition, unless compelled to do so by the impossibility of suggesting a mode in which the transformation of species can take place without its aid. (See Appendix IV, p. 310.) It is obvious that the unconscious conviction that we need the aid of acquired characters has hitherto securely maintained the assumed axiom of the transmission of such features. It was believed that we could not do without such an axiom in order to explain the transformation of species; and this was believed not only by those who hold that the direct action of external influences plays an important part in the process, but also by those who hold that the operation of natural selection is the main factor. Individual variability forms the most important foundation of the theory of natural selection: without it the latter could not exist, for this alone can furnish the minute differences by the accumulation of which new forms are said to arise in the course of generations. But how can such hereditary individual characters exist if the changes wrought by the action of external influences, during the life of an individual, cannot be transmitted? We are clearly compelled to find some other source of hereditary individual differences, or the theory of natural selection would collapse, as it certainly would if hereditary individual variations did not exist. If, on the other hand, acquired differences are transmitted, this would prove that there must be something wrong in the theory of the continuity of the germ-plasm, as above described, and in the non-transmission of acquired characters which results from this theory. But I believe that it is possible to suggest that the origin of hereditary individual characters takes place in a manner quite different from any which has been as yet brought forward. To explain this origin is the task which I am about to undertake in the following pages.