"The origin of a variation is equally independent of selection and of amphimixis, and is due to the constant recurrence of slight inequalities of nutrition in the germ-plasm which affect every determinant in one way or another, and differ even in the same germ-plasm—not only in different individuals but also in different regions. These variations are at first infinitesimal, hut may accumulate; and, in fact, they must do so when the modified conditions of nutrition which gave rise to them have lasted for several generations. In this way deviations may occur in the structure of single determinants or of groups of them—never, perhaps, in all ids at once, but at any rate in several or even many of them. A doubling of certain determinants of the germ-plasm may originate in the same way. The process of amphimixis has an important share in the accumulation of these modified determinants, for it may raise the minority previously existing in the two parents to a majority by combining their halved germ-plasms. Then, and then only, does selection begin to take place."[1]
After all this it is certainly surprising that he should still cling to his former declaration that acquired characters are not transmissible. After abandoning all his premises he still adheres to his conclusion. Dr. J. G. Romanes, who has been one of his most liberal critics, after characterizing the latter part of the Germ-Plasm as "a right-about-face manœuvre," says that his first impulse "was to cancel all the criticisms which I had written of the Weismannian theory,"[2] and it really seems as though it were time to drop this prolonged discussion. Its further continuance must certainly be chargeable to his own course as pursued in Chapter XIII of his Germ-Plasm, and in his reply to Mr. Spencer in the face of these concessions. It is somewhat difficult to understand how he is able to reconcile these apparently conflicting views. That he does not limit the influence of external conditions to the germ-plasm proper, or fertilized germ cell, is apparent from his cheerful acceptance of Nägeli's "opinion that all variations are slowly prepared in the idioplasm in the course of generations before they become apparent," and we must suppose him to admit that it is the hereditary units themselves that are undergoing these transformations. In my address before the Biological Society I had referred to this in the following language:
"You will understand that I am speaking of variations which take place in the germ cells and sperm cells of parental organisms before they blend in the fertilized ovum. Most of Weismann's argument is directed to show that the fertilized ovum itself can not be affected by any transforming influence acting upon the
