1 78 THE MUTATION THEORY 



when crossed is decisive proof that they have not arisen by muta- 

 tions, the reproduction of which is alternative, for had they so 

 arisen the cross-breds would also display alternative reproduction. 

 As we see, artificial varieties, which, admittedly, have arisen by 

 mutations, display, not only alternative reproduction, but latent 

 characters in abundance. It does not follow, however, that all 

 characters which alternate in cross-breeding have arisen by 

 mutation. So far as has been observed, reproduction is alternative 

 only when characters are sharply contrasted ; and sharp contrasts 

 may arise as well by the piling up of fluctuations occurring during 

 many generations as by single mutations. It would seem, however, 

 that alternative reproduction is much more the rule when the 

 contrast has been reached by mutation than when it has resulted 

 from the accentuation of fluctuations. 



296. On the other hand, if it be contended that mutations 

 occurring in wild nature have been preserved, not because their 

 reproduction was alternative, but because the mutants were sterile 

 with the parent type but fertile inter se, then it must be shown 

 that mutations, not only useful but similar in kind, occur in such 

 numbers that physiological isolation l without extinction is possible. 

 The fact that human races are fertile when crossed is proof that 

 physiological isolation can have played no part in their differentia- 

 tion. Unless, then, heredity is different in plants and lower 

 animals from what it is in man, we must believe that the natural 

 varieties of the former were similarly inter-fertile at their origin, 

 but that, in many of the cases which have been examined, differen- 

 tiation is now physiologically greater than in the case of human 

 races. Doubtless most natural varieties are more ancient than 

 those of man ; for few species can have so quickly invaded diverse 

 environments, where selection differed, as the human being, who 

 is omnivorous, and therefore not dependent for nutriment on 



types are mongrels (in a real sense) between male and female. Again, the human 

 races of North- Western Europe, where travel and interbreeding have been so 

 common, and which in the past has been so often invaded by alien conquerors, 

 are mongrels between several races, as is shown by the fact that, unlike all other 

 animal races, they have variously coloured eyes. Yet, again, the albino rabbit 

 and mouse are mongrels between the white and the ancestral grey types. In 

 this connexion it matters not, of course, whether a variety is composed of many 

 individuals, or of only a single individual who has mutated from the parent 

 type. De Vries states that latency is of universal occurrence (Species and 

 Varieties, p. 222), but he gives no evidence. He mentions instances of retrogres- 

 sion in plants, and assumes (p. 631) without further proof that it implies, not total 

 loss, but latency. He assumes, in effect, that while the germ-plasm may gain 

 characters, it cannot lose them, see 188. 

 1 See 332. 



