Nature oi" Heredity. 339 



union, and that later on. when new germ-cells are separated from 

 the soma, the descendants ^ot" the original male and female 

 ■chromosomes separate otit from one an<jther and form pure germ- 

 nuclei. 



The Mendelian cases, and also those mentioned under the 

 second heading, may be regarded as intermediate between cases of 

 ■complete sterility (due to total want of amalgamation between the 

 male and female pronuclei owing to excessive differences) and cases 

 ■of complete amphimixis or sexual union (due to successful co-opera- 

 tion of the male and female pronuclei). 



It ap2>ears then that in cases where the ancestral stimuli stored 

 lip in the male and female pronuclei exhibit too great differences, 

 it will be impossible for these pronuclei to unite in one segmentation 

 nucleus for the control of the developing organism. The sterility of 

 distinct species when crossed is thus probably due tO' the confusion 

 and disruption of the systems of forces in the pronuclei of the germ- 

 cells by antagonizing ancestral stimuli. 



Here also, it appears to me, is to be found the explanation of 

 the well-known fact that crossing frequently leads to reversion. When 

 two individuals are crossed together which are varietally or even 

 specifically distinct from one another, but not toO' distinct to be 

 fertile together, the offspring very frequently ' reverts " to some 

 ancestral condition, as, for example, in the classical instance of the 

 production of the ancestral blue rock pigeon by crossing of distinct 

 domesticated races. Here we may suppose that the very different 

 ■characters of the immediate ancestors of the mongrel (represented 

 in the germ-cells by stored stimuli), tend to cancel one another, whde 

 the ciosely similar or identical characters derived from more reivote 

 ancestors (also represented in the germ-cells by stored stimuli but 

 usually over-mastered by more recently acquired stimuli) ten 1 to 

 augment one another, with the result that ancestral characters 

 appear in the offspring while more recently acquired characters are 

 .absent. 



According to Weismann and his school amphimixis is to be 

 regarded as a process whereby variations are produced b\ the per- 

 mutation and combination of characters deri^•ed from two distinct 

 lines of ancestry. These variations form the material upon which 

 natural selection operates. 



According tO' our view sexual reproduction is by no means 

 necessan- for the production of variations, which are due primarily 

 to the influence of the environment. Indeed it seems more rea.sonable 

 to suppose that the result secured by sexual reproduction is that of 

 •checking excessive modification and preventing the organism from bje- 

 coming too specialised and stereotyped. If a species or variety 

 loses the power of adapting itself to its environment it is liable 

 to extermination when that environment changes. The power of 

 adaptation may be lost by over-specialisation in anv one direction, 

 and this over-specialisation is checked by amphimixis, which must 

 tend to maintain the average characters of a species in equilibrium 

 ^vith the average characters of the environment. 



