460 THE GERM-PLASM 



Although reversion to ?nore rejiwte ancestors is also brought 

 about by the same factors, — viz., the 'reducing division' and 

 amphimixis, — it recjuires further elucidation. The theory of 

 selection requires that only a majority, and not all, the deter- 

 minants of a part which is to be modified shall undergo a 

 corresponding change. Old unmodified determinants of various 

 parts are therefore retained in the germ-plasm of a species, and 

 can only be removed from it very gradually by fortuitous 

 'reducing divisions.'' This renders reversion to the characters 

 of very remote ancestors possible; its occurrence, however, 

 depends upon the reducing division and amphimixis taking 

 place in a favourable manner. If the reduction causes similar 

 groups of ancestral determinants to be brought together in 

 several ids, and this germ-plasm, in the process of amphimixis, 

 unites with that of another germ-cell, which also contains 

 similar ancestral determinants in several ids, these may gain 

 the victory over the modern determinants in the struggle of 

 the ids during ontogeny. This, however, will chiefly depend 

 on the kind and strength of the modern determinants which are 

 opposed to them ; and thus reversion to ancestral characters 

 occurs very frequently in crosses between races (pigeons) and 

 species (mules), in which the modern determinants are heterody- 

 namous ; — they do not co-operate, and their forces counteract 

 one another, while the ancestral determinants are similar and 

 their forces cumulative. 



Numerous phenomena of reversion in plants and animals 

 may be explained in a very simple manner on these principles, 

 and from this point of view it is also possible to understand 

 that form of reversion which occurs in gemmation and partheno- 

 genesis. The more remote the ancestors to the characters of 

 which reversion occurs, the more rarely will it take place. 

 Reversion to the three-toed ancestors of the horse, for instance, 

 is of extremely rare occurrence, for it is due to a retention of 

 the ancestral determinants in question — which have certainly 

 disappeared from all the ids in the germ-plasm of most existing 

 horses — in single individuals of certain series of generations, 

 and to the chance of the coming together of two germ-cells 

 containing such ancestral determinants. 



The remarkable phenomenon of dir/iorphisni, which has been 

 introduced so extensively — more especially into the animal king- 

 dom — by means of sexual reproduction, must be due to the 



