THE PHENOMENA OF REVERSION 337 



of reversion. It can easily be understood that the crossing of 

 different species and races is liable to lead to a preponderance of 

 those ancestral idants, ids. or determinants which are common 

 to the two parents. We miglit. however, also infer from the 

 theory that certain of the offspring of hybrids are verv liable to 

 revert in the next generation to one of the ancestral species ; 

 for the germ-cells of the hybrids are very dissimilar as regards 

 the germ-plasm they contain, in consequence of the reducing 

 division which every germ-cell undergoes at its formation : most 

 of them will contain idants of both parents in every conceivable 

 proportion, while in others, idants of only one of the parents 

 will be present. The • reducing division ' is therefore one of the 

 most efficacious of the primary causes of reversion, for it renders 

 possible the uneven distribution of the different qualities of the 

 idants which were contained in the germ-plasm of the germ-cells 

 of this parent. This principle applies equally to the cases in 

 which not a single entire idant remains untransformed in the 

 germ-plasm, which then only contains a minority of unmodified 

 ids or even merely of determinants, scattered amongst several 

 idants. 



Reversion in any degree therefore depends, in the first instance, 

 on the process of amphimixis, for without sexual reproduction the 

 reducing division would not have been introduced into the organic 

 world, and the second extremely important factor in reversion 

 — viz., the crossing of different germ-cells — would also not exist. 

 Reversion is, however, not of necessity connected with actual 

 amphimixis, but may, as I shall show later on, occur in con- 

 nection with parthenogenesis and gemmation ; although this is 

 only true in the case of those organisms in which amphimixis 

 formerly occurred, and the germ-plasm of which may therefore 

 contain ancestral ids or ancestral determinants. It is evident, 

 however, that the chance of reversion occurring must be mucli 

 greater when amphimixis takes part directly in tlie process of 

 reproduction ; for the relative proportion of modified and un- 

 modified units of the idioplasm of this order may be quickly 

 or suddenly altered in favour of the unmodified units, and thus 

 the ancestral units which were originally present mav undergo 

 accumulation, and predominate over the somewhat dissimilar 

 homologous units of the modified kind. 



Our comprehension of the problem just discussed is certainly 

 increased by the above considerations, and the following passage 



