336 THE GERM-PL.'^SM 



or races, reversion is sometimes more and sometimes less marked 

 — in some cases restricted and in others very extensive. In 

 those cases, therefore, in which ritdimentapy and insignificant 

 characters only are concerned, such as stripes in mules, pelorism 

 in flowers, and rudimentary nipples in the liitman race, indi- 

 vidual variations must frequently occur in all degrees, from the 

 complete absence of reversion to the most marked form of it. 



Conversely, it is just as easy to explain whv reversion occurs 

 without exception in other cases, such as in tliat of Daiura- 

 hybrids, if we assume that entire groups of ids or idants, as well 

 as certain determinants in individual ids, have remained unmod- 

 ified. For if the former are present in such numbers as to pre- 

 ponderate in a hybrid over the two kinds of modified ids, which 

 counteract one another, they must be contained in nearlv equal 

 numbers in the germ-plasm of every individual. 



We can in principle understand the whole series of the 

 phenomena of reversion, if we represent the transformation of 

 the germ-plasm as being due to the majority of the ids in most 

 of the idants becoming modified first, while a minority of them 

 remain unmodified : in the course of generations the number of 

 these unmodified ids then gradually becomes smaller, owing to 

 the action of natural selection, until the ancestral species is only 

 represented by a number of scattered ids ; and finally, by con- 

 tinued selection, these ids also become modified in the same 

 direction, to such an extent that eventually only the determinants 

 of those individual characters escape modification which are less 

 important or quite meaningless for the life of the organism. We 

 know that complete reversion to the ancestral form may occur 

 in young species in the case of favourable crosses between 

 allied species (^Datura) ; and that in the further course of 

 pliylogenv, that is, in older species, total reversions can no 

 longer occur, although reversion to single characters, or even to 

 entire groups of characters, may still take place, and does so 

 with certainty under certain conditions — as is exemplified by the 

 reversion of certain hybrids between diflFerent races of pigeons. 

 In its last stage, then, reversion occurs as an entirely uncertain 

 and apparently capricious re-appearance of an individual an- 

 cestral character, such as, for instance, that of the occasional 

 striping in mules. 



By means of our theory much may also be deduced and ren- 

 dered comprehensible in principle concerning the external causes 



