THE PHENOMENA OF REVERSION 305 



species of which Zths is represented, although the individual 

 specimens still display considerable differences in form.'* The 

 result of these observations also agrees with our theory, for even 

 in the second generation of hybrids, eight was the highest pos- 

 sible number of ids of the ancestral species A in the mature 

 germ-cells of the hybrid, and in the process of fertilisation these 

 met with the same number of idants of the ancestral species B, 

 which was used for recrossing. The germ-mother-cells of the 

 f hybrid cannot contain more than 8A X 8B, and, as a rule, a 

 smaller number of the idants of A will be present. The reducing 

 division then again halves these sixteen idants in various ways ; 

 and in the case which is most favourable for the preservation of 

 the hereditary substance of A, the halving of one of these germ- 

 mother-cells, which was most favourable for A, might again lead 

 to the formation of germ-cells with eight idants of A and eight 

 of B. But in all other 'l hybrids the germ-mother-cells already 

 contain more idants of B than of A, as is shown in Fig. 22, III., 

 and the reduction therefore results in the germ-cells containing 

 either idants of B only, or, at any rate, some of B in addition 

 to those of A. If these egg-cells are then fertilised by germ- 

 cells of the pure ancestral form B, the largest conceivable 

 number of idants of A which can be contained in the fertilised 

 egg-cell will again be eight. Hence the form exactly inter- 

 mediate between the two ancestors may possibly also appear in 

 the third generation, but this will occur m^^c/i 7iwre rarely than 

 in the second generation. On the other hand, those individ- 

 uals the idioplasm of which contains only a few idants of A 

 will preponderate ; and in others, even idants of B only may 

 be present, for the idants of B were in the majority in most of 

 the mother-germ-cells of the second generation : the halving 

 due to the reduction must therefore have frequently resulted in 

 the distribution of idants of B only to one of the germ-cells, and 

 these then once more combined in the third generation with 

 idants of the pure species B. Theoretically, therefore, the 

 \ hybrids must in part revert to the ancestral form B. This 

 conclusion is supported by facts, inasmuch as Kolreutter and 

 Gartner, who made a large number of experiments on recross- 

 ing, found ' that three to six generations (four to five as a 

 rule) were required for the complete transformation of the 



* Focke, p. 485. 



