164 AMPHIMIXIS OR ESSENTIAL MEANING OF [XII. 



For a long time I waited in vain for the converse result, viz. 

 the appearance of light individuals of the variety A among 

 those of the dark sub-species j5, and I was coming to the 

 opinion that the latter was the original form of both varieties, 

 when, in the winter of 1890-91, a few typical individuals of ^ 

 were found in a colony of the sub-species B^ which had bred 

 true for many years. This colony had arisen from a single 

 dark individual which, in the course of seven years, had pro- 

 duced many hundreds of descendants all of the typical dark 

 variety. 



We might perhaps refer to the changing influence of external 

 circumstances as an explanation of these divergences from the 

 type, but any such interpretation is entirely excluded, because 

 both forms made their appearance side by side in the same 

 aquarium and under precisely the same external conditions. 

 These remarkable phenomena must certainly be ascribed to 

 internal causes, viz. to changes in the composition of the germ- 

 plasm. The required explanation is by no means difficult 

 when the subject is studied from the point of view afforded by 

 the theory of idants : in fact these observations seem to me 

 almost a proof of the validity of the opinion expressed above 

 that a 'reducing division' occurs in parthenogenetic develop- 

 ment, and that by its means a fresh combination of idants is 

 brought about. 



The fact that the variety A passes into B and conversely, 

 B into A, leads to the conclusion that both types originated at 

 a time when parthenogenesis was not the exclusive method of 

 reproduction : had this been the case, the ids a could not have 

 been included in the germ-plasm of animals of the type B, and 

 conversely the ids b could not have existed in the type A. The 

 explanation of the existence, side by side, of both kinds of ids, 

 is only to be found in sexual reproduction which must have 

 taken place at no very distant time. 



Let us assume the simplest possible relationship, viz. that 

 there are only four idants in the germ-plasm, of which three are 

 wholly composed of ids of the type A, and one of ids of the 

 t3''pe B. The four idants, aaab, of the primitive germ-cell 

 become doubled in the mother-germ-cell by longitudinal split- 

 ting, and give rise to the eight \^diWi'=,, aaa aa abb. Let us 

 further assume the most favourable case for reversion towards 



