XII.] CONJUGATION AND SEXUAL REPRODUCTION 1. 5 7 



of the cessation, for all future time, of the mingling of the 

 idants of the ovum with those of the spermatozoon, it would 

 by no means follow that the offspring of a female would 

 necessarily become ' identical twins.' With tAventy different 

 idants, if there are not the 377 miUion different combinations 

 which calculation indicates, there would be nevertheless such 

 a vast number of different combinations of idants, that two ova 

 produced by the same mother could only rarely be identical. 

 Among all the possible combinations, that very one might arise 

 which existed in the original egg-cell of the mother herself 

 and became expressed in her somatic cells. Such a combination 

 would contain one idant of every kind, and such an ovum would 

 give rise to an individual ' identical ' with the mother, that is, to 

 one similar to the mother in all respects except as regards 

 those modifications of the inherited developmental tendencies, 

 which are called forth by external circumstances. 



We need not consider the unlikely suggestion, that all com- 

 binations are equally probable ; if only it be conceded that 

 any degree of difference is possible for the combinations of 

 the germ-plasm, remarkable consequences follow. In the 

 first place it appears that, in persistent pure parthenogenesis, 

 the mmiber of different idants contained in the idioplasm must 

 steadily diminish, although perhaps at a very slow rate. If the 

 number did not diminish new combinations could never arise, 

 and that of the first parthenogenetic mother {A) would be re- 

 tained indefinitely, — thus if there were twenty different idants 



{a, b, q d, e t) the whole series would persist unchanged. 



If, however, another combination arose in the daughter (5), for 



example a ah c d e /, this would be brought about by one 



of the idants {a in this instance) becoming double, and then 

 inasmuch as the total number of idants must remain the same, 

 it follows that one of the others must be absent (for example /), 

 or the number would be twenty-one instead of twenty. As a 

 result of this the idant / would be wanting in all the descendants 

 of B. If now we suppose that such a new combination, arising in 

 this way by the omission of one idant and the reduplication of 

 another, would not be formed in each generation, but only in 

 every tenth, it follows that at the end of each series of ten 

 generations, a fresh combination will arise by another omission 

 and another reduplication, and so on, so tliat after a hundred 



