EVIDENCE OF GERM CELL SELECTION 409 



original experiment was smaller than that used in the supple- 

 mentary test, it does not seem likely that there was a dearth of 

 sperm at any time during the experimental periods, but more 

 probable that some of the eggs were entered by sperm too badly 

 injured to develop a pronucleus or at least a viable conjugation 

 nucleus. An egg 'fertilized' by such a sperm would very likely 

 give no other reaction than an egg that had been reached by 

 no sperm whatever, and would be recorded as non-fertile. 



Pearl's statements as to the non-effect of alcohol on the trans- 

 mission of Mendelian characters are clearly meant to apply 

 only to the data he presents, and those data were derived from 

 experiments obviously neither intended nor adapted for the 

 solution of the problem attacked in this paper. Since he dealt 

 with crosses between pure-bred homozygous strains in which 

 all the Mendelian characters were in stable equilibrium, there 

 was no chance for selection except on the basis of such characters 

 as vitality and vigor. With reference to these characters, 

 Pearl found selection to be possible. In the experiments now 

 being reported the material afforded a chance for selection be- 

 tween other traits, in this case Mendelian, and the results are 

 believed to show that here also selection is possible. 



The question still remains as to the precise nature of the 

 selection that takes place. Cole and Davis ('14) have pro- 

 duced, evidence that with rabbits the sperm of one male may 

 have greater fertilizing capabilities than that of another, even 

 when conditions would seem to be more favorable for the latter. 

 They have also shown that the fertilizing power of the sperm 

 can be influenced by poisons administered through the male 

 soma. But in their work also, homozygous males seem to have 

 been used and no evidence is presented as to whether or not 

 two kinds of sperm produced by the same male could be dif- 

 ferentially affected. These and other results obtained by Cole 

 and his colaborators, Stockard's findings, and the conclusions 

 reached by Pearl, all tend to suggest that the effect of poison- 

 ous reagents is lethal rather than stimulating, if such is the 

 case, we may assume that a germ cell or nucleus bearing a 

 determiner for brachydactyly is more resistant to the effects of 

 alcohol treatment than one not bearing such a determiner. 



