182 CHARLES R. STOCKARD AND GEORGE N. PAPANICOLAOU 



dently on the developing individuals. The selection in our case 

 is a continuous selection of individuals, eliminating, no doubt, 

 certain of the least resistant germ cells, but continuing to act on 

 the embryonic population to eliminate the most defective of 

 these during very early developmental stages, and so on until 

 the individuals born are a mixture of strong specimen and others 

 only sufficiently strong to have reached birth and possibly to 

 survive in a subnormal fashion for a shorter or longer time. This 

 continuous, both germinal and individual, selection seems to us 

 more to be expected than the abruptly broken germinal selec- 

 tion advocated by Pearl, which completely eliminates all weak 

 germs, and therefore no weak individuals begin development. 

 We must admit that the data from Pearl's double alcoholic 

 matings considered alone strongly suggest only a germinal selec- 

 tion, but the results from our double alcoholic matings, while 

 leaning in the same direction, still show a greater late prenatal 

 and postnatal mortality than do the control matings, and in 

 addition present much evidence to suggest a very high early 

 embryonic elimination. This same early embryonic elimination 

 may be included among the high percentage of infertile eggs re- 

 sulting from the matings of two alcoholic fowls, and in the 

 case of the fowls it may be so much more severe that the later 

 mortality records compare favorably with the control. This 

 again would lead us to an abrupt break after the high very 

 early prenatal mortality and might be thought to vitiate our 

 entire supposition, yet the guinea-pig records show almost all 

 gradations up to the condition for the fowls. 



Our results show that in the alcoholic lines the higher the 

 early prenatal mortality and consequently the smaller the aver- 

 age-size litter, the lower the late prenatal and postnatal death 

 rate, much as Pearl also finds for fowls. These findings will be 

 still further discussed in connection with the sex ratio, table 6. 



