No. 592] 



SIIOBTER ARTICLES AND DISCUSSION 



249 



more ready to concede the existence of genotypic variation in 

 this character than Pearl has shown himself to be. And I have 

 been reluctant to accept at its face value Pearl's statement that 

 at the conclusion of his fecundity selection experiments he had 

 more good winter layers than at the beginning, but none better. 

 For in our selection experiments with rats it is very clear that 

 when high-grade individuals grow common, a few individuals of 

 higher grade are sure to put in an appearance. Genotypic varia- 

 tion seems to me to be of such wide occurrence that it is difficult 

 to believe that it is ever wholly absent, that absolutely pure lines 

 really exist. I quite agree with Pearl's conclusion that somatic 

 character is not a sure index of genetic constitution and that it 

 was therefore entirely logical and necessary for him to make 

 progeny tests in order to classify his pullets genetically. To 

 establish the point it is not necessary for him, as he observes, "to 

 be fussily nasty ' ' by citing page after page from my Mendelian 

 writings. I had granted the point years before it was raised. 



This brings us again to what Pearl considers "the most serious 

 phase of Castle 's attack, namely that in which he denies the va- 

 lidity of my conclusions respecting the inheritance of the char- 

 acter fecundity in fowls." Let it be made very clear at the out- 

 set what is attacked. Not the idea that fecundity is inherited. 

 I think that I am even more ready than Pearl to admit that 

 fecundity is a quantitatively variable character and that its va- 

 rious quantitative conditions are inherited. This is merely to 

 state in another way that genotypic as well as phenotypic varia- 

 tions in fecundity occur. If they occur, it is possible to isolate 

 them and thus to produce families characterized by them. The 

 conclusion which I "attack" is this, that the observed variations 

 in fecundity depend upon two and only two differential factors, 

 both of which are Mendelian, one sex-linked and the other not sex- 

 linked. Several possibilities are conceivable, which this conclu- 

 sion does not include, as for example that more than two genetic 

 factors are concerned in the variation, that one or other or both . 

 of the supposed factors are quantitatively variable and so capa- 

 ble of gradual change under selection. I am not advocating or 

 defending any of these possibilities. I am merely attacking the 

 conclusion outlined substantially as I understand Pearl to hold 

 it. There are really several distinct points in this conclusion, 

 some of which seem to be better grounded than others. If I were 

 asked either to accept or to reject it as a whole (and Pearl's pub- 



