658 THE AMERICAN NATURALIST [VOL. XLIV 



Applying no correction at all, however, practically 90 

 per cent, of the offspring agree with the rule that they 

 are not darker than their darker parents. Nevertheless, 

 the hypothesis that intermediate skin color is sometimes 

 due to the imperfect dominance of the simplex deter- 

 miner must be admitted as plausible. 



(e) Intermediate X Blond.— The results of this mat- 

 ing are given in Table V, showing the distribution of 

 skin color in the offspring. If all intermediates were 

 simplex in skin pigmentation we should expect blonds 

 and intermediates in equal numbers. Also on the hy- 

 pothesis that the higher grade of pigmentation is epi- 

 static we should expect blonds and intermediates, but no 

 brunets. The actual distribution agrees nearly with ex- 

 pectation on either hypothesis: where the term ' ' blond" 

 is used 54 per cent, of the offspring are blonds; and 

 among the "fairs" 56 per cent, are fair. Most of the 

 remainder are "intermediate," the exceptions consti- 

 tuting only about 6 per cent, of all offspring. In some 

 of the most aberrant families, like Dar. — 1, one finds in 

 the hair color reason for doubting if classification was 

 always made with judgment. It seems probable that 

 when the parentage really falls into this class brunet 

 offspring rarely, if ever, occur. 



(/) Intermediate X Brunet (Table VI).— If intermedi- 

 ate skin color is simplex then brunet is duplex and ex- 

 pectation is that half of the offspring shall be brunet, 

 half intermediate and there shall be no blonds. But if 

 intermediate and brunet represent two different stages 

 of pigmentation either of which may be epistatic to 

 blond, then a certain proportion (sometimes less than 25 

 per cent.) of the offspring should be blonds. Actually 

 there are many blonds (about 24 per cent, of all off- 

 spring) and consequently the second hypothesis is fa- 

 vored again. 



Of the 80 families 39 have no blond offspring; we may 

 inquire if the ancestry of such families differs in the 

 proportions of the blonds from those that produce blond 

 offspring. We find that in the families with blond off- 



