No. 573] SHORTER ARTICLES AND CORRESPONDENCE 571 



individual in a thousand will have the most extreme dark or 

 light grade of hoodedness possible. However, by selecting the 

 more extreme individuals, and mating them together, a still more 

 extreme grade of hoodedness may be obtained in F ; , (both as to 

 average and limiting values), and the same process may be con- 

 tinued for a good many generations. The number of generations 

 during which effective selection is possible depends on the num- 

 ber of factors concerned, the rigor of selection, and the amount 

 of inbreeding of brother to sister. 



In regard to the latter point, since brother and sister are much 

 more apt to be alike in their genetic constitution than are other 

 individuals, offspring from such a mating are more apt to be 

 homozygous and alike, or. we may say. such offspring will tend to 

 be homozygous and alike in a larger number of factors; then, 

 mating two individuals homozygous for these factors together, 

 there will be much less variation and so less opportunity to con- 

 tinue selection among their progeny. In the case of Castle and 

 Phillips's experiments, however, no such attempt at inbreeding 

 was reported. Here, then, the individuals mated together would 

 be more apt to differ genetically, even though they looked alike 

 (thus, one might he AA bb, the other aA bB), and their 

 descendants would therefore present a larger number of different 

 combinations of factors for the selector. Often a greater effect 

 may he eventually produced in this manner than by inbreeding, 

 for a larger number of combinations of factors are thus pro- 

 duced, some of which may he of more extreme type. The effect 

 would usually be slower, however, since such matings tend to 

 keep the strain heterozygous and are often steps backwards. 

 Cross-breeding, then, will help to explain the relatively slow but 

 long-continued and eventually large effect of selection in Castle 

 and Phillips's experiments, although such a result could also be 

 obtained without cross-breeding if the factors were numerous 

 enough. 



The " return selections" also are easily explicable on the 

 multiple factor view. Due to the original difference in so many 

 factors, and the fact that crossdt reeding diminishes the tendency 

 to homozygosis which selection favors, the rats were presumably 

 heterozygous even after generations of selection. They would 

 not be as heterozygous as before, of course, and, correspondingly, 

 Castle and Phillips did find less variation in the rats after selec- 

 tion. Yet there would still be a good chance for recombination, 



