104 



THE AMERICAN NATURALIST 



[Vol. LI 



Dowell emphasizes the high variability of these early generations, 

 few in individuals, and attaches importance to the relatively 

 smaller variability of later generations. It seems to me fairer to 

 compare the first half of the series with the second half. Con- 

 cerning the point I have said (Publication 241, p. 172) : 



The amount of the variation as measured by the standard deviation 

 is less in the last half of the experiment than in the first half. It is 

 also steadier, owing in part doubtless to the fact that the numbers are 

 larger, and in part to a more stable genetic character of the selected 

 races. But the genetic variability is plainly still large enough to per- 

 mit further racial modification and there is no indication that it will 

 cease until the hooded character has been completely selected out of 

 existence, producing at one extreme of the series all-black rats, and at 

 the other end of the series black-eyed white rats. 



It should be noted that these conditions have already been 

 approximated in individual cases. 



The New Evidence 



1. The progeny of plus selected crossed with wild rats. 

 (Quoted without change from Publication 241, pp. 163-168.) 



In 1914 Castle and Phillips published a report on breeding experi- 

 ments with hooded rats, in which it was shown that the hooded color- 

 pattern— itself a Mendelian recessive character in crosses with the 

 entirely colored (or "self") coat of wild rats— is subject to quantitative 

 variation, and that different quantitative conditions of the hooded 

 pattern are heritable. (Compare fig. 36, plate 7.) It was also shown 

 that by repeated selection of the more extreme variations in the hooded 

 pattern (either plus or minus) it is possible gradually to modify the 

 racial mean, mode and range as regards these fluctuations, without 

 eliminating further fluctuation or greatly reducing its amount. We 

 concluded that the unit-character, hooded color-pattern, is a quanti- 

 tatively varying one, but were at that time unable to decide whether the 

 observed variability was due simply and exclusively to variation in a 

 single Mendelian unit-factor or partly to independent and subsidiary 

 modifying Mendelian factors. 



Since publication of the above I have been engaged in further experi- 

 ments designed to show which of the alternative explanations is the 

 correct one, and these are now sufficiently advanced to indicate definite 

 conclusions! Previous experiments had shown that when a race of 

 hooded rats, whose character has been modified by selection (either 

 plus or minus), is crossed with wild rats, the extracted hooded animals 



