No. 602] SHORTER ARTICLES AND DISCUSSIONS 111 



TABLE 145 



of tables 142 to 144, in which the second F, 

 to the grade of their first hooded grand- 



Classified according to the grade of the (first FJ grandparent, they 

 show a correlation between grade of grandparent and grade of grand- 

 child. The lower-grade grandparent has lower-grade hooded grand- 

 children, and the higher-grade grandparent has higher-grade hooded 

 grandchildren. This shows that the variation in grade is (in part at 

 least) genotypic. As the experiment yields no evidence that the varia- 

 tion in the hooded character is due to independent modifying factors, 

 there remains no alternative to the conclusion that the single genetic 

 Mendelian factor concerned fluctuates in genetic value. Fluctuation 

 accordingly is not exclusively phenotypic, as DeVries and Johannsen 

 have thought, but may be genetic also. Hence racial changes may be 

 effected through selection by the isolation of genetic fluctuations, as well 

 as by the isolation of mutations. Moreover, genetic fluctuation makes 

 possible progressive change in a particular direction, repeated selec- 

 tion attaining results which it would be quite hopeless to seek by any 



2. The progeny of "mu 

 Publication 241, p. 173.) 



Castle and Phillips described, under the name oi 

 the plus-selection series of extremely high grade 

 heterozygotes between the average condition of 

 at that time, about -f 3.75, and a new condition, 

 m our hooded races, but resembling that seen in " 

 black all over except for a white spot on the belly 

 on our grading scale as about In Intor 



animals homozygous for the darkor .•(.ii,liti<i!i ,ju 

 Irish rats). The homozygous "mutant" raco pn 

 in color pattern, varj'ing only from T)!.! to ^T,. w 

 Rials graded 5^2. Attempts to alter the modal 



crossed ivith wild rats. (From 



I (that of 



