52i ; 



THE AMERICAN NATURALIST [Vol. LII 



in the opinion of the writer, is the mutation theory as out- 

 lined by DeVries in so far as it may account for progres- 

 sive evolution. 



Inasmuch as it seems probable that the results obtained 

 by Castle are to be explained upon the same basis as 

 those of DeVries, it will be well to consider them in this 

 connection. Here it is assumed that a continuously 

 variable heritable gene is involved, and that progressive 

 results are obtained through the selection of the "unit 

 characters" produced by such a gene. Castle, however, 

 stands almost alone in vigorous support of such a varia- 

 tion, while opposed to him are the Hagedoorns, Morgan, 

 Pearl, Punnett, McDowell, Muller and others equally in- 

 istent that genes once having originated pass on from 

 one generation to another unchanged except in compara- 

 tively rare instances where so-called "mutations" occur. 1 

 It is maintained by those advocating this view that the 

 results in connection with hooded rats on which Castle 

 bases his contentions, are due to an uncertain number of 

 modifying genes not in themselves variable, and that the 

 existence of such genes has been demonstrated in other 

 organisms presenting results similar to those obtained 

 in rats. The work of Little (1917) with mice where 

 three segregating types of spotting were found to pro- 

 duce varying degrees of color pattern, indicates that 

 multiple genes are involved. Furthermore, the analysis 

 by Little of the data obtained by Castle, Phillips and 

 Wright, points decidedly to the interpretation of their 



