56 Heredity and Eugenics 



dark or extremely light individuals, until a pure race was 

 obtained. P^urther, if the black character should become 

 attached to additional material bodies in the cell (chromo- 

 somes or the like), so that it would be represented by addi- 

 tional units, then the occurrence of light-colored progeny 

 would become still rarer, and deeper intensities of black- 

 ness than before existed would now occur. Thus selection 

 would become a means for the modification of a character 

 really dependent upon the inheritance of unchanging units. 

 Xow this is perhaps what occurs when one seeks to modify 

 size by selection. 



There are strong reasons for believing that mendelizing 

 characters can be modified by selection, though this idea 

 is vigorously denied by many Mendelians, as for example 

 by Johannsen. In Johannsen's \iew, selection can do 

 nothing but sort out variations already existing in a race. 

 I prefer to think with Danvin that selection can do more 

 than this, that it can heap up quantitative variations until 

 they reach a sum total otherwise unattainable, and that 

 it thus becomes creative. I will describe briefly certain 

 exi)eriences of my own which support this idea. 



In several cases I have observed characters at first feebly 

 manifested gradually improve under selection until they 

 became established racial traits. Thus in guinea-pigs, the 

 hind-foot commonly bears three toes (Fig. 30, .4). But 

 several years ago I observed an individual which had an 

 imperfectly developed fourth toe, similar to that shown in 

 Fig. 30, C. From the descendants of this animal, obtained 

 by inbreeding and selection, was formed a race having 

 well-developed fourth toes (Fig. 30, B) on both hind 

 feet. The extra toe made its apj^earance poorly de\'el- 

 oped on the left foot only. About 6 per cent of the 



