Negative Results of Selection i8i 



have always yielded negative results. There have been, to 

 be sure, both traditions and systematic experiments among 

 breeders and horticulturists that seem to justify the assump- 

 tion that consistent selection will result in improvements. 

 The success of artificial selection was in fact the foundation 

 of Darwin's theory. 



Some thirty years ago an experiment was started at the 

 University of Illinois designed to modify the proportion of 

 protein in corn grain by selection. Starting with a grain in 

 which the average amount of protein was about 1 1 per cent, 

 they produced two extreme strains so that in the course of 

 twenty years of selection, one had i4/4 per cent protein and 

 the other just half as much. At the present time, however, 

 the results of this experiment would be interpreted as indi- 

 cating something other than mere selection. It is now ap- 

 parent that while the experimenters were not aware of doing 

 more than segregating extreme phenotypes, they were actu- 

 ally segregating genotypes, just as Johannsen had done with 

 his beans. 



The selection of variants generation after generation 

 does not modify the hereditary capacity of strains. To this 

 extent, therefore, our experimental evidences fail to sup- 

 port the theory that new varieties (and a fortiori, new spe- 

 cies) arise by consistent selection for a particular character. 

 Indeed, if we had no further evidence than experiments of 

 the kind mentioned, we might be strengthened in our be- 

 lief that species are actually fixed entities, however much 

 intergrading there may be between one species and another. 

 We could accept these intergrades as merely phenotypical 

 overlappings of one species and another with the certainty 

 that each individual, however much he may resemble an 

 individual of another species, belongs truly and permanently 

 to its own kind. But the kind in this case does not corre- 

 spond to what we commonly think of as a species, but to a 

 hereditary strain which manifests the character in question. 

 The ordinary species may have in its composition an in- 

 definite number of genotypes. 



