86 Outline of Genetics 



tative characters seems to justify artificial selection, but it does 

 not justify the old blind method of selection. It emphasizes the 

 need of intelligent selection, and shows how such selection can be 

 made. In order to do this, one must understand the mechanism 

 of inheritance involved, and must make his selection on the basis 

 of genotype rather than phenotype. All along the line, strains 

 must be discarded which, though recommended by the phenotype 

 of one of their ancestors, are to be condemned on the basis of their 

 breeding capacity. Selection is always to be made on the basis 

 of breeding capacity, that is, genotype. 



The situations just considered enable one to understand two 

 phenomena which have been baffling scientific plant breeders for 

 some years. The races of plants improved by artificial selection 

 have usually reverted to type when selection ceases. This fact 

 was recognized for a long time, but was first pointed out clearly 

 by De Vries (3) . Since then we have always expected this 

 result, that no improvement will maintain itself, but will run back 

 unless the selection is continuous. When a practical breeder 

 announces that he has developed by selection a new race which 

 continues to breed true without further selection, we are inclined 

 to disbelieve him, for we know that only elementary species breed 

 true. We explain that the practical breeder bases his selection 

 on fluctuations, and therefore his new race is bound to revert to 

 type. It is obvious now that there is a flaw in this argument. 

 The practical breeder may be basing his selection on fluctuations, 

 but at the same time he may be piling up cumulative factors in 

 the right direction. Thus he might eventually secure a race con- 

 taining all the cumulative factors. Such a race would be a homo- 

 zygote and could not help breeding true. Most of the claims of 

 artificially improved races that breed true may be false, but it 

 should be remembered that such a thing is possible, and may be 

 "stumbled upon accidentally," even with unscientific breeding. 



There is another phenomenon which has been much discussed, 

 and which can now be explained in the same way. This is the 

 so-caUed "fixation of hybrids." For years breeders have made 

 promiscuous crosses and then begun artificial selection with the 

 F2 generation. Eventually they have secured a pure-breeding 

 new type. It will be remembered that it was in this way that 



