Inheritance of Quantitative Characters 85 



The selection proposed is preceded by an intelligent hybridiz- 

 ing, and after that genotypes rather than phenotypes are se- 

 lected; that is, the selection is made on the basis of germ plasm 

 rather than body plasm. 'J'his would be a sufficient reason for 

 the superiority of the new method of artiiicial selection as compared 

 with the old. A little further analysis will make the difference 

 clearer. 



In the old method of artificial selection, the breeder, in the 

 first place, is dealing with such germinal variations as happen to 

 appear in his crop; and, in the second place, he is dealing with 

 those fluctuations which appear as responses to the environment. 

 When he selects a large plant to use for seed, that plant may be 

 large on account of its germinal constitution; but, on the other 

 hand, it may be large because it is growing in a less crowded place 

 or a place more heavily fertilized than the others. In that case, 

 the large plant might not furnish good seed. The plant breeder 

 of the old method undoubtedly made such unfortunate selections 

 frequently; that is, he selected on the basis of external appearance, 

 and external appearance is very often a poor index of hereditary 

 capacity. Furthermore, he would not keep his lines pure, but 

 would deal constantly with an unmanageable mixture of good and 

 mediocre types. Intelligent selection is based on germinal con- 

 stitution only^by keeping careful pedigree records a selection of 

 genotypes is possible — and therefore its results are quicker and 

 surer. It is really a pedigree culture rather than a mass culture 

 method. 



Another phase of the subject should be considered. When a 

 plant breeder is trying to improve his crops by selection for quan- 

 titative characters, although he uses the old method of selection, 

 he is likely to be making some gain, as the experience of hundreds 

 of years has shown. The germinal constitution of his crop plants 

 is masked by fluctuations, of course, but this mask is not complete. 

 Most of the plants he selects are bound to possess high numbers 

 of factors of the right kind, and he probably rejects most of the 

 plants with few factors. In any event, he has generally succeeded 

 in the long run in getting a somewhat improved race. 



A summarized statement of this situation may be helpful. 

 Our recently developed knowledge of the inheritance of quanti- 



