1922] Setchell-Gaodspeed-Clausen: Nicotiana Tahacum 515 



another and from the parents than the original parents are from each 

 other. Moreover, our experiments show that as a result of simplifi- 

 cation of the factor differences the derivative strains crossed with each 

 other or with the parents give Fo progenies which often exhibit clear- 

 cut segregation in characters which showed intergrading series in the 

 original F, population. In other populations, however, from crosses 

 between derivatives, the populations still exhibit perplexing com- 

 plexities which make classification difficult and uncertain. In such 

 cases we could again resort to the method of establishing intermediate 

 derivatives from them ; but if the number of factors concerned in a 

 given character is even moderately large, as is certainly the case with 

 many of these quantitative characters, the number of genotypically 

 different derivatives which may be secured becomes so great as to make 

 the method impracticable. 



Our experience indicates that the successful factor analysis of these 

 quantitative character differences depends not only upon getting what 

 Castle (1919) has called the residual heredity equivalent throughout 

 the population, but also in establishing the proper kind of residuum 

 which will most emphasize the character differences associated with 

 the pair of factors or pairs of factors under investigation. The prob- 

 lem may be illustrated crudely by considering the pair of flower color 

 factors Rr. If the residuum should contain PP, the effect of which 

 is described below, segregation would give PPRR, PPRr, and PPrr. 

 In character expressions these three different genotypes would doubt- 

 less all be of various shades of dark red, difficult or impossible of ac- 

 curate separation. With such a residuum, therefore, it would be im- 

 possible to investigate satisfactorily inheritance in the factor pair Rr. 

 But if we should substitute pp for PP in the residuum, the segregation 

 products would be ppRR and ppRr, which would be pink, and pprr, 

 which would be red. Here the segregation would be sharp and dis- 

 tinct, and there would be practically no difficulty in classification. 

 How complex such interrelations can be has been shown most clearly 

 by Bridges (1919) in his account of specific modifiers of eosin in 

 DrosopJiila. As Bridges shows it would easily be possible to obtain 

 populations of Drosophila defying classification, but by keeping the 

 factors separate and studying their character effects with known 

 residual genotypes, it has been possible to determine and locate the 

 factors involved. Doubtless much of the extraordinary success of 

 Mendelian analysis in Drosophila has been due to the fact that factor 

 differences arose under conditions such that the residual genotype 

 gave no difficulty ; whereas in crop plants, the geneticist starts with 



