HYBRID VIGOR IN DROSOPHILA 487 



ments, the environment is sufficiently important to the yield of the particular 

 race to account for nearly as much of the total yield variation as race alone. 

 With widely ranging environments, given genotypes may show much more 

 variation in phenotypic expression. The interaction terms show that genes in 

 quantitative inheritance are not stable in their effects. In one condition the 

 phenotypic reaction, in some degree, could be such as to suggest recessive 

 action; in another dominant, in another additive or epistatic. 



These interpretations may be brought out by another analytical approach. 

 In analyzing data of this kind it has been customary to neglect the genotypic 

 environmental interactions. This neglect finds expression in the models 

 adopted to explain the yield. For the data above it is sometimes assumed that 

 yield, yuk, may be accounted for by a basic value common to all crosses, the 

 mean; deviations due to additive general combining ability of the different 

 races ga, gb, etc.; deviations due to specific combining ability, Sab, Sac, etc., 

 such as dominance variations, and epistatic effects common only to members 

 of that particular cross; reciprocal effects, Tab vs tba, etc., of any differences be- 

 tween members of reciprocal crosses; and a term representing residual varia- 

 tions, Bahk, Back ctc, duc to unknown causes. These variables are set up in the 



linear equation: 



yuk = ni + gi + gj-\- Sij-\- rij-\- Cijk 



Analysis of the data for the contributions of these variables to the yield 

 variajice gives these results: 



EXPERIMENT 35 

 General combining ability. . . . 11.3% 

 Specific combining ability. ... 9.7 



Reciprocal effects 2.3 



Residual effects 76. 6 



Two sets of experiments are available. One is for five and the other for six 

 inbred line hybrids. The test as presented above shows that 11.3 per cent of 

 the variance is due to differences in general and 9.7 per cent to differences in 

 specific combining ability. Differences in reciprocal crosses account for 2.3 

 per cent. Experiment 36 shows similar contributions attributable to general 

 and to specific combining ability, but the effect of reciprocal crosses is insig- 

 nificant. The two experiments are concordant in showing that general and spe- 

 cific combining ability account for most of the variation attributable to known 

 causes. In both cases general combining ability is somewhat more important 

 to productivity than specific combining ability. 



These results from Drosophila are entirely without any previous selection 

 for combining ability. They are comparable to the observations which were 

 obtained in corn when combining ability was tested for the early crosses of 

 inbred lines. It is significant that Sprague's analyses of such crosses show gen- 

 eral combining ability twice as important as specific. This difference is like 

 that of Drosophila but gives even more emphasis to general combining 

 ability. 



