ESTIMATION OF AVERAGE DOMINANCE OF GENES 



497 



employed earlier in the Heterosis Conference. The symbols u and an are 

 used here for consistency with usage in articles by Comstock and Robinson 

 (1948) and Comstock et al. (1949). Let the number of segregating gene pairs 

 that affect a particular character be symbolized as N, and a numerical sub- 

 script to w or a specify the locus to which the symbolized quantity is relevant. 

 Thus 2^3 is the difference in effect of the two homozygous genotypes of the 

 third locus and a^Uh is the dominance deviation for the fifth locus. 



Now granting validity of several assumptions (to be listed and discussed 

 later) a;,, aj, af„f, and a'lt have genetic meaning as set out in Table 30.4. 



TABLE 30.3 

 VARIANCE ANALYSIS (EXPERIMENT III) 



* at is progeny variance arising from genetic differences among F2 (male) 



parents 

 a? 



i,i is progeny variance arising from interaction of genotypes of F2 and in- 

 bred parents. 



THE ESTIMATE OF AVERAGE DOMINANCE 

 The magnitude of a measures the degree of dominance in the action of any 

 one pair of genes, being related to qualitative classification of dominance as 

 follows: 



However, a problem arises concerning the best way to represent the average 

 dominance for all loci with a single number. An obvious possibility is the 

 unweighted mean of a's for all gene pairs. On the other hand, it can be argued 

 that a mean in which individual a's are weighted relative to the importance 

 of loci would be more useful. This in turn raises the question of how the rela- 

 tive importance of loci should be measured. However, the matter will not be 

 pursued further, since the experiments under consideration offer no choice of 

 measure to be estimated. 



The estimate that can be made is of 



a- = 





2^2 



or 





