496 



R. E. COMSTOCK AND H. F. ROBINSON 



unit in this case is a randomized block arrangement {2n entries and r replica- 

 tions) of r plots of each of the progenies of a set. 



ANALYSIS OF DATA 



The appropriate variance analyses for the data of the three experiments 

 are outlined in Tables 30.1 to 30.3. The expected value (the value that 



TABLE 30.1 

 VARIANCE ANALYSIS (EXPERIMENT I) 



* a^ is the error variance among plots of the same progeny (due in part to 

 soil variation among plots in the same block and in part to variation among 

 plants of the same progeny). 



aj is progeny variance arising from genetic differences among female 

 parents. 



al, is progeny variance arising from genetic differences among male parents. 



TABLE 30.2 

 VARIANCE ANALYSIS (EXPERIMENT II) 



* cr^m is progeny variance arising from interaction of genotypes of male 

 and female parents. Other symbols are defined in Table 30.1. 



would be approached as a limit if the amount of data were made infinitely 

 large) is listed for each mean square to be used in interpretations. 



In order to specify the significance of components of the total variance of 

 which estimates can be used for inferences about dominance, some additional 

 symbolism must first be established. Consider the three genotypes possible at 

 a locus where there is segregation between two alleles. Let the difference in 

 effect of the two homozygous genotypes on a measured character be 2u and 

 the deviation of the effect of the heterozygous genotype from the mean effect 

 of the homozygous genotypes be au. Note that u and au have the same sig- 

 nificance as d and h, respectively, in the symbolism used by Fisher et al. 

 (1932). Also, they have the same significance as d and k in the symbolism 



