498 



R. E. COMSTOCK AND H. F. ROBINSON 



a^ is a weighted average of all a-s, weighting being relative to the square of u 

 (one of the possible measures of the importance of loci), a- and a can exceed 

 unity only if one or more individual a's are larger than one, but values of a^ in 

 excess of one do not exclude the possibility of partial dominance at numerous 



TABLE 30.4 



GENETIC NATURE OF COMPONENTS 

 OF PROGENY VARIANCE* 



* Summation is in all cases over loci, i.e., 

 2;«2 = («?-F«i+ . ..ujf) and 

 Zaht^ = (af «f +4ul+ ... a%Uff) 



loci. On the other hand, a- will not be less than one unless dominance is less 

 than complete at one or more loci, but values less than one do not insure 

 absence of overdominance at all loci. 



Experiment I 



In accordance with the mean square expectations of Table 30.1 we can 

 estimate 



al^ as (Mn-Mii) / rn 



and 



0-2 as iMi2—Mu)/r 



and from Table 30.4 we see that in this experiment 

 and 



(r2 = 121*2 



in o 



1 m 16 



Hence, 



2[(«+l)Mi2-Mu-wMi3] . ^. ^ ,2a2«2 ., 

 IS an estimate of .„ ^ = a^ 



Mn —Mn 



Experiment II 



Note first from Table 30.4 that in this experiment a^ = o-/. If the experi- 

 ment is designed with m = n (this will be assumed in what follows since it is 

 a rational procedure where possible) this means that the expectation of Mn 



