168 



HAROLD H. SMITH 



indicates that the other 12 possible combinations do not introduce any sig- 

 nificant non-additive effects. 



Another indication that epistatic effects were unimportant in the in- 

 heritance of plant height, leaf number, and leaf length was afforded by the 

 evidence that the means of the double crosses did not differ significantly from 

 each other (Table 10.1). 



The average variance of the double crosses was greater than that of the 

 parents or Fi's (Table 10.1), as would be expected from segregation. 



Partitioning Phenotypic Variance, Heritability, and 

 Number of Effective Factors 



Estimates of the magnitude of the non-heritable variation (0%), in popu- 

 lations involving Olson 68 and 34753 (1949 data), were obtained by taking 



TABLE 10.6 



ESTIMATES OF COMPONENTS OF \\\RIABILITY, NUMBER OF EFFECTIVE 

 FACTORS (Ai), HERITABILITY, AND GAIN FOR PLANT HEIGHT, LEAF 

 LENGTH, AND NUMBER OF NODES IN THE X. RUSTICA CROSS OLSON 

 68 X BRASILIA, STRAIN 34753* 



* 1949 data. 



an average of the total within plot variance of the non-segregating families — 

 Pi, P2, and Fi. As shown in Table 10.6, the values obtained were 25.76 for 

 plant height, 0.71 for leaf length, and 7.38 for number of nodes. 



The following symbols are used for the components of heritable variance 

 (total phenotypic minus environmental) : alb = variance depending on addi- 

 tive gene effects, a|) = variance depending on dominance. The heritable 

 variance of the F2 was calculated and equated to: 1/2o-g + l/4o-|). The 

 pooled heritable variance of the two first backcrosses was equated to 

 l/2crG + l/2o-z). Solving for 0%, the values obtained were 67.32 for plant 

 height, 1.04 for leaf length, and 8.16 for number of nodes. Values for 0%, 

 as calculated by substitution, were 113.20 for plant height, 0.22 for leaf 

 length, and 2.20 for number of nodes. 



In view of the influence on node number of a differential interaction of the 

 two parental genotypes with environment, an additional way of approach- 

 ing an analysis of the data on this character was tried. If a simple relation 



