THE HIOMETRICAL ANALYSIS 



When we omit the summed variances of backcrosses and the 

 mean variance of Fs's from the Antirrhinum results, we find: — 



D = 20-369 

 H= 8-209 

 El = 3 -426 



Eg = 0-887, 



and again expected vakics can be calculated for the various statistics 

 as shown in column five of the Table. 



The analysis is now much more satisfactory (Fig. 18). In the first 

 place a positive value has been obtained for H, though it is too small 

 to afford any reliable indication of dominance. Secondly, the 

 discrepancies between observation and expectation have been very 

 much reduced in Table 7. This is true, not merely for the summed 

 variances of backcrosses and mean variance of Fg's, for which linkage 

 allows us to assume a perfect fit, but also for the other statistics, 

 notably E^ and the variance of F2. The results indicate that at least 

 some of the polygenes are not rccombining freely. Furthermore, 

 the mean variance of F3 is larger than would be expected from 

 the F2, so that the linkage must be preponderantly in the repulsion 



phase, i.e. of the kind — • Large, therefore, as was the phenotypic 



difference in height between the species, it did not represent the fuU 

 difference between the genotypes, for some of the genes must have 

 been counter-balancing one another's effects in these repulsion 

 linkages. 



Randomly Breeding Populations and Correlations between Relatives 



The types of family which were used in this experiment on shoot 

 length are not, of course, exhaustive. In particular, we can raise 

 families of the third generation by intercrossing pairs of individuals 

 taken at random from Fg. These biparental families will, in fact, be 

 the only type available in the third generation of experiments 

 with most animals, where true Fg's are obviously impossible to 

 obtain. Tlie same range of statistics is available from biparentals 

 as from Fg's, viz. variance of biparental means, mean variance of 

 biparentals, and covariance of F2 parents and their biparental 



