GENIC INTERACTION 181 



EBc d c a P at birth and as adults there is reciprocal reversal of order of effect, the order of 

 intensity being adult sepia (highest), newborn brown, newborn sepia, and adult 

 brown. The case of rrSt and Rstst, both with strong forehead rosettes, RSt with mere 

 irregularity on the forehead, and rrstst, smooth, {mm present in all) represents an example 

 in which reciprocal inhibition by two dominants causes reversal of direction of effect 

 in two pairs of loci. 



SELECTIVE VALUE 



Interaction effects of this fourth type are undoubtedly less common among simple 

 characters than are those of the second or third type. There is, however, a complex 

 character in which such interaction effect must be almost the rule. This is selective 

 value, which, in relation to the total environment of a given population, is the character 

 that is all important in the evolution of the latter, with the qualification that an array 

 of genotypes may be even more important than any one genotype. 



The reasons for the prevalence of this sort of interaction are partly indicated in the 

 following quotation from an earlier paper 1422 which is cited because a recent account of 

 the history of population genetics 860 attributes to it the assumption of absolute selective 

 values for each gene and thus of no interaction. The actual assumptions in this respect 

 (and in most others) were the opposite of those stated. 



"Selection, whether in mortality, mating or fecundity, applies to the organism 

 as a whole and thus to the effects of the entire genie system rather than to single genes. 

 A gene which is more favorable than its allelomorph in one combination may be less 

 favorable in another. Even in the case of cumulative effects, there is generally an 

 optimum grade of development of the character and a given gene will be favorably 

 selected in combinations below the optimum but selected against in combinations above 

 the optimum." 



The sort of case referred to last is illustrated in the upper part of figure 29 in which 

 it is assumed that a character varying quantitatively is determined by four pairs of 

 alleles with equivalent effects and no dominance. It is supposed that selective value 

 falls off from the optimum at the midgrade in proportion to the square of the deviation. 

 There are six optimal genotypes in this case, each distinguishable from each of the 

 others by at least two homozygous replacements. With more loci and multiple alleles 

 at each locus, the number of optimal genotypes in this sort of case may be enormous. 1423 



It is assumed, for simplicity, that the effects of the genes are additive on the under- 

 lying physiologic character. It is probable that, in any real case, the situation would 

 be complicated by specific interactions as in the case of coat and hair direction. The 

 case of coat color may, indeed, be cited as such an example. The cream agouti color of 

 the wild ancestor of the guinea pig (Cavia cutleri) is presumably the optimum in nature. 

 It is intermediate in intensity and can be simulated more or less in at least four 

 independent ways, none of which are similar to it genetically. 1427 



The quotation above does not refer to a point that was taken up later in the 1 93 1 



