254 



not be demonstrated. As we have be- 

 come better acquainted with individual 

 genes and their functions, the assump- 

 tion that genes, as a rule, are individu- 

 ally essential to life has lost its plausi- 

 bility. 



Mutation to an intermediate allele is 

 sometimes considered evidence against 

 loss mutation. This involves another 

 assumption, that of the unitary nature 

 of the gene— an assumption made con- 

 sciously and with careful consideration 

 in the early development of gene 

 theory, but one that must be seriously 

 questioned in the light of later evi- 

 dence. It is only on the hypothesis that 

 multiple alleles are variant forms of a 

 single unit that we may exclude the 

 possibility of their occurrence by loss 

 mutation. On the hypothesis that they 

 represent different mutations in a com- 

 plex of closely linked genes, we could 

 account for mutation to different levels 

 by the loss of different segments of the 

 chain. 



The basis for the choice of the uni- 

 tary hypothesis is perhaps best shown 

 in the considerations underlying the 

 classical criterion, of allelism. These 

 were stated by Morgan in 1919 (23) 

 as follows: 



Probably the most important evidence 

 bearing on the nature of the genes is that 

 derived from multiple allelomorphs. Now 

 that proof has been furnished that the 

 phenomena connected with these cases 

 are not due to nests of closely linked 

 genes, we can probably appeal to these 

 as crucial cases. . . . The demonstration 

 that multiple allelomorphs are modifica- 

 tions of the same locus in the chromo- 

 some, rather than cases of closely linked 

 genes, can come only where their origin 

 is known. . . . 



1 • 



2« 



A 3» 



4« 



Fig. 108 [in part]. Diagram illustrating muta- 

 tion in a nest of genes so closely linked that 

 no crossing over takes place. 



Let the five circles of Fig. 108, A rep- 

 resent a nest of closely linked genes. If a 

 recessive mutation occurs in the first one 

 (line B, a) and another in the second gene 

 (line B, b), the two mutants a and b if 

 crossed should give the atavistic type, 

 since a brings in the normal allelomorph 

 (B) of b, and b that (A) oi a. . . . Now 

 this is exactly what does not take place 

 when members of an allelomorphic series 

 are crossed — they do not give the wild 

 type, but one of the other mutant types 

 or an intermediate character. Evidently 

 independent mutation in a nest of linked 

 normal genes will not explain the results 

 if the new genes arise directly each from 

 a different normal allelomorph. 



It will be noted that the test rules 

 out the existence of the nest of closely 

 linked genes only on the assumption 

 that each mutation must be an altera- 

 tion of a single number of the group. 

 If, instead, each mutation were a loss 

 of one or more contiguous numbers of 

 the group, the fact that crosses be- 

 tween them might commonly show 

 them to be allelic would not rule out 

 the "compound gene" as the basis of 

 the multiple allelic series. This is illus- 

 trated in the following diagrammatic 

 arrangement: 



The "compound gene" is in a sense 

 a contradiction in terms, for the hypo- 



