laughnan: nature of mutations 15 



this hypothesis is valid, the frequency of nonrecombinant AT and sli, 

 colorless derivatives, should equal the frequency of nonrecombinant 

 T and Sh, alpha derivatives. 



Since a number of independent isolations of beta derivatives 

 were employed in these experiments, it will be helpful before pro- 

 ceeding to the results to consider the criteria that were employed in 

 establishing that an isolated beta was really dealt with. In all cases 

 prospective beta isolates were tested for pericarp phenotype and only 

 those which had lost the dominant brown effect and hence the alpha 

 component in the crossover event were used in the present experi- 

 ments. Incidentally, Ave have never obtained an alpha derivative from 

 a crossover isolate whose pericarp test indicated a loss of the brown 

 phenotype. A more serious problem is posed by the fact that a con- 

 siderable number of beta isolates (lacking the alpha component) are 

 new complexes of the ft : a type in which the a element from the 

 homologue has taken the place of alpha in the complex. Detailed 

 analyses of this phenomenon will be published elsewhere and it will 

 suffice here to point out that this substitution is expected if a is paired 

 to the right (see Figure 2A) of the crossover which takes place between 

 beta and alpha. The ft : a crossover derivative has a pericarp pheno- 

 type indistinguishable from that of ft, and since it is established inde- 

 pendently that homozygotes of the ft : a complex yield both crossover 

 and noncrossover a derivatives much as the ft : a complex yields 

 crossover and noncrossover alpha derivatives, the unwitting use of 

 ft : a in the present experiment would lead to the isolation of non- 

 crossover a derivatives which are not attributable to mutation of the 

 beta component. Consequently, we have adopted the criterion that 

 any beta isolate that has yielded crossover a derivatives among its 

 progeny is a ft : a complex and is not legitimately included in the data 

 to be presented. At the same time it is apparent that this runs the risk 

 of mistakenly including data from those ft : a complexes whose prog- 

 enies, speaking statistically, were too small to yield even a single 

 crossover a case, even though they might produce occasional noncross- 

 over a cases, which in that event would be mistakenly taken to repre- 

 sent mutation of beta to beta . 



One particular beta extraction (designated isolate #1) has had 

 extensive testing. The absence of the alpha component in isolate #1 

 has been independently established by an analysis of over 50,000 



