408 RADIATION BIOLOGY 



that are already supplied with "visible" alleles of certain chosen genes 

 and cross them with individuals having the dominant alleles of the same 

 genes, and then note how often offspring appear which show one of the 

 given recessive characters. Such offspring ordinarily represent muta- 

 tions of the dominant gene to one of its recessive alleles. These studies, 

 although not yet carried out on an adequate scale for spontaneous muta- 

 tions in Drosophila, nevertheless agree, as far as they go, with the esti- 

 mate, arrived at by the above over-all method, of one spontaneous muta- 

 tion per gene per generation being found among 100,000 to 200,000 germ 

 cells. This is an average rate, since there are indications that the (detect- 

 able) rate is not the same for different loci. In contrast with this, the 

 results obtained by the specific-locus method when radiation is applied 

 are thus far discrepant by a factor of some 2 times with those of the over-all 

 method. They have indicated that, on application of some 5000 r to 

 spermatozoa, about one gene mutation is on the average (again with 

 differences from locus to locus) induced per locus among some 3000 germ 

 cells, rather than among the 1 in 1250 to 2500 indicated by the other 

 method. 



There are various possible reasons for this discordance. Among these 

 are the considerable uncertainty regarding the number of genes, possible 

 differences in induced mutation frequencies between the different stocks 

 used, undoubted differences in mutation frequencies between different 

 loci, and major technical difficulties in the detection both of detrimental 

 and of visible mutations, as well as in the large-scale maintenance of 

 stocks of them. Probably as important a difficulty as any of these arises 

 from the fact that heavy irradiation of spermatozoa produces many 

 minute deletions and other structural chromosomal changes. Many of 

 these, in the case of lethals especially, are confused with the gene muta- 

 tions. In the work with visible mutations at specific loci, on the other 

 hand, far more of these cases were sifted out, both by breeding tests 

 and cytological observations, and thus excluded from the count of muta- 

 tions. This source of error is a much smaller one in the Drosophila 

 studies on mutations produced by X or 7 irradiation of ordinary inter- 

 phase stages and in the studies on either ultraviolet or spontaneous 

 mutations. In all these studies the great majority of the phenotypically 

 recognizable types of genetic changes, including the lethals, fulfill present 

 criteria for gene mutations. In such studies, however, the specific-locus 

 technique has as yet been little used, because of the large-scale operations 

 required for these applications of it. 



A source of error of a different kind, which is present when the fre- 

 quency of lethals arising in the X chromosome of ordinary interphase 

 nuclei in premature (gonial or primordial) germ cells of the male are being 

 studied, is that called germinal selection. This term refers to the fact that 

 some of the lethals, especially those involving deficiencies, kill or retard 



