No. 631] FACTORS OF HEREDITY 117 



neously. (It is for this reason that Sturtevant was first 

 able to hit upon the general fact of linear linkage, on the 

 basis of numerous careful experiments involving only 

 two factors at a time.) 



7. Owing to the inherent inconsistencies of the methods 

 that were used to construct the solid models, it is to be 

 expected that any predictions regarding separation fre- 

 quencies which are deduced from them would be ex- 

 tremely unsafe. Castle states, however, that if any newly 

 discovered gene has been located in the model, by obtain- 

 ing its frequencies of separation from any three of the 

 other genes contained therein, then the relation of the 

 new gene "to all the others could be predicted by direct 

 measurement from the model." In the case of two of 

 the four predictions which Castle has made in this way, 

 some evidence concerning the distance between the loci 

 involved is already in existence. 



One of the frequencies of separation in question is that 

 between the loci of the recessive mutant factors glazed 

 eye and rugose eye, in Drosophila virilis. Castle pre- 

 dicts, on the basis of his model, that the per cent, of 

 crossing over between them should be found to be 4 or 5, 

 or "probably a little greater." The work of Metz, and 

 unpublished work of Weinstein, have shown, however, 

 that hybrid females which carry both mutant factors 

 exhibit the somatic character sterility possessed by the 

 more extreme mutant type. When this dominance of an 

 ordinarily recessive character in F x is taken together 

 with the close similarity between the unusual effects pro- 

 duced by the two mutant factors (both produce a similar, 

 peculiar effect on the eye, which is sex-limited, being 

 more marked in the males), and with the fact that there 

 is also a third mutant member of the series, with similar 

 peculiar effects and similar linkage relations, it becomes 

 highly probable that these factors are all allelomorphs. 

 In that case they occupy "identical loci," and the fre- 

 quency of separation between them must be 0. A direct 

 determination of the per cent, of crossing over between 



