211 



THE AMEBIC AX XATURALIST 



[Vol. L 



it C) which reduces the frequency of separation, and 

 which is dominant, since it produces an effect in the hybrid. 

 I have found that two other races of flies, one having the 

 factor for spread wings (also in group III), and the other 

 showing no " visible" mutant factors, also contain C, as 

 they behave in the same way as ebony. However, the hy- 

 brids produced when these races are crossed with ebony 

 give high frequencies of recombination again ! This re- 

 sult shows that, as in group II, these races do not really 

 contain a factor which normally reduces separation fre- 

 quency, for, when both homologous groups of an individ- 

 ual contain the factor C— i. e., when it is homozygous— 

 the frequency of separation is high again. (This also 

 explains an irregularity observed by Dexter, who obtained 

 a high frequency of separation in a cross involving ebony 

 flies.) 



It happens, however, that these high separation fre- 

 quencies obtained when C is homozygous are even higher 

 than those occurring in crosses not involving C at all, and 

 so presumably homozygous for its allelomorph, c. By 

 analogy with Sturtevant's findings in the second group, 

 this would mean that in most crosses hitherto made not 

 involving C there has nevertheless been another factor 

 heterozygous, which has a similar, but lesser, effect on 

 the regions of the group studied. Some support for this 

 interpretation is found in the fact that occasionally higher 

 frequencies are obtained in these crosses not involving C, 

 which appear to overstep the limits of chance variation. 

 The evidence thus far secured on groups II and III 

 points to the conclusion that the highest frequency ob- 

 tained is that which should be regarded as the normal 

 value, and that very marked departures from this, which 

 affect only a particular group, are generally due to heter- 

 ozygosis in special factors of that group. If it should be 

 found that marked differences affecting the total fre- 

 quencies of particular groups do occur, in cases where 

 the flies are homozygous for whatever factors influencing 

 linkage they may contain, we might naturally expect that 



