J. B. S. Haldane 



49 



whilst for the other factors together the cross-over value is 22*7 ± 8 °/^. 

 The probable errors given are those due to random sampling only. The 

 divergence may probably be explained on the following hypothesis, which 

 is substantially that of Dexter (2). The factors B, C, D, etc. are incom- 

 pletely dominant over their absence, which is represented by A, and 

 are not allelomorphic to one another, but very strongly repelled. This 

 is undoubtedly true of B and /, since they may become coupled. We 

 may then more accurately represent the zygote A®.B6 as bcS . BcO 

 B(d . CO as Be® . bC0, and so on. Now it has been shown by Sturtevant(3) 

 and Muller(4) that in Drosophila melanogaster certain factors, when 

 present in the heterozygous condition, cause great increases in linkage, 

 while Gowen (5) has shown that numerous factors cause small increases. 

 If then the factors B, C, D, etc. have this effect in a slight degree, we 

 should expect the linkage to be greater where the zygote is triply 



