ARE RECESSIVES LOSSES OF GENES? 91 



It follows that in both types the genes not only keep 

 their identity but also their sequence. From the way in 

 which the two types, fBB'fu and fB'Bfu, were made up, 

 the sequence of the genes is known, and in all cases the 

 breaking apart of B and B' agrees with the sequence pre- 

 viously determined. 



(a) 



i_B_Bjfu_ BarlnfraBar f B Bar 



Norma.1 B'~fu~ Ilif"<'<=^Bar 



(b) 



f_B^B_fu. Infra Bar Bar i ^' Ir^fraBar 



Normal B f^ Bar 



Fig. 52. 



Diagram, in a, of a mutation of forked bar, infra-bar fused and 

 in b, a mutation of forked infra-bar, bar fused. 



These results furnish crucial evidence in favor of the 

 correctness of the theory that reversion in bar is due to 

 crossing-over. This is, at present, a unique case. There 

 would seem to be some peculiarity in the X-chromosome 

 at the bar locus that allows crossing-over between allelo- 

 morphic factors to occur. Sturtevant speaks of this as 

 unequal crossing-over.^ 



This result raises the question as to whether all muta- 

 tions may not be due to crossing-over. There is explicit 

 evidence in Drosophila that this is not the general expla- 



1 Several curious problems concerning the bar locus are involved in these 

 relations. For instance, when bar crosses over what is left in the bar locus? 

 Is it an absence of bar? Did the original bar arise by mutation in a wild 

 type gene, or was a new gene created? These questions are still under 

 investigation. 



