134 



firms the sequence of the component 

 parts of bar-infrabar. It was tested, and 

 all the results from double-infrabar 

 reported below were obtained from 

 flies descended from it. 



Examination of the data in table 1 1 



STURTEVANT 



shows that the double form obtained 

 there must have had infrabar to the 

 left of bar; that is, it was infrabar-bar 

 instead of bar-infrabar. The tests made 

 with it appear in table 15. 



These results show, in fact, that the 



Table 15 

 fBJB 



fu 



9 Xffu $ 



sequence is infrabar-bar as supposed. 

 Except for this difference the mating 

 is the same as in table 12. It will be 

 observed that the two not-forked not- 

 fused single types in that table were 

 both infrabar, while the two obtained 

 here were bar; the one forked fused 

 there was bar, here it was infrabar. 

 There is a total of 13 single-type 



mutants in tables 12 to 15, all of them 

 agreeing in indicating that the tw^o ele- 

 ments of double types maintain not 

 only their individuality but their se- 

 quence. 



The double-infrabar obtained in ta- 

 ble 14 has been tested against round 

 (tables 16 and 17). 



The results in these tables show that 



Table 16 



fBm 



-r 9 xff^S 



I u 



double-infrabar was correctly iden- 

 tified, and that it behaves as was to be 

 expected, giving infrabar by both 

 kinds of crossing over,— just as double- 

 bar gives bar in both cases. 



Double-infrabar over infrabar has 

 also been tested, in the hope of obtain- 



ing triple-infrabar (table 18). 



No triple-infrabar was detected; but 

 its absence is not surprising, since the 

 corresponding round occurred only 

 once, and since it is not at all sure that 

 the triple form could be distinguished 

 from the double one. 



