460 WALTER W. MARSHALL AND HERMANN J.MULLER 



flies were homozygous for balloon but heterozygous for the 

 other factors and they were identified by the phenotypic appear- 

 ance of the balloon and streaked factors. The other combinations 



were discarded. Next, males of this type — were crossed 



oa 



to wild (normal) females in order to eliminate the S dj c. Half 

 of the Fs flies received from their father the chromosome with 

 S dj cha and half received the chromosome with ha. Both 

 types of course received from their mother the normal chromo- 

 some. The two sorts of flies could be distinguished by the 

 presence or absence of the character streak. Those flies con- 

 taining streak were discarded. The F3 males and females that 

 were normal in appearance were bred together and their resulting 

 offspring (F4) that showed phenotypic balloon were homozygous 

 for this character and contained no other mutant factors. 



Now that this balloon factor which had been in a state of 

 continued heterozygosis was freed, it remained to be compared 

 to the balloon in flies which had been kept homozygous for a con- 

 siderable length of time. The flies used for this comparison had 

 been kept homozygous for an even longer time than the others 

 had been kept heterozygous, the latter having in fact been 

 derived from this homozygous stock. 



The first problem that was encountered here was to arrange 

 the variations of the character in a consistent series. After a 

 large number of wings had been observed seven grades were 

 established which approximately accounted for all the varieties 

 that had appeared. The accompanying diagrams (fig. 2) will 

 serve to show the nature of these grades, which are based upon 

 the wing venation. The sorts of variation used as standards for 

 these classes were fairly typical, and although all intermediate 

 conditions were found there was seldom doubt in deciding to 

 which grade a particular wing belonged. The different grades 

 may now be considered in detail. 



1. Some few wings showed no modification whatever; these 

 were classed as normal. In all others the balloon character made 

 its appearance near the posterior margin of the wing by an 



