60 



THE AMERICAN NATURALIST 



[Vol. LV 



viduals have then to be tested singly to determine what recessive 

 characters they carry and hence to what crossover category they 

 belong. 



Thus, by improvements (1) in the type of experiment planned, 

 (2) in the culture media and methods used, and (3) in the 

 method of calculation, disturbances in the ratios can usually be 

 held within negligible amounts. 



To these indirect methods of obtaining accurate values is to 

 be added one still more important — namely, the discovery of 

 new mutants in which viability is practically normal, and which 

 can be substituted for mutants less satisfactory in that regard. 

 Many of the loci are represented by several mutant allelo- 

 morphs, which often are different in viability as in other char- 

 acteristics. Thus, of the eight cut allelomorphs, or appearances, 

 eut-6 is distinctly the most nearly normal in viability. Likewise, 

 of the five or six allelomorphs of the truncate locus, "dumpty" 

 is the most satisfactory. 



In most of the more complex linkage problems, especially 

 those involving linkage-variations or coincidence, the behavior 

 of particular regions of a chromosome is being examined, and 

 the particular loci utilized are only indices of the behavior. 

 What is most essential, therefore, is that there be workable 

 mutant loci distributed rather evenly over all regions of the 

 chromosomes. As the number of mutants in a particular region 

 increases, there is a greater range of choice and greater prob- 

 ability that one or more of the mutants of that region will have 

 normal viability. Thus, bifid (at 7.3) was long the only work- 

 able mutant at a favorable distance from the left end of the 

 X-chromosomes. More recently ruby (at 7.5) because of its 

 better viability has displaced bifid from general use, and this in 

 spite of the fact that ruby interferes with the classification of 

 several other eye-colors (especially prune and garnet) while 

 bifid is workable with nearly all other mutants. 



In regions less well represented in numbers of mutant loci, a 

 mutant with excellent characteristics may be used rather than 

 one whose position is more favorable but whose other character- 

 istics are poorer. Thus, ''hunipty" is very favorably located in 

 the second chromosome in the middle of the long region from 

 curved (73.5) to plexus (98.5), but its viability is so poor that 

 in most experiments it is better to leave this section unfollowed 

 than to introduce humpty. There are at present few regions 



