374 CALVIN B. BRIDGES 



same point. This was a striking revelation of a relationship 

 vaguely realized before — that the modification by each of these 

 genes is proportionately as great in the male as in .the female, 

 although the actual modification is much greater in the case of 

 the female. 



Sex Iwiitation. The above relation does not hold in the case 

 of pinkish, for the connecting line cuts the base line at a point 

 only about three fourths as distant as the others. The dilution of 

 the pinkish male is proportionally greater than that of the pinkish 

 female. It is just in this respect that eosin itself differs from 

 the other eye colors. The combination of these two 'sex-limited' 

 characters, both of which differ in the same direction, increases 

 the sexual dimorphism of the flies to a striking extent. 



Specificity. The primary characteristic of all these modifiers 

 is that their visible effect upon eye color is dependent entirely 

 upon, or is greatly increased by, the coaction of eosin. Of the 

 eight modifiers, six (creams a, II, b, c, dark, and whiting) are 

 entirely dependent, giving no visible effect whatever in the 

 absence of eosin, while pinkish gives a very shght dilution and 

 cream III gives a dilution which is strong enough so that by 

 avoiding certain adverse conditions there is the possibility of 

 using it without the complication of eosin. In the case of pinkish 

 the dilution is so slight that only in about a third of the flies is 

 there any detectable difference. With both cream III and pink- 

 ish the ease of separation and the sharpness of the difference is 

 vastly greater in the presence of eosin. 



The term 'disproportionate modifier' might perhaps be better 

 for such cases as those of cream III and pinkish. Most of our 

 mutations are what may be called 'general modifiers,' since their 

 effects seem to be independent of one another, and the combined 

 effect is cumulative and roughly proportionate. 'General modi- 

 fiers' may be represented by the familiar parallelogram, in which 

 the initial condition (wild-type) is one corner, the effect of each 

 gene acting independently is represented by the length and di- 

 rection of the two adjacent sides, and the combined effect (double 

 mutant type) is the opposite corner (diagram 2, a). In the case 

 of the completely specific modifiers, such as cream II, the length 



