i 9 IQ] SHU LL INHERITANCE OF SEX IN LYCHNIS 123 



Aside from the remarkable difference in the value of the means, 

 the two curves are skewed in opposite directions, the strong negative 

 skewness and suggested bi modality of the sex curve, as contrasted 

 with the nearly normal character of the curve for color, being par- 

 ticularly noteworthy. The variability represented by these two 

 curves is properly compared by means of the standard deviations, 

 and not by the coefficient of variability, since we are dealing here with 

 groups of variates having similar theoretical limits of range, and 

 similar expected mean values. On this basis the sex ratios show 

 considerably greater variability than those of the color character. 

 This is especially interesting in view of the fact, pointed out elsewhere 

 (SiiULL 10), that crosses between heterozygous purple and white 

 Lychnis dioica may frequently result in other ratios than 1:1, owing 

 to the part taken by several distinct genes in the production of the 

 purple colors in this species, and that therefore the data for the color 

 curve may be somewhat heterogeneous, while the sex data are at 

 present assumed to be homogeneous. 



Summary 



Hermaphrodite mutants occur in pure-bred families of Lychnis 

 dioica L., the ratio of mutability being somewhat less than i : 1000. 



None of these hermaphrodites was diseased and several were 

 fully functional both as females and males, thus negativing the sug- 

 gestion of STRASBURGER that the hermaphrodites in this species may 

 always be due to the attacks of the smut, Ustilago violacea. 



In their hereditary relations these hermaphrodites were of two 

 kinds, some behaving like normal males, others having the capacity 

 of transmitting the hermaphrodite character to their male offspring. 

 This second class of hermaphrodites when self fertilized, or crossed 

 with normal females, gives progenies consisting of normal females and 

 hermaphrodites, and when fertilized by normal males gives normal 

 females and normal males. 



The hermaphrodite individuals are modified males, for in all 

 families in which these hermaphrodites were the pollen parents the 

 offspring showed the same ratios of females and hermaphrodites as 

 would have been expected of females and males had a normal male 

 been used as the pollen parent. 



