124 BOTANICAL GAZETTE [FEBRUARY 



These results agree in a general way with those of Bryonia in that 

 they demonstrate the homozygous character of the females and the 

 heterozygous character of the males in Lychnis. 



Whether the females are positive or negative homozygotes cannot 

 be determined from the F I? as both assumptions can be made to fit 

 the facts by the aid of simple correlative hypotheses whose correctness 

 or incorrectness can be determined only by further breeding. 



A fundamental difficulty in either case is found in the fact that the 

 egg cells of the hermaphrodites are apparently of a single type, all 

 possessing the female-producing gene. The assumption that the 

 hermaphrodites are heterozygous leads us to expect equal numbers 

 of two different types of egg cells. Cytological studies may perhaps 

 explain this difficulty. The male germ cells of the same plants are 

 of two types, as required by theory. 



The occurrence of two hermaphrodite individuals in a progeny 

 produced by a cross between hermaphrodite and normal male sug- 

 gests the possibility that the hermaphrodite character may also be 

 transmitted through the female. This fact, together with the occur- 

 rence of two genotypes among the hermaphrodites, is held to be 

 slightly favorable to the view that the female is a positive homozygote. 



The demonstration that the hermaphrodite individuals of Lychnis 

 are modified males indicates that STRASBURGER was mistaken in 

 assuming that his hermaphrodites were diseased females. They were 

 probably diseased males in which the dominance of the male character 

 was modified by the fungus. 



The sex ratios in Lychnis do not accord well with a theory of sex 

 which requires males and females to be present in equal numbers. 

 The ratios found in my cultures are in accord with those found by 

 STRASBURGER, the average for the past year being 1.32 females to 

 i male, with a very wide difference in the ratios of different families. 

 The significance of these ratios is not yet understood. 



When the variability in the sex ratios is compared with that in 

 ratios produced by crossing heterozygous purple with white-flowered 

 individuals, it is found that not only is there an undue departure in 

 the sex ratios from the expected ratio i : i, but also that there is greater 

 variability in the sex ratios than in the color ratios, and that the curve 

 is strongly negatively skewed and possibly not monomodal; while the 



