122 BOTANICAL GAZETTE [FEBRUARY 



zygous with respect to this character, would demand equality in the 

 average numbers of the two sexes; but each year Lychnis dioica has 

 produced on the average a considerable excess of females, while the 

 ratios in different families range all the way from less than 5 per cent, 

 female to nearly 90 per cent, female. The percentages of females 

 in 135 families reared during the summer of 1909 are shown in the 

 form of a variation curve in fig. i. It is difficult to believe that a 

 family of 4 females and 100 males or of 87 females and 10 males is 

 theoretically referable to a ratio of 1:1. In 1909 my cultures of 

 Lychnis dioica, taken collectively, consisted of 6366 females and 

 4831 males (including hermaphrodites), or a ratio of 1.32:1, which 

 agrees very well with ratios between 1.20:1 and 1.40:1 reported by 

 STRASBURGER (n). After carrying on extensive experiments on 

 the influence of various environmental factors, with negative results, 

 STRASBURGER reached the conclusion that the sex ratio is determined 

 by inherent factors. This is only another way of saying that it is 

 hereditary in some sense. I have been attempting for several years 

 to test the heredity of the sex ratios, but have not yet found the key 

 to the situation. It is hoped that the numerous crosses which have 

 been made will in the near future throw some light upon the signifi- 

 cance of these exceedingly variable ratios and allow the causes which 

 determine them to be understood. 



In order to compare the variation in these sex ratios with that in 

 a character known to be Mendelian, I have plotted curves representing 

 the percentage of females and the percentage of purple-flowered 

 individuals in all families (94 in number) raised during the years 

 1907, 1908, and 1909, which were produced by the union of hetero- 

 zygous purple with white (see fig. 2) . The variation constants of the 

 two curves appear in the following table : 



TABLE IV 

 VARIATION CONSTANTS or CURVES CONTRASTED IN fig. 2 



