328 



THE AMERICAN NATURALIST [Vol. L 



his Tables G-U). This includes 231 families classed by 

 Hoshino as "constant." They are descended from 15 

 different F 2 individuals ranging in flowering time (in 

 1912) from 48 to 65 days, this being practically the entire 

 F 2 range. The combined range of the F 3 families raised 

 in 1913 extended from 45 to 71 days. From certain indi- 

 viduals, selected at intervals throughout the ranges of the 

 fifteen F 3 families so as to represent their complete vari- 

 ability seeds were planted which produced the F 4 genera- 

 tion. 421 such F 4 families were reared and studied and 

 of these Hoshino regards 231 as "constant" because of 

 the limited range of variation in flowering time of each. 

 The others are regarded as still heterozygous. The mean 

 flowering time of each of the "constant" families has been 

 calculated by Hoshino and these means have the distribu- 

 tion shown in Table IV (omitting fractions from the class 

 magnitudes which would make them .5 greater than those 

 given in the table, but would not affect their distribution). 

 It should be noted that on account of the peculiar weather 

 conditions of 1914, the flowering time came about 10 days 

 earlier than in the two previous years, the range of the F 4 

 means extending from 34 to 59 days, whereas the F 3 range 

 of flowering time was from 45 to 71 days. Since both 

 upper and lower limits of the range are displaced by like 

 amounts and in the same direction, the general character 

 of the distribution is not affected thereby. 



TABLE IV 



Classification of Means of F 4 "Constant" Families from Hoshino 's 

 Tables G-U 



Table IV shows unmistakably that the F 4 constant 

 families fall into three natural groups, not four as 

 Hoshino 's hypothesis would lead one to expect. The 



