FLAX AND TOBACCO 163 



The Broadleaf variety is commonly grown in one section of the 

 Connecticut Valley and is especially valuable for cigar wrappers. 

 Sumatra which is an imported variety produces many leaves 

 per plant but they are small. As may be seen from an examina- 

 tion of the table, the FI had an intermediate number of leaves. 

 Segregation occurred in F 2 and selected F 2 plants gave F 3 families 

 which differed in the average number of leaves. -1-14, showed 

 the lowest coefficient of variability of any F 3 family. Progeny 

 of this same F 2 plant were also grown at another locality and 

 they proved uniform in number of leaves, the calculated coeffi- 

 cient of variability being 6.44 0.27. 5-1-10 gave a low coeffi- 

 cient of variability and a mean leaf number which was about the 

 same as in the FI generation, i.e., intermediate between the 

 parents. 



A cross was studied between Connecticut Havana, which is 

 grown as a wrapper and binder tobacco both in the Connecticut 

 Valley and in Wisconsin, and Cuban, a variety commonly grown 

 under shade. The parents and FI gave about the same number 

 of leaves but in F 2 there was a great increase of variability, forms 

 being obtained with a higher and lower leaf number than in either 

 parent. The inheritance of size and shape of leaf was likewise 

 investigated. The Cuban variety gives a short broad leaf and 

 the Havana a longer leaf which is proportionally narrower than 

 the Cuban. Lines were obtained in F 3 which bred true, respec- 

 tively, to the parental leaf shapes. 



East (1916a) has listed eight requirements, most of them inde- 

 pendent mathematically, which should be met if size inheritance 

 is typically Mendelian, when all populations succeeding the 

 original cross are obtained by growing progeny of single self- 

 fertilized plants. These are: 



"1. Crosses between individuals belonging to races which from long 

 continued self-fertilization or other close inbreeding approach a homozy- 

 gous condition, should give FI populations comparable to the parental 

 races in uniformity. 



"2. In all cases where the parental individuals may reasonably be 

 presumed to approach complete homozygosity, F 2 frequency distribu- 

 tions arising from extreme variants of the FI population should be prac- 

 tically identical, since in this case all F t variation should be due to 

 external conditions. 



"3. The variability of the F 2 population from such crosses should be 

 much greater than that of the FI population. 



