22 
Research Bulletin No. 2 
in F 2 . The reasons for such a phenomenon are these. Two 
plants are the same size. The size in plant No. 1 is due to the 
influence of factors "AABBCC" the size in plant No. 2 is due to 
the influence of factors "DDEEFF." Three interchangeable 
factors in the homozygous condition have produced the same size 
effect in each case. But no one of these factors in plant Xo. 1 is 
allelomorphic to a factor in plant Xo. 2. The F x generation is 
therefore theoretically about the same size as either parent, for 
there are six factors present in each case. The fact that in one 
case these six factors are made up of three homozygous allelo- 
morphic pairs, and in the other case by six heterozygous allelo- 
morphic pairs, makes no difference in the size produced (leaving 
out of consideration the stimulus due to heterozygosis). In F 2 , 
on the other hand, recombination gives a frequency distribution 
far above and far below this typical size. It is the distribution 
obtained by expanding the binomial (ir + -J) 12 . In fact, it is 
exactly what would have occurred if a plant "aa~b~bccddeeff v had 
been crossed with a plant "AABBCCDDEEFF." 
Recently Hayes (1912) has reported a fulfillment of this pre 
diction. He crossed two varieties of Nicotiana tabacum and 
studied the inheritance of number of leaves. In both varieties 
the mean number of leaves was about 20 (see Table 2), and the 
model class about 20. The variability of each variety was small; 
for the "Cuban" variety it was 7.53 ± .293 in the first year 
studied and 5.29 ± .227 in the second year studied, while for the 
"Havana" variety it was 6.98 ± .272 the first year and 8.87 ± .345 
the second year. The mean, the mode, and the variability of the 
F x generation were about like those of the parents. The coeffi- 
cient of variability was 6.10 ± .237. The F 2 generation, however, 
showed a range of variability so much greater than the F t genera- 
tion that it can hardly be questioned that segregation and recom 
bination of several "leaf-number" factors has occurred. The 
coefficient of variability was 15.84 ± .549. 
Cases of the same kind are reported later in this paper. ( See 
Table 33 and Table 38 I . 
