126 Hayes. 
7. Intermediates as well as extremes may breed true. Of the 
three F, generations which bred true for leaf number, giving no 
3 §' Stags 
greater variation than the parents, two were intermediates. 
Interpretation of Results. 
The above results are a statement of the behavior of parents 
and crosses under careful observation. 
These results in the light of the discovery of multiple factors for 
qualitative inheritance seem most easily and helpfully interpreted as 
due to the interaction of several independently inherited units or 
genes. Each of these independent interchangeable units, or genes, 
allelomorphic to its own absence, is capable of adding to the character, 
and the heterozygous condition of any unit is half the homozygous. 
Let us apply this to our tobacco crosses. Suppose, for example, 
in cross (403 >< 40I) that each of our parents is pure for the same 
basal condition of 20 leaves and that in addition the Sumatra parent, 
No. 403, has some inherited properties which result in a production 
of 26 leaves, Let us suppose this is due to three interchangeable 
allelomorphic character pairs, each inherited separately, and that the 
heterozygous condition is half the homozygous condition. 
If we follow the usual Mendelian method and represent our 
characters by A, B and C and their absence by a, b and c, we get a 
condition in F, of AaBbCe or 23 leaves. In F, we will get a range 
of germinal variation of from 20 to 26 leaves. 
In a total of 64 Fy, individuals, 8 will breed true in F,, and of 
these eight ı will breed true for each parent form or for 20 and 26 
leaves, 3 will breed for 22 leaves, and 3 for 24 leaves. Of the 
remaining 56, 8 will give as great variability as the F, generation, 
24 will give a variation for one character pair, and 24 for two 
character pairs. 
Of the 13 F3 generations of cross (403 >< 401), Table 2, two gave 
no greater variation than the parents, 1. e., on the average, I in 
61% bred true, while the expectation, if three allelomorphic pairs are 
involved is that ı in 8 will breed true in Fs. 
Considering now our cross (40I >< 405), in which both parents 
averaged about 20 leaves per plant, we find a great increase of 
variability in Fy. This condition is very easily explained by our hypo- 
thesis. If we suppose the parent forms to be represented by gametic 
values for leaf number of 16 AABB and 16 CCDD we will obtain in 
F, 16 AaBbCeDd, or 20 leaves, and in F, a greatly increased variability. 
