414 
Journal of Agricultural Research 
Vol. V, No. io 
On the hypothesis of factors acting as multipliers, the range of the 
long-podded plants should be about 1.5 times that of the short-podded 
plants, as it is in the more reliable 1912 results. 
To sum up, the results of investigation of the second generations agree 
with the hypotheses that all the factors act as multipliers; that factor E 
is completely dominant; that the minor factors show zero dominance; 
that the minor factors act symmetrically with regard to each of the two 
grandparental lengths, which is not the case in a cross of the Florida 
velvet bean by the Yokohama bean {Siizolobium hassjoo). 
THIRD GENERATION 
Table VI gives all the third-generation families, grown in the elimi¬ 
nation field, which segregated measurable short podded; and also all 
which did not, but had eight or more measurable survivors. Because of 
the crowding, these results are not so reliable as those given in Table VII, 
which include all the families grown on poles in 1913. 
Table VI .—Frequency arrays of the average lengths of ripe pods of the third generation 
Florida velvet bean X Lyon bean {classes of 3 mm.) 
[The asterisk (*) shows the pod length of the parent plant of the family.] 
a The averages for the first nine families refer to the long-podded plants alone. 
& Grown in the elimination field in 1911. 
c Grown on poles in 1912. 
