YUZO IIOSHINO 241 
sprouting in this year was different i'vom that in the previous year. The 
sprouting then occurred gradually. In this year, however, we can divide all 
individuals into two groups, one consisting of those which sprouted during 
the 3 days from the 15th to the 17th of May, and the other, those which 
sprouted from the 3 ist of May to the 4th of June. In Table 5, the frecjuency 
distribution of tlie number of days from sowing to flowering in the two groups 
is shown. The black-faced types indicate the frequencies of the early-sprout- 
ed group. 
Erom Table 5, we can affirm the following facts: — 
1. The variation types of all families descending from M- P. are alike and 
uniform. 
2. Among the families descending from G- P- there are two distinctly dif- 
ferent types of variation. All families from I, XII and XVI show the 
early type of variation, and those from VI and VIII the late type of 
variation. Among the families from C., the first 12 families are of the 
early type and the remaining 8 families of the late type. 
3. In I'aising the offsprings of one common grand-parent, we took special 
care to select those parents plant which differed in their flowering times ; 
and in raising those of C family, we selected those parents which flower- 
ed in different grades within the variation range of the family. But all 
families descending from a common grand-parent showed the same type 
of variation; and the variation types of all families from C fell into two 
distinct categories. 
From the facts cited above, it can safely be concluded that the character 
of the flowering time in peas is a fixed and stable one, and that the population 
of the G- P- variety contains two different pure lines. By assuming the pre- 
sence of two pure lines in the G- P- population, we can easily explain the 
iri'egularity and wide range of the varia,tion in the G> P- offsprings which we 
have seen in several previous raisings (Tables i, 2, 3 and 4). 
