YUZO HOSHINO 237 
variable (heterozygous) parents (Nos. 15-20 in the early; Nos. 21-35 the 
late). The variation ranges and means of the would-be constant families 
descended from the early or the late F., are not in exact coincidence with 
those of the corresponding parent varieties. For example, the variation 
means of the families No. i to No. 4 are very near to, but those of the 
families No. 5 to No. 14 are larger than, those of I. P., especially that of 
No. 14. And in the variation types of the variable families, there is much 
iri'egularity and complicacy ; for example, in the variable families from the 
selected late F^, No. 21 and No. 22 vary with the combined ranges of both 
parent varieties, while other variable families vary with different narrower 
widths of range. 
Thus we see, as Tschermak (1904) already observed, that among those 
F^ plants which flowered within the variation range of the early parent variety, 
as well as among those which flowered within the variation range of the late 
parent variety, onl\' one part was constant and the other part showed segrega- 
tion. As to the interpretation of such complicated phenomena, we shall treat 
it fully later in the present paper. 
Experiments Proper with Peas. 
For the reason stated in the preceeding pages, we selected a new variety 
for the early parent in the present experiments. A short description of the 
variety is as follows : — 
Half dwarf, 86.02 cm. high (measured in igio); white-flowered; tough- 
podded; flowering time about the same as I. P.; cultivated for over 10 years 
in the College Vegetable Garden under the name "Mans". We shall denote 
this variety as "M- P". 
We selected some 50 seeds of both G- P- ^I'ld M- P- from the stock 
seeds harvested in 1909 in the sample section of the Garden, and sowed 32 
seeds of G- P. on Dec. 2, 1939, in a forcing house. From the 9th to i ith 
of the same month, all the seeds sprouted, and after the plantlets grew to 
about 2 inches high, one half of them were thinned out. The first sowing of 
