200 ON THE INHERITANCE OF THE FLOWERING TIME IN PEAS AND RICE 
for those described in (4), we shall try to interpret them later on, by the con- 
tamination hypothesis. 
So far as variation types are concerned, we could interpret approximate- 
ly all experimental results in the Fg raisings by the proposed two-flictor 
hypothesis, admitted that some gametic contamination did take place. Now, 
we shall consider the numerical I'atios among Fg families assumed to be in 
different zygotic constitutions. As it is quite difficult to distinguish clearly 
between the progenies of aabB and aAbb, of aAbB and aABB, and of aABB 
and AAbB, we shall take the progenies of aabB and aAbb as one group 
(variable families descended from the early flowered Fg), and those of aAbB, 
aABB and AAbB as another (variable families from the late flowered F^). 
The ratios between four constants and two groups of variables in Tables 9 and 
10 are as follows : — 
aabb 
aaBB 
AAbb 
AABB 
Early group 
variables 
Late group 
variables 
Table 9 
4 
6 
5 
4 
25 
55 
Table 10 
3 
5 
3 
5 
1 1 
27 
Sums 
7 
1 1 
8 
9 
36 
82 
Actual 
Ratios 
0.8 : 
1-3 
: i.o 
: I. r 
: 4-5 
: 10.2 
Expected 
Ratios 
I ; 
; I 
: I 
: I 
: 4 
: 8 
Here we do not see much difference between actual ratios and expected. 
In Table i i, however, we meet with a very singular distribution of constant 
families. There are only two families of the late constant, the early constant 
being entirely absent ; while there are seven families of the early intermediate 
constant (No. i — No. 7), and two families (No. 8 and 9) with peculiar varia- 
tion types. To attribute to chance such a deviation of actual distribution 
from the expected, seems to be rather conventional, but to deny absolutely that 
such a deviation might occur by chance may be dogmatic. So, we believe that 
1) Including ''pseudo-early" 
2) Including "pseudo-late" 
