Yuzo HosniNo 253. 
were no frequencies, but in classes 52 and 53 the frequencies increased. In J 
we see that the variation types of all families are not so uniform as we have 
seen in Table 12 A- The 18 families, from No. i to No. 18, are very simi- 
lar in their variation types, but the remaining 8 families have different varia- 
tion types. The variation means of the latter are greater than those of the 
former. From this we can assume that the grand-parent plant was not 
constant in a strict sense, but somewhat variable in its nature. 
K- The parent family of this raising was also one of those which were 
to be designated as the early intermediate constants. Its variation range was 
wider than that of the parent family of J. Its maximum frequency class was 
51, one day later than in the case of the parent family of J. By examining 
Table K, however, we can assume that the wider range of variation in the 
parent family was not genetic but environmental, because the variation types 
of all the families in K are narrower in range and very similar to one another, 
suggesting genetical canstancy of the F.^ grand-parent plant. But the differ- 
ence in the position of the maximum frequency class in the parent families 
of K and J might have genetic meaning, because the average variation means 
in the K raising is larger (later) than that of the majority of the families in 
J. From this inequality of the variation means in both raisings, we have to 
conclude that all F3 families, which we have designated as the early inter- 
mediate constant, are not equal genetically. 
L. The parent family of this raising, together with the family No. 9 in 
Table 1 1 (PI, X^'^), had a peculiar variation type (see page 248). They were 
very likely to be the families which vary within the combined range of the 
early intermediate and the late intermediate constant, having their maximum 
frequency class in the latter constant. Now, by the actual laising of F^ 
progenies, we see : — • 
1. All F^ families are more or less different in variation type, suggesting 
that the F^ grand-parent was not a strictly constant one. 
2. All families, however, vary only within the combined range of the pre- 
vious two raisings (J and K), suggesting that the occurrence of the maxi- 
