Ikeno. — On Hybridization of some Species of Salix. II. 177 
Each number in the above table includes plants derived from one cat- 
kin, and all catkins are taken from one tree. The table shows that very few 
plants are produced from each catkin— very often only one. This is first of 
all to be ascribed to the fact that each catkin is very poorly fertile, owing 
evidently to the great difficulty of the hybridization under discussion. 
Many more seeds were actually obtained than might be inferred from 
a simple inspection of the table ; many of them, however, failed to germi- 
nate, or else many seedlings perished in very young stages. In the table 
Nos. 3, 9, 12, 21, 24, 25, and 27 are wanting, because no adult plants were 
available from these pedigree numbers, owing to one of the two causes 
stated above. In one plant from No. 11, ranked under the column GM, 
some branches bear G-type catkins, and others M -typed ones. Nos. 26 and 
29 have each produced one female multinervis (placed under the column 
MM), to which I will refer later in this paper. 
If we add together the F 1 hybrids from the crosses done in 1910 and 
have : 
. Table II. 
Year. 
G-type. 
M-type. Multinervis. 
GM. 
Totals. 
1910 
1 1 
3 
H 
1918 
59 
11 2 
1 
73 
Totals 
7 o 
14 2 
1. 
87 
% 
83-3 
16.7 
If we take simple G- and M-types into consideration, the three remain- 
ing plants being left out of account, we have eighty-four offspring in all, 
composed of 83-3 per cent. G-types and 16-7 per cent. M-types. 
The important question in respect to the hybridization under dis- 
cussion is : Why will it yield the two types G and M already in F v despite 
the fact actually observed that each of the two parents breeds true respec- 
tively ? Have we here to deal with a segregation ? In my former paper I 
was led to the conclusion that we have in this case no segregation at 
all, and I tried to explain the appearance of two types in F lt as well as 
the behaviour of the F 2 progeny, by the interference of invisible factors X and 
Y . 1 The results of hybridizations between F x hybrids and either of the 
two parents (i. e. back-crossing) which were performed in 1917 and 1918 
served for testing the validity of my hypothesis. They were by no means 
particularly favourable for it. Especially, the results from G-type F x x 
gracilistyla were contrary to my theoretical expectation : thus, if we denote 
by X the factor which is able to cause Gg to develop to G-type, as was 
done in my former paper, 2 we have G-type F 1 = XXGg, and gracilistyla = 
GG, and the cross XXGgx GG or XXGg xxxGG should give rise to 
2 (XxG G + Xx Gg), i. e. to G-type progeny exclusively, which was not 
1 1. c., p. 48 ff. 
2 1. c., p. 50 ff. 
