ON CROSSES WITH DISSIMILAR HEREDITY. 
253 
In this way, for instance, in my trial garden there have originated 
from CE. Lamarckiana two new species, those cited above as CE. lata and 
CE. nanella, which differ from it more or less in every organ and are of 
entirely different habit of growth. If we cross these, four types may 
result — viz., the two parental ones, one which is simultaneously lata and 
nanella, and one in which these two characters are absent. These last 
individuals, however, display the character of CE. Lamarckiana, which is 
latent in the parents. 
The trials showed that lata-nanella examples certainly appear but 
are always rare, and that the three other types appear in varying though 
in often nearly equal proportions. The crosses are so similar to the 
parent or grandparent that they cannot be easily discriminated. I 
found, for instance, in the seedlings of a crossing effected in 1896 the 
following proportion of examples with the characters of 
CE. lata 30 per cent. 
CE. Lamarckiana 18 „ 
CE. nanella 47 „ 
And so with other trials. I then tested a number of the Lamarckiana 
and the nanella examples for constancy, and so far I have found them to 
come always quite true from seed. 
It is worth while to add another consideration regarding this trial. 
Through the crossing the latent character of the common progenitors 
reappeared. If the origin of CE. lata and CE. iianella had been unknown 
it is clear the result of the crossing would have betrayed their pedigree. 
And it appears to me that we have here a very important principle, 
which later, perhaps, may afford an experimental basis for the study of 
plant origin. 
So far we have only considered such crossings the product of which 
does not separate after self -fertilisation. Cases, however, occur, even 
among the false crosses, of separation. They appear, however, to be 
very rare, and since the individuals concerned are not differentiated by 
any character from the non- separating ones, it is a matter of chance 
whether precisely these are chosen for self-fertilisation. Only very 
extensive cultures could lead to any definite result.* 
The finest character in the genus CEnothera by which the separating 
false crosses may be studied is that which is termed cruciata. For a 
very long period this has been known as a widely spread North American 
species, CE. cruciata, Nuttall, but it appears now also as a sub-species of 
CE. biennis and CE. Lamarckiana. The four petals are narrow, ribbonlike, 
and form together a small cross, whence the name. In 1895 I have 
effected two crosses between CE. Lamarckiana and CE. biennis cruciata. 
The first generation of the cross progeny was, as regards the petals, 
uniform; the cruciata flower was entirely absent. In one of the two 
trials I self- fertilised four examples taken at random, excluding foreign 
pollen. Three of them yielded in the second generation a constant 
progeny without a trace of cruciata. The fourth example, however, 
when sown separately, yielded in 1897 about 50 per cent, of individuals 
with broad cordate petals and about an equal number with narrow 
linear ones. The actual numbers were 57 per cent, of the first and 
* Last summer I planted out about 200 square metres with crossings of CEnothera. 
