248 
A lien . — Nuclear Division in the 
maintained that the two parental idioplasms fuse to form a new idioplasm 
different from either ; but he conceived this process as involved in the 
fusion of the germ nuclei, instead of being, as we now see it must be, 
postponed to a much later period. 
Such a complete fusion of the parental idioplasms, with a subsequent 
equational division, would result in the production of only one kind of germ- 
cells. If the individual producing these germ-cells be a hybrid, then 
the generation produced by fertilization between this and other hybrids 
from the same cross will consist of individuals all of one sort. I shall speak 
of the generation resulting from the inbreeding of hybrids from the same 
cross as the second hybrid generation. Now, with regard to this second 
generation, while its members all resemble each other, two possibilities arise 
as regards a comparison between them and their parents, the hybrids of the 
first generation. First, the two generations may be exactly similar ; in 
other words, the original hybrids may breed true, and so constitute a variety 
which, having once appeared, is constant. But this is not, at least theoreti- 
cally, necessary. We may imagine that the set of characters resulting from 
the intimate fusion of two idioplasms is something quite different from the 
set produced by the joint action of the same idioplasms existing separately 
within the same cell. So the possibility arises that the hybrids of the 
second generation, while resembling each other, may differ from their 
parents, the first generation hybrids. 
So far as I am familiar with the literature, no instances have been 
reported which seem to correspond to the latter-mentioned possibility ; 
hybrids which breed true, however, have been described in a number of 
cases. Well-known instances are the Hieracium hybrids of Mendel (’69) ; 
those of Salix reported by Wichura (’65) ; de Vries’ (’01) hybrid between 
Oenothera rubrinervis and O . nanella ; and Burbank’s cross (Swingle and 
Webber, ’98) between Rubus ursinus and R. crataegifolius. In this class 
too may possibly be included Millardet’s ‘ false hybrids ’ already mentioned. 
In these there occurs apparently a complete dominance in the first hybrid 
generation of all the characters of one parent ; and this complete dominance of 
one set of characters persists in the offspring of the second and later genera- 
tions. Cases of constant hybrids are numerous enough to justify the assump- 
tion that a complete fusion of the parental idioplasms is possible ; and that, 
therefore, the heterotypic division may, in certain instances, be an equational 
one. But their comparative rarity is evidence that this method of chromo- 
some reduction is not usual ; although the possibility remains that hybrids 
do not furnish a fair test in this matter, that a complete fusion may more 
readily occur between the idioplasms of parents of the same race, and that, 
therefore, such a fusion with later equational division may be commoner 
than now seems probable. On the other hand, de Vries (’03) points out 
that constant hybrids, so far as known, result from the crossing of parents 
