somewhat to bring them within the pale 
of this law, if it proves to be universal. 
I he important fact is that in the germ 
ceils of hybrid pigeons an actual separation 
of paternal and maternal germ plasm ap 
parently occurs. When germ cells are to 
be matured, before the real reduction, there 
is in most forms a so calied false reduction, 
in which the chromosomes fuse in pairs so 
that there appears to be only half the nor¬ 
mal number present, though in reality each 
is double (bivalent) and equivalent to two 
of the simple (univalent) type. The doub¬ 
ling of chromosomes which normally occurs 
at such times is frequently incomplete, or 
lacking, in hybrids. This is especially true 
if the hybrids are from widely separated 
species. Instead of a normal spindle bear¬ 
ing the usual number of bivalent chromo¬ 
somes, multipolar spindles, or two separate 
spindles may appear, thus apparently per¬ 
mitting the two kinds of parental chro¬ 
matin to remain apart. In the most ex¬ 
treme cases a complete separation may 
occur subsequently, the entire chromatin 
of one parent occupying one cell, that of 
the other a different cell. Such visible 
separations, however, only occur exten¬ 
sively in sterile hybrids from markedly 
different parent species. Fertile hybrids 
from closely related forms, for the most 
part, display spindles normal in appear¬ 
ance. We may suppose, however, that 
there is the same tendency for the respec¬ 
tive parental characters to separate, though 
the incompatibilities of the plasmas are 
not sufficient to prevent the formation of 
bivalent chromosomes and normal spin¬ 
dles. It does not necessarily follow that 
in the ensuing division all of the charac¬ 
ters of one parent will be set apart in a 
separate cell, as was the case in the abnor¬ 
mal mitoses. If such a separation were 
actually to occur, then the offspring which 
returns to a grandparent type must revert 
in not only one given character, but in all 
characters ; that is, the reversion would be 
complete. The Mendelian law, however, 
confines itself to a given character, and if 
any other character is chosen, although it 
will follow the same law, it does so with¬ 
out any reference to the first character, so 
that offspring may be pure with respect 
to a particular character, yet also possess 
other characters of a mixed nature, or even 
pure characters of the other parent. In 
the case of these milder fertile crosses, 
then, where reversions follow the Mende¬ 
lian law, the germinal incompatibilities 
must be narrowed down to the qualities 
themselves rather than confined to the 
respective germ plasms as a whole. These 
qualities must separate and each take up 
its abode in a different germ cell irrespec¬ 
tive of whether the other qualities of that 
particular germ cell are of a different 
parentage or not. The cases in which the 
entire plasmas are segregated are then 
probably but magnified images of what 
occurs among the specific qualities of the 
milder crosses. The interesting possibility ] 
arises that if fertile hybrids can be secured | 
from widely different species the plasmas : 
of which must be more incompatible than ! 
those of nearly related forms, such hybrids J 
will give rise to offspring in which there j 
is reversion, not only of one character, but ] 
of many or all characters in the same in- ,j 
dividual, due to a more thorough segre- 
gation of the parental germ plasm as a 
whole. In other words, the farther apart j 
the parent species are, the more complete j 
will be the return in anv given offspring \ 
which shows reversion. To the mind of j 
the writer, the nearness or the remoteness 
of the relationship between the two ances- j 
tral plasmas will be found to be an impor- ] 
tant factor in determining the amount j 
(number of characters) of reversion in \ 
any given offspring. 
