92 2 Gates.—Pollen Formation in Oenothera gigas. 
which is characteristic of the heterotypic anaphase appears, and also in 
its direction, which apparently may be either longitudinal or transverse. 
The result is a very wide variation in the shape of the chromosomes as 
they pass to the poles of the heterotypic spindle. Davis states (TO, 
p. 644) that the chromosomes of the heterotypic mitosis ‘ in both biennis 
and grandiflora have the form of thickened Vs and are not subglobular, 
as described and figured by Gates and Geerts \ With further observa¬ 
tion he would have found that all the variations I have described occur in 
O. biennis , as well as in O. gigas and the other races of Oenothera. The 
split of the chromosomes for the homotypic mitosis may occur in the 
very early anaphase of the heterotypic, or it may be delayed so that 
the chromosomes retain a globular or subglobular shape with no evidence 
of fission even after the daughter nuclei of the heterotypic telophase have 
been formed. As will be seen from PL LXVIII, Figs. 24-26, the V-shape 
of the chromosomes in the heterotypic anaphase is frequently much less 
common than a more nearly globular shape. 
When the chromosomes pass to the poles of the heterotypic spindle 
they at first form a very close group in contact with each other. The 
nuclear membrane very soon appears, as I have described elsewhere (Gates, 
’09 £, pp. 184-7), and as the karyolymph begins to increase the chromo¬ 
somes continue to separate, and they remain hugging the nuclear mem¬ 
brane as the nucleus grows in size. They usually show clearly their bivalent 
character at this time, though the halves of these chromosomes are held 
tightly together and never appear in loose contact, yet the ends of the 
arms of the two halves are, usually at least, widely separated, as though 
they repelled each other like the strips of gold leaf in a Leyden jar. In 
my extensive observations of this interkinesis stage in a number of 
Oenothera forms, I have never found two chromosomes in contact at this 
time. They appear to be always equidistantly spaced just within the 
nuclear membrane, and it is probable that this manner of spacing is due to 
their mutual repulsion at this time. 
During interkinesis there appears to be much variation in the extent to 
which the chromosome bivalents for the homotypic mitosis lose their com¬ 
pact condition and anastomose with each other. It is probable that in 
some cases the chromosome bivalents retain their compact condition with 
little change between the heterotypic telophase and the homotypic pro¬ 
phase. In other cases the chromosomes may become irregular in shape, 
with a considerable amount of anastomosis between chromosomes. Fig. 27 
shows such a case in its earliest stages. In O. gigas this process frequently 
goes further, giving a series of coarse threads in which the boundaries 
of individual chromosomes can no longer be determined. In Fig. 27 
the last of the heterotypic spindle is just disappearing. I have observed 
a number of times in O. gigas that an evanescent cell-plate may be formed 
