Pollen Mother-cells of Lilium canadense. 245 
The stages in the heterotypic division which succeed the longitudinal 
splitting of the spirem thread show many peculiarities, most of which 
are, I think, explicable in the light of the facts already described. Seg- 
mentation of the thread into one-half the somatic number of chromosomes 
results naturally from the fusion of the two threads side by side. Seg- 
mentation occurs at every point in the thread at which two chromosomes 
are joined ; and there is thus no such difficulty as is involved in the notion 
of ‘ bivalent 5 chromosomes which result from the segmentation of the thread 
at only half the points of union between its constituent segments. Each 
segment of the thread has been formed by a material and a paternal 
chromosome lying side by side ; but whether these parental chromosomes 
are to be thought of as corresponding respectively to the two visible 
segments resulting from longitudinal splitting is a question which, as I have 
said, cannot be answered from direct observation. However this may be, 
from its method of origin it is to be expected that each heterotypic chromo- 
some in its mature form will have twice the thickness of a somatic chromo- 
some ; and it is well known that the chromosomes in this mitosis are much 
thicker than those that appear in any other division. They are, in fact, 
somewhat shorter and thicker, though not greatly so, than would be 
accounted for by the lateral apposition of two somatic chromosomes ; this 
is doubtless to be explained by a slight increase in the amount of contrac- 
tion which the chromosomes normally undergo during the prophases. The 
process of growth in mass of the nuclear material which occurs during the 
resting stage is followed during each somatic mitosis by the longitudinal 
splitting of each chromatic segment. A similar increase in mass occurs, as 
my figures show, in the stages succeeding the anaphases of the last pre- 
heterotypic division. As a result of this growth and of the fusion of 
the threads in synapsis, the spirem must undergo two longitudinal fissions 
in order to produce segments which shall approximately correspond in size 
and chromatin content to the daughter chromosomes of a somatic mitosis. 
This, as we have seen, is what actually happens. The longitudinal splitting 
which occurs before segmentation seems to be the one necessitated by the 
fusion of the two threads ; and the splitting which is completed in the 
metaphases is the one provided for by the growth of the chromatin which 
has occurred since the anaphases of the last preceding division. 
Evidences of the second splitting appear in the lily, as I have shown, 
shortly after segmentation. In many cases described by other authors, 
no trace of this fission appears until after the separation of the daughter 
chromosomes. It is quite possible that further study may show that 
the fission in these species also appears at the earlier stage ; but, in any 
event, the thick chromosomes of the heterotypic mitosis are at least 
potentially four-parted, instead of being, like those of any somatic mitosis, 
two-parted ; and in order that daughter chromosomes may be produced 
