428 Wilson. — Spermatogenesis in the Bryophyta . 
at this stage only one division of the nucleolus takes place, one of the bodies 
produced then passing into the cytoplasm. Or, of the three bodies formed 
from the nucleolus, only one becomes separated and two remain in contact 
with the nucleus for a considerable time. The same result is, however, 
arrived at in all cases ; a transference of the chromatic material from the 
nucleus to the cytoplasm takes place either in the form of one or of two 
bodies. These bodies, which are found scattered throughout the cytoplasm 
frequently in the neighbourhood of a vacuole, probably increase in number 
by division, as more than two can often be discovered (PI. XXXVII, Fig. 18). 
Some or all of these now give rise to short irregular rod-like structures which 
are usually grouped together within or close to a vacuole (Fig. 19). At this 
stage the nucleus is only distinguishable as a denser mass which takes 
up the stain rather more strongly than the surrounding cytoplasm. The 
nucleolus is frequently found in a dividing condition, generally partially 
separated into two portions, but occasionally into three (Fig. 19). During 
these divisions of the nucleolus the rod-like bodies already described are 
always present in the cytoplasm, and on this account this stage can be dis- 
tinguished from the earlier period of division, although the processes in the 
two cases are apparently identical. Later on the bodies produced by the 
division of the nucleolus become completely separated (Fig. 20) and pro- 
bably pass into the cytoplasm, becoming associated with the rod-like bodies 
and perhaps giving rise to some of them. 
The rod-like bodies now increase in length, becoming irregularly curved, 
and in this condition exhibit a remarkable resemblance to chromosomes. 
Their number is not constant, but three or four can often be distinguished, 
although counting is rendered difficult by their irregular form (Figs. 21-4). 
A further consideration of the results obtained by J. and W. Docters 
van Leeuwen-Reijnvaan is of interest at this stage. In both Polytrichum 
( 41 ) and Mnium ( 42 ) these investigators state that the spermatogenic cells 
shrink and become rounded before the final division. In Polytrichum 
a process of constriction goes on, and in the daughter-cells so produced 
chromosomes can still be distinguished. Although no constriction is 
described in the case of Mnium , chromosomes are similarly represented 
in the daughter-cells. It would seem probable that these investigators 
have seen spermatids containing the rod-like bodies and have considered 
that this is a stage in the final division. The number of the rod-like bodies 
as already pointed out is three or four, and this would correspond roughly 
to the number of chromosomes demanded by a double reduction. The 
partiaLsuperposition of two spermatids will account for the constriction 
shown in Fig. 56 of their communication ( 41 ). 
The sequence of the following stages in Mnium hovnum is very difficult 
to determine. During the early stages of development of the rod-like 
bodies the nucleolus is undivided (Figs. 21 and 22), but when these have 
