268 Cartel \ — Studies on the Chlor op lasts of Desmids. III. 
consideration of specimens of any one particular species, because of the 
varying factors in each individual cell. Thus there is considerable varia- 
tion in the amount of green material contained in different individuals 
of the same species, irrespective of size, and, again, there is also great 
variation in the size of the pyrenoids themselves ; cf. Figs. 78 and 80. 
Besides this, in all species of Cosmarium which have hitherto been regarded 
as having one or two pyrenoids in each semi-cell, there is, as in other 
genera, the possibility of the division of one or both the original pyrenoids 
to form a group which occupies the same relative position in the cell 
(Figs. 56, 57, and 69). Thus there is normally the possibility of consider- 
able variation in the number of pyrenoids. The factors which influence the 
division of the original pyrenoids are very obscure, and cannot simply be 
external, since the pyrenoids in one part of the cell may divide to form 
such groups, whilst in other parts of the cell they do not (Figs. 56 and 69). 
For this reason it is quite impossible to draw any relationship between the 
number of pyrenoids and the size of the cell. 
In the living condition the compact groups of pyrenoids resulting from 
the division of the original ones cannot be distinguished from single 
pyrenoids, and so this variability of number has not generally been 
observed. 
The position of the pyrenoids in the axile chloroplasts of Cosmarium 
is dependent on the shape of the chloroplast, and so the points at which 
they may occur are usually fixed. Thus, although the actual number of 
pyrenoids is not constant, the number of pyrenoid groups is, since, as a rule, 
the shape of the axile chloroplast is particularly constant. 
Thus the old idea that all species of Cosmarium having axile chloTo- 
plasts were provided either with one or two pyrenoids, although not strictly 
true, was not without foundation, for there are many species which, whilst 
not having invariably either one or two pyrenoids in each semi-cell, yet 
usually have either one or two groups, each group consisting of one to four 
pyrenoids, and appearing to have been formed from the division of an 
original pyrenoid (Figs. 1-23, 28-32, 37-57, and 62-72). The reason that 
so many species have either one or two points of pyrenoid formation is that 
two corresponding types of chloroplast structure happen to be very common 
in the genus. Occasionally there are more than two points of pyrenoid 
formation in a semi-cell. Thus in C. pseudoconnatum there are four (Figs. 35 
and 36). It is only very rarely that the pyrenoids do not occur in definite 
positions in the axile chloroplasts of Cosmarium , the only examples 
encountered during this work being C. Ralfsii and C. ornatum (Figs. 33, 34, 
and 58-61). In the former species particularly the pyrenoids are both very 
numerous and scattered. The opposite is the case with those species 
having parietal chloroplasts, however, for here it is the rule for the pyrenoids 
to be indefinite in number and scattered (Figs. 73-88). 
