THE PHYLOGENY OF THE GENUS BRACHIOMONAS 8I 
by its own cilia. Bohlin states that the first division is longi- 
tudinal, and certainly this is regularly the case. Chodat 
presents a figure (2, f. 66), and I have seen two or three such 
cases, where the first division appears to be transverse, but it 
is quite possible that such an appearance is due to an unusual 
shifting of the position of the cell contents after a proper longi- 
tudinal division. The subsequent divisions have not been 
described or figured by previous observers, and are ordinarily 
difficult to follow because of the continual active movement 
of the mother cell. I was so fortunate as to find one very 
favorable case in B. submarina f. obtusa, and later many similar 
ones in the type form, which I believe to be entirely representa- 
tive. This individual became quiescent after the first division 
was completed (FIGs. 22, 23) ;it presented its anterior pole upward, 
so that it was possible to see easily that the second plane of 
cleavage is also longitudinal and perpendicular to the first. 
Immediately after the second cleavage a slight stretching and 
shifting of the daughter portions took place, so that one of them 
appeared somewhat underneath two others (Fic. 24). The 
third cleavage appears to cut each of the four daughter portions 
across its longer diameter, that is the plane seems to be essentially 
transverse to the previous divisions and to the axis of the mother 
cell. The pyrenoid disappears before the first cleavage occurs, 
and the stigma fades out during this process. A new and 
rather small pyrenoid appears in each daughter portion after 
the second cleavage, and probably another pyrenoid arises de 
novo in four of the eight daughter cells. About an hour after 
the completion of the third cleavage the eight daughter cells 
had acquired the typical form of the arms, and also the eye- 
spot and cilia (Fic. 27); after about another hour they escaped 
from the mother cell-wall. In certain cases it is clear that the 
cilia of the mother cell remain connected by a strand of proto- 
plasm to one of the daughter portions during division (FIG. 19) 
or even until the eight zoospores acquire their final form. 
Whether this is regularly the case appears doubtful. In mate- 
rial identified as Brachiomonas gracilis Bohlin, at Aalesund, 
Norway, I observed numerous cases of division which followed 
the course outlined above. 
In the case of the production of only four daughter cells 
it is clear that the cytoplasmic polarity of the mother cell 
would be directly transferred to each of the new zoospores. 
But when division proceeds further it is less clear how the 
