CELLULAEINE AND OTHER POLYZOA. 323 
two other zooecia to be intercalated between them. These two zocEcia are 
denoted E and F ; E being derived from A and lying on the inner side of 
C ; and F being derived from B and lying on the inner side of D. E is 
usually in close relation with the axil of the bifurcation, and for this reason 
I term it the axillary zooecium. It may be the proximal inner zooecium of 
■one of the branches and it is succeeded by Gr, the second inner zooecium. 
F and H are the corresponding members of the other branch. 
A careful study of various Cellularine genera, with particular reference to 
the relations of the zooecia above enumerated (and in some cases of one or 
two additional zooecia) has led me to the conclusion that the mode of bifur- 
cation may be used as an important generic or specific character. I do not 
overlook the danger of relying exclusively on a single character, but the use 
I make of the bifurcation seems to be justified by finding that species thus 
assorted appear, on the evidence of other characters as well, to form natural 
groups. It appears to me probable that during the evolution of these genera 
particular methods of bifurcation were adopted at an early stage, and can be 
recognized, with modifications, in most if not in all of the species which 
constitute the genera. It must be added that in most of the species here 
considered the proximal end of the distal zooecium of a longitudinal row over- 
laps its predecessor on its basal side. The diagrammatic figures are all repre- 
sentations of basal views, and the distal ends of the zooecia are thus bidden by 
the proximal ends of their successors. The zooecium A always lies to the 
right of the figure ; and, in species in which A and B alternate, A typically 
lies on that side of the parent-internode which is external in relation to the 
preceding bifurcation. It thus follows that if the branches EGG and 
F D H were followed further to the points where new bifurcations occurred, 
the " A " zooecium of each of these would be found on the outer side. The 
right hand bifurcation would be a repetition of the bifurcation actually 
figured, and the left hand bifurcation would be its looking-glass image, 
A lying on the left or outer side. Didymozoum fornis an exception to this 
rule, as the intercalation of a median ovicell-bearing zooecium takes place by 
a division of the inner row of a biserial branch into two rows. 
Before bifurcation occurs, the zooecia of opposite sides of the branch are 
in lateral contact with one another, and are connected by rosette-plates or 
communication-pores. In Biigula and some of its allies a peculiar modifi- 
cation of this relation is found at the bifurcations (PI. 16. figs. 3-5). Before 
becoming completely disjoined, the two branches remain as a rule united with 
one another by a special communication, formed in the axil by two of the 
lower members of the branches. The connecting process {c. p.) is formed in 
several different ways, which are remarkably constant within the limits of a 
species, which maj thus be distinguished from others with which it might be 
confused. In only one or two species I have found, however, that the more 
proximal bifurcations of a colony may belong to one type, and the more 
