324 SIR SIDNEY F. HARMER ON 
distal ones to another. This is the case, for instance, in a colony of Bicel- 
larina alderi, in which two successive tj'pes are represented ; but, so far as 
my observations go, this is exceptional. It is well known, however, that tlie 
proximal end of a colony may show characters regarded as juvenile, as com- 
pared with the more differentiated distal parts ; and even if it should be 
found that differences in the mode of bifurcation occur at opposite ends 
of the colony, in other species, it will not, I think, affect the correctness of 
the statement that each species of Bugula has a practically constant method 
of bifurcation, in its fully developed condition. It may be added, finally, 
that the arrangement assumed is dependent, in the main, on the extent to- 
which the inner zooecia remain in lateral contact with one another, or, in 
other words, on the distance to which the split forming the bifurcation 
extends towards the zooecia A and B. 
The position of the rootlets, with regard to the bifurcations, has previously 
been recognized as a character of importance, particularly by Waters (1897, 
1913, cited on p. 321). In the series including 3Ien/pea and its allies, these 
structures are given off from a pore-chamber which projects into the body- 
cavity, sometimes on the distal side of the jointed region (fig. 15), and some- 
times on its proximal side (figs. 9-11). The difference may appear a slight 
one, but the general flexibility of the colony must be affected by the position 
of the rootlets. It seems probable that, when a joint has been evolved, the 
relation of the rootlets to it would not be easily altered during subsequent 
modifications of the species. Thus one series of species, represented by 
Meni'pea and Notoplites, may be supposed to have started with their rootlets 
on the distal sides of the joints, and at the proximal ends of the internodes ;. 
and to have retained these relations during their later evolution. Tricellaria 
may be similarly supposed to have originated from a condition in which th& 
rootlets were on the proximal sides of the joints and at the distal ends of 
the internodes. 
Levinsen (1909, p. 132 n.) has stated that vestigial vibracula may be recog- 
nized in the pore-chambers of the rootlets of Tricellaria ternata. I think 
this view is correct, and in fig. 11 I have shown structures which admit of 
this interpretation in T. peacJiii, on the proximal segments of the zooecia 
and D. In Scrupocellaria each zooecium is typically provided with a vibra- 
culum, which lies on the basal surface of the branch, usually near its external 
border. The vibraculum belongs to the proximal end of the zocecium, 
although it is in close relation with the external or marginal avicularium of 
the preceding zocecium of the same longitudinal row. The vibraculum is 
constantly provided, in this genus, with a chamber, cut off by calcareous walls 
from the cavity which contains its muscles, and this chamber gives origin to 
a rootlet, in the more proximal members of the colony at least ; although in 
many of the other vibracula the rootlet is merely represented by an oval 
fenestra in the outer wall of the rootlet-chamber, corresponding exactly with 
