CELLDLAKINE AND OTHER POLYZOA. 325 
the fenestra occurring in a vibraculum which actually develops a rootlet. 
I cannot agree with Levinsen's statement (1909, p. 134) that the rootlet- 
chamber is not thus cut off in Scrupocellaria. In Tricellaria peachii (fig. 11) 
the structure from which the rootlet originates is a chamber cut off by a wall 
from a more distally placed portion which shows a slight longitudinal groove, 
apparently representing the rostral groove which receives the seta of a normal 
vibraculum. I regard the structure in question as a vestigial vibraculum, 
situated in its proper place at the proximal end of a zooecium and on its basal 
surface. The rootlet-chambers thus situated are doubtless the rootlets pro- 
duced by " a definite dorso-lateral chamber situated just above the lateral 
avicularium," mentioned by Miss Robertson (1905, Univ. California Publ., 
Zool. ii. p. 250) \n ^^ Menipea" {Tricellaria}. As pointed out by the same 
author, the rootlets given off by these chambers may pass distally and form 
the tendril-like structures known to occur in Tricellaria. The tendrils are 
not always formed in this way, however ; as I find that in Emma tricellata, 
Busk, one of these structures may replace a branch at a bifurcation. I believe 
this to occur by the suppression of the proximal segment of F or G at a 
bifurcation, as in fig. 13, and by the hypertrophy of the chitinous tube which 
would otherwise have formed a joint, unaccompanied by the production of 
the calcareous parts constituting a zooecium. In other cases, as in fig. 12, 
a rootlet is developed from the outer side of a zocEcium which is not con- 
cerned in a bifurcation. The pore-chamber of the rootlet is still in the 
proper place for a vibraculum, at the proximal end of a zocecium ; and 
although greatly reduced may still represent that structure. 
The rootlets which do not become tendrils pass proximally, sometimes 
along the basal surface, but in many cases along the lateral margins of the 
branches. The formation of lateral bundles of parallel rootlets is specially 
characteristic of Amastigia, JSotoplites, and Menipea. 
Type 1 (PI. 16. fig. 1). — The zooecia occur in pairs^ and not alternately on 
the two sides of the branch. A and B are thus at the same level, and the 
split separating the two branches reaches their distal ends. E, C and F, D 
thus form symmetrical pairs^ E and F remaining in contact with one another, 
at their proximal ends, on the basal- side of A and B. — Didymozoum. (The 
arrangement is slightly modified in an internode which has become triserial 
distally by the development of an ovicell.) 
Type 2 (fig. 2). — Bifurcation occurs before a doubling of the number 
of zooecia takes place, the proximal end of each branch being thus uniserial. 
C remains in contact with the inner lobe of the proximal fork of D ; and the 
doubling of the number of zooecia takes place at the distal ends of C and D,, 
or of one of their successors if more than a single uniserial zocecium occurs- 
at the proximal end of the branch. — ■Stirpariella, sp. (' Siboga ' Collection) . 
LINN. JOURN. ZOOLOGY, VOL. XXSV. 24 
