THE COLLAR-CELLS OF HETEROCGLA. 15 
structure can be seen in profile, though generally less easily ; 
but in a few cases the fibrils have become separated in the 
course of preservation, and stand out like the fringe of a tassel. 
I find that in Naples I once observed fresh preparations of 
Leucosolena primordialis in which the endocytes showed 
no collars or flagella, but appeared as if set with short cilia; 
the conditions were probably pathological. 
As to Sollas’s membrane, the statements of Vosmaer and 
Pehelharing! (19), which I originally went to Plymouth to 
confute, I can now only confirm. 
The sponges were examined alive from rising tide, from 
ebbing tide, from deep tide-pools; after hours in a small vessel, 
after days and weeks in the aquarium. Many sections were 
watched on the slide until absolute death ensued. In nosingle 
instance was Sollas’s membrane observed in a sea-water pre- 
paration. 
As mere accident it would seem that often two neighbouring 
collars must be in contact, yet I only succeeded in observing 
with certainty two or three cases of this. There is never any 
membrane whatever in a plane at right angles to the axes of 
the collars. Neighbouring collars never pass into one another 
in a continuous curve. In every case that I have yet ex- 
amined, where I had reason to believe that the sponge was 
thoroughly healthy, the collar-cells had the form shown in 
figs. 1,2, and 19. In perfect health the collar is very little, 
if at all, expanded from the cylindrical ; it is never trumpeted. 
After suffocation, as detailed below, the collars become coni- 
cal, expanding distally ; probably this is the explanation of my 
observations on S. raphanus (figs. 114 and 11c,—noted in 8, 
p. 630), especially as in this species F. E. Schulze (3) states 
that “unter Umstiinden kann eine solche Erweiterung des 
1 In their otherwise complete summary of the literature these authors 
have omitted Topsent’s statement (9, p. 27) that in Cliona celata “les 
cellules sout unies entre elles par les collerettes. Collerettes et cils sont 
rétractiles comme les pseudopodes de cellules amiboides.” Quite recently 
(24, p. 282) he writes, ‘‘ Les choanocytes d’une méme corbeille peuvent rester 
libres de toute adherence entre eux, ou bien ils se soudent, a l’occasion, par 
les bords de leurs collerettes,” 
