56 SPONGES 



distal extremity is often difficult to see, especially the actual opening. 

 Hence these cells were at first described and figured by Carter and Dobie 

 Its having three Hagella — a larger median and two smaller lateral. 

 More usually the collar is found retracted down to the level of the second 

 (distal) hoop, which then appears thickened and easy to make out. 

 Frei[uently the collar is found still further retracted, and it often ends 

 at the first hoop (Fig. 52, i>, i). In the extreme case of contraction of 

 the sponge, no collar is to be made out at all. 



In Clathrina coriacea there appears to be but a single hoop, correspond- 

 ing probably to the proximal hoop of .4. falcittu : and in Sycou, according 

 to Bidder, the collar is fiuted, being supported by about thirty vertical 

 rods or thickenings (Fig. 52, C, «). Within the collar, at its base, Bidder 

 describes a sphincter-like thickening (hoop ?). In Leucosohnia the collar 

 cells are very similar to those of Sijcon (Fig. 52, C, 6). 



In ChoanoHagellata, Franco (1897) describes the collar as originating 

 by the folding round of a protoplasmic membrane or band, which runs 

 up the side of the body and is twisted in a spiral round the base of the 

 tlagellum. Its structure could be imitated by twisting one end of a broad 

 paper band or ribbon into the shape of a funnel. Nothing of the kind 

 has been described in collar cells. 



The details of cell division in the case of the collar cells have not as 

 yet been studied in full, but in Clathrina coriactu this process is initiated 

 by the nucleus travelling to the summit of the cell and taking up a 

 position beneath the tlagellum. The nucleus then divides, one half 

 parses down, and the cell divides transversely to its long axis. The 

 upper half, bearing the original collar, grows a new basal portion, into 

 which its nucleus travels ; the lower portion forms a new collar and 

 tlagellum. 



In ontogeny the cuHar cells — that is to say, the ciliated Cflls of the 

 embryo, which become the collar cells of the adult — have idways the 

 nucleus near the distal extremity, and the tlagellum arising directly from 

 the nuclear membrane, and passing out through the cell. This condition 

 is retained in the Ltucosoleniid((e and most Heterocoela, and is probably the 

 primitive state of things. When it occurs the larger vacuoles are found 

 at the base of the cell, not at the summit. In Clathriniihte, however, the 

 nucleus loses its connection with the flagellum, becomes attached to the 

 side of the cell, and finally travels down to the base, leaving behind it at 

 the upper extremity the distinct granule from which the tlagellum arises, 

 representing, perhaps, a centrosome ^cf. Fig. 58, 5\ It is interesting to 

 note that, as described above, each collar cell in this family when about 

 to divide commences by placing its nucleus in the primitive position at 

 the apex of the cell. 



In siliceous sponges the collar cells are much smaller than in Calcarea 

 and often excessively minute. In HulichondrM the nucleus is apical, as 

 in LencosoUmia ; in i>pomjilhy on the other hand, it is Uisal, as in 

 VUithrina (NTosmaer and Pekelharing\ 



Much discussion has been carried on as to the existence of a membrane 

 uniting the margins of the collars, described by Sollas in many Deniu- 

 siHjngiao, and hence termed ".">ullas's nienibrano.'' It was asserted by 



