SPONGES 53 



from the epithelium, which, as has been said, is not always the case, 

 are usually at first rounded cells, within which a minute sjucule 

 appears as an intracellular concretion (Fig. 49, h). As the spicule 

 increases in size it outgrows the secreting cell, which assumes the 

 form of a fusiform or stellate corpuscle apposed to the shaft, or 

 attached to the tip, of the growing spicule, and sometimes sending 

 out processes towards other cells (Fig. 49, h, i). If the s^jicule 

 formed is of large size, the cell, or at least its nucleus, commonly 

 divides to furnish two or more formative cells. In Calcarea, where 

 the scleroblasts migrate inwards from the external epithelium, they 

 at first resemble the epithelial cells in being very granular, but as 

 the spicule grows the granules gradually disaj^pear, and at the same 

 time the nucleus decreases slightly in size. In SjMngilla the spiny 

 naicroscleres are formed within cells of the flat epithelium which 

 have the usual granular nucleus, but the macroscleres are formed 

 within larger cells of the skeletogenous layer, of which the nucleus 

 is at first vesicular in structure, but afterwards becomes granular 

 (Evans). More than one scleroblast may combine together to form 

 a compound spicular system, as in Calcarea (see below, p. 108). 



In Calcarea the scleroblast, or at least one of the two formative cells 

 derived from it, remains attached to the fully formed spicule. In 

 siliceous sponges, on the other hand, no cells have as yet been described 

 attached to the full-grown spicules, and hence it is pi'obable that the 

 scleroblast leaves the spicule when its task of secretion is completed, as 

 occurs also in the case of one of the formative cells in the triradiates of 

 Clxthrinidau. This fact may perhaj^s be correlated with the develoi^ment 

 of a distinct connective tissue system in siliceous sponges, and its absence 

 in the Calcarea. In the latter the formative cells that quit the spicules 

 appear to go back to the external epithelium again [17]. 



(4) The Gastral Lcujer consists in all sponges of one kind of cell 

 and one only, the so-called collar cells, aggregated to form an 

 epithelium of a very peculiar and characteristic type, which fur- 

 nishes a continuous lining to all but a small part of the gastral 

 cavity, as in Ascons, or is broken up into discontinuous cell groups 

 lining the flagellated chambers (see above, p. 32), as in all other 

 known sponges. Each collar cell resembles, as has been said, a 

 single choanoflagellate monad, their most striking characteristic 

 being the possession of a protojilasmic collar surrounding the flagel- 

 lum, as described above for the Oli/nthus (cf. Figs. 52 and 53). 



The variations of the collar cells or choanoajfex of difi'erent sponges are 

 limited in their range as compared with the free living, and therefore 

 more adaptable Choanoflagellata. Differences are seen chiefly in the 

 position of the nucleus, in the relative size, shape, and structure of 

 the collar, and in the size of the cell as a whole. The largest collar cells 

 are founil in the Calcarea, and especially in the family Cluthrinidae, 



