DESCRIPTION OF EPHYDATIA BLEMBINGTA. 95 



the yolk-cells, the trophocytes pass in among them. Thej^ 

 are incapable of passing thi'ough the columnar layer after it 

 has been completely formed, but seem to be able to push 

 their way through when the cells in qiiestion are arranging 

 themselves and becoming elongated. Not all of them pass 

 among the yolk-cells, some, as it appears, only entering among 

 the developing columnar cells and turning back. The majority 

 of them, however, seem to pass among the yolk-cells. As a 

 rule, they pass through the developing columnar layer singly, 

 but occasionally groups of several cells are witnessed making 

 their way in. After the trophocytes have entered among 

 the yolk-cells they distribute nutritive material to them, pro- 

 bably in solution. They take no part in the formation of the 

 reproductive portion of the gemmule further than to supply 

 it with nutritive material whicli the yolk-cells store up in the 

 yolk-bodies. When the inner chitinous layer is about half 

 formed (PI. 3, fig. 13^), the few remaining trophophores are 

 seen travelling towards that part of the gemmule where the 

 pore will appear. They pass out and become scattered about 

 round the gemmule (PI. 3, fig*. 13c). It is not difficult to 

 understand why the trophocytes travel all in the same direc- 

 tion, i. e. away from the portion that is already formed of the 

 inner chitinous layer, for it is undoubtedly the direction of 

 least resistance. 



5. Summary of Conclusions. — (1) Four classes of cells, 

 each of which is derived independently from the sponge, 

 take part in the formation of the gemmule ; first, the mother- 

 cells of the yolk-cells which, alone, constitute the reproduc- 

 tive portion of the gemmule ; secondly, the mother-cells of 

 the columnar cells which pass back to the sponge; thirdly, 

 the mother-cells of the amphidiscs, " scleroblasts," which 



' Tlie sections represented in figs. 13 — 13a? (PL 3) were cut from material 

 preserved in Eleniming's weak solution, while those represented in ail I he 

 other figures were from material preserved either in absolute alcohol or in a 

 mixture of 92 parts of saturated solution of corrosive sublimate and 8 parts 

 of glacial acetic. This explains the absence of the dirty-looking granules 

 from all the trophocytes except those represented in figs. 13 — ]'3d (PI. 3). 



