STRUCTURE OF THE LARVA OF SPONGILLA LACUSTHIS. 393 



mass, and in this way give rise to the chamber-like groups 

 found in the latter. The chief difference between the elements 

 of the cell groups in type B and the flagellated cells at the 

 surface is the lack of a flagellum in the former and its presence 

 in the latter. But the cells which lack the flagellum in type B 

 surpass their rivals, so to speak, in type C, and develop not 

 only a flagellum, but a collar as well. 



In order to lay more emphasis on the arguments which favour 

 a homology between the cell groups and the flagellated cells, it 

 seems advisable to summarise them briefly. First, the cells in 

 question are almost exactly of the same size ; secondly, the 

 nuclei of the cell groups iu type B are identical in characters 

 with those of the flagellated cells; thirdly, when the cells with 

 vesicular nuclei cease to give rise to flagellated cells they 

 begin to give rise to cell groups; and fourthly, though at first 

 devoid of a flagellum, the individual elements of the cell 

 groups develop both a collar and a flagellum, becoming the 

 collar-cells of type C. 



The relationship of types A, B, and C to one another seems 

 to be a simple problem to solve, for the passage from A to B 

 and from B to C is very gradual ; but when type D is taken 

 into consideration, the problem of the relationship of the 

 larvae to one another becomes much more diflficult to solve. 

 Type D cannot, apparently, be fitted anywhere into the above 

 series of larvae, either in an intermediate position or at the end, 

 but can only have been produced directly from type A. The 

 almost complete absence of cell groups in type D is conclusive 

 against its origin from types B and C. On the other hand, it 

 cannot possibly have given rise to type B, in which many of the 

 cells with vesicular nuclei still retain the structural characters 

 of blastomeres. It would be equally impossible to imagine 

 type D giving rise to type C, for were the cells with vesicular 

 nuclei to proceed to divide and to produce cell groups, the 

 result would be a larva possessing quite different characters 

 from those of type C. There would be a great number of cells 

 with granular nuclei on the one hand, and of cell gioups on the 

 other, but hardly any cells with vesicular nuclei. It mav there- 



