STRUOTUliE OF THK LARVA OF SPONGILLA LACUSTRIS. 433 



and the formation of new ciliated chambers in a growing 

 sponge are scarcely known, and no satisfactory observations 

 have been recorded with reference to this point." In the 

 present appendix it is my purpose to give a brief description 

 both of the multiplication of collar-cells and of the formation 

 of new chambers. 



The collar-cells, when they have been fully formed either by 

 the fragmentation of a cell with a vesicular nucleus or by the 

 immigration of the flagellated cells during metamorphosis, 

 divide by mitosis. It is a rather remarkable thing that these 

 cells, after having been developed, in one case at least, from cells 

 with vesicular nuclei by fragmentation, begin to divide by 

 mitosis. 



The nuclei of the collar-cells are situated at the base of the 

 cell, but when they are going to divide the nuclei travel to 

 the other end of the cell, and are seen situated close to the 

 collar, which is gradually withdrawn (fig. 33 b, the cells c. c). 

 The nucleus soon loses its ordinary structure, and travels into 

 the middle, and there goes through all the changes of mitotic 

 division. The longitudinal axis of the nuclear spindle is placed 

 tangentially to the wall of the flagellated chamber. Conse- 

 quently the cell must divide radially with respect to the 

 chamber (fig. 35, the cell c. c). The presence of the spindle 

 in the collar-cell marked c. c. in fig. 35 proves, beyond any 

 doubt, that the collar-cells divide, and consequently increase 

 in number. It might therefore be argued, on a priori ground 

 alone, that the chambers also must divide, because the number 

 of cells which go to form a chamber has its limits. But there 

 is no need to have recourse to a priori reasoning, for flagel- 

 lated chambers which are actually dividing have been often 

 observed in sections. One such chamber has been drawn in 

 fig. 29 a, and is marked C. The chamber in question is 

 actually being constricted into two, the collar-cells having 

 already become arranged in two groups round two different 

 points as centres, and the cells with granular nuclei, especially 

 from the outer side, are making their way in with a view to 

 the formation of the lining of an exhalant canal, which will 



