44 ARTHUE DENDY. 



nected at their margins by a very distinct membrane^ which 

 appears in section as a thicker line running from one to the 

 other, but interrupted by the mouths of the collars^ as shown 

 in the figure. The structure and relations of this membrane 

 (SoUas's membrane) appear exactly as I have described them 

 in Stelospongos. The collared cells nearest to the exhalant 

 opening of the chamber are shorter than those farther away, 

 so that the membrane gradually approaches more and more 

 closely to the gelatinous ground-substance around the chamber, 

 and finally seems to run into it at the opening itself (fig. 

 1, c). The fact that Sollas^s membrane appears here as a 

 thicker line than the outlines of the collars is, I believe, 

 simply due to the thickness of the sections causing us to see a 

 little more than the mere cut edge of the membrane. 



Professor Sollas, in his valuable ' Report on the Tetracti- 

 nellida of the Challenger Expedition,' observes, " I have never 

 yet seen the flagella of the concrescent choanocytes, though I 

 have never failed to find them in the case of choanocytes which 

 are not concrescent. It might be explained on the supposition 

 that the flagella are retracted in the former case; but that 

 naturally leads to the inquiry as to why they are not retracted 

 in the latter." In Stelospongos I was also unable to detect 

 the flagella, but I expressed my belief in their existence in the 

 living sponge, and gave a diagram showing them. My researches 

 on Halichondria panicea fully justify this view. Fig. 1, 

 which is a careful drawing of an actual preparation, shows the 

 flagella plainly, projecting from the bodies of the collared cells 

 through the collars and into the cavity of the chamber (a). 

 Thus the question as to the coexistence or otherwise of Sollas's 

 membrane and the flagella of the collared cells is settled. 



If we now look down upon the wall of a chamber from the 

 outside, instead of examining it in section, another important 

 point is brought to light. The collared cells, which, it will be 

 remembered, stand well apart from one another, are connected 

 at their bases by broad protoplasmic processes, so that the wall 

 of the chamber appears to be made up of a network of broad 

 protoplasmic strands with nuclei at the nodes of the net (fig. 



