REPORT ON THE KERvVTOSA. 15 



are shown in PI. VIII. figs. 4 and 5. What I have to communicate concerns their origin, 

 provided 1 am right in identifying the dumb-bell shaped bodies whose description will be 

 given later on along with the filaments. Bowerbank, Schmidt, and Kolliker found in 

 many of the specimens characterised by the presence of filaments certain round bodies, 

 and it has been supposed by the last named naturalist that these bodies give rise to the 

 filaments. F. E. Schulze, on the contrary, maintains that there can be no question as to 

 any such connection ; he finds these bodies too, but is inclined to consider them to be 

 unicellular algffi, the more so as he has observed them in phases of division more or less 

 complete. I am uncertain whether the bodies I am about to describe, which are 

 represented on PL VIII. fig 2, are identical with those of F. E. Schulze and Kolliker. I 

 believe them, however, to be so, since the corresponding description of Schulze is thoroughly 

 applicable to them, and also because I found some of them on the point of dividing into 

 two halves ; but even if identical, I am yet more inclined to adopt the opinion of Kolliker, 

 and to ascribe to them a certain connection with the filaments. I observed these round 

 bodies in my Cacospongia dendroides, and comparing them with the heads of its 

 filaments, was struck by their mutual resemblance. Not in every case, however ; for, 

 beginning with forms characterised by the thickness of their walls and indistinct central 

 difi'erentiation through numerous intermediate stages, I came to forms with walls far 

 thinner, and, in their central contents, recalling very much the drawings Schulze gives of 

 his conjectural alg^ in Hircinia (loc. cit, pi. iv. fig. 15). An attentive examination 

 of their walls shows that they are provided with a small thickening, which appears to 

 correspond exactly with the spot where the thread of the filament separates from its 

 head. In a couple of cases, I think, I have also seen these bodies, when grown thin-walled, 

 6tni in connection with the thread; but even if this were not an optical illusion, it seems 

 to occur pretty seldom, and we have also to suppose that the further difi'erentiation of the 

 filamental heads — provided that they are identical with our round bodies — takes place 

 only after the head has separated from the thread. The final phase of this development 

 consists in the bursting of the wall so that the internal contents escape. It is represented 

 by corpuscles which, when out of the capsule, present a great variety of size, some so 

 small that they appear to be merely points, others so large that they are readily to be 

 discerned l)y a magnifying power no higher than 200-250. In some cases these corpuscles 

 appear to have linear or even fusiform outlines. This, however, is very rare; they almost 

 always have a very definite dumb-beU shape, so that if such a body were to be imagined 

 as growing in length a typical filament would result. But if this reaUy occur, it apparently 

 does not take place immediately, the corpuscles having the faculty of multiplying previously 

 by division; sometimes, but not often, a cross division has been observed, sometimes a 

 star-like one, sometimes a longitudinal one. As the result of this division, a new 

 generation of corpuscles of the same form may again occur. As to their internal structure, 

 I can state that, as iii the filamental heads, their walls consist of concentric layers ; a 



