618 W. J. SOLLAS. 



sarily be a modification of the process of invagination, but a 

 wholly different method of arriving at the same end and re- 

 sulting from the presence of a mesenchyme. For a hollow 

 blastula is evidently in a very unstable state, and folding might 

 almost be predicted for it ; but in a planula containing mesen- 

 chyme the internal layer is well supported^ and not so advan- 

 tageously conditioned for folding. 



If, however, in the planula the flagellate layer grows more 

 rapidly than the mesenchyme can keep pace with it, and yet 

 at the same time maintains its coherence, and remains adherent 

 to the mesenchyme, tension will arise within the latter, from 

 which on account of its extreme tenderness it would readily 

 relieve itself by splitting. 



The fissure once formed, whether for this reason or any 

 other, soon becomes lined by a layer of flagellate collared cells, 

 the appearance of which is remarkable, but readily intelligible, 

 if we adopt the hypothesis that the cells of the mesenchyme 

 have a flagello-choanate infusorian ancestry. 



In conclusion I would offer some comment on the classifi- 

 catory relations of Halisarca. As a genus it is distinguished 

 by the comparatively undifferentiated character of its cells ; the 

 ectodermal cells retain their flagella — though not the collar — 

 throughout life, and the mesodermal cells exhibit none of those 

 modifications which are so characteristic of the higher Sponges, 

 such as the Geodise. There is also a conspicuous absence of a 

 skeleton, siliceous or calcareous. This striking simplicity of 

 both species of Halisarca, and the absence of any features in 

 their development which would tend to show that they are 

 degraded forms, leads me to regard them as amongst the lowest 

 of existing Sponges. On the other hand the Chondrosiadae 

 which have been associated with them appear to occupy a much 

 higher position. They suggest affinities with the Tetractinellidse, 

 of which, perhaps, they may be regarded as degraded de- 

 scendants. 



Consequently, however much simplicity of classification 

 may suffer, it appears to me advisable to remove the Chondro- 

 siadae from the order Myxospongise, which will then contain 



