I go NATURE AND AFFINITIES OF THE SPONGES. 



Islands examples, and as exhibited in their various developmental phases 

 at PI. IX. Figs. 18-2 1, are identical with the so-called rosette-cells lately 

 described by Metschnikoff ; the descriptions and delineations of them as 

 given by him being, however, deficient in one important point — that of the 

 possession by each constituent unit, in its matured condition, of the charac- 

 teristic collar. This oversight is, however, altogether what might have been 

 expected when it is found on examining his text and plates illustrating 

 the histology of various calcareous and siliceous sponge-forms, that that 

 important structure is not, except in one doubtful instance, either figured 

 or alluded to. 



The true significance of the special rosette-like aggregations remains 

 to be discussed. Their derivation through the coalescence and segmenta- 

 tion of a greater or less number of amoebiform units, as in the case of the 

 larger ciliated gemmules, and that of the ampullaceous sacs, was ascertained 

 by the author, their close correspondence with the former of these two struc- 

 tures being thus made especially apparent. Their only distinction from the 

 ciliated gemmules subsists, in fact, in their much smaller size, through being 

 derived from the fusion of a small number of zooids only, and their retention 

 of a spheroidal uvella-like contour, without any development, by separation 

 from each other, of an extensive cleavage cavity. It would seem to be by 

 no means improbable that the rosette-shaped aggregations, thus derived 

 from the fusion and segmentation of a smaller number of the typical 

 collared cells, represent in the one direction a more primitive form of the 

 ciliated reproductive gemmule, and in another the more primitive mode of 

 grouping of the typical collared zooids in the colonial sponge-stock. As 

 already stated, and as shown at PL IX. Fig. 2i, these spheroidal groups 

 were found attached to and projecting from the cytoblastema into the 

 adjacent canal-systems in the case of Halisarca, and contributed largely, in 

 combination with the ampullaceous sacs, to the composition of the general 

 substance of the compound body. It is quite probable that sponge-forms 

 exist, or have existed, in which such a disposition alone of the collar-bearing 

 zooids represents the normal and characteristic type of structure. A close 

 approach to this has in fact already been observed by the writer in con- 

 nection with a siliceous-spiculed sponge, further investigation, however, 

 being desirable for its complete exposition. Should anticipations in this 

 direction be confirmed, it will certainly admit of a yet closer approxi- 

 mation of the sponges to the independent Flagellate Infusoria than is now 

 attempted. For in such a case, a sponge-stock having its essential collared 

 zooids so disposed that they project simply in the form of uvella-like 

 clusters into the interstitial canal-system, and whose motile reproductive 

 bodies consist simply of similar, but detached, uvelloid clusters, would in 

 all ways be comparable to a colony of Codosiga botrytis or Anihophysa 

 vegetans, in which, instead of developing a branching stem, the spheroidal 

 clusters or "coenobia" of associated zooids extended around them, and 



