1884.] 93 [Hyatt. 



stage of Gummina minosa in his " Eponges de la Manche," p. 57, 

 leaves hardly a doubt that this sponge is a calcispongian and, pos- 

 sibly one of this class entirely devoid of a skeleton. This impres- 

 sion is supported by his figure (38, pi. 14) which is that of an 

 undoubted calcispongian embryo. His figure of the later larval 

 stage is valuable in another aspect, since, as stated by Barrois, it 

 is a gastrula, or as it has been called since he wrote, a transient 

 gastrula. 



Metschnikoff (Zeitschr. Wissen. Zool., vol. xxxn, 1879, pi. 33, 

 f. 1-3) gives the earliest segmentation stages of Ascetta primor- 

 dialis among the Ascones, and shows that they closely resemble 

 the same stages in other forms. The ova have first two then 

 four and finally seven cells originating through vertical segmen- 

 tation in a placula. The subsequent stages of the multicellular 

 larva confirm Oscar Schmidt's observations upon Ascetta primor- 

 dialis, and Ascetta clathrus. Schmidt (Arch. Mikr. Anat., vol. 

 xiv, pi. 15-16) describes the free larva in these species as having a 

 single layer of ciliated cylindrical cells, which change at one pole 

 into a few, large, granulated cells. These, after losing their cilia, 

 become rounded and amoeboid in aspect, and, making their way 

 inwards as wandering cells, give rise to the endoderm. Notwith- 

 standing the distinguished author's opposition to the gastrula, it 

 is possible that this inwandering of differentiated cells may be a 

 primitive stage in the formation of the gastrula. Schmidt and 

 Metschnikoff both describe two kinds of cells in the parenchy- 

 mula, — the set described above derived from the inwandering 

 cells and another of smaller cells arising by delamination from 

 the internal parts of the endoblast or the ectoblast cells. The 

 habit of building by delamination is a common one in the amphi- 

 blastula of sponges, and also the free production of granules by 

 disintegration of cells. When Metschnikoff (op. cit., p. 364.), 

 therefore, supposes the small and clearer cells derived from delam- 

 ination to be the element of the future endoderm and the large 

 granular wandering cells to be the generators of the future mes- 

 oderm, it is evident that one can just as readily consider the 

 smaller cells as destined to build up the mesoderm and the wan- 

 dering cells as the probable generators of the future endoderm, 

 as they were first considered by their discoverer, Schmidt. This 

 would appear reasonable, if, as we have suggested above, the for- 



