ON A FREE-SWIMMING HYDROID. 13 



together into a thick rounded blob or drop, containing many- 

 nuclei (fig. 13), but this condition appears to be of rare 

 occurrence. Probably the nuclei multiply by division, as 

 indicated in fig. 14, at x. This peculiar tissue appears to 

 originate, in part at any rate, from the inner walls of the 

 endodermal canals.^ The mesogloeal portion of these walls 

 may be very thick, and occasionally little groups of cells (fig. 

 12, End. Bud) may be seen growing into it from the endo- 

 dermal lining of the canal. These cells have very finely 

 granular contents and small nuclei. Irregular cavities (fig. 

 12, D. F. G.) are apparently developed between them, 

 and gradually enlarge until the nuclei become widely 

 separated, while the mesogloea is reduced to thin sheets 

 separating adjacent cavities from one another, and the proto- 

 plasm of the endoderm cells becomes spread out over these 

 sheets in the form of a granular syncytium. 



Sometimes, where a comparatively thin layer of mesogloea 

 lies behind the endoderm of the inner wall of an endodermal 

 canal, threads of finely granular protoplasm may be seen 

 stretching at right angles through the mesogloea from the 

 one surface (covered by the finely granular syncytium) to the 

 other (covered by the endodermal cells of the canal wall). 



Thus it appears that the supporting membranes of the float 

 originate in a peculiar manner from the endoderm. It is not 

 certain, however, that they do not receive cells from the 

 external ectoderm also, for thread-cells in various stages of 

 development may sometimes be observed in places where the 

 mesogloea is thick, beneath the external ectoderm and doubt- 

 less derived from the latter. This inward migration of the 

 cnidoblasts can hardly be looked upon as normal, but if they 

 are able to migrate inwards it seems equally possibly that 

 other ectoderm cells may do the same, and possible eventually 

 take part in the formation of the supporting membranes. 



1 Professor Ray Lankester has pointed out to me tliat a somewhat similar 

 method of tissue formation lias been observed in the " laminar tissue" of 

 Amphioxus (vide Pouchet, "On the Laminar Tissue of Amphioxus," 

 ' Quart. Journ. Micr. Sci.,' vol. xx, u. s., p. 421, pi. xxix). 



