124 I. IJIMA : HEXACTINELLIDA. I. 



In connection Avitli tlie ectosome let me lieve say a word 

 about the endosome. I consider this as being in its fullest 

 development in those Hexactinellida, as, e.g., most Rosseliids, in 

 which the gastral skeleton is so highly developed as to form a 

 continuous latticework covering the inner apertures of the ex- 

 current canals. In such cases, the main spicules of the lattice- 

 work — the autogastralia — are generally hexactius disposed in much 

 the same way as the hexactin-dermalia of the Euplectellid 

 ectosome ; consequently, the eudosomal trabeculse connected with 

 the autogastralia likewise exhibits an arrangement more or less 

 similar to that in the Euplectellid ectosome. Now, in Eupledella, 

 as also in certain other genera, the gastralia are far too few to 

 form a continuous latticework; so that the inner apertures of the 

 excurrent canals remain perfectly open. Moreover, the gastralia 

 present are pentactins lacking the freely projecting proximal 

 ray ; and what here exists of the endosome between the excurrent 

 apertures, is represented merely by the trabeculie delimiting the 

 internal trabecular layer from the gastral cavity, veiy much in 

 the same way — mutatis mutandis — as the ectosome (dermal mem- 

 brane) is represented in those species in which the freely out- 

 standing distal ray is wanting to the dermalia. The said 

 trabecular (not excluding those of E. marshalli), though often 

 cobweb-like and indistinguishable from those more deeply 

 situated, are at places spread out into a more or less extensive 

 film-like membrane, a circumstance which makes the name 

 gadral membrane appear all the more applicable to them inasmuch 

 as they make up the lining of the gastral surface. The gastral 

 membrane is continued into the excurrent canals as the canalar 

 membrane. 



The iutertrabecular lacuna) underlying the gastral and the 



