REPORT ON THE HEXACTINELLIDA. 147 



sheath of finely spinous oxyhexacts (PL LVI. fig. 12), in wliich the radial ray, directed 

 towards the centre, projects into the lumen of the gastral space, while the opposite radial 

 ray intrudes into the subgastral space, and the tangential rays, intersecting at right 

 angles, lie in the gastral membrane (PL LVI. fig. 2). 



The medium-sized oxydiacts, which project freely from 3 to 5 mm. outwards and 

 upwards beyond the external surface, are unfortunately for the most part broken. They 

 occur somewhat irregularly and are not ver}^ numerous, being probably to a large extent 

 lost. I have not discovered any pcntact prostalia similar to those which occurred in 

 Rossella, and must therefore conclude that they are absent. 



2. Acanthascus duhius, u. sp. (PL LVIL figs. 8-13). 



To the south of Puerta Bueno, in Patagonia (Station 310, lat. 51° 27' 30" S., long. 

 74° 3' W.), from a depth of 400 fathoms and a blue- mud ground, a Hexactinellid was 

 trawled, of which unfortunately only the basal part, which had grown on the ramifications 

 of a colony of Lopliolielia, was preserved, and that imperfectly. What remains is about 

 the size of a man's fist, and represents the basal portion of a cup-shaped sponge, which 

 in its entirety must have been about four times as long. The remnants of the external 

 surface still preserved seem to be approximately smooth, but it is impossible to decide 

 as to the presence or absence of elevations with freely projecting spicules. On the 

 interior gastral surface of the wall (2 to 3 cm. in thickness), there are large round 

 apertures, 3 to 8 mm. and more in diameter, — the openings of the efferent canals into the 

 gastral cavity. The latter must have been at least 5 cm. wide. 



The tissues are a good deal macerated, so that the skeleton is somewhat loose, and would 

 have altogether fallen to pieces were it not that the larger spicules of the parenchyma 

 are thoroughly bound together by means of synapticula and siliceous cementing masses. 

 The somewhat thick parenchymal skeleton consists of numerous oxydiacts, either isolated 

 or in strands, and often bound together as above noted. They vary in length, but 

 attain no considerable thickness. Especially in the lower portions of the sponge which 

 had grown on the Lophohelia, the parallel, straight, rod-like spicules are very thoroughly 

 bound together by numerous transverse synapticula (PL LVIL fig. 12). The terminal 

 portions of these spicules are usually tuberculated or rough, and are either gradually 

 narrowed or conicaUy pointed, or else rounded ofi", and sometimes even somewhat clubbed. 

 The regular oxyhexacts which occm- here and there between the numerous long 

 diacts, are under medium size, either spinous or smooth, and in rare instances exhibit 

 thickened or tubercled central nodes (PL LVIL fig. 13). Here and there small smooth 

 oxyhexacts occur, with slender, straight rays, which do not exceed in size the numerous 

 rosettes about to be described. Among these there is a special abundance of oxyhexasters 



