REPORT ON THE HEXACTINELLIDA. 149 



apex a tuft of divergent pointed spicules, which project freely for 2 cm. or more. The 

 general form of the body, and the occurrence of these spherical elevations at approxi- 

 mately equal intervals on the outer surface, and especially the tufts of spicules pro- 

 jecting from their apices, produce a close resemblance between this sponge and 

 certain prickly cactuses. I have for this reason given this form the specific designation 

 cactus. 



While the whole external surface of the body is covered with a hue meshed dermal 

 network, the simple, smooth internal surface is covered with a firm gastral mem- 

 brane, which is abundantly perforated by numerous round apertures, about 1 mm. in 

 breadth. 



The parenchymal skeleton exhibits numerous thickly distributed oxydiacts in 

 variable disposition. These are, for the most part, surrounded by oxydiact comitalia 

 of but slight development. Near the conically pointed ends these diacts are usually 

 somewhat rough, but are otherwise smooth. In the middle they are usually quite 

 smooth, though less frequently exhibiting a median thickening, or else four cruciate 

 or two opposite elevations. Besides the slender comital oxydiacts, numerous isolated 

 spicules of the same sort occur. 



I have not observed in the parenchyma any simple hexacts of medium size, but 

 only small, slender, isolated oxyhcxacts, resembling in size the rosettes about to be 

 described. 



There is an abundant occurrence of the familiar oxyhexaster forms with short principal 

 rays, and long, smooth, diverging terminals, which are here somewhat strongly developed 

 (PI. LVII. fig. 3). Between these, spicules occur of similar size and structure with 

 six undivided rays, each of which exhibits, at the same distance from the common 

 node of intersection as the point where the principals divide in the oxyhexasters, a 

 sharp flexure, succeeded by a small curve, and finally by a straight portion continued 

 on to the point. Spicules thus modified may be regarded as reduced derivatives of the 

 above described oxyhexasters. 



There is besides a not unfrequent occurrence of irregular rosettes, which seem at first 

 sight to be quite distinct from the Hexactinellid type of spicule, since they almost 

 always exhibit eight principal rays. These somewhat thick, but not exactly cylindrical 

 rays, which spring from a thickened central node, arc often beset with knot-like protrusions. 

 Each bears a bundle of four to eight thin straight terminals of equal length, which 

 diverge slightly from one another, without being sharply marked ofi' from the principals. 

 Each terminal ray bears on its extremity a minute disc (PI. LYII. figs. 4, 5 ; c/l 

 PI. LXV. fig. 3). I am of opinion that some of these apparent principals have arisen 

 by the splitting of real or primary principal rays. This is suggested, for instance, by 

 the fact that at their base they are not cylindrical, but somewhat convergent and 

 flattened. From the base of one ray a smooth link may be seen passing to the base of 



