362 THE VOYAGE OF H.M.S. (JHALLENGEl^. 



;ire furnished with numerous small tubercles and pointed elevations. The portions of the 

 main beams, on the other hand, which lie within the space enclosed by the buttresses are 

 perfectly smooth, and are not inconsiderably thinner than the outer portions of the 

 beams (PI. CIV. fig. 3). As Marshall and Meyer have noted in detail in the Mittheil. 

 d. Zool. Museums, Dresden (1877), the oblique buttresses round about the nodes of 

 intersection are really spine-like processes of the main beams, growing out in a given 

 direction from the external surface of two adjacent beams, until they finally meet and 

 fuse with one another. That they are in fact simple spines is further demonstrated 

 by the fact, which Marshall and Meyer have emphasised, that they have no axial canals, 

 which can be distinctly detected, on the other hand, in all the beams of the main frame- 

 work, and even on those thin portions which are enclosed by the lantern -like arrangement 

 of the buttresses. 



It need hardly be noted that here also the entire quadrate lattice-work is built up of 

 single hexacts, in which the corresponding parallel rays are enclosed in a common 

 siliceous sheath, and thus united. 



Here and there, especially near the bounding surfaces, there are also simple, solid, 

 slightly thickened nodes of intersection in the dictyonal framework. 



The covering plate which encloses the sponge as in a capsule, apparently 

 arises from the outward bending of the tube-wall at right angles, or in a trumpet-like 

 curve. This is distinctly seen in the macerated specimen represented in PI. CIV. fig. 2. 

 In the two larger specimens it contains no connected skeleton, while in that trawled off" 

 Banda Islands (PI. CIV. fig. 2), it is supported over a large extent by a dictyonal frame- 

 work, which corresponds exactly to that of the tubular network, and is in fact a direct 

 continuation of the latter. 



In this form, better than in those obtained off" Little Ki Island, a double principal 

 canal can be seen, which seems to have opened with a free margin at the narrower end 

 (PL CIV. fig. 2). 



Sections through the wall of the tube show distinctly in many cases the structure 

 of the soft parts. The deeply folded chamber layer is connected to the dermal 

 membrane by a somewhat uniformly developed external trabecular framework, which 

 extends also into the afferent clefts and passages. The internal trabecular, on the other 

 hand, only extend between the sieve-like gastral membrane and the internally pro- 

 jecting folds of the chamber-layer (PL CIV. fig. 3), without being to any marked extent 

 continued into the eff'erent canals. The chambers themselves are comparatively small 

 and simply thimble-shaped. 



The capsule has an average thickness of 0'3 mm., and exhibits between the outer 

 and the inner porous limiting membrane numerous passages and vesicular cavities, but 

 in the Little Ki specimens, at least, no chambers. 



As to loose spicules, I found in the parenchyma between the chambers in the tra- 



