314 THE VOYAGE OF H.M.S. CHALLENGER. 



communication is established between two or more adjoining diverticula, while they are 

 united at the base into a common simple tube or are in open connection throughout 

 their length. If this union occurs between the diverticula belonging to the longitudinal 

 row, it may result in extreme cases in the formation of a longitudinal fold, on which 

 the individual diverticula are only indicated as short boss-like swellings. On the other 

 hand, the long diverticula W'hich occur here and there are sometimes forked, and in 

 this there lies the tendency to form branches. 



Where the upper terminal opening with its natural margin is preserved quite uninjured 

 it is closed, just as in Aphrocallistes beatrix, Gray, by a transversely stretched narrow 

 meshed lattice-like plate. The latter is usually somewhat concavely incurved and becomes 

 united to the honeycomb-like lateral wall in a com^^act, somewhat tuberculate margin. 

 As already reported by Oscar Schmidt and Marshall, several such thin lattice-like trans- 

 verse partitions usually occur in the interior of the tube, but I would call attention to 

 some points of distinction between these internal diaphragms and the terminal sieve-plate 

 of other Hexactinellids. While the narrow-meshed terminal sieve-plate of other Hexac- 

 tinellids is united all round to the body-wall so that (apart from the sieve-like meshes) 

 a complete closure of the tube results, in this case, a semicircular marginal portion of 

 the internal diaphragms remains unclosed wherever a lateral diverticulum opens into 

 the large lumen of the tube (PI. LXXXIII. fig. 2). With regard to the occurrence, 

 number, and arrangement of these transverse septa I have found great diflFerences. While 

 some specimens well preserved in other respects possess, apart from the terminal plate, 

 no trace of septa, others show three or more internal diaphragms, but no constant relation 

 to the whorls of diverticula can be recognised, so as to suggest the reduction of the 

 entire tube to series of metameres. I regard it as most probable that during the growth 

 of the tube a temporary provisional occlusion is effected by a transverse sieve-net, and 

 that only after growth has ceased is a terminal regularly constructed lattice-work formed 

 which entirely closes the lumen. While the latter consists of tolerably similar thick round 

 beams, varying from 0'3 to 0"5 mm. in diameter, which surround rounded polygonal meshes 

 of tolerably uniform size, and while freely terminating rays only project here and there 

 into the lumen of the meshes, the transverse septa in the interior of the tube have 

 a somewhat different character, inasmuch as they consist of beams of very various 

 thickness which meet one another to form a network at very diverse angles, in which the 

 mesh-spaces are not rounded but have sharp angles (PI. LXXXIII. fig. 2). 



The ihicroscopic structure of Aphrocallistes hocagei agrees essentially with that 

 of the corresponding skeletal parts of Aphrocallistes beatrix, as represented by Bower- 

 bank in his excellent figures.^ The dense network of beams which forms the di\4diug 

 septa of the six-sided prismatic meshes of the wall consists, as in the case of every 

 dictyonal Hexactinellid framework, exclusively of amalgamated hexacts. These do not 



' Proc. Zool. Soc. Loncl, 1869, pi. xxi. figs. 2-4, and pi. x.xii. fig. 2. 



