28 HOLASCUS EDWARDSII. 



with only one end-ray being usually associated with bifurcate, bifurcate with 

 trifurcate, and trifurcate with quadrifurcate ones. In the simple rays, that is 

 in those consisting of a main-ray and a single end-ray, the point of demarcation 

 between main- and end-ray is clearly marked by a thickening of the distal end 

 of the former (Plate 19, figs. 4, 5). I consider this thickening a rudiment of the 

 other, reduced, end-rays. The end-rays arise steeply from the main-ray, but 

 immediately curve outward, and are, farther on, usually nearly straight. Occa- 

 sionally the proximal part of the end-ray, beyond the basal curve, is irregularly 

 bent. The end-rays are conic, gradually attenuated to a fine point, 57-83 n 

 long and 1.6-2.8 ^ thick at the base. The bases of the simple rays have the 

 same thickness as the main-rays. All parts of the spicule are perfectly smooth 

 (Plate 19, fig. 11). 



The graphiocomes (Plate 19, figs. 12, 13) have main«rays which enclose 

 angles of 90° with their neighbours and are equal in the same spicule. The 

 main-rays are 11-17 m long and 2.5-3.5 ti thick. The single end-ray brush 

 measured was 15 m long. 



The ring-shaped sigms (Plate 19, figs. 14—21), which, as above stated, may 

 be skeletal structures foreign to the sponge, are rods, 1-2 ju, rarely 2.8 ti, thick in 

 the middle, attenuated at both ends to fine points, regularly and uniformly curved 

 so as to form a whole low spiral turn or, more rarely, a part of such a spiral. 

 Lying flat they usually appear as circular rings with an interruption at one point. 

 The rings formed by them are 17-57 pt in diameter. The ends are usually simple 

 and sharp-pointed (Plate 19, fig. 14) ; rarely they bear on the concave side one 

 or two small, cylindrical, terminally rounded spines (Plate 19, fig. 18). Near 

 the middle of the rod a slight irregularity can usually be made out, but this does 

 not appear to be a thickening which could with any probabihty be considered 

 as the rudiment of another ray. 



Although the upper end is missing in all the specimens and it must therefore 

 be left undecided whether they possessed terminal sieves or not, I think that the 

 want of parietal apertures, the spiculation, and the other characters described 

 above show clearly that they belong to the genus Holascus. From the nine 

 hitherto described species of this genus they differ by possessing ring-shaped sigm 

 microscleres. Since, however, these ring-sigms may not be homologous to the 

 sigms of H. fibulatus, but foreign to the sponge, I shall not consider them in the 

 following systematic discussion. 



Apart from this, Holascus edwardsii differs from five of the nine Holascus 

 species by the absence of calicocomes. Of the remaining four, one, H. undulatus, 



