92 



oxyhexasters first. The main-rays of these spicules are about 8 m long and 2 i^ thick 

 and bear on the margin of the thickened end 4, more rarely 5, 6 or 3 branch-rays. 

 The latter are somewhat bent outwards at the base, but further up they are quite 

 straight and considerably divergent. The branch-rays are 26-30 ^ long and gra- 

 dually attenuated to a sharp-pointed end (pi. XV, f. 5). If five branch-rays are 

 attached to each main-ray, one branch-ray of each of the six groups usually lies 

 In the continuation of the main-ray, whilst the four others diverge, forming, if 

 viewed from above, a cross around the central one. The simplest derivative of 

 this regular form is a hexaster in which two opposite main-rays are longer than 

 the four others. The branch-rays of the latter are reduced to three or two, or 

 they are absent altogether, these main-rays remaining simple (Hemibxyhexaster). 

 Many different forms of such irregular hexasters are met with, and it would lead 

 too far to describe them all. Some, however, are so frequent as to deserve a more 

 accurate description. Such are the hexasters with two differentiated, equal, 

 opposite, elongated main-rays each of which bears four or five branch-rays. If 

 five branch-rays are present, these are arranged as described above. The four 

 other main-rays remain simple, are about 80 /* long, pointed and straight (pi. 

 XV, f. 12), In other frequent forms the two opposite, differentiated main-rays 

 are unequal in length, one of them being so reduced that its branch-rays become 

 nearly sessile and concentric with the other main-rays. A considerable difference 

 in the dimensions of the two opposite, differentiated main-rays is also observed in 

 those hexasters which are termed " porrecto-multiradiate " by the English authors 

 and which have always been considered as typical of the species Aphrocallistes 

 heatrix J. E. Gray. These spicules are hemioxyhexasters, attain a length of 

 150 a* and are generally pretty stout. The four lateral main-rays, which are 

 vertical to the axis formed by the two differentiated ones, are usually unbranched, 

 whilst each of the two axial main-rays generally bear five branch-rays, one of 

 which lies in the continuation of the main-ray to which it belongs, that is in the 

 axis of the spicule (pi. XV, f. 6, 7). Sometimes the axial branch-ray is absent. 

 More rarely spicules are met with in which one of the lateral main -rays is divided 

 at the base into two straight diverging branches as long as the other, simple, 

 lateral main-rays. 



The terminal sieve-plate extends transversely and covers the entrance to the 

 cavity of the caliculate sponge. Sometimes a second transverse sieve-plate, situat- 

 ed further down, is met with. The skeleton-net supporting the sieve-plates has 

 polygonal meshes, rounded at the corners and 1-2 mm. wide. J. E. Gray has 

 figured it iu natural size and magnified 3 diameters {I.e. pi. XI, f. 1,2). These 

 drawings illustrate its macroscopic appearance very correctly. Microscopically 

 the skeleton-net and the isolated spicules of the sieve-plate do not essentially 

 differ from those of the body-wall, where however meshes of the skeleton-net 

 are a little smaller and more regularly hexagonal. 



