81 6 w. j. sollas on tiie structure and 



with it and normal to the direction of the fibres themselves: some- 

 times this curve is made more apparent by the presence of a number 

 of accompanying minute granules ; and it appears to indicate a line 

 of growth. The fibrous tufts, or hemispherical bosses, do not, 

 however, nearly fill up the whole of the interstice, but always 

 leave a larger or smaller space in the middle, which is also filled 

 with crystalline silica, giving either a granular colour-pattern with 

 polarized light, or a fibrous one — but usually the latter, in which case 

 these crystalline fibres are bolder than those of the hemispherical 

 bosses. Here, when the fibres from two adjacent centres of crystal- 

 lization meet and oppose one another, they terminate abruptly in a 

 sharply defined straight line of demarcation ; and if the adjacent and 

 opposing groups be more than two in number, the respective lines 

 of demarcation intersect each other at angles of 60°, and hence they 

 may be regarded as sections of the sides of incompletely formed 

 crystals of quartz. The fact that we find solid siliceous spicules in 

 the uninfiltrated network, and mere empty spicular casts when the 

 interstices are filled with silica, seems to me very significant, and 

 certainly suggests the idea that the spicules have to some extent 

 furnished the silica with which the sponge has become mineralized, 

 and thus, up to a certain stage at least, the sponge has fossilized 

 itself. 



3. Specimen of S. costata from Wilts. — This specimen is solid 

 throughout, the canals are filled with quartz sand, glauconite grains, 

 and other foreign bodies, cemented together by crystalline silica. 

 The interstices of the skeleton are filled up by transparent and crys- 

 talline silica ; and the spicules have become absorbed, leaving only 

 hollow casts in their place. These casts are lined internally by 

 white opaque material, and frequently contain certain curious black 

 linear bodies or acerate spicula, which, on treatmeut with nitric 

 acid, lose their dark colour, turn faintly yellowish, and become per- 

 fectly transparent, while the resulting acid solution jdelds a blue 

 precipitate with potassium fcrrocyanide : their composition would 

 thus appear to be mainly siliceous, while their black colouring- 

 matter consists of some salt of iron. It is just possible that these 

 spicules may be the remains of some Cliona-like sponge, which 

 entered the Siphonia some time after its death, and specially in- 

 habited the enlarged axial canals in the spicules of its skeleton. 



The exterior of the arms of the skeletal spicules is sometimes 

 covered with a number of hemispherical bosses, very regular in 

 shape, and with a sharply defined contour. If these were integral 

 parts of the original spicules, they would remove S. costata from its 

 alliance with S. pyriformis and the recent species Discodermia 

 p oltj discus, and place it in some other group of Lithistids. I have 

 consequently given great attention to their examination, and find, 

 first, that they are inconstant : in some cases a group of spicules 

 does not exhibit a single one ; in others the spicules are covered with 

 them, and occasionally so thickly that the bosses appear piled one 

 on another in thick clusters. IS T cxt, the walls of the spicular casts 

 are subject to other, though related, peculiarities, sometimes be- 



