AFFINITIES OF THE GENUS SIPHONIA. 815 



fibres is too significant to be overlooked. A comparison of the two 

 shows a correspondence of a very exact kind : wherever the canal 

 expands, there the fibre becomes thicker; when the canal branches, 

 the fibre bifurcates with it, and both undulate together in common 

 curves : the two structures agree together, in fact, just like a mould 

 and a cast which has contracted after its first formation. The ex- 

 planation which therefore commends itself to my mind, and which 

 at least satisfies all the facts of the case, is, that the silica of these 

 threads infiltrated the canals while in the collodial state, and com- 

 pletely filled them up with a siliceous jelly; a subsequent and 

 gradual loss of water of hydration caused this jelly to become dry 

 and simultaneously to shrink till it attained its existing solidity and 

 dimensions ; finally, after a great lapse of time, the unstable colloid 

 passed into the crystalline condition. The possibility of the latter 

 transformation has been already proved by numerous observations 

 on the existing crystalline condition of fossil siliceous spicules ; and 

 any one who has experimented with colloidal silica must have been 

 struck with the wonderful amount of contraction which this sub- 

 stance experiences on desiccation, a contraction quite sufficient to 

 account for the shrivelling of the silica in the canals of the Haldon 

 SiphonicB. 



Network. — The interstices of the network are sometimes empty 

 spaces, and sometimes occupied wholly or in part with crystalline 

 silica. When empty, the spicules of the net retain their siliceous 

 composition, but do not exhibit the central hollow canal, which 

 frequently puts in an appearance in recent specimens of Lithistid 

 spicules ; they are solid throughout, and, since they give colours 

 with polarized light, have evidently exchanged their originally 

 colloidal state for the crystalline condition. In a single instance, 

 to which I have not been able to recur, a spicule was observed in a 

 section of one of these specimens with a smooth surface, and shining 

 with a vitreous lustre, like the spicule of a sponge only just dead ; 

 but in all other cases the spicules are excavated all over with small 

 hemispherical pittings, similar to those which Carter has described 

 as affecting the deciduous spicules of recent sponges ; they have a 

 bluish opalescent appearance and a feeble resinous or gum-like 

 lustre. 



When the interstices are filled with silica, the spicules are repre- 

 sented by empty casts, the walls of which are coated interiorly with 

 an opaque white substance, which appears black by transmitted 

 light. Frequently the cast has entirely failed to preserve the original 

 shape of the spicule, and the skeletal network presents the appear- 

 ance of a number of fragments of moss irregularly scattered through 

 a transparent ground of crystalline silica; these dendritic fragments, 

 however, are, as we have already stated, not solid, as they appear 

 to be by transmitted light, but merely hollow empty spaces. 



The silica of the interstices presents a fibrous crystalline arrange- 

 ment, the fibres generally radiating* from the spicules, or what 

 remains of them, in little tufts, each of which is defined at no great 

 distance from the centre of crystallization by a curve concentric 



