437 
“ The wall of the sponge is trarersocl by numerous canals, which open on both sides 
of the wall, and extend inwards, either at right angles or obliquely to the surface of 
the sponge. The canals bifurcate or branch within the wall ; they do not penetrate 
through it, but terminate just below the opposite surface to that on which their oscules 
open. There is no definite arrangement in the course of the canals, which occasionally 
appear to anastomose within the vvall. Tlie canals are evenly bounded by the ordinary 
spicular tissue, and do not possess a special lining. The oscular apertures are either 
circular, ellijjtical, or trumpet-shaped, and in some instances divided by the extension 
over them of the spicular fascicles. 
“The skeletal mesh composing the substance of the sponge-wall is close, resistant, 
and intricate. It consists of slender fascicles or bundles of straight, elongate, linear 
spicules, disposed parallel with each other, and united together by a common coating 
of silica throughout the greater part of their length. These spicular fascicles have no 
definite arrangement; they overlap each other in different directions, though for the 
most part they are generally parallel with the sponge-wall. They are more abundant 
on the exterior surfaces, though they also traverse the interior of the wall in various 
directions. Between these fascicles there are rod-like spicules, either nearly parallel 
with each other or disposed irregularly, crossing and intercrossing each other in all 
directions. At their points of contact these spicules arc definitely fused together, and 
they are further united by the development of short transverse rods and siliceous pro- 
cesses which connect adjoining spicules, and thus form a strong and complex spicular 
mesh-work with irregular interspaces. Here and there in the interspaces of the mesh 
are solitary regular hexactinellid spicules, their rays being fused with the adjacent 
mesh-work, and in many places also there are small groups of minute regular 
hexactinellid spicules, intermingled closely together without any order, and their rays 
fused at the points of contact with each other. 
“There is no apparent specialised dermal layer. On both surfaces of the sponge- 
wall the spicular fascicles extend over the oscular vents, and the spaces between the 
fascicles exhibit mueh the same spicular character as the interior of the wmll. Further, 
the canals are not lined by any distinctive spicular tissue. 
Ols. “ Mr. H. J. Carter has truly remarked of this genus that its minute spicular 
structure is totally different from that of any other hexactinellid sponge. Its peculiar 
characters render its relationship somewhat doubtful. At first sight, indeed, one would 
hardly recognise it as a hexactinellid sponge, since only the fascicles of linear spicules 
are exposed on its surface, and the intermediate scaliform or iiregular spicular tissue 
more resembles a combination of monactinellid than hexactinellid spicules. The pre- 
sence, however, of definite six-rayed hexactinellid spicules, cither singly or forming 
small groups in the interspaces of the mesh-w'ork, at once determines its relations to 
this order. Its position within the order is less easy to define. The entire absence of 
any regular arrangement of the spicular mesh, or of any regular tissue formed by the 
fusion of the rays of regular hexactinellid spicules with the corresponding rays of 
adjacent spicules, removes the genus from the Hictyonine Group of the Ilexactinellidse. 
fhe definite fusion of the sjiicuhir rays with each other, thus binding them firmly 
together, is, on the other hand, a character which would tend to exclude this genus from 
the Lyssakine Groupi, in which, as a rule, the spicules are only held in position by the 
®oft animal structures of the sponge. Notwithstanding this fact, the general 
spicular characters of Piirisiphonia appear to me to indicate a closer relationship to the 
Lyssakina than to the Dictyoniua. 
“ Thus the fascicles of sjncular rods may be compared with those of the existing 
genus Huplectella, Owen, though they are not continuous as in the recent form, and a 
