terminations (Pl. III. fig. 2); but now and then these terminations are wanting, and the quadrilateral meshes from which they spring lie level or nearly level with the surface. From the spinose ends, or from the quadrilateral meshes, an exceedingly fine network of delicate, glassy, pullulating fibrelets is given off, which fills up the interstices of the secondary rete (PÍ. III. fig. 4, Pl. IV. figs. 4, 5, 6, 7, Pl. V. fig. 4); frequently it is wholly irregular, but in numerous instances exhibits the true sexradiate arrangement. Its meshes and fibrelets vary in size, the average measurement from node to node being to inch, and the diameter of the fibres 1500 to 300. Thus the latter are, as a rule, not appreciably thicker than the spines of the secondary rete: and this suggests that some of these spines may be, after all, nothing but the attached parts of fibrelets, which have been broken off or dissolved away; and often a series of gradational forms can be traced, proving that some are of this nature; but many, from their smooth sides, regular conical form, and abundance in places free from fibrelets, must, as we have already considered, be true spines.

From the minuteness and proximity of the sexradiates one would conclude that they have been coated merely with a thin film of siliceous material, or are only soldered together at their ends; and the same characters would also lead us to infer that they do not afterwards come to form a part of the interior skeleton, but remain as a surface-coating, which must be regarded as an aftergrowth creeping over the oscules of the anterior plate, as this becomes overgrown by the posterior

mass behind.

Though this network is in general collected only about the ends of the radii from the secondary rete, beneath or between the meshes of this rete, it yet also happens occasionally, especially near the base, that it accumulates in patches to a much greater extent, burying up the network below, so as to completely conceal it from sight (Pl. IV. fig. 4), and forming a low but distinct mound above the general surface, and even, in one case, producing a series of rounded ridges (Pl. I. fig. 2, Pl. II. fig. 2, r) which pass straight across the anterior face of the oscular plate, horizontally from one side to

the other.

The superficial network, where it covers up the oscules, descends some distance into the excurrent canals, as may be well seen by breaking a specimen across the oscular plate, etching the fractured surface, and then examining it by reflected light. The skeleton-fibres, with their projecting spines, are then exposed; the superficial network is seen

covering over the oscular opening, and giving off one or two pendent processes into the excurrent canal; and, moreover, the skeletal fibres which surround the canal are also produced into outgrowths of delicate reticulation and irregular fibres which straggle across the canal from side to side (Pl. III. fig. 2); and the tendency of the fibre to pass into secondary growths thus manifested is carried so far that, even in the normal smooth network not immediately surrounding the canal, an occasional spine puts in an unexpected appearance.

The superficial network does not frequently occur over the posterior mass; and its rarity in this position appears, in some cases, to result from the wear and tear to which a convex surface like that of the posterior mass is especially exposed; in other cases it is due to a less favourable state of fossilization than obtains in the anterior plate; while in others still it would appear to be absent because the posterior surface has never been furnished with it, which last, indeed, is only what one would expect on the view that the posterior mass is an aftergrowth which increases behind while the aftergrowth of fine network is extending itself in front. It is only when the posterior mass has, like the attached anterior plate, ceased to grow, or, at all events, when its growth has for a time been arrested, that one would expect to find a final overgrowth of fine network on its surface. Such a laver I have met with in one case only, though whether it is, in this particular instance, exceptionally produced or exceptionally preserved, is of course impossible to say. This network, under a magnifying-power of 50 or 60 diameters, appeared to be without a sexradiate arrangement, its meshes not having any very regular form, and each of its fibres seemed to be pitted or perforated with a number of minute holes (Pl. V. fig. 1); but when a power of from 100 to 140 diameters was applied, it was found that these minute holes were the intermeshes of a delicate net, and that each fibre was itself a complex reticulation of exceedingly delicate fibrelets (Pl. V. fig. 2), which, where most perfectly preserved, showed a regular sexradiate disposition, with nodes distant 1250 to 2006 inch from each other, and fibre 150 to go inch in diameter. Where a sexradiate arrangement could not be detected, the defect appeared to be owing to the disappearance of some of the fibrelets necessary to the arrangement, by solution or otherwise. The cylinders of network exhibit sometimes a central axis of solid fibre from which the finer rete is given off all round; and sometimes they pass into a solid fibre ornamented with projecting fibrelets—a transformation apparently due to the fusion of the compound network-fibre into a solid one by the further deposition of

siliceous matter. Between the open meshes of this most exquisite net (which, in the delicate and complex tracery of its transparent fibres, surpasses almost every thing I have seen. amongst the Hexactinellida) one observes either an intermesh perfectly open and leading to the interior of the skeleton, or else a multitude of minute glistening fibrelets, which pass from fibre to fibre of the secondary rete below, and weave across its meshes a transparent vitreous web (Pl. V. fig. 3). The secondary rete passes in its turn into the skeletal network below, which, at first beautifully spined, soon becomes, as it leaves the surface, perfectly smooth.

The foregoing facts could be observed by examining the surface of the etched fossil by reflected light; but by splitting off a few fine chips with a scalpel, treating them with acid in a watchglass, washing with distilled water, and finally drying, the network could be obtained in a state fit for mounting in Canada balsam and other media, and for observation with transmitted light.

Traces of the network with complex fibres may be detected along the lateral edges of the oscular plate in the specimen where it occurs; but further on, over the anterior face, it quite vanishes, and only the ordinary superficial reticulation prevails (Pl. V. fig. 4).

Flesh-spicules. The perfect manner in which the superficial network is preserved led me to think that some rosettes or other flesh-spicules might perhaps be seen in the sponge; and the most likely places to look for them appeared to be, first, in the residue set free in suspension on treating the fossil with acid, and, next, in the open meshes of the skeleton. A careful examination of the former proved altogether unsuccessful, while in the latter iron pyrites was observed under a variety of forms. In this there was hope, since I have slides showing minute coccoliths and delicate radiolarians perfectly preserved in this material: therefore I made a long search in the expectation of finding some form of iron pyrites which should display evident traces of the rosette form; but, with a few very unsatisfactory exceptions, my search was quite in vain. The flesh-spicules of the Hexactinellidæ have yet to be found in the fossil state.

Other Spicules.-I have, however, met with two spicules other than sexradiates in this fossil. One is a completely erectly spined cylindrical form (Pl. V. fig. 5) with one part hidden in the network, from which the other portion projects freely, making an acute angle with the oscular surface as it points upwards from the base. This spicule bears

a close resemblance to that figured by Bowerbank * from Aphrocallistes (Iphiteon, Bk.) beatrix.

The other spicule occurs in a parallel section of the oscular plate, as a cast, partly hollow, partly filled with iron pyrites; it is simple, not spined, terminates so obscurely that its ends cannot be made out, and is imbedded in skeletal fibre in company with the ordinary sexradiate spicules (Pl. V. fig. 5).

Modes of Fossilization. The fossilizing material is usually crystalline transparent carbonate of lime, or calcite, which fills up the meshes of the network, and occupies the sexradiate canals of the siliceous fibre; where it occurs in large quantity, as in the meshes and excurrent canals, it is traversed by numerous cleavage-planes; and it is usually impure from the presence of a little aluminous matter. The fibre thus enclosed consists of silica, and in a few cases is almost as homogeneous and purely siliceous as when it existed in the living state; but even in this, its most perfect condition it generally exhibits the marks of decay, not only by the absorption of its interior spicule, but in the presence of numerous hemispherical pits excavated from its exterior to various depths, like those described by Carter as affecting recent spicules †; from this condition it soon passes through a series of changes, the final result of which is to leave it wholly converted into carbonate of lime. The first step in the process is a granulation of the fibre about the internal canal, which soon extends itself, chiefly by eating its way from within outwards, till at length it reaches the outer boundary of the fibre; and this, which during the process of change has retained its definite outline, often its transparency as well, yields at last, and the fibre becomes. granular all through. The granulation, however, also frequently appears at the outside and the inside of the fibre at once, and proceeds from each direction till it meets in the interior. While the granulation is thus progressing, a process of absorption is set up about the interior canal, accompanied by a replacement of the fibre in carbonate of lime; this change takes place from within outwards, and continues till at length. a mere shell of rounded granulations of silica separates the calcite without from that within the fibre; finally this shell itself disappears, and the exterior and the interior calcite become one. But even then, with this extreme mineralogical change, the original structure is not obliterated: the calcite which fills the internal canal and the interspaces of the meshes is transparent and usually colourless, or with a faint yellowish

Proc. Zool. Soc. 1869, pl. xxii. fig. 9.

+ Ann. & Mag. Nat. Hist. ser. 4, vol. xii. p. 457.

tinge; while that which replaces the siliceous fibre is, by reflected light, of a milky blue colour, and by transmitted light brownish, less transparent, and granular with dark spots. And thus while the fundamental spicule has become absorbed, and its hollow cast filled with crystalline calcite, and the same material has replaced the siliceous fibre and the sarcode between the meshes-while, in fact, the whole of the metamorphosed net consists of one material, carbonate of lime, the structure is yet left as definitely recorded as in a sponge with its natural composition only just dead; and from this striking fact is forced upon us the conclusion that in determining the characters and affinities of fossil sponges, the mineral composition is an argument of but fifth-rate value, and the form and structure here, as in most other anatomical questions, is the one thing important.

It frequently happens that while the sponge towards the exterior is preserved in calcite, it is fossilized with silica in the interior; and between these two conditions one can often trace a series of transitional changes. Thus in one specimen the sharp outline of the siliceous fibre soon disappears as it proceeds inwards, and is replaced by a botryoidal surface of hemispherical bosses (p. 18. fig. 6, a; p. 19. fig. 7, a), each with a corresponding cavity on the inside; from the botryoidal exterior a fibrous crystallization of silica radiates towards the middle of each intermesh *, filling it up; the interior of the fibre, on the other hand, is occupied with clear transparent calcite exhibiting cleavage-planes, and the sexradiate canal is filled with silica, crypto-crystalline, and exhibiting patches of colour when polarized light is passed through it. Thus the original siliceous spicule is, after a cycle of changes, restored again to the siliceous state. And here one may notice the very important fact that these pseudomorphic spicules are not continuous with each other, but remain perfectly distinct, with their rays overlapping, precisely as they do in Farrea and Aphrocallistes (fig. 5, a). In one or two instances (fig. 5, b) four spines equally distant from each other have been noticed surrounding the proximal end of each ray, and pointing towards the centre of the spicule-thus indicating that in these cases a hollow process, now converted into a spine, once proceeded from the central canal and entered the thickening of fibre which fills up the angles at the nodes of the network. If, as might easily happen, these canals underwent an extension so far into the thickening as to meet one another, and become continuous, we should have a structure singularly homoplastic with that of

* "Intermesh," the space included between a mesh..

Ann. & Mag. N. Hist. Ser. 4. Vol. xix.

2