New individuals are produced from the union of ova and spermatozoa, which develop from wandering amoeboid cells in the mesoderm. The walls of Ascetta are strengthened by calcareous scleres, more especi ally designated as spicules, which have the form of tri-

radiate needles. If we make abstraction of these we obtain an ideal sponge, which Haeckel has called Olynthus (6), and which may be regarded as the ancestral form from which all other sponges have been derived. To give greater exactness to our abstraction we should perhaps stipulate for the Olynthus a somewhat thicker mesoderm and more spherical form than a decalcified Ascon presents.

Canal System.— We shall now trace the several modifications which the Olynthus has undergone as expressed in the different types of canal system.

The simple paragaster of Ascetta may become complicated in a variety of ways, such as by the budding off from a parent form of stolon-like extensions, which then give rise to fresh individuals, or by the branching of the Ascon sac and the subsequent anastomosis of the branches; but in no case, so long as the sponge remains within the Ascon type, does the endoderm become differentiated into different histological elements. The most interesting modification of the Ascon form occurs in Homoderma sycandra (12 ), in which from the walls of a simple Ascon caecal processes grow out radiately in close regular whorls, each process reproducing the structure of the parent sponge (figs. 2, 3). From this it is but a short step to the important departure which gives rise to the Sycons.

In the simplest examples of this type the characters of Homoderma sycandra are reproduced, with the important exception that the endoderm lining the paragastric cavity of the original Ascon form loses its primitive character, and from a layer of flagellated cells becomes converted into a pavement epithelium, not in any distinguishable feature different from that of the ectoderm. The flagellated cells are thus restricted to the caecal outgrowths or radial tubes. Concurrently with this differentiation of the endoderm a more abundant development of mesoderm occurs. In some Sycons (Sycaltis, Hk.) the radial tubes remain separate and free; in others they lie close together and are united by trabeculae, or by a trabecular network, consisting of mesodermic strands surrounded by ectoderm (fig. 4). The spaces between the contiguous radial tubes thus become converted into narrow canals, through which water passes from the exterior to enter the pores in the walls of the radial tubes. These canals are the “intercanals” of Haeckel, now generally known by their older name of incurrent canals. The openings of the incurrent canals to the exterior are called pores, a term which we have also applied to the openings which lead directly into the radial tubes or paragastric cavity; to avoid ambiguity we shall for the future distinguish the latter kind of opening as a prosopyle. The term “pore” will then be restricted to the sense in which it was originally used by

Grant. The mouth by which a radial tube opens into the paragaster is known as a gastric ostium. In the higher forms of Sycons the radial tubes no longer arise as simple outgrowths of the whole sponge-wall, but rather as outgrowths of the endoderm into the mesoderm, which, together with the ectoderm, exhibits an independent growth of its own; and this results in the formation of a thick investment, known as the cortex (fig. 5), to the whole exterior of the sponge. The radial tubes may branch, Heteropegma (fig. 4). If the branches are given off regularly, as the radial tubes were in the first plan, and if at the same time the original radial tube exchanges its flagellated for a pavement epithelium, a structure as shown in fig. 6 (Polejna connexiva, Pol.) will result. This form might also be brought about by unequal growth of the gastral endoderm leading to a folding of the inner part of the sponge-wall. Very little direct evidence exists as to which of these two plans has actually been followed. Phylogenetically the transition from a simple Ascon to the most complicated Sycon can be traced step by step; and ontogeny shows that such a Sycon form as Grantia raphanus passes through an Ascon phase in the course of its larval development.

Returning to the ancestral form of sponge, Olynthus, let us conceive the endoderm growing out into a number of approximately spherical chambers, each of which communicates with the exterior by a prosopyle and with the paragastric cavity by a comparatively large aperture, which we may term for distinction an apopyle ; at the same time let the endoderm lose its flagellated character and become converted into a pavement epithelium, except in the spherical chambers. Such a form, called by Haeckel “dyssycus,” may be more briefly named a Rhagon from the grape-like form of its flagellated chambers, which differ from those of a Sycon both by their form and their smaller dimensions. The Rhagon occurs as a stage in the early development of Plakina monolopha (Schulze) and Reniera fertilis (9 ) (fig. 7); a calcareous sponge which appears to approach it somewhat is Leucopsis pedunculata, Lfd. By the folding of the wall of a Rhagon, or by its outgrowth into lobes, a complicated structure such as that of Plakina monolopha (20) (see fig. 26 f ) results. This is characterized by the chambers retaining their immediate communication with the incurrent and excurrent canals, opening into the latter by the widely open apopyle and receiving the former by one or several prosopyles. This may be termed the eury- pylous type of Rhagon canal system. The fold ing of the sponge-wall may be simple, as in the example given, or too complex to unravel. In higher forms of sponges (Geodinidae, Stellettidge) the chambers cease to open abruptly into the excurrent canals: each is prolonged into a narrow canal, aphodus, or abitus, which usually directly, sometimes after uniting with one or more of its fellows, opens into an excurrent canal. The prosopyles, now restrict ed to one for each chamber, may remain unchanged in character, or at the most be prolonged into very short tubes, each a prosodus or aditus (fig. 8). This may be termed the aphodal or racemose type of Rhagon system, since the chambers at the ends of the aphodi radiating from the excurrent canal look like grapes on a bunch. As Haeckel, however, has used “racemose” in a different sense, we shall adopt here the alternative term. By the extension of the prosodal or adital canals into long tubes a still higher differentiation is reached (fig. 9). This, which from the marked presence of both prosodal and aphodal canals may be termed the diplodal type of the Rhagon canal system, occurs but rarely. Chondrosia is an example.

The following scheme will render clear the foregoing distinctions :—

1. Ascon type: simple, ex. Ascetta, Hk.; strobiloid, ex. Homo derma, Lfd.

2. Sycon type: simple radial tubes, ex. Sycetta, Hk.; branched radial tubes (cylindrical chambers), ex. Heteropegma, Fl.; chamber-layer folded, ex. Polejna, Pol.

3. Rhagon type: eurypylous, with several prosopyles to each chamber, ex. Spοngelia; with a single prosopyle to each chamber, ex. Oscarella, Thenea ; aphodal, aphodal canals well developed, ex. Geodia, Lmk.; diplodal, with both aphodal and prosodal canals well developed, ex. Chondrosia, O.S.

In the case of the calcareous sponges Polejaeff has argued forcibly that the eurypylous type arises directly from the Sycon and not from the Rhagon. It is therefore doubtful how far the Rhagon in other sponges is a primitive form derived directly from an Olynthus, or whether it may not be a secondary larval state resulting from the abbreviated development of a former Sycon predecessor. Whatever may have been its past history, the Rhagon serves now at all events as a starting-point for the development of the higher forms of canal system.

In the higher Rhagons, as in the Sycons, further complications ensue, owing to an independent growth of the external ectoderm and the adjacent mesoderm. While the endoderm, with its associated mesoderm, is growing out or folding to form the excurrent canal system, the superficial mesoderm increases in thickness, and the ectoderm, extending laterally from the sides of the incurrent sinuses, burrows into it, parallel to the surface of the sponge. Thus it forms beneath the skin (i.e ., the layer of superficial mesoderm and investing ectoderm) cavities which may be either simple and spacious or be broken up into a number of labyrinthine passages by a network of mesoblastic strands (invested with ectoderm) which extend irregularly from roof to floor of the chamber. These cavities are known as subdermal chambers.

With the appearance of subdermal chambers the sponge becomes differentiated into two almost independent regions, an outer or ectosome and an inner or choanosome, which is characterized by the presence of flagellated chambers. The ectosome forms the roof and walls of the subdermal chambers, and is in its simplest form merely an investing skin; but in a large number of sponges it acquires considerable thickness and a very complicated histological structure. It is then known as a cortex. The thickening which gives rise to a cortex takes place chiefly beneath those parts of the skin which are not furnished with pores. Beneath the pores—in this case collected into sieve-like areas—dome-like cavities are left in the cortex; they open freely into the subdermal cavities below and their roof is formed by the cribriform pore membrane above. In many sponges (Geodia, Stelletta) the cortical domes are constricted near their communication with the subdermal cavity (subcortical crypt) by a transverse muscular sphincter, which defines an outer division or ectochone from an inner or endochone (fig. 10), the whole structure being a chone. The endochone is frequently absent (fig. 10). The early development of the cortex has scarcely yet been studied. In Stelletta phrissens (Soll.), one of the “Challenger” Stel lettidae, an early form of the sponge (fig. ll), shows the choanosome already characteristically folded within the cortex, which forms a com plete not-folded envelope around it. The roots of the incurrent sinuses form widely open spaces imme diately beneath the cortex and are the rudiments of subcortical crypts. Again, in some sponges a part of the endoderm and asso ciated mesoderm may like wise develop independ ently of the rest of the sponge, as in the Hexac tinellida, where the choa nosome forms a middle layer between a reticulation of ectosome on the one side and of endoderm and mesoderm, i.e ., endo some, on the other. Fin ally, the attached or lower half of a Rhagon may de velop in an altogether dif ferent manner from the other or upper half, the endoderm not producing any flagellated chambers. In this case the upper portion alone is characterized by the flagellated chambers, which are the distinctive mark of a sponge, and hence may be called the spongomere ; the lower half, which consists of all three fundamental layers, may be called the hypomere.

The form and general composition of sponges are exceedingly various and often difficult to analyse, presenting, along with some important differences, a remarkable general resemblance to the Coelentera in these respects. Like them, some sponges are simple, and others, through asexual multiplication, compound. The only criterion by which the individual sponge can be recognized is the osculum; and, as it is frequently difficult, and in many cases impossible, to distinguish this from the gastric opening of a large excurrent canal, there are many cases in which the simple or compound nature of the sponge must remain open to doubt. The oscule may also fail (lipostomosis), and so may the paragastric cavity (fipogastrosis); the problem then becomes insoluble. The loss of the oscule may in some cases be due to the continued growth of several endodermal folds towards the exterior, with a corresponding absorption of the mesoderm and ectoderm which lie in the way, till the folds penetrate to the ectoderm and open at the exterior, thus giving rise to excurrent openings, which are not readily distinguishable from pores. At the same time the original osculum closes up and entirely disappears. Lipogastrosis, on the other hand, may be produced by the growing together of the roots of the choanosomal folds, thus reducing the paragastric cavity to a labyrinth of canals, which may easily be confounded with the usual form of excurrent canals. While in some sponges the original oscule is lost, in others secondary independent openings, deceptively like oscules, are added. This pseudostomosis is due to a folding of the entire sponge, so as to produce secondary canals or cavities, which may be incurrent (vestibular) or excurrent {cloacal), the opening of the latter to the exterior being termed a false oscule or pseudostome. The faulty use of the term oscule for what is neither functionally nor morphologically a mouth is here obvious, for in one sense the oscule is always a pseudostome; it would be better if the term pseudoproct could be substituted.

Skeleton.— All sponges, except three or four genera belonging to the Myxospongiae, possess some kind of skeletal structures. They may be either calcareous or silicious or horny scleres, the latter usually having the form of fibres, which sometimes enclose silicious needles (spicules) or foreign bodies introduced from without. Foreign bodies also contribute to the formation of the skeleton of some silicious sponges, and occasionally form the entire skeleton, no other hard parts being present.

Mineral scleres usually occur in the form of spicules. The spicules of calcareous sponges consist of carbonate of lime, having the crystalline structure and other properties of calcite (29 ). Each spicule, so far as its mineral component is concerned, is a single crystal, all the molecules of calcite of which it is built up being similarly oriented. On the other hand, its form and general structure are purely organic. Its surfaces are always curved, and usually it has the form of a cone or combination of cones, each of which consists of concentric layers of calcite surrounding an axial fibre of organic matter,—probably of the same nature as spongiolin or spongin, the chief constituent of the fibres of horny sponges. A thin layer of organic matter, known as the spicule sheath, forms an outer investment to the spicule and is best rendered visible as a residue by removing the calcite with weak acid. Silicious spicules consist of colloid silica or opal, and hence can be distinguished from calcareous by having no influence upon polarized light. Structurally the two kinds of spicules present no important difference. The spicules of different sponges differ greatly both in form and in size. They may be conveniently divided into two groups,—minute or flesh spicules, which usually serve as the support of a single cell only {microscleres), and larger or skeletal spicules, which usually contribute to the formation of a more or less consistent skeleton {megascleres). The distinction is not one that can be exactly defined, and must so far be regarded as of a provisional nature. There is usually but little difficulty in applying it in practice, except in some doubtful cases where large spicules do not form a continuous skeleton, or in others where flesh spicules appear to be passing into those of larger size. It is indeed highly probable that all large spicules have originated from flesh spicules (12 ).

(1) Monaxon Biradiate Type {rhabdus).— By far the commonest form is the oxea, a needle-shaped form pointed at both ends and produced by growth from a centre at the same rate in opposite directions along the same axis. It is therefore uniaxial and equibiradiate (fig. 12 a ). (2) Mon axon Uniradiate Type (stylus).— By the suppression of one of the rays of an oxea, an acuate spicule or stylus results (fig. 12 b). (3) Triaxon Tmradiate Type.— Linear growth from a centre in three directions inclined at an angle of 120° to each other gives rise to the primitive form of triradiate spicule so eminently characteristic of the calcareous sponges, but by no means confined to them (fig. 12 c). (4)

Tetraxon Quadriradiate Type *Calthrops).— Growth from a centre in four directions inclined at about 110° to each other produces the primitive quadriradiate form of the Tetractinellida and of some calcareous sponges (fig. 12 d). (5) Sexradiate Type.— Growth in six directions along three rectangular axes produces the primitive sexradiate spicule of the Hexactinellida sponges (fig. 12 e). (6) Multiradiate Type.— Extensions radiating in many directions from a centre produce a stellate form (fig. 12 f). (7) Spherical Scleres.— Concentric growth of silica about an organic particle produces the sphere, which occurs as a reduction of the rhabdus in some species of Poecillastra, or as an overgrown globule (flesh spicule) in Caminus.

Usually conical, the spicular rays often become cylindrical; usually pointed (oxeαte) at the ends, they are also frequently rounded off (strongylate), or thickened into knobs (t ylotate), or branched (cladose). Their growth is not always rigorously confined to a straight line: frequently they are curved or even undulating. They are also liable to become spined, either by mere superficial thickening or by a definite outgrowth involving the axial fibre (fig. 13 g, h).

The rhabdus if pointed at both ends is known as an oxea (fig. 13 c ); if rounded at both ends as a strongyle (fig. 13 a ); if knobbed at both ends as a tylote (fig. 13 b ); the tylote if pointed at one end is a iylotoxea (fig. 13 d ); the strongyle similarly becomes a slrongyloxea. These last two forms are with difficulty distinguished from the stylus, which is usually pointed at the end, and strongylate (fig. 13 f ) or tylotate (fig. 13 e) about the origin. A particular case of the cladose rhabdus, but one of the most frequent occurrence, is the triaene ; in this form one ray of a rhabdus ends in three branches, which diverge at equal angles from each other. The rhabdus then becomes known as the shaft or rhabdome, and the secondary rays are the arms or cladi, collectively the head or cladome of the spicule. The arms make different angles with the shaft: when recurved a grapnel or anatmaene is produced (fig. 13 k), when projecting forwards a protriaene (fig. 13 l ), and when extended at right angles an ortho-triaene (fig. 13 m ). The arms of a triaene may bifurcate (dichotriaene) once (fig. 13 n ), twice, or offener, or they may trifurcate. Again, they may extend laterally into undulating lamellae, or unite to form a disk, the triaene character of which is indicated by the included axial fibre. The shaft may also become trifid at both ends, arnphitriaene (fig. 13 p), and the resulting rays all bifurcate, or the cladome may arise from the centre of the rhabdome, eentrotriaene (fig. 13 o). Amongst one group of Lithistid sponges {Rhabdocrepida) the normal...