III.—On the Mode of Formation of the Spicular Skeleton in the Pluteus of Echinus esculentus

The material on which I worked was procured during my stay at the Plymouth Marine Biological Laboratory in April, 1904, and was produced by carefully rearing larvæ (derived from artificially-fertilised ova) in large jars, each provided with a Browne plunger, and daily supplied with fresh sea water obtained from outside the limits of Plymouth Sound. There is no necessity for me to describe in detail the methods adopted in rearing the larvæ of E. esculentus since the whole process has already been admirably described by MacBride in his important paper relating to the development of this species (15, pp. 288—292), and to this account I refer the reader. I wish to add, however, one or two remarks supplementary to MacBride’s description. Prof. MacBride carried out his work at Plymouth five years previously to my own visit there, and I was, on this account, able to benefit by his experiences as communicated to me by Mr. Smith, assist-ant at the Laboratory, who kindly initiated me into the practical methods of larva rearing. In my own case I was fortunate enough to secure at the first haul three or four ripe males and females out of about a dozen specimens of E. esculentus, and seeing that this was in April—a month earlier than the time selected by MacBride, when, as he says “one was lucky if one obtained a single ripe pair out of a haul containing over a hundred specimens “—I think the fact is note-worthy. Again, I may mention that during the early stages of development it was found necessary, by means of Bolton silk, to strain off from the daily supplies of fresh sea water the flagellate alga Phaeocystis globosa, since this was liable to become entangled with the larvæ. Later on, however, when the plutei were about half grown, the same alga, broken up so as to release the spores, served as food, and, indeed, the plutei principally subsisted on this diet. As re-gards MacBride’s tabular scheme of the succession of events given on page 293, I found it in general correct, so far as my observations of the earlier stages went, but Mr. Smith writes me that “some of the plutei (the eggs were fertilized on April 20th, 1904) hung on till the beginning of September, never looking robust but still healthy”—a period after fertilisation of nineteen weeks, i. e. 133 days—whereas in MacBride’s experience, metamorphosis occurred about the fiftieth day. This abnormal prolongation of the larval stage was, without doubt, due to the fact that pressure of work at the laboratory prevented very much attention being bestowed upon the larvæ during the summer months—the somewhat stale water and scarcity of food resulting from this inattention retarding the normal rate of development; however, by September 12th all the plutei had undergone metamorphosis, and one jar alone contained “quite a hundred” small Echinus. As might have been expected, most of these died later through sheer lack of food, and by December 12th only four Echinus were left. Had it been possible to properly superintend the later development of these specimens, I have no doubt but that they would have attained a very considerable size before succumbing to laboratory conditions of life.¹

The only method of efficiently preserving the plutei at different stages of development—the desideratum, of course, being the preservation of the skeleton—which I found practicable was as follows:—A dozen or more larvæ were removed from one of the large bell-jars employed by means of a small pipette (the Browne plunger having, for the purpose of collecting, previously been removed for about half an hour or less to enable the larvæ to aggregate in the upper layers of water), and placed in a test-tube. The problem now was to get rid of the great bulk of the water in which the dozen or more larvæ were suspended, and this was easily solved by means of an ordinary centrifuge—the larvæ settling at the bottom of the tube, and so allowing all but one or two drops of the water to be poured off. The test-tube was then filled with absolute alcohol, and this was usually a sufficient quantity to render negligible the small volume of sea water originally present; when, as occasionally happened, the mixture thus formed contained a conspicuous precipitate of calcium sulphate, the centrifuge was again used. The larvæ being thus fixed and preserved in absolute alcohol; were then simply packed in small tubes duly labelled until it was convenient to examine them. I found that it was of no use adopting the method of preparation usually employed in the study of spicule formation, viz. fixing the tissues with osmic acid and staining with picro-carmine, since these reagents destroyed the skeleton within a very short time. Indeed, like the spicules of most calcareous sponges, the pluteus skeleton is very susceptible to traces of acid, and the utmost care must be taken to exactly neutralize all reagents employed in its preparation. The larvæ preserved in neutral absolute alcohol were doubly stained with safranin (the larvæ remaining for at least a week in a saturated solution in absolute alcohol) andnigrosin (the larvæ remaining half an hour or more in a similarly saturated solution), though, judging from my later experience in preparing holothurian spicules, I have no doubt but that licht grün (about fifteen minutes’ immersion in the same saturated solution) would form a satisfactory substitute for the latter stain. The result was that the nuclei assumed a pink colour and contrasted well with the slaty hue of the cytoplasm.

Before proceeding to the subject proper of the present paper I wish to express my warmest thanks to Dr. E. J. Allen, who very kindly afforded me all facilities for carrying out ray work at Plymouth, to Prof. Minchin for general advice, and to the Council of the British Association who, at the kind suggestion of Mr. Garstang, granted me a free table.

As is well known, the mesenchyme cells, the great majority of which are skeletogenous in function, arise in the blastula (text-fig. 1) of E. esculentus, as in Echinoderm blastulæ generally, by the migration into the blastocœle of cells at first situated in the flattened posterior wall of the larva (text-fig. 2), and later at the convexity of the hypoblastic invagination (text-figs. 3 and 4). This internal budding commences as soon as the flattening of the posterior wall has become apparent and continues up to the formation of the enteroccele or true mesoderm. It has been stated by Boveri (1) from observations made on the pigmented and unpigmented blastomeres of Strongylocentrotus livid us that these free mesenchyme cells are all derived from that portion of the blastula wall which extends from the posterior pole as its centre upwards for about a quarter of the distance to the equatorial line, and the researches of numerous other observers on this and other genera—Echinus, Echinocyamus, Toxopneustes, Ophiopholis, Echinarachnius, Arbacia, etc.—tend to confirm this conclusion. These budded-off mesenchyme cells, though free in the sense that they are not attached to adjacent structures, are yet not free in the sense that they are at liberty to wander into all regions of the blastocœle; on the contrary all observations show that they take up a very definite position in this cavity. In E. esculen tus, e. g. as the accompanying semi-diagrammatic figures (text-figs. 3—6) indicate, they first aggregate in two positions, situated one on each side of the larva, where they form two elongated strands lying close to the body-wall and running parallel to the long axis of the gastrula. Later these two longitudinal cellular strands, which start from the posterior end of the larva, become connected by the formation of two chains of cells (in single series), which, extending round the archenteron on both sides and in each case joining on to the bases, i. e. posterior extremities, of the strand, thus form a complete circle (blastoporic ring) of cells lying immediately underneath the gastrula wall where it bends in to form the archenteron (text-figs. 3, 4).

From the fact that these mesenchyme cells thus take up their position at all those regions of the body-wall where sharp curvatures are formed—the two longitudinal strands, e. g. being situated in the two corners, one at each end of the flat side, of the gastrula when this is viewed in a transverse plane (text-figs. 5 and 6), and the ring of cells being placed, as above mentioned, just where the posterior wall bends in to form the archenteron, i. e. round the blastopore,—one who had not previously observed how strictly localised the area of immigration is, would with reason suppose that the cells had been budded off in these regions—the cells being squeezed in, so to speak, by the foldings of the wall. But this is certainly not the case, as all observers agree. At the same time, it is an undisputed fact that in E. esculentus, and most, if not all, other genera, all these free mesenchyme cells migrate into all the corners of the gastrula, thus giving rise to the conformation described above. As Théel (21), who has observed the process in living larvae, says: “It is a sight of the greatest interest to follow these cells, to see how they move towards these two places [the positions of the two longitudinal strands] as by word of command.”

The cause of this directed migration is not easy to ascertain. Herbst attributes it to the desire for oxygen on the part of the cells, and certainly, if we imagine oxygen as uniformly diffusing through the gastrula walls, those cells situated in corners will doubtless obtain the greatest share. Driesch (3) also advocates a chemotactic solution to the problem, and he performed an interesting experiment in connection with this subject. Stated briefly, this experiment consisted of shaking up Echinoid gastrulæ in a test-tube in such a manner as to displace the mesenchyme cells forming the blastoporic ring and two longitudinal strands from this their normal disposition and to cause them to become irregularly dispersed throughout the blastocœle: the result was that, after the lapse of a certain time, Driesch found that the cells had returned to their original position in the blastocœle, and, as if nothing had happened, proceeded in their-further development quite normally.

One interesting fact concerning the cells of the blastoporic ring which I do not think has been definitely described before is the protoplasmic continuity existing between them. This protoplasmic continuity is shown in Pl. 18, figs. 8, 11, 12, 15, etc., in which the cells form a single series, the members of which are conspicuously joined together by long protoplasmic threads into a continuous circle. These connections between the cells of the blastoporic ring are primarily and not secondarily formed, since the ring originates by the outgrowth from each longitudinal strand of two chains of cells (i. e. the cells are already connected), one on each side, which extend round the archenteric invagination and meet in the median ventral plane of the larva the corresponding two of the opposite strand. Until one of these chains of cells has met its fellow of the opposite side its foremost cell often has the appearance of protruding in front of it a long pseudopodial process which perhaps acts as a feeler. Connections (though not elongated) between some of the adjacent cells situated in the longitudinal strands doubtless also exist, though it is not always easy to see them.

Starting, then, from this disposition of the mesenchyme cells in the blastocœle, viz. a blastoporic ring attached at two points to two longitudinal strands (text-figs. 3 and 4), the young spicules first make their appearance at the centres of the two longitudinal strands of cells, and in the form of one or more spherical granules contained in one or more cells (Pl. 18, figs. 1—5, 8, 15). In E. esculentus, as elsewhere, “the simplest form of spicule is a minute granule, generally more or less spherical” (Sollas, 18), and thorough agreement on my part with this general statement leads me to here protest against that prevalent dogma which, in my opinion,. most unwarrantably asserts that in all cases the calcareous spicule first, or, at least, early in the development, assumes the shape of a tetrahedron. In E. esculentus there is not at any stage a trace of such a crystalline form of the young spicule, nor have I detected such either in the young spicules of Cucumariidæ, of A. digitatum (25), or of Sycon sponges (24). Semon (20) himself, who is the chief authority for the assertion that the tetrahedron is the primordial form of calcareous deposits in Echinoderms, does not figure any such stage in the development of the wheel- and-anchor spicules of Synapta inhaerens, and, indeed, admits that such does not here occur. Fewkes (5) doesnot describe such in Ophiopholis aculeata, nor does Gerould (7) in Caudina arenata, nor Garman and Colton (6) in Arbacia punctulata, and many other cases might be cited. Minchin (16) is equally positive on this point for Ascon sponges, Chun (2) for Auricularia larvæ, and von Koch (12) for the Alcyonarian genus Clavularia. On the other hand, Semon (19) describes most distinctly the young tetrahedron spicule for the holothurian Chirodota venusta, also Herouard (11) for certain other holothurians, and Théel (21) in his excellent paper on the development of Echinocyamus pusillus. Now all of these last three authors admit that calcareous spicules are deposits of mesenchymatous cells, and, indeed, Herouard and Théel endeavour to trace the form of the later deposits to the disposition of the secreting cells, and yet, this being the case, it is deemed necessary to have initially a visible crystalline structure as the basis upon which the future spicule is to be developed ! Probably having in mind the “integrant molecule ”or crystal particle of crystallography, the young stage of the triradiate spicule, which, from a crystallographer’s point of view, might at first sight be mistaken for a tetrahedron (see text-fig. 7 below), is deemed to be such. Now this assumption is a very doubtful one for two reasons: (1) because Semon (with Fewkes, Selenka, Ludwig [14] and others) admits that the calcareous deposit originates in the interior of a cell as an approximately spherical grain, and that this grain forms the nucleus of the future spicule: hence the tetrahedron can at most only be the first “definite “form to be assumed by this grain,¹ from which it is to be inferred that a geometrical, i. e. crystalline, structure is interposed between two stages of spicule growth which are non-crystalline in form—the spherical granule and the adult triradiate. This is evidently grossly improbable à priori. There is nothing very objectionable in commencing with a crystal as a nucleus for the future spicule, but to interpose one in an otherwise uniform series of non-crystalline meta-morphoses is a proceeding to be regretted; and (2) because of the rarity of the occasions on which this tetrahedron is stated to have been observed. For these two reasons, then, I think it more feasible to suppose that the observers named mistook the small three-cornered mass which represents the transition from the spherical granule to the young triradiate (B in text-fig. 7; also in figs. 9, 11, 22) for the crystalline form in question, and that this last does not occur in the development of any calcareous spicule.

To continue the description of spicule formation in E. esculentus. As already stated, the skeleton of one side of the larva not infrequently (in say 20 per cent, of larvæ) arises in several centres, i.e. as several spherical granules situated in more than one cell, and seeing that the longitudinal strand usually consists of from half a dozen more or less closely-apposed cells, this is only what we should expect, but normally a single spherical granule is produced in the interior of one of the centrally-situated cells.¹ This spherical granule thus situated becomes after a short time three-cornered, and these three corners later assuming the form of three rays, there is thus produced a young triradiate spicule ² (figs. 6, 7, 13, 16, etc.).

When we consider the conditions under which calcareous spicules are in general developed—when we study their development in distinct groups of animals (12, 16, 24, 25, etc.)—it will seem probable that the triradiate form is corre-lated in the pluteus larva, as in other instances, with the disposition of the secreting cells. That this correlation roughly exists in the present instance is obvious from the most superficial inspection of the figures provided. In the region of a lateral longitudinal strand it is evident that, given a position situated just above the junction of the strand with the blastoporic ring as a centre, the cells are arranged in groups diverging from this centre in three directions—one group (the strand) extending longitudinally upwards and away from the blastopore, and two series of cells stretching round the hypoblastis invagination on both sides. Hence it follows merely from this general disposition of the cells that further growth of the initial deposit must occur more or less in these three directions, i. e. in the directions which the three rays of the triradiate actually take. In other words, the form of the triradiate, viewed from a general standpoint, undoubtedly, though roughly, does correspond to the disposition of the scleroblasts.¹ However, there is another possible factor which must be considered. It is easily observable that in the growth of the four spicule rays (two belonging to each triradiate) which extend some distance round the region of the blastopore, the actual deposition occurs in the course of the long protoplasmic processes connecting the adjacent cells (figs. 12, 20, 25, 26), or, in other words, the deposition takes place in a mould already formed for it. Since this is the case for this distal portion of the skeleton, it is therefore possible that the moulding takes its origin from the cell which contains the initial granule, i. e. centrally, and that therefore the initial triradiate form is immediately due to this. It is certainly possible on occasion to trace a faint streak directly continuous through the mass of cells from the tip of the ray of the young triradiate to the protoplasmic “cord” on which the cells of the blastoporic ring are “strung,” and therefore it would seem that the protoplasm in the course of the growing spicule is modified in connection with its further extension. At the same time, it happens on occasion that several triradiate spicules (only one of which persists as a part of the mature larval skeleton, the others becoming gradually absorbed) are developed from the several granules initially deposited in the longitudinal strand of cells (figs. 17, 20, 22), and if the single spicule possesses a triradiate mould in which to deposit its substance, so also must the several. From these facts, then, it would seem that, instead of the mould determining the form of the spicule, the contrary is rather the case, since the number of moulds is determined by the number of spicules. The elimination of this factor of a mould leaves us the disposition of the secreting cells as the only known cause to which we can attribute the triradiate conformation. We must imagine that the cells extending in any one direction from the centrally-situated granule or granules co-operate to exert a species of tractive influence on the freshly-deposited lime, tending to cause this to be deposited in a direction which is that of the resultant of their individual “pulls,” and which the cells eventually place themselves in line with: such a conception, indefinite though it may be, seems at present alone capable of accounting for the facts. Compare the development of a triradiate spicule in a calcareous sponge (16, 24). Here the triradiate, unlike that of the pluteus, originates in three centres—in three pairs of cells, three cells of the three pairs being closely apposed centrally (the “trefoil”), and three being more distally situated—and each of the three lime deposits is, owing to the fact that it originates in one cell with another in close apposition, elongated from the commencement—the influence of an adjoining cell is obvious. Again, in the sponge triradiate, owing to each deposit, i. e. each ray, having only one cell in its vicinity in addition to that in which it originates, this is fairly straight; on the other hand, in the pluteus the initial deposit has, in any one of the three directions, not one, but many cells in its vicinity, and the resulting ray is correspondingly curvilinear. It will be seen from what has just been stated how impossible it would be to assume that in the pluteus triradiate the direction of a given ray is at any point determined by the cells in its immediate neighbourhood,¹ for, if such were the case, the curves of the ray would be numerous and sharp, whereas, in actuality, they are comparatively few and gradual. It may be added in connection with the hypothetical tractive influence of the cells on the growth of the initial calcareous granule that the undoubted influence which the basal actinoblasts of the tri-radiate spicule of Calcarea exert on an adjacent pore-cell or cell of the oscular rim (16, 24) supplies a very fair analogy for my supposition.

The above seems to me to be the only possible explanation, in our present state of knowledge, as to the cause of the triradiate form of the pluteus spicule. Consider the argument: triradiate spicules very alike in form, but not exactly alike, arise in all plutei in a certain position; the disposition of the secreting scleroblasts is in all larvæ substantially the same en masse, but is very different in different larvæ as regards the arrangement of individual cells—from which premises the only logical conclusion to draw is that which I have stated above.

Théel suggests an explanation of the triradiate form which, though it seems to me wholly untenable, I think is worth while quoting, since it contains several very true observations. “I could never observe,” he says, “that the tetrahedron [the three-cornered granule] becomes visible before at least three calciferous cells have arranged themselves in a heap close to the blastoderm. But after this is done the formation of the tetrahedron takes place in a clear pseudopodial plasm situated between these cells and the ectoderm [see my fig. 15, e. g.] and evidently derived not from one cell but from all three cells, the pseudopodia of which have united into a small clump … Thus, according to my opinion, the calcareous tetrahedron is a result of the activity of several cells, which deposit calcareous salts in a liquid state in the common pseudopodial clump, where the formation of the tetrahedron afterwards takes place … The minute tetrahedron grows rapidly to a small star with three very short arms, acquiring a shape almost completely corresponding to the interspace between these [three] close-lying calciferous cells. Thus one is almost tempted to think that there exists a certain relation between the form of the deposit and the interspace in question. The calciferous cells having placed themselves close to the ectoderm, the deposit becomes pressed between them and the latter. If it be so, that this interspace decides the outline of the star, one would expect always to find it in its early developmental stage placed just between the calciferous cells. This seems, however, rarely to be the case. Mostly I have noticed the star situated by the interspace with the arms upon the three cells and not between them, and sometimes I have seen the star itself somewhat displaced. Notwithstanding this, I cannot free myself from the thought that the cells mechanically exercise influence on the outline of the tetrahedron, and the star in the earliest stages of the development.”

There yet remains, of course, a third possible supposition, viz. that the form of the triradiate spicule is due to some agency at present unknown, as e. g. that complex of physical causes termed heredity, but since this supposition certainly does not constitute an explanation—does not affiliate the phenomenon under discussion to other phenomena whose mode of production is known—I am not prepared to discuss it.

TO proceed once more with the description. Though, in E. esculentus, the skeleton of the larva primarily originates in two centres—the two triradiates of the longitudinal strands —yet it often happens that isolated granules, rods, and (occasionally) triradiates arise independently in other regions of the skeletal area marked out by the presence of scleroblasts, i. e. in the course in which further extension of the arms of the original triradiates will occur. These isolated deposits, which mostly originate in the pseudopodial processes connecting the cells of the blastoporic ring, are shown in figs. 8, 12, 25. The rod in fig. 25 has doubtless originated between two cells once closely apposed but which have since separated—the apposition of two cells being, at least in calcareous sponges, an essential condition to the. production of an elongated structure. In fig. 25 also the triradiate, in all probability, arose in a clump of cells—at least three—which have since dispersed to join the adjacent longitudinal strand. It is probable that these independent deposits largely contribute to the small processes and other irregularities characterising the adult larval skeleton.

The three rays of each of the two triradiate spicules once produced gradually elongate in their three respective paths. The curved transversely-disposed rays (the “recurrent apophyses”) ultimately meet ventrally in the median plane, but are not in the majority of cases, contiguous in the same line (figs. 27. 28); the rays pointing anteriorly—away from the blastopore—extend upwards to form the supports of the antero-dorsal arms; the rays pointing downwards, or towards the blastopore, extend posteriorly for a short distance and then curve suddenly dorsally (tending to thus complete dorsally the calcification of the blastoporic ring already accomplished ventrally by the recurrent apophyses) to form later the “baguettes de corps ”or main terminally-thickened posterior supports of the larva; thus the blastopore becomes (relatively) shifted to the ventral side. At the point where these postero-dorsal arms of the original triradiate bend dorsally a pair of new arms originate to form the supports of a pair of postero-ventral arms.

The projection of a new pair of arms from any point of the previously formed skeleton is always preceded at that point by a small cluster of scleroblasts. As Théel says, “As a rule, calciferous cells are present at such parts of the increasing spicule where branches protrude, apparently caused by them, but this seems not always to be the case in the formation of unimportant spines.” I do not know whether the fact that the rods of the postero-ventral arms appear at the bends of the previously deposited skeleton, cells having previously collected there, is of any significance, but it may at least be pointed out that such is the case; moreover, in the other pluteus which I have carefully examined (that of E. miliaris), the same thing happens, the skeletal rods of the dorsal arms arising from sudden bends which here occur in the recurrent apophyses. The same phenomenon occurs in other plutei, though whether secondary branches always arise in such positions¹ I am unable to say.

That the formation of spicular processes is essential to the production of the arms characteristic of plutei is now well established. Herbst (8), e. g. showed that arms are not developed in plutei reared in sea water deprived of calcium salts—a skeleton necessarily being absent under such conditions. He also showed (9) that if, by immersion in lithium water, the calcareous needles are displaced from their normal position in the pluteus, arms are produced from portions of the ectodermal surface which do not normally share in the formation of these outgrowths —a result confirmatory of the same conclusion.

As regards the relation between the scleroblast and the lime deposited, I can amply confirm the statements of Théel as to the deposition occurring in the “clear homogeneous hyaloplasm or ectoplasm” of the cells, and not in the vacuolated endoplasm, and the young triradiate originating in a “clear pseudopodial clump ”formed by the fusion of the ectoplasms of three or more cells. Also, as Théel remarks, it is “on account of the transparency of the pseudopodial plasm and the opacity of the granular [very much vacuolated in E. esculentus] main portion of the cells [that sometimes] one gets the impression that the tetrahedron is extra-cellular in position.” This fact explains why it is that the outline of the pluteus scleroblast bears such a very different relation to the spicule ray compared with that which the actinoblasts of calcareous sponges assumes. In Calcarea the actinoblast is elongated in the direction of the ray-length, and is in most cases cylindrical in form, i. e. envelopes a certain length of the ray on all sides; also on one side the cytoplasm forms a mound of gradual slope containing the nucleus (text-fig. 8, A). In the pluteus larva, on the other hand, the actinoblasts are spherical in contour, and are attached to the skeletal rod by only a very small portion of the entire circumference, i. e. the base of attachment is very limited (text-fig. 8, B). In other words, the deposition of lime in Calcarea occurs in the mass of the cell-substance, whereas in the pluteus larva it solely occurs in the ectoplasm or thin peripheral layer; hence the difference of cell-outline in the two cases. As the skeleton of the larva (both in E. esculentus and miliaria, and doubtless all other plutei) assumes its adult form, however, the scleroblasts become more closely applied to the spicule ray (fig. 29), and then resemble the actinoblasts of the Calcarea in form: this change in contour being probably correlated with the ex-haustion of the cell-substance, the vacuolated cytoplasm becoming gradually absorbed and changed in constitution as the active deposition of lime proceeds.

Illustrating Mr. Woodland’s paper, “Studies in Spicule Formation.” III.

All figures magnified 1000 diameters except Fig. 5 (× 2000) and Figs. 25 and 26 (× 500).

PLATE 18.

FIGS. 1—3 show the initial granules deposited in the cells of the longitudinal strand which is viewed from the surface.

FIG. 4.—The granule in the longitudinal strand when this latter is viewed edge on, as also in Fig. 15. In Fig. 8 the strand is viewed end-on.

FIGS. 4, 8,10,11,12,15 (and 19 and 20 of Plate 19) show well the pseudopodial processes connecting adjacent cells in the blastoporic ring.

FIG. 5.—The scleroblast containing the granule, magnified 2000 diameters.

FIGS. 6, 7, 10, 11, 13, 14, 16.—The young triradiates derived from the initial granules.

FIG. 9 shows well the young “tetrahedron”—the three-cornered granule; also seen in Figs. 11 and 22 (Plate II).

FIG. 15 shows well the position of the spicule relative to the wall of the larva.

FIG. 16 shows two young triradiates dissimilarly aurientated ed (p. 316); also Fig. 7.

PLATE 19.

FIGS. 17, 20, and 22 illustrate the production of two large triradiafes in the longitudinal strand. In Fig. 25 as many as three triradiates have been formed.

FIG. 19 shows half a dozen granules produced at the centre of the longitudinal strand.

FIGS. 21 and 25 illustrate the rare occurrence of small triradiate spicules in the course of the blastoporic ring. In Fig. 25 the cells have deserted the spicule after forming it. In Fig. 22 they are still present.

FIGS. 22, 24, and 26 show the typical fully-formed triradiate spicules with their actinoblasts. As is evident, the precise position of the scleroblasts is in no way constant, at least at this stage, for the given form of spicule—the relation is only a general one.

FIG. 25 is remarkable for the number of independent deposits which have arisen in the blastoporic ring. The ring is viewed from the posterior end of the larva.

FIG. 26.—Spicule formation well advanced. The recurrent apophyses have not yet met in the mid-ventral line.

FIGS. 27, 28 show the junction of the recurrent apophyses. Despite the completeness of the blastoporic ring, the apophyses are not contiguous in the same line.

FIG. 29 shows the “mound” form which the actiuoblast of the adult larval skeleton assumes (p. 320).