Dr. dyster (6) appears to have first published observations upon the development of Phoronis (P. hippocrepia). The anatomical part of his paper is principally concerned with the alimentary and vascular systems, and parts of this will be referred to later on.

He watched the whole process of oviposition, the passage of the eggs through the nephridia (the “hollow ridge “) and their deposition in the inner space of the lophophore. To the wall of this ridge they adhere by a glutinous exudation. “They are voided alternately through each ridge (nephridium) and form a compact white mass, separable only with considerable difficulty on each side of the space in the concavity of the lophophore, shadowed over by the interlacing extremities of the inner tentacles.” He states, further, that the larva “quits the parent nest when about forty-eight hours old.” He figures three larvæ at different stages, the latest having one pair of tentacles. Both the figures and description clearly show that he mistook the anal end for the oral, and vice versa. The eggs after deposition measured inch in diameter.

Kowalevski (9) in 1867 published the results of his labours upon Phoronis at Naples. Some observations upon the anatomy are followed by an account of the early development. He states that fertilisation takes place in the ccelom. Total cleavage leads to the formation of a hollow sphere, which invaginates. The mesoderm is formed by delamination from the ectoderm. The blastopore shifts from a terminal to a ventral position. Further development results in the formation of the hood over the mouth and the tentacles at the posterior end. Between the latter the anus breaks through. On becoming set free from the egg membrane the larva is uniformly clothed with cilia. He succeeded in definitely identifying the Phoronis larva as being none other than. Actinotrocha. Finally, he was led to consider that Phoronis had no connection with the Grephyrea or the Bryozoa. In fact, he doubted the correctness of placing it amongst the worms, himself inclining rather towards the Mollusca as its true phylum.

Metschnikoff (13) in 1871 gave an account of early Actinotrocha larvæ and their structure in so far as could be ascertained from external observation. He figures and describes a free-swimming larva with one pair of tentacles, and follows this up through stages with two and three pairs of tentacles. His description of stages later than this do not concern the present subject.

In 1882, however, in a paper (14) dealing generally with gastrulation and the origin of the mesoderm in Metazoa, he follows these processes in Phoronis from the blastula onwards. He appears to have relied solely upon an examination of the entire embryos for his results. He shows large mesenchyme cells present in the blastoccele cavity before gastrulation, and smaller ones scattered during this process in the reduced cavity, more especially in the preoral lobe. In this region they are figured as arranging themselves as a layer lining the cavity. From these observations be is led to conclude that the mesoderm of the adult arises from mesenchyme cells. In this opinion he is supported by Fœttinger (7), who finds what he considers early mesenchyme cells (“le premier élément mésodermique”) in as early a stage as that of eight blastomeres. He followed the segmentation processes from the earliest condition to gastrulation. Like Metschnikoff, he relied upon optical sections of the entire embryos, and also subjected them to a course of treatment with acetic acid.

In 1885 appeared a short paper on the early stages of Phoronis by Caldwell (3). This observer was the first to apply the method of sections to the elucidation of the subject in hand. His paper is almost entirely concerned with the origin of the mesoblast. Briefly, he dismisses Kowalevski’s statements by supposing that he was deceived by the ectoderm cells being darker at the base, and those of Metschni-koff and Fœttinger by explaining their “mesenchymatous” cells as being “amoeboid processes of the endoderm cells growing into the segmentation cavity,” or possibly other non-nucleated bodies which are to be found in the blastocœle. He concludes, “I have observed them frequently, but it is certain that they have nothing to do with the true mesoblast.”

His papers and figures have been introduced into numerous text-books, and are well known, so that we need merely repeat here his summary (p. 25).

1. The blastopore gives rise to both mouth and anus.

2. The mesoderm arises in an anterior pair of entoblastic modified diverticula, and in a posterior pair of ectoblastic diverticula connected by a few mesodermic cells derived from the middle of a primitive streak.

3. The nephridial openings to the exterior are parts of the blastopore.

In 1890 Boule (15) attacked the subject, and formed no exception to the others in differing remarkably from his predecessors. His material was derived from Phoronis saba.tieri,

Roule’s paper is unaccompanied by any figures, and is of the nature of a résumé. He has followed the segmentation to the stage with thirty-two blastomeres, in which he states there is no blastocœle. The spherical blastula spreads out laterally to a discoidal shape, which then invaginates and hence becomes globular. The blastopore is at one of the poles of the gastrula, but becomes eccentric by differential growth. He finds no mesoblast till the mouth and preoral hood over it are established, thus differing from Metschnikoff and Fœttinger, and indeed from Kowalevski. From this stage onwards he finds primary mesench y matous cells in the blastocœle. Later still, the “mésendoblast” in the neighbourhood of the anus proliferates to form the initial “mésoblastiques.” The cells of these break up partly into single cells, which join the mesenchyme already referred to, and partly form two compact masses which Roule considers to be the mesoblastic bands (homologous to those of the trochophore). Roule does not state his methods of investigation, though it would be difficult to determine with any certainty the points above described without serial sections of the embryos.

Lastly, Schultze (16) has quite recently (1897) published a short paper upon the development of the Phoronis found in the Black Sea. He found that the blastula became bilateral. As in the case of Roule and Metschnikoff, he describes the mouth as formed from the blastopore, the anus being a new formation. He, like Metschnikoff, finds single mesoderm cells in the blastocœle of the blastula stage. They arise from the endoderm, and later become pressed into the preoral lobe and the anal region, eventually arranging themselves into somatic and splanchnic layers.

He illustrates this process by figures, but does not in any way refer to his methods. He attempts to account for Roule’s and Caldwell’s descriptions as being due to the personal element and the influence of Hertwig’s coelom theory. He also remarks that Caldwell evidently mistook the ventral invagination which gives rise to the adult body for a pair of posterior cœlomic sacs. This cannot possibly have been the case, for Caldwell’s embryos were cut in the tentacles of the parent, and in all the species of Phoronis whose development has been known up till now the larvæ leave the parent at the stage with only one pair of tentacles long before the invagination commences. Apart from this, Caldwell’s figures, drawn by camera from the sections, scarcely allow of such an explanation.

Lastly, he refers to my comparison of Phoronis and Balanoglossus, and objects to it on the score of the different fate of the blastopore in each case.

Reference to this will be made later.

Thus the ontogeny of Phoronis is almost unique in this respect, that with regard to it there is universal disagreement on such an important point as the development of the mesoderm.

In a former paper I have investigated the structure of the late Actinotrocha larva, but was debarred from following the early stages because neither Phoronis nor the early Actinotrocha occurs in St. Andrews Bay. The later larvæ appear to be brought by the tide currents from elsewhere, probably from the Frith of Forth, in which Dr. Stret-hill Wright first found the adult.

Although a renewed investigation of the early stages might furnish fresh proof of the correctness of the view which was suggested of the natural affinities of Phoronis, and could hardly reveal anything prohibitive of it, more especially as the origin of the mesoblast in Tornaría is still considered as undecided (cf. Spengel and Morgan), it seemed justifiable to publish at once the facts already found, such as the five cœlomic sacs, and draw the legitimate inferences from them rather than keep back the whole work for years. Such, however, does not appear to be the opinion of some, to judge by recent criticism, so that I determined to attempt an investigation of the early stages in Phoronis Buskii, although the specimens kindly given me by Professor McIntosh were preserved in about 1870, and even then more with regard to preservation of the adult than of the embryo. This author has himself made a few observations upon the larvæ and their presence in the tentacles of the parent (12).

The animals themselves were well preserved, probably in corrosive sublimate followed by spirit, but some little difficulty was experienced in the case of the embryos, so that the study of them has failed to elicit all that might have been expected, especially with regard to the origin of the mesoderm.

The embryos were cut in series mostly with a thickness of 1 · 25 to 2 μ, and were then deeply stained with hæmalum followed by aqueous eosin. A very beautiful differential staining resulted. Not only were the nuclei sharply defined, but the stages in development were clearly indicated, all the earlier stages staining with eosin, and the later stages, especially after gastrulation, becoming clearer. At the same time the hypoblast always took a darker eosin stain than the epiblast. Such a method of staining very thin sections when applied to suitably fixed embryos of Phoronis kowalevskii could not fail to settle the vexed question of mesoderm formation in this species, with regard to which every observer has up till now disagreed with his predecessors.

Some of the sections were made with paraffin alone, and others with the celloidin-paraffin method ; the latter were the more satisfactory, mainly on account of the possibility of more accurate orientation.

The Early Development of Phoronis Buskii, Mcl.

It appears as far as is known to be a general character of the Phoronidea to retain the eggs in the lophophore until the early stages are passed through. In the case of P. Buskii the same arrangement holds as was noticed by Professor McIntosh in his monograph upon the species. The doubts which have been thrown upon the claims of this form to specific distinction do not seem to be well founded, as it has well-marked differences which cannot be explained as variations. Apart from those which concern the comparative length of the adults, the number of tentacles, and the differences in pigmentation, it is noteworthy that the size of the embryos in Phoronis australis is greater than that of P. Buskii; so much is this the case that it is easy to distinguish the embryos of each species by their relative size in a mixture of the two.

In P. Buskii the embryos are found in enormous numbers in the outer coil of the lophophore. They are all enclosed in a thin egg-membrane, which is intact in .the latest stage found in the tentacles. Rupture of the egg-membrane would appear to be synchronous with escape of the larvæ from the lophophore of the parent.

Fig. 59 is a transverse section of the lophophore and surrounding partsof Phoronis Buskii, seen from above. As is well known, the lophophore is composed of a nearly complete ring of tentacles surrounding the mouth. This ring, at first nearly circular in those species in which the young stages are known, is drawn out laterally into an ellipse, and then the extension on each side is rolled up dorsally (towards the anus) into a spiral.

The inner ‘(oral) surface of all the tentacles presents a ciliated epithelium of long cells, which probably cause currents of food and water to pass downwards towards the mouth. The outer surface of the tentacles has a non-ciliated epithelium of cubical cells, which are much flatter than the cells of the inner surface. In Fig. 59 the distribution of the two layers is shown as seen in a section about halfway up the lophophore.

Thus the inner coil of the lophophore leads down to the mouth, whilst the outer coil eventually opens out opposite its fellow in the median line between mouth and anus. Through the greater extent of the lophophore the apertures of the outer coils upwards are freely open, but in a section at the very base of the lophophore they are seen to be partially blocked by a pair of large organs, one on each side (see fig. 61).

These problematical organs have had various functions assigned to them. Caldwell (1) claimed for them a sensory function as indicated by the name “ciliated pit,” and they have been compared to certain sensory pits in some of the Gephyrea. McIntosh (12) also observed them in the species under consideration, and noted that they were simple and rudimentary in the young individual. He also noted the presence of mucus in the central cavity of the organs.

Benham (2), working upon Phoronis australis, was led to doubt their sensory function, and considered them to be glandular in nature. He remarked that a similar glandulai-modification extends as a ridge throughout the lophophore, lying along the face of the inner series of tentacles. I have carefully examined these organs, and can confirm the observations of Benham upon their structure and extent. The “ciliated pit “itself is really only the involuted terminal portion of the ridge of ciliated glandular epithelium extending along the base of the outer coil. Benham has named the whole organ, including the portion extending throughout the length of the outer coil, the “lophophoral gland.” My observations of its structure and relationships in P. Buskii do not differ sufficiently from his to justify further figures.

With regard to the functions of this lophophoral gland, I believe there is sufficient evidence to assume that it secretes mucus, which, driven up the coil by the cilia, forms an adherent surface for the eggs and embryos to be carried in the same direction. Thus these glands would come under the category of subsidiary reproductive organs, or nida-mental glands.

The proof for this assertion is partly direct and partly indirect.

Upon dissecting the lophophore, the mucus lying in the gland may often be removed as a long band with a great number of embryos adhering to it. The embryos usually present a more or less progressive series as regards development (fig. 20). This band is also found in Phoronis australis.

If we are justified in assuming that the cilia on the inner surface of the tentacles cause a food- and water-current downwards, the water must tend to filter between the tentacles into the outer coils, and down them into the median space, where the current would be reinforced by the water directed by the epistome out of the inner coil through the dorsal gap in the tentacles, which I called the branchial fissure This current of water would pour outwards between the two spirals of the lophophore, and over the apertures of the nephridia and the anus. I have not had any live specimens of Phoronis, so cannot demonstrate with certainty that these currents actually exist, but I fail to see any other way of interpreting the structure of the lophophore. This current would tend to bear away from the animal both the fæces from the auus and the excretory products from the nephridia, but in the breeding season the sexual products would in like manner be carried away from the parent. Whilst this may or may not be a desirable consummation in the case of the spermatozoa, further provision would be necessary to retain the eggs in the lophophore (see figs. 61 and 62). If we may assume that by a contraction of the base of the lophophore the Openings of the nephridia could be approximated still more to the mouth of the lophophoral gland, the eggs would come in contact with the mucus immediately upon extrusion, and adhering thereto must be carried up the outer coil in the teeth of the down-coming water-current, in which there would doubtless be spermatozoa from another individual. Sheltered in the tentacles of the parent, the eggs are in this situation not only in the best position for fertilisation, but the continuous stream of water makes the lophophore, like the branchiæ of Lamellibranchs, an ideal “nursery “for the early stages (see fig. 61).

Again, the fact that the lophophoral glands develop very late in life indicates a probable connection with the sexual function.

These remarks concerning fertilisation are contrary to the statements of Kowalevski that it takes place inside the body-cavity, but this does not appear to be the case in P. Buskii or P. australis. Like Benham, I have failed to find any mature spermatozoa in the ccelom. In like manner the eggs are never segmented in the cœlom, whilst the unsegmented eggs are found in great numbers in the tentacles. In any case all are agreed that the nephridia act as genital ducts.

Thus the structure, function, development, and relationship to surrounding parts of the lophophoral gland indicate alike that it acts as an accessory gland in connection with reproduction with a nidamental function.

Arrived at the apex of the lophophoral spiral the embryos appear to become detached from the mucus, and are then carried round the outer coil, probably by the water-current, which passes not only downwards, but from the apex of the spiral to the opening of the outer coil. Thus the forward movement of the embryos to the centre of the coil is reversed, and they travel round the spiral till they are discharged in the water-current to the exterior between the two spirals of the lophophore, dorsally to the lophophoral gland (see figs. 60, 61, and 62).

A suitable section will show the spiral with every stage of development from the segmented blastula at the apex to the advanced larvæ With three pairs of tentacles in the outer coil (cf. fig. 59).

We may regard the increase in number of tentacles and their coiling as being adaptations connected with nutrition and growth, and these features have been seized upon for the protection and “nursing “of the young in an analogous manner to similar phenomena in the Lamellibranchiata. At the early phyletic stages of the group, when the lophophore was a ring, it could have afforded little or no protection to the young; the lophophoral glands were probably absent, and the eggs were discharged direct into the water. As the lophophore commenced to coil the eggs were retained for a longer and longer portion of the ontogeny in the shelter of the tentacles. Phoronis australis and Phoronis Buskii form the head of the series in this as in nearly all other features of anatomy and physiology. Thus the eggs and young are found in the tentacles, and as a rule still enveloped in the egg-membrane till as late as the Actinotrocha stage with three pairs of tentacles.

The earliest unsegmented ova are found in the lophophore, which fact militates against the view that fertilisation and part of the segmentation is effected in the ccelom of the parent, as was stated to be the case in the Phoronis examined by Kowalevski (9). Cori (4) has been led to doubt this observation, and I have certainly failed to find either in P. Buskii or P. australis a single ovum uudergoing segmentation within the ccelom. The ova are perfectly round, and surrounded by a very delicate and pellucid membrane, which usually has a small pedicle by which it was attached to the mucus band (see fig. 20).

The first segmentation results in the formation of two equal blastomeres, which appear to be almost symmetrical. The second furrow is at right angles to the first, and results in four quadrants which now differ slightly at the two poles. Each tapers to one end, so that the whole egg appears to be slightly pointed and broader at the base (figs. 2 and 3). This condition is foreshadowed in the two-cell stage, but becomes well marked here. A view of the base (fig. 4) shows the arrangement in which two opposite blastomeres form a cross-line. All the eggs of this and the eight-cell stage which I have seen show this arrangement, though, as in the case of the frog, there are probably variations, and the point is one to which little importance need be attached. At the apex the two furrows form a regular cross. A transverse section of this stage shows (fig. 21) that there is a well-defined blastocœlic cavity, slightly compressed from side to side. Houle (15) states that the blastocœle does not develop till much later.

The third furrow is at right angles to the first two, and is almost exactly equatorial. The fact that the first four blastomeres taper results in the upper four being less in bulk than the lower (fig. 5). The base and apex of this stage appear similar to those of the four-celled. The further stages in segmentation are difficult to trace in the specimens to hand. The cells of the blastula stage are very nearly if not quite equal -in size. The cells of the lower hemisphere, however, appear to retain a slightly greater size. Fig. 22 is a median section through an early blastula, with five upper cells and four lower cells in the median plane. The. blastula is still perfectly spherical, but in further segmentation both the blastula and the blastocœle appear elongated in a median section (fig. 23). In this figure the lower cells are distinctly larger and rather longer than the upper ones. The elongated appearance is not due to an elongation in one axis of the blastula (the horizontal axis), but to a gradual flattening of the whole blastula in the vertical plane, so that it becomes disc-shaped (cf. fig. 11). Fig. 24 shows an even later stage of the blastula, in which there isa great increase of cells, and the lower cells have become slightly flattened on their outer surface. This is the commencement of gastrulation. As segmentation proceeds, the nuclei, which in the four-celled stage lie at the centre of the cells, gradually move outwards to the peripheral part, and in fig. 24 come to lie almost under the limiting cell wall.

The external appearance of the several stages here dealt with is shown in figs. 6 to 15. The embryos not having been observed alive, one can only follow the changes by the examination of a selected series. The whole process of gastrulation and closure of the blastopore appears to be extremely rapid if one may judge by the small proportionate number of embryos to be found at these stages, and the very small extent of the coils which is occupied by embryos at this stage. The spherical blastula appears to become more or less hemispherical by a complete invagination of the lower and rather larger cells, so that the gastrulation is typically embolic. Fig. 11 shows the view from below of an embryo in which the invagination is nearly completed ; fig. 6 is a lateral view of the same embryo. The blastopore at this stage is large and circular, with a diameter little less than that of the whole blastula. In the next stage which I have found (fig. 7) the embryo has commenced to elongate slightly in an axis perpendicular to the principal axis of the blastula, and corresponding very nearly with the long axis of the late larva. Hence in ventral view (fig. 12) the embryo appears to be oval in outline, and the blastopore is correspondingly oval. At the part which we may now distinguish as the posterior end, the oval blastopore is seen to taper off into a groove which extends as a furrow very nearly to the posterior border of the embryo. These appearances certainly lead one to the inference that the blastopore is closing up by an approximation of its lips in the posterior region.

A reference to the lateral view of the same embryo (fig. 7) shows that the anterior part of the embryo, mainly in front of the blastopore, is increasing in size rapidly, and is tending to bend over ventrally.

At the next stage (fig. 13) this increase of the anterior region has caused the outline of the embryo as seen from below to apparently taper at the posterior end. The blastopore has now further contracted in extent, and narrows off posteriorly in connection with the median groove, which has increased in length. In comparing this stage with the last we note that the anterior lip of the blastopore appears to be further back than in the latter, which is probably due to the growth of the anterior region and to its further ventral flexion (cf. fig. 8). On the other hand, the appearances here still further confirm the result arrived at above, that the blastopore is undergoing a process of closure by an approximation of the lips in the median line. The closure proceeds from behind forwards, and the line of closure is indicated by the faint median groove.

At the next stage the growth, both laterally and ventrally, of the anterior region has altered the outline of the embryo (figs. 9 and 14). The blastopore has closed for the greater part of its extent, and it is now rather broader from side to side than before. The closure having almost ceased, the blastopore, or rather its anterior end which is still open, broadens out and soon after this becomes circular ; it is definitely to be identified as the larval mouth, or more accurately the opening into the “stomach.” The ventral groove can now be barely distinguished except by careful arrangement of the light, and after this stage is quite indistinguishable.

The ventral flexure of the anterior part of the embryo is very marked (fig. 9), and in the next stage (figs. 10 and 15) has proceeded so far that the mouth is completely covered over by this part, now to be identified as the “hood” or preoral lobe.

The period included between these two last stages appears to be a fairly long one, during which some important internal differentiations are in progress.

So far as external observation can show, we may notice that the gastrulation is embolic, the blastopore closes from behind forwards, its anterior part persisting as the “mouth.” We may here anticipate a consideration of the work of other observers by saying that in no embryo have I been able to find either in external view or in section any trace of the “posterior pit” of Caldwell (3) as a posterior persistent portion of the blastopore.

Fig. 25 is a median vertical section through fig. 11. In it the larger granular cells are seen to be invaginating. Throughout the whole series of sections the histological characters of the hypoblast are markedly different from those of the epiblast. Apart from the differences of size and number, the hypoblast cells contain far more yolk, which stains well with eosiu, the cell walls are much fainter, and are in some cases indistinguishable (due to method of preservation), whilst the nuclei are slightly larger and less regularly arranged.

In fig. 25 the blastocœle cavity is still seen, and it is usually quite destitute of any structural contents. Occasionally I have noticed a few fine strands of cytoplasm crossing it, or even gathered into small masses, but in no case have I seen a nucleus, or any body which stained with the dark blue tint of hæmalum present in the blastocœle.

Fig. 26 shows a median sagittal section of a stage a very little later than fig. 12. Here the invagination is completed, and the blastopore even in its length has considerably contracted. The epi blast cells are slightly longer and more abundant at the anterior end than elsewhere. The blastocœle cavity is still distinguishable and empty as before. Up to this stage there is nothing which could be construed into a trace of njesodermic elements in any part of the embryo.

Fig. 27 is a median sagittal section of the stage represented by fig. 13 or slightly earlier. Here we may note that the hypoblast is in close contact with the epiblast in every direction. Whether due in any degree to a post-mortem contraction or not, the blastocœle cavity has disappeared altogether. The blastopore has further narrowed in extent, and the epiblast at its lips has commenced to grow inwards, following in the wake of the invaginated hypoblast.

Figs. 29 and 30 are median sagittal sections of figs. 14 and 15 respectively. In them the further formation of the stomodæum is shown, fig. 30 indicating that the whole “oesophagus “of the larva is formed in this way from epiblast. The gradual ventral growth of the preoral hood is also clearly shown. One feature to be noted is the entire absence of any indication in section of the ventral groove. In transverse sections it appears merely as a superficial depression of the cells, and it would appear that concrescence of the respective hypoblast and epiblast of the two lips takes place as rapidly as the formation of the groove itself, so that there cannot (in this species) be found a part in which these two layers are fused; at least, I have failed to find a section exhibiting this condition. The ventral groove would thus appear to be a structure merely indicating the line of fusion of the blastoporic lips, and not correlated with an internal line of junction between hypoblast and epiblast. Fig. 27 shows that internal separation of the two layers proceeds forwards with the fusion of the blastoporic lips.

A determination of the true formation of the mesoblast in Phoronis was the primary object of this investigation. Unfortunately, owing to the fact that the material was not preserved with a view to embryological work, I am not in a position to demonstrate every detail of the mesoblast formation in P. Buskii to the extent which I could wish, and I was tempted to withhold these results till the opportunity of working out another species with properly prepared material presented itself ; but as the lesser details may as likely as not differ in the two species, I propose to give a brief account of the results to which I have been led by the examination of sections of P. Buskii.

We have already seen that up to the stage depicted in fig. 12 and in fig. 26 in section, there is no trace of mesoblast in any part of the embryo.

In fig. 27 the archenteron is seen to be extending into the anterior part of the embryo, which is at this stage rapidly growing.

The hypoblast level with this forward extensiou of the archenteron bounding it laterally and anteriorly is thinner than the rest (fig. 27, p. c.), and in places has more nuclei. At a slightly later stage, in coronal section, this anterior part of the archenteron is seen to have still thinner walls, and to grow backwards as a pair of lateral boms. Its cavity is still in continuity with the general cavity of the archenteron, but is connected only by a narrow aperture (fig. 28). At a later stage (fig. 31) complete separation is effected, and the preoral lobe then contains a cavity separated from the archenteron, and lined by low flat cells which have been directly derived from the hypoblast. The cavity forms the preoral ccelom lined by its mesoblast. In fig. 30 the same cavity is shown with its walls growing back dorsally between epiblast and hypoblast, and in figs. 35 and 36, transverse sections of the same stage, it can also be recognised as growing down laterally. The preoral body-cavity (or protocœle) Would thus appear to arise directly from the archenteron, and its mesoblastic walls from the anterior hypoblast.

At the stage seen in fig. 28 the cells of the hypoblast at about the middle of the embryo are seen on either side to have a group of massed nuclei (msc), whilst here and there a cell-wall can be seen separating them. In transverse section of this region (fig. 33) it is seen that the masses are really ventro-lateral in position, and that the inner surface of the archenteric cavity is indented just opposite the massed nuclei. These indentations or grooves correspond in position and appearance with similar grooves described by Caldwell (3) in Phoronis Kowalevskii. The cells of the hypoblast surrounding them appear to be segmenting off their inner ends, and cell divisions proceeding rapidly, a pair of ventrolateral masses of mesoblast result. The separation of these two mesoblastic masses must be effected later than that of the preoral ccelom, and they do not appear at first to contain a cavity. Their future development will be followed ; they give rise to the collar cavities or the cavities of the lophophore. In the stage of fig. 28 there is an arrangement of cells at the posterior end of the archenteron much as seen in the figure (mtc.). After careful search I have not been able to discover any more definite condition than this. A little later on, in fig. 31, there can be recognised a pair of posterior accumulations of nucleated cells, as depicted.

There can be no question that these cells are segmented off from the hypoblast in this region, and in fact could even be regarded as being still integral parts of the hypoblast. I cannot say for certain whether these two masses of cells have archenteric grooves opposite them or not. It is possible that this may be the case, and even that they are formed by two archenteric diverticula which have been disguised by shrinkage and imperfect preservation. The development of these posterior mesoblastic masses is later than that of the collar, for in fig. 32, a section a little ventral to fig. 31, the collar cavities have expanded to form a median ventral cavity. The posterior masses still lie in a dorso-lateral position above the gut, and very little further developed in the latest stages; there can be little question that they give rise to the ccelom of the trunk, as found in late Actino-tr ocha.

Figs. 34—39 are a selected series of transverse sections through the stage depicted in fig. 15. Figs. 34 and 35 show the preoral cœlom (pc.) containing its cavity, and giving off its two lateral horns. These horns proceed backwards on either side of the mouth, and their terminal portions are cut across in fig. 37 level with the mouth. Further back, in fig. 38, the collar mesoblast has fused ventrally, and formed a spacious collar cavity (msc.) extending up laterally towards the dorsal surface.

In fig. 39 the posterior tip of the collar ccelom is cut ventrally, and the paired masses of trunk ccelom (mtc.) are seen lying dorso-laterally to the gut. They contain no cavity, and are small.

Returning to fig. 15, we find that further external differentiation results in (fig. 16) the bifurcation of the posterior end to form the rudiments of the two first tentacles and a further growth of the preoral hood, Round the edge of this hood a slight ridge may be traced, and on either side this ridge passes downwards till it is lost on the surface of the tentacles. It does not appear to be ciliated, though this may be due to the inadequate preservation. Lastly, there appears at the hind end posterior to the two tentacles a single median protuberance. This is the anal papilla, and its presence gives the embryo the appearance of bearing at this stage three tentacles.

Figs. 40 to 45 are selected from a series of transverse sections through this stage. In fig. 40 may be seen the large stomodæum which is forming the oesophagus, and is spreading out laterally to form the atrial grooves. Surrounding the oesophagus and part of the stomach is the spacious preoral coelom. The stomach still consists of long granular cells with more or less irregular nuclei. In fig. 41 the atrial grooves are still wider, a mere tip of the preoral coelom is cut in the hood, and there is no mesoblast present in the body.

Lower down, however (fig. 42), the two collar elements (msc.) appear in a ventro-lateral position, and these may be traced through figs. 43,44, and 45 as gradually expanding out, and containing a collar cavity within them. The epiblast in figs. 44 and 45 can be seen to grow out into the two tentacular rudiments with great numbers of elongated cells. In fig. 45 the collar mesoblast can be seen to grow out into these rudiments.

In figs. 44 and 45 the trunk mesoblast (míe.) can be recognised as lying dorsally to the stomach and more or less united into one mass. The anus does not appear to open at this stage, nor do any of the coelomic cavities open to the exterior.

Figs. 17, 18, and 19 illustrate the external appearance of two later stages. In fig. 17 the first pair of tentacles has increased in size, and the rudiments of the second have grown out on either side posterior to the first.

The ridge round the hood can still be noticed, and it now passes down to the second tentacle on each side. The anal papilla has increased in size. The ridge now passes to the latest developed pair, and has a prominent bay at the spot at each corner of the hood where the atrial grooves emerge. The tentacles may be regarded as arising on this ridge one by one. If the ridge prove in future investigations to be ciliated, the early Actinotrocha at this stage would have a single postoral ciliated band comparable to that of Bipinnaria, which in later life breaks up like that of this Echinoderm into a pre oral and a postoral band.

Under any circumstances the preoral band edging the preoral hood, and the postoral band following the course of the tentacles, are connected at these early stages by a thickened ridge which probably indicates a phyletic unity. The common origin of these two bands from one “architroch “has been suggested by Lankester (9a).

In fig. 18 the tentacles spread out in such a way as to hide the anal process altogether, but it may still be recognised by reversing the larva.

Fig. 19 shows a front view of the same stage ; the large bell-shaped preoral lobe is conspicuous, and it extends out laterally beyond the body. This is the latest stage found in the tentacles of the parent, and it is probable that rupture of the egg-membrane is effected after its attainment, the larva then leaving the lophophore of the parent.

The internal structure of the stage with two pairs of tentacles is intermediate in character between that with one pair which has been described, and that with three pairs (figs. 18 and 19). The internal structure of this latter stage is indicated by the sections shown in figs. 46, 47, 48, and 49, and by the restoration in fig. 50.

There is little change in the epiblast, except that the central nerve-ganglion (ng.) can be discerned in figs. 46 and 49 as a thickening of the epiblastic cells over the oesophagus. The stomodseum (oesophagus) is now very long and curved round into the stomach, which is still simple, and does not appear to have yet given rise to the pleurochords. The hind end of the alimentary canal is constricted off to form the intestine (fig. 47), and in most the anus opens to the exterior, though the aperture is very minute. The intestine seems to be hypoblastic in origin.

In the mesoblast there are considerable changes. The mesoblast cells have now formed the thin “lining-membrane “type of endothelium,—in fact, typical coelomic endothelium, not to be distinguished from the coelomic endothelium of the free Actinotrocha with six pairs of tentacles. The several portions of the ccelom have come together and formed typical mésenteries.

The preoral ccelom or protocœle does not otherwise differ in extent from the stage with one pair of tentacles, except that the two horns have reached back to meet the mesocœles or collar cavities on either side of the oesophagus, which also grow forward from their former position to effect the junction. At the point of junction are formed a right and left mesentery (fig. 49), but in the middle line dorsally the protocœle and mesocœle do not meet, but leave a hæmoccele space just under the nerve-ganglion (figs. 49 and 46, sns.). This is the subneural sinus, differing only in size from that of the later Actinotrocha. Othei’ hæmocœle spaces can be observed, such as that below the oesophagus in fig. 49, but the extent of these vascular spaces is difficult to trace, and probably varies to a considerable extent. The mesocœles(wise.) have extended dorso-laterally round the stomach, to meet dorsally just on the fore-part of this organ (fig. 47), whilst they expand widely in a lateral direction, and dorsally at their posterior part, giving off branches to each tentacle on either side. In fig. 48, a transverse section at the level of the postoral ring of tentacles, the two mesocœles may be seen pushing dorsally, their walls forming a pair of conspicuous mesenteries with the walls of the metacœles (mtc.). On either side of the anus and slightly ventral to it the mesocœles open to the exterior by a small pore (fig. 47), which is evidently thé mesocœlic pore or collar pore. There can be little doubt that this is later metamorphosed into the collar nephridium of the later Actinotrocha, probably by the invagination of an epiblastic portion carrying the actual mesoblastic pores into the interior of the mesocœle.

The metac œ les appear to have fused, and to lie asa shallow (from side to side) cœlomic sac dorsal to the stomach and intestine. They are surrounded in front and on either side by the mesocœles, and their walls form with them a conspicuous pair of dorso-lateral mesenteries (fig. 48), and a median transverse mesentery (fig. 47).

Fig. 50 is a semi-diagrammatic representation of a half-larva at this stage, cut in the median sagittal plane. It serves especially to illustrate the position and inter-relationships of the cœlomic cavities at this stage.

It is clear that very few changes are necessary in order to change this larva into the free Actinotrocha withfive pairs of tentacles, the structure of which has been described in a previous paper. In the epiblast the preoral senseorgan and the subneural gland have yet to appear, - and possibly a proctodæum, whereas a pair of protocœlic pores do not seem to have been yet formed. In the hypoblast the pleurochords and the partial separation into pharynx and stomach are still required. The later growth is largely a protuberance of the anal papilla and the surrounding parts, and with it an increase in size and extent of the metacceles until they would meet ventrally to form a ventral mesentery, whilst the perianal band would form later.

Metschnikoff (13) found pelagic Actinotrocha larvæ with no more than one pair of tentacles, and traced them up to the later stages, so that the presence of a proctodæum and the fuller development of the collar nephridia are among the few points still requiring elucidation. Metschnikoff’s larvæ belong to one of the smaller species at Naples, and it is important to note that they leave the tentacles of the parent at a much earlier stage than is the case in Phoronis Buskii or P. australis.

397

In comparing these facts with the previous work of others, one can divide them into two series. The first of these consists of facts which have been observed by external examination and connected with the external differentiation of the embryos, including the phenomena of segmentation, gastrulation, and the fate of the blastopore ; the second are the changes undergone by the tissues within the embryo, especially connected with the phenomena of gastrulation and the formation and subsequent fate of the mesoblast. With regard to the first series all observers agree very closely. Segmentation has in all cases been recognised as being nearly equal, and gastrulation embolic. The blastopore appears to survive (in part at least) as the mouth in all the species, but all do not recognise the fusion of the blastoporic lips posteriorly. Metschnikoff (14) followed and figured the fate of the blastopore in detail, and he notes that it is at first round, then becomes reduced in size, and then presents a pointed oval outline. He also describes a median longitudinal furrow running backwards from the reduced blastopore to the posterior end of the embryo. Caldwell (3) followed with another account, which describes a similar ventral furrow (his so-called primitive streak), but in addition he finds a posterior pit, which he regards as the posterior part of the blastopore. Metschnikoff’s account would be quite as true for Phoronis Buskii as for his own species. Schultze (16), working upon a species in the Black Sea, finds no trace of this ventral furrow.

The difference may all be probably accounted for as due either to defective observation or to specific variability.

Coming to the second series of phenomena a very different condition prevails. All previous observers with the exception of Caldwell appear to have relied upon the observations of entire embryos in optical section, and this method, when applied for the elucidation of important internal changes, can be regarded as neither reliable nor conclusive. Of these Kowalevski (9) held that the mesoblast arose by delamination from the hypoblast, and, so far as it goes, this agrees with the present results.

Metschnikoff (14) described and figured mesenchyme cells present in the blastocœle cavity at an early stage before gastrulation, aud Fœttinger (7) went even further, and claimed to recognise the mesoblast cells at an even earlier stage. Caldwell (3), iu applying the method of sections to the question, failed to corroborate these observations, and attempted to explain Metschnikoff’s figures on other grounds. He failed to find any trace of mesoblast till well on in the process of gastrulation. Roule (15), who followed him, also failed to detect any mesenchyme cells until the preoral lobe and the mouth were established. As stated above, there are no figures, and Roule gives no account of his methods. Additional interest in the question has been roused by a recent communication of Schultze (16), in which he reasserts the origin of the mesoblast from mesenchyme cells. At the present stage of the paper it will be sufficient to note that he claims to find a mass of mesenchyme cells scattered throughout the blastocœle cavity at a very early stage before gastrulation. Again we are led to inquire, what were the methods pursued ? Schultze gives three woodcut figures which may or may not have been drawn from sections, and this is all.

Such being the case, special care was taken in this instance to find out whether there are or are not mesenchyme cells present in the early stage. -The sections were stained very deeply with eosin and hæmalum, and the nuclei were depicted with remarkable clearness. In the great number of embryos I have examined I have failed to find a single nucleus in the blastocœle cavity till after gastrulation. The most one ever finds are a few cytoplasmic strands and frag-ments, which are stained with eosin alone, and show no trace of nuclei. In all the stages up to the spherical blastula all the nuclei which are present are arranged symmetrically in the ectoderm cells as in the figures. The blastula is simply a single-layered sphere with each nucleus in its place, enclosing a spacious blastocœle space. In this respect Phoronis Buskii appears to resemble the species examined by Caldwell (P. Kowalevskii) J

The results of Kowalevski (9) and Roule (15) are in accordance with what an exact observer would see of the development of mesoblast in Phoronis Buskii by an examination of transparent embryos. Roule probably observed the anterior mesoblast cells after their origin from hypoblast, and the posterior masses (metacceles) which he called meso-blastic bands.

In the case of Fœttinger (7) his figures and remarks upon them would lead us to suppose that the bodies he describes are neither cells nor nuclei, and do not give rise to the mesoblast; whilst the entire absence of mesenchyme at early stages in sections as shown by Caldwell, and confirmed by myself above, cast grave doubt upon the results of Metschnikoff and Schultze. It seems more natural to make an appeal to the evidence of sections in a case of this kind.

With regard, however, to the actual development of the mesoblast from the hypoblast there are considerable discrepancies between Caldwell and myself. He describes and figures a pail’ of “modified archenteric diverticula,” which are evidently identical with the rudiments of the collar ccelom as described above. From them he derived the preoral cœlom. In this I am inclined to think he was misled by the growth backwards of the two horns of the preoral cœlom, which he figured in precisely the same position as above in fig. 37. The lateral masses of mesoblast, instead of growing forwards as he stated to form the preoral cœlom, grow backwards and ventralwards to give rise to the collar or tentacular cœlom.

The rest of the coelom he derived mainly from a pair of posterior pouches which arose in connection with the “posterior pit”—a remnant of the posterior part of the blastopore. Schultze emphatically denies the existence of these posterior pouches, and in this, as far as P. Buskii goes, I can corroborate him. Many weeks’ search amongst hundreds of embryos whole and in section has failed to bring to light a single structure bearing the slightest resemblance to either the posterior pit or the posterior ccelomic pouches. Schultze (16) suggests that Caldwell was misled by the ventral invagination which gives rise to the adult trunk, but this cannot be the case, as this organ does not appear till long after the larvæ have left the parent.

Caldwell (3) figures these appearances not only in P. Kowalevskii but in P. australis as well, and although I have failed to find them in the latter I merely state the fact, and defer further consideration till I am able to examine the other species with fresh material.

In the first of this series of papers on the Diplochorda the structure of the late stage of Actinotrocha was compared in some detail with that of Balanoglossus. The facts above recorded of the early development serve to emphasise this comparison. There are at present known at least two types of development in the Enteropneusta, the direct development of Balanoglossus Kowalevskii (?) known to us through the work of Bateson (la), and the so-called indirect development with Tornaría larva. The early stages of this indirect development do not appear to be known, though one would expect to find the development of Phoronis with its free-swimming larva Actinotrocha conform more closely to this type. However, taking Bateson’s type for comparison we find that the stages of segmentation are identical. In both cases there is total seg-. mentation with very slightly pronounced inequality in the blastomeres, or, in other words, subequal segmentation. Gastrulation is in each case total, and Bateson’s fig. 5 would serve equally well for a stage of Phoronis. The subsequent behaviour and fate of the blastopore is markedly different. In Balanoglossus it closes up gradually and completely, the hypoblast becoming completely separated from the epiblast. Subsequently the anus opens at about the same spot, as far as can be judged, as that at which the blastopore closed, and the mouth appears ventrally as a new structure. In the case of Tornaría the young stages do not appeal’ to have been followed prior to Goette’s larva. In Phoronis it is certain that the mouth is a persistent anterior portion of the blastopore, and it is pretty clear that the anus opens subsequently at the posterior closed portion of the blastopore.

Schultze (16) emphasises the different fate of the blastopore in criticising my comparison of Balanoglossus and Phoronis. Those who accept the view that the blastopore becomes phyletically both the mouth and anus of the Cœlomafa will not regard the fact that it persists in a special ontogeny as mouth or anus, or both, or neither, as having any bearing upon the question ; but for those who hold other views one may recall the behaviour of the blastopore in other groups. In the Echinoderm at a the blastopore usually becomes the anus, but does not survive as mouth or anus in Antedon, nor in Asterina. Again, within the group of Gastropod Mollusca the blastopore becomes the mouth ip Patella (Patten), Chiton (Kowalevski), and Nassa (Bobretsky), whilst it survives as the anus in Paladina (Lankester) and numerous others.

In the face of such examples as this we can hardly attach phyletic relationship or otherwise to a comparison of the behaviour of the blastopore. In Chiton the blastopore would appear to migrate from a terminal position, where the anus opens later, to the position where it survives as the mouth. In this, as in Balanoglossus Kowalevskii, it is possible to follow this migration owing to the fact that there is present in each case a ring of cilia. Such a migration would be very hard to follow, except by direct observation of a developing embryo, in such a type as Phoronis. The larva of Balanoglossus Kowalevskii differs from Tornaría in developing the perianal band very early alone, whereas the latter develops this band after the postoral band. In this respect Actinotrocha resembles the free-swiinmiug Tornaría, as the perianal band has not appeared in P. Buskii even in the larva with three pairs of tentacles.

I hope to show later elsewhere that in Tornaría and other larvæ, such as those of Echinoderms, the ciliated bands can be classified by their primary functions into trophic and motor. Of these the postoral band of the Echinoderms, Tornaría, and Actinotrocha is trophic, whereas the perianal band of Ternaria, Actinotrocha, and the circular bands of Auricularia and Antedon are motor. Thus in Tornaría and Actinotrocha, with free larval life at a very early stage, the trophic bands are essential, whilst the perianal motor band is added in due course. In the demersal type of Balanoglossus Kowalevskii the mouth and anus are closed till late ; no pelagic food is required, and the postoral band does not appear, whereas the perianal band is developed early for the locomotion of the larva. The same considerations apply to Antedon and pupal Auricularia.

After closure of the blastopore the next essential comparison is in the development of the mesoblast. In Phoronis, as we have already seen, there has been a great difference of opinion upon this point. The following facts appear to be indicated.

1. The mesoblast arises from five separate parts of the archenteric hypoblast, of which one is preoral and unpaired, and the other four are paired and postoral.

2. The preoral mesoblast is in the form of a portion of the archenteron which is pinched off from it, the cavity being part of the archenteric cavity shut off in course of development.

3. The first pair of postoral elements (collar-somites) are formed of masses of cells segmented from the distal ends of the hypoblast cells which lie opposite to slight depressions in the archenteric walls. These cells at first have no cavity.

4. The second pair of postoral elements (trunk-somites) arise at the anal extremity in a similar way to the former, but they may consist of invaginations of hypoblast in which there are two walls in close contact. Under any circumstances these mesoblastic elements come to lie between hypoblast and epiblast as paired masses with no cavity.

As regards 1, there is an identity in type with Balano-glossus.

2. The development of the preoral ccelom in Tornaría does not appeal* to have been followed except in the case of Goette’s larva, in which it is indicated as in course of invagination from the gut. Morgan (14a), however, has thrown doubt upon this interpretation. In any case the preoral ccelom arises much earlier in Tornaría than the rest of the mesoblast, and this agrees exactly with Phoronis, as has been shown above.

In the case of the demersal larva of Bateson (la) the data are definite, and the preoral ccelom appears to arise in a manner precisely similar to that of Phoronis. In both cases it commences to be separated from the archenteron at the time when the proboscis ou the one hand, and the preoral hood on the other, are commencing to be differentiated externally. In both cases this cœlomic pouch, after separation from the archenteron, grows backwards laterally and dorsally, forming a pair of lateral horns. “The pouch of mesoblast grows backwards, surrounding the gut except on the ventral surface, but especially forming the hollow horns lying in a horizontal position, one on each side of the gut” (Bateson, loc. cit., p. 220).

The only difference appears to lie in the earlier separation of the coelomic pouch from the archenteron in Phoronis, and in this feature it resembles Tornaria.

3. The development of the collar-somites (mesocœles) in

Tornaría has been carefully followed by Morgan (14a). He finds that there is “a process of proliferation of the walls of the stomach, so that the wall at this point, by division of its cells, becomes two-layered.” The mesoblastic mass, thus derived, has at first no cavity. The whole process is precisely similar to that of Phoronis, as described above.

In the demersal larva Bateson states that the collar mesoblast arises by a modified form of archenteric diverticula. “This pair of cavities is bounded on the inner sides by the cells forming the wall of the gut, and the external boundary is made up of a single layer of cells continuous dorsally and ventrally with the hypoblast.” As a matter of fact, Bateson’s reasons for regarding them as archenteric diverticula appear to be two. Firstly, there is a connection of their cavity with that of the archenteron which occurs in “very few larvæ secondly, he notes that “the middle mesoblastic tracts in Tornaría are said to be archenteric diverticula.” Under any circumstances the inner walls would appear to arise from the hypoblast in a similar manner to the mesocœle in Tornaría, according to Morgan (14a), and the mode of formation is really a type intermediate between the formation of archenteric diverticula and delamination. Bateson points out that these mesoblastic pouches extend principally posterior to their point of origin, which is also the case in Phoronis.

4. The third pair or trunk somites arise in Tornaría (Morgan) as a pair of hypoblastic outgrowths, two-walled from the first, but containing no cavity till after separation from the hypoblast. In the demersal larva they arise as a pair of archenteric diverticula. In Phoronis I am unable to state with certainty that they arise either as hypoblastic outgrowths actually involving the wall of the archenteron, or whether they are formed by active division from the distal ends of archenteric cells, but very probably by one of these two methods. In this respect Phoronis resembles Tornaría rather than the demersal type. Bateson (la) states that the wall separating the trunk somites from the archen-teron is last closed on the dorsal side, and in Phoronis the trunk somites are dorso-lateral in origin, whilst his section (fig. 33) shows the collar somites to be ventro-lateral, in this respect also resembling the homologous structures in Phoronis.

It will thus be seen that the formation and relationships of the several parts of the mesoblast in Phoronis resemble even in minute particulars that of Balanoglossus, the resemblance in mode of origin being rather closer to Tornaría than to the demersal type, though in many respects the latter differs more from Tornaría than from Phoronis.

My comparison of Actinotrocha with Balanoglossus and Cephadiscus receives very little modification with further work upon the subject. Professor L. Houle (15a) has published one or more papers in which he states that he is unable to find in the Actinotrocha of Phoronis sabatieri certain structures which I have described in the Actinotrocha at St. Andrews; perhaps further research may clear up this difficulty. He corroborates the description given of the chordoid structures in Actinotrocha, but is inclined to regard this larva as closely allied to the trochophore of the Annelida. My views of the relationship of the trochophore to Actinotrocha are sufficiently clearly indicated in a recent paper (10a), but it may be noted here that Roule makes no reference to the five coelomic cavities which I have figured and described in Actinotrocha, though the mesenteries separating them can be seen with a hand lens in the living or mounted larva !1 His paper has no figures, and his methods are not described. Roule’s account of the early development of the mesoblast in Phoronis so differs from mine and from that of others that it is not surprising that there should be a like discrepancy in description of the later stages.

The anatomy of Phoronis Buskii and that of Phoronis australis, a closely allied species, have been described by more than one worker, and it is here proposed merely to add a few fresh observations to the facts already known.

Preservation in alcohol does not appear to affect to any extent the black pigment of Phoronis australis. The arrangement of this is somewhat peculiar. The lophophore or crown of tentacles is banded. Figs. 51 and 52 are dorsal views of the entire animal (natural size). The tentacles appear to be black throughout the greater part of their length, except for a light unpigmented area extending horizontally across each coil at about three fifths of the distance towards the base. Another narrower band is found at the base of the lophophore, under which lies the nerve-cord. Fig. 54 is a ventral view of another specimen, in which the white band on the tentacles is seen to extend down to the base of the lophophore, thus meeting the lower band.

There is, however, considerable variation in the coloration, though these three illustrate the general rule. Fig. 54 shows an abnormal specimen, in which the whole lophophore is inky black, with no bands.

The trunk is pigmented for a very variable extent of its length, as is seen by figs. 51—54, in which a dotted line indicates the transition from the pigmented to the unpigmented areas. It is possible that the unpigmented portion is that part which during life was embedded in the skeletal tube of Cerianthus, which also appears to be dark in colour. All the pigment is deposited as minute black granules in the cells of the ectoderm, but whilst the pigment in the trunk is uniformly distributed in the cells, that in the lophophore is confined to the ectoderm of the cubical type, which is found only on one side of the tentacles, Ectoderm.

The ectoderm covering the trunk is corrugated (fig. 55) into circular ridges, which have been supposed to be due to contraction caused by spirit. An inspection of the figure would lead one to doubt this, as the reduplications appear to be very regular, and the thin cuticle runs from tip to tip of the corrugations instead of following the course of the corrugations. The cells are elongated and have elongated nuclei. On the left hand the presence of black pigment spots is indicated. At the base of the ectoderm cells is seen the fine plexus of nerve-fibres which is very characteristic of Phoronis as of Balanoglossus. Below this again is the chondroid tissue.

At the junction of collar and trunk the nerve-plexus becomes hypertrophied into a massive postoral ring of nerve-cells and nerve-fibres (fig. 56). The inner ends of the ectoderm cells can be traced in some cases into nerve-cells, and in others into nerve-fibres. The nerve-cells rather tend to accumulate close under the chondroid tissue, but every transition stage can be selected from the densely crowded long ectodei’m cells to the nerve-cell removed from the surface. Fig. 57 shows some of the transition stages ; they indicate how intimately the nervous system still is bound up with the ectoderm in this group.

The ectoderm of the collar or lophophore is of two types, which I have elsewhere described as branchial and atrial epithelium respectively.

From other work upon the alimentary processes of Echinoderm and other larvæ I have been led to more generalised names for these types of ectodermal epithelium. A great number of facts seem to point to the supposition that the method of food ingestion, by the activity of ciliated areas causing currents of water, which in their turn carry microscopic food particles, is the primitive method for the early Metazoa from the gastrula stage onwards. This method of food ingestion may be termed cilio-trophic, and is found very generally in the lower types of Metazoa, especially sedentary types in which the cilia, no longer motor, are entirely trophic in function. The first important differentiation in cilio-trophic alimentation is the separation and retention of food particles and the removal of the water-current. The active areas which are trophic have a characteristic form of epithelium which may be termed tropho-phoral, and the areas along which the return water-current is removed have another type of epithelium which may be termed hydrophoral. Trophophoral epithelium is made up of densely crowded elongated cells, covered with actiW cilia, and very commonly having glandular cells added to them. Hydrophoral epithelium has flat or cubical cells, usually non-ciliated or with few cilia, and often pigmented or even actively excretory.

For the separation of the food particles from the water-current at least two important general methods are resorted to. The first is the principle of filtration, in which water is allowed to pass through certain apertures and food retained, or vice versa; and the second is by an entanglement of the food particles in mucus strands which allow the water-currents to pass in other directions, whilst they themselves are conveyed into the alimentary canal.

The first system is probably the most primitive, and in almost every case the second is superadded to it. I have attempted elsewhere (10b) to show that a great number of the characteristic organs of the Chordata, including noto-chord, hypochorda, pharyngeal clefts, thyroid gland, and hypophysis, may be traced to an origin which they subserved in either one or other of these methods of food ingestion, and the matter will be further dealt with later.

In Phoronis both methods are exemplified very perfectly. In the case of water-filtration it is evident that the greater the surface of contact which is effected between the tropho-phoral areas and the hydrophoral areas, the greater will be the efficiency of the filtration attained. The simplest relationships obtain in the larvæ of Echinoderms in which the whole body-surface is divided into trophophoral and hydro-phoral areas, and the pinnæ or “arms” and processes may be all, as I hope to show later, accounted for as contributing to the efficiency of filtration. In Actinotrocha the line of contact between the two areas is at first multiplied by a simple row of tentacles, throughout the length of which the line of contact passes (cf. figs. 59—62).

After fixation the young Phoronis is provided in like manner with a simple row of tentacles round the oral aperture, and the subsequent history of the Phoronidea may be traced as a successful attempt to increase the efficiency of filtration by a gradual multiplication of the number of tentacles.

The known species may be placed in order from Cori’s (4) table :

From this it will be seen that the increase in size corresponds very closely with the number of the tentacles and their thickness. Probably the comparative length of the tentacles follows the same rule. As the tentacles seem to serve the ingestion of food, the increase of ciliated surface on the one hand, by increase in number, length, and thickness of the tentacles, and the increase in bulk of the body, on the other hand, are probably directly connected with each other.

The tentacles in transverse section show that their inner or oral surface is covered with trophophoral epithelium and their outer surface with hydrophoral, and bear a remarkable structural resemblance to the ctenidia of Lamellibranchiata, the tentacles of numerous sedentary Annelida and Brachiopoda, Cephalodiscus, and even the branchial filaments of Tunicata and Amphioxus. These resemblances depend upon a similar method of food ingestion, the cilio-trophic, in each of the groups, and it seems fairly cleai’ that in each case the branchial function has been only secondarily added. In Phoronis not only does this filtration method hold, but the trophophoral epithelium has glandular cells distributed in it. As the result of the activity of these cells fine strands of mucus pass down the oral side of the tentacles and serve to entangle minute algoid bodies. In fig. 8 these strands with their contents may be seen passing along the oral side of the epistome into the oesophagus.

Lastly, the epistome itself is so arranged in relation to the rest of the lophophore that it acts as an organ to remove the water from the oral aperture.

Figs. 60, 61, and 62 are intended to illustrate this point. The coils of each half of the lophophore are arranged spirally, and the tentacles are widely separated from each other for about their distal two thirds and united together laterally for the proximal one third. A transverse section of the lophophore may illustrate the arrangement at different heights.

Thus in fig. 59 the outermost half-coil shows the tentacles cut in transverse section, each having the trophophoral epithelium on its oral surface and filtration spaces between contiguous tentacles. For the rest of the course of the coil the deeper parts are cut in which the tentacles are fused together laterally, so that the tentacular filtration method no longer acts at this depth. But not very far up, at the beginning of the second coil in the figure, the epistome may be seen in transverse section projecting across the floor of the oral aperture for a little over a coil in length. After this the section is too deep to cut the epistome completely with its ectodermal epithelium, but its cœlomic cavity can be seen in transverse section up to the very tip of the spiral. Benham (1) criticises Allman’s description of the epistome, and remarks that Allman’s figure “conveys quite a wrong idea of the organ.” The same remark would be equally applicable to his own. He described the epistome as extending “right across the oral side of the animal from right to left,” and figures it in the same way (fig. 7). As a matter of fact it is continued as a conspicuous ridge on either side to the very tip of the spiral, continuous throughout the three coils of the lophophore. If the tentacles be regarded as forming the incomplete lateral walls of a spirally coiled chamber, then the epistome forms the floor of this chamber except for a fissure between it and the outer row of tentacles through which the mucus strands pass down into the oesophagus (fig. 58). This floor descending from the apex to the base of the spiral must necessarily serve to prevent access of the greater portion of the water-current into the oesophagus, and to cause it to flow out in the median line through the gap in the inner row of tentacles where it would join the current along the hydrophoral area of the outer surface of the tentacles.

The figures 60, 61, and 62 should make this clear. In fig. 60 the lophophore is viewed from above, and the epistome (shaded) is seen projecting over the mouth. The currents are shown by the arrows descending the coils and passing through the tentacular gap to escape over the anal area.

Fig. 61 is a section through fig. 60 at A, showing the course of the water first separated by the intertentacular filtration, and further down the course of the water impinging upon the upper surface of the epistome. In fig. 62, a section through B, this water is seen to pass out through the tentacular gap, whilst the mucus strands continue their course down the oesophagus. The epithelium of the trophophoral area is indicated by thick lines, that of the hydrophoral by thin.

My labours upon the Diplochorda (Phoronis and Cephalodiscus) have now extended over some time. We may here refer to some of the results.

The first of these is that Actinotrocha has a pair of organs which have at least as great a claim to be regarded as rudimentary chordoid organs allied to the notochord of the Vertebrates as have the so-called notochord of Balano-glossus and Cephalodiscus. Roule (15a) has confirmed the chordoid structure of them, although he differs in other points.

In 1897 (10) it was pointed out that the chordoid structure of the gut in Actinotrocha was not confined to the two diverticula, but was also found in the mid-ventral portion of the stomach just opposite the sac-like invagination which later gives rise to the trunk of the adult.1 This alone showed the need for caution in drawing too close homologies, and although the hypothesis was made that the two pleurochords represented the paired rudiments of the single notochord of the Vertebrates, it was with some reservation.

A study of Balanoglossus itself shows that the chordoid metamorphosis of the gut epithelium is very widely present in the various parts of the alimentary canal, especially in the pharynx, in parts of which it extends completely round the whole pharyngeal wall. This is so easily demonstrated (it is referred to by Spengel) that it is surprising, firstly, that the fact is universally ignored in text-books ; and secondly, that in the face of it the small preoral diverticulum alone should have been selected for comparison with the notochord of the Vertebrata (cf. Bateson). Considering that the organ, lies preorally, its great claim for notochordal recognition must rest in the fact that its cells are metamorphosed into vacuolated tissue similar to that of the vertebrate notochord, and the value of this comparison is considerably lessened when it is considered that the dorsal part of the gut in the collar region, and even beyond it, is chordoid, and is in a position relative to the nervous system more closely resembling that of the vertebrate notochord. The structure of Actinotrocha and Bala oglossus points out clearly that chordoid modification may occur in any part of the hypoblastic organs. On the other hand, in Ce phalo-discus, and, as I hope to show elsewhere, in Tornaría, the chordoid metamorphosis is restricted to a pair of lateral grooves along the pharynx, so that we are justified iu supposing that this represents the earliest condition of the Archichorda, and that the Diplochordate condition has become disguised in adult Balanoglossus by a further spreading of the chordoid metamorphosis. Hence in Balanoglossus there does not appeal* to be any one organ which can be compared directly to the notochord of the Euchorda.

The phyletic history of an organ appears to consist of the following steps :

1. The function and structure are co-extensive with the organism itself, ox* very early limited to eithei* the endoderm, ectoderm, Or mesoderm.

2. The function becomes concentrated in a certain part of the primary layer, and the part itself then becomes differentiated structurally from the rest of the layer.

3. The function is so fully defined as apart from that of the primary layex* as a whole, that the organ also becomes organically separated from the parent layer. In a natural group it is to be expected that the lower members will exhibit stages 1 and 2 of organs which are in stage 3 in the higher.

It has been suggested to divide the Chordata into two main groups, Archichorda and Euchorda, the former comprising Balanoglossus, Cephalodiscus, Phoronis, and possibly the Echinodermata ; whereas in the latter there are included the Urochorda, Cephalochorda, and Vertebrata.

The Arçhichorda are regarded as being the lower members of the group, and we may inquire, How do they stand the test of the organismal relationships above referred to ?

In the case of the nervous system, stages 1 and 2 are prevalent in all the Archichorda, a diffuse nervous plexus (stage 1) and sundry definite nerve tracts being characteristic of the group ; whilst in the Euchorda the central and peripheral nervous systems are alike organically separated from the parent ectoderm.

Applying the same test to the notochord, we find that in the Archichorda a large and more or less indefinite part of the endoderm is metamorphosed into chordoid tissue in accordance with the needs of each particular type. (A cross-section of the pharynx of Balanoglossus shows this clearly.) At the same time there are indications of stage 2. The chordoid tissue becomes definitely located at certain parts, and these become more or less distinct structurally as organs (pleurochords) still in organic continuity with the rest of the endoderm.

In the Euchorda the notochord has completely separated from the endoderm in the adult.

This method of looking at the matter will show clearly the relationship of the chordoid organs of the Archichorda to those of the Euchorda. It may be possible with further research to indicate an actual organ among the former which may be regarded as the direct homologue of the notochord, but it will probably be in some larval form with an unpaired dorsal rudiment, not in adult Archichorda.

The same relationships might be pointed out in the meso-dermic organs. The mesoderm in the Archichorda merely shows archimeric segmentation, and does not show a complete separation into various organs. In the Euchorda the mesoderm is not only metamerically segmented, but early divided up into separate organs, such as nephrotome, myotome, sclerotome, etc.

Lastly, in the case of pharyngeal clefts, I have attempted to indicate elsewhere (10b) that the origin of these organs must be sought for in the grooves at the corner of the mouth which, very early even in the history of the Archi-chorda, become segmented backwards from the mouth. In this way I would regard the pharyngeal clefts as originating primarily from the mouth, the inverse to Dohrn’s hypothesis, which regards the Vertebrate mouth as a pair of fused pharyngeal clefts, a view which does not appear to take sufficient account of physiological differentiation.

In part 1 (10) it was shown that Actinotrocha has five coelomic cavities, with many of the relationships of the five cavities in Balanoglossus; aud they were compared not only with these, but with the five cavities which appear to be au integral part of the constitution of a great number of the lower Coelomata. In this last part of the work the origin of these cavities is shown to be similar to that of the same organs in Balanoglossus.

Roule (15a) in a recent paper disagrees with my conclusion on the relationship of Phoronis to the Hemichorda (so called), and attempts to regard Actinotrocha as a trochophore. The body-cavity of a trochophore is a hæmocœle, whereas the hæmocœle is restricted in Actinotrocha to small spaces between the coelomic sacs. Until Roule is in a position to demonstrate the absence of these five cœlomio sacs one need hardly attach much importance to his comparisons of Actinotrocha with a trochophore nor to his refusal to accept my enlignment of the former with the Enteropneusta.

ST. Andrews ; March, 1899.

Illustrating Mr. Arthur T. Masterman’s paper “On the Diplochorda.”

ABBREVIATIONS.

arch. Archenteron. bp. Blastopore, be. Blastocœle. ent. Enteren, m. Mouth, wise. Mesocœle (collar cavity). mtc. Metacœle (trunk cavity). ng. Nerve ganglion, p. Preoral hood. pc. Protocœle. sus. Subneural sinus. st. Stomodæunl. t’, P, t3. Pairs of tentacles.

PLATE 18.

FIG. 1.—Embryo in egg-membrane.

FIG. 2.—Embryo with two blastomeres; side view.

FIG. 3.—Embryo with four blastomeres; side view.

FIG. 4.—Embryo with four blaslomeres; ventral view.

FIG. 5.—Embryo with eight blastomeres; side view.

FIGS. 6—10.— Lateral views of series of larvæ.

FIGS. 11—15.—Ventral views of series of larvæ.

FIGS. 16—18.—Lateral views of late larvæ.

FIG. 19.—Ventral view of Fig. 18.

FIG. 20.—Mucus band with embryos.

PLATE 19.

FIG. 21.—Transverse section of Fig. 3.

FIGS. 22—24,—Longitudinal sections of embryo later than Fig. 5.

FIG. 25.—Longitudinal sagittal section of Fig. 11.

FIG. 26.—Longitudinal sagittal section of Fig. 12.

FIG. 27.—Longitudinal sagittal section of Fig. 13.

FIG. 28.—Longitudinal coronal section of Fig. 13.

FIG. 29.—Longitudinal sagittal section of Fig. 9.

FIG. 30.—Longitudinal sagittal section of Fig. 10.

FIGS. 31, 32.—Longitudinal coronal sections of Fig. 10.

FIG. 33.—Transverse section of Fig. 10.

FIGS. 34—39.*—Series of transverse sections of Fig. 16 (early).

FIGS. 40—44.—Transverse sections through late stage of Fig. 16.

PLATE 20.

FIG. 45.—Transverse section through larva of Fig. 16.

FIGS. 46, 47.—Longitudinal sagittal sections through Fig. 18.

FIGS. 48, 49.—Transverse sections through Fig. 18.

FIG. 50.—Semi-diagrammatic half-larva at stage of Fig. 18.

ARTHUR T. MASTERMAN.

FIGS. 51—54.—Lateral views of Phoronis australis, natural size.

FIG. 55.—Longitudinal section through body-wall of trunk region oi Phoronis Buskii.

FIG. 56.—Transverse section through nerve-ring of P. Buskii.

FIG. 57.—Isolated elements from Fig. 58.

FIG. 58.—Longitudinal section through epistome of P. Buskii.

FIG. 59.—Transverse section of lophophore of P. Buskii.

PLATE 21.

FIG. 60.—Diagram of transverse section through lophophore of P. buskii, showing currents.

FIG. 61.—Longitudinal sagittal section through lophophore of P. buskii, showing currents. The section is supposed to pass through A. in fig. 60, but the coil of end coil in fig. 60 should be twisted a little more, so as to be cut once more by the line. (Cf. fig. 59.)

FIG. 62.—Median sagittal section through lophophore of P. Buskii, showing currents. The section is supposed to pass through B. in fig. 60.

1.
Benham
,
W. B.
—‘
Quart. Journ. Mier. Sei.,’ July, 1889. la. Bateson, W
. —
‘Quart. Journ. Mier. Sei.,’
vol.
xxiv
,
1884
.
2.
Caldwell
,
W. H.
‘Proc. Royal Soc.,’
London, xxxiv
,
1882-3
.
3.
Caldwell
,
W. H.
‘Quart. Journ. Mier. Sei.,’
vol.
xxv
,
1885
.
4.
Com
,
I.
“Inaugural Dissertation,”
Prague
,
1889
.
5.
Com
,
C. J.
‘Zeitschrift wiss. Zool.,’
Bd.
li
,
1891
.
6.
Dyster
,
F. D.
‘Trans. Linn. Soc.,’
London
, vol.
xxii
.
7.
Fœttinger
,
A.
‘Arch, de Biol.,’
tom.
iii
,
1882
.
8.
Kowalevski
,
A.
‘Mém. Acad. Impér.,’
St. Petersburg
, t.
x
.
9.
Kowalevski
,
A.
“Inaug. Dissertation,”
St. Petersburg
,
1867
.
9a.
Lankester
,
E. R.
“Polyzoa,” ‘Encyclopædia Brit..,’
1885
.
10.
Masterman
,
A. T.
‘Quart, Journ. Mier. Sei.,’
vol.
xl
,
1897
.
10a.
Masterman
,
A. T.
‘Proc. Royal Soc.,’ Edinb
.,
June
,
1898
.
10b.
Masterman
,
A. T.
‘Report Brit. Assoc.,’
1898
.
11.
McIntosh
,
W. C.
“Challenger Reports,”
vol.
Ixxv
,
1888
.
12.
McIntosh
,
W. C.
“Challenger Reports,”
vol.
xxviii
,
1888
.
13.
Metschnikoff
,
E.
‘Zeitsch. f. wiss. Zool.,’
Bd.
xxi
,
1871
.
14.
Metschnikoff
,
E.
‘Zeitsch. f. wiss. Zool.,’
Bd.
xxxvii
,
1882
.
14a.
Morgan
,
T. H.
‘Journ. of Morph.,’
v
,
1891
.
15.
Roule
,
L.
‘Comptes Rendus Ac. Sci.,’ Paris
, tom.
ex
,
1890
.
15a.
Roule
,
L.
‘Comptes Rendus Ac. Sci.,’
Paris
,
Oct
.,
1898
.
16.
Schultze
,
E.
‘Imp. Soc. Nat..,’
St.. Petersburg
, vol.
xxviii
, pl. 1.
1

I am indebted to Mr. A. E. Shipley, of Christ’s College, Cambridge, for kindly sending me some original drawings of the earliest stages of P. Kowalevskii, which appear to resemble closely the similar stages as given in this paper.—A. T. M.

1

As already noted (‘Quart. Journ. Mier. Sei.,’ August, 1897), they were correctly figured by R. Wagener as early as 1847, but were supposed by him f,o be nerves,—A. T. M.

1

This appears to be in the same position as the so-called “pygochord “of Willey, occurring in certain species of Enteropneusta.—A. T. M., March, 1900.