This work was undertaken at the suggestion of Professor W. K. Brooks for the purpose of getting, if possible, some further hint upon the significance of the larval form in Echinoderm phylogeny. It was carried on from October, 1889, to April, 1891; from June to October, 1890, at the Laboratory of the United States Fish Commission at Wood’s Hall, Massachusetts, where the living animals were studied, and for the remainder of the time in the Biological Laboratory of the Johns Hopkins University, where the work was upon the preserved material. My heartiest acknowledgments are due to Professor Marshall McDonald, U.S. Fish Commissioner, for the advantages furnished at the Government Laboratory; and to Professor Brooks, of this university, who has so kindly placed at my service the preserved material, the preparations, and drawings, which formed the basis of his recent paper before the National Academy (4).

The material was obtained by means of the surface net, and was supplemented by that obtained by artificial fertilisation. The larvae reared by the latter process were kept in glass beakers containing fronds of Ulvaceæ, which served both for aeration and by the liberation of zoospores furnished to some extent food for the developing larvae. The larvae were daily transferred with a pipette to beakers of fresh sea water. The time most favorable for finding an abundance of sexually ripe starfish at Wood’s Hall is the month of June and in early July. The larvæ are to be found in considerable numbers at the surface during June, July, and August. Those thus obtained were immediately examined, figured as live objects, and then killed separately and hardened for sectioning. All the stages of larval development have been studied (1) in the living condition ; (2) as total preparations; (3) the results confirmed by sectioning. For killing I found that Kleinenberg’s picric salt gave the most satisfactory results, particularly in the younger stages. Flemming’s, followed by Merkel’s fluid, gave excellent results, as did also Perenyi’s fluid. Oil of cedar or of origanum proved most satisfactory for clearing.

—The ovary is a very large compound racemose gland, with a great number of spherical alveoli. When sexually mature it completely fills up the cavity of the arm. Its colour at this time is a delicate tint of salmon. In a cross-section of an alveolus is shown an external covering, the peritoneal membrane, consisting of a single layer of cubical cells (fig. 1,p. e.). Next, internal to this, is a muscular layer of considerable thickness (m.). This is made up of fibres running in every direction; the outer and inner parts, however, are made up chiefly of fibres running nearly at right angles to each other. Within this muscular layer, and lining the lumen of the alveolus, is the germinal epithelium (g. e.), showing cells in various stages antecedent to their separation to form ova. In the earlier stages the size of the nucleus is large in proportion to the cytoplasm, but later the cytoplasm increases very rapidly. In fig. 1 the lumen is completely filled with mature eggs, which have assumed various shapes from mutual pressure. The spaces between the ova are filled by branching cells and fibres of connective tissue. Each egg has a thin gelatinous external membrane, which upon contact with the water swells by imbibition to a considerable thickness.

The condition of the ovary a few weeks after the discharge of the eggs is shown in fig. 2. The lumen is seen to be traversed by the radiating branched cells, while the germinal epithelium is very much thicker, and crowded with nuclei of various sizes, which will form the next crop of eggs.

—The testis (fig. 3) differs histologically from the ovary only in a less development of the muscular layer (m.) of its wall, and in the smaller size of the cells of the germinal epithelium (i. e.). These cells separate from the epithelium, and by the increase in size and final separation of the germinal epithelium-cells behind them are pushed towards the centre of the lumen. These separated cells are the sperm mothet-cells of the spermatozoa (fig. 3, s. m. c.). Their diameter is many times less than that of the ova at a corresponding stage. Each sperm mother-cell, in its progress towards the centre of the lumen, divides into two smaller cells. In the further progress towards the centre of the lumen each of these smaller cells divides again into two. Thus from the sperm mother-cells are formed four very small cells (fig. 3, sp.), each of which, without further division, is directly changed into the form characteristic of the spermatozoon. Each sperm mother-cell gives rise to four spermatozoa, and not to a large or indefinite number. The entire sperm mother-cell apparently passes into the four spermatozoa, with no traces of the “residual corpuscles” supposed to be the homologues of the polar bodies. This fact is of interest when compared with Hertwig’s recent results upon Ascaris (15).

—The facts of the process of maturation, fertilisation, and cleavage have been carefully studied by others. Sufficient here to add that, as in many other groups, the four cells arising by the first two divisions become pressed together, so that two have their apices directed towards the centre, truncated, so to speak, while the other two have sharp points; and the arrangement is such that the two opposite cells are alike, while the two adjacent cells are unlike.

The plane of bilateral symmetry is plainly indicated in the 8-cell stage (fig. 4). In the 16-cell stage the difference in size between the cells of the ectodermal and entodermal area is conspicuous (fig. 5). Throughout the process of cleavage I watched carefully for some particular cell to which the origin of the mesenchyme could be referred, but with negative results.

—About twelve hours after fertilisation cleavage is completed, and results in a ciliated cœloblastula, which spins around within the egg membrane; soon by the rupture of the egg membrane the blastula becomes free-swimming, and immediately seeks the surface of the water. Then appear the first traces of mesenchyme formation. In the region of the more columnar cells, the future entoderm, one and then more cells push out into the segmentation cavity, and become amœboid mesenchyme-cells. Usually the entire cell pushes out from the entoderm, but frequently there is a transverse division, and only the inner half becomes amoeboid (fig. 6). Somewhat later this portion of the sphere becomes flattened (fig. 7) and is gradually invaginated. During the progress of the invagination amoeboid mesenchyme-cells in great numbers wander into the segmentation cavity from the walls of the invaginated portion (fig. 8); some of these amœboid cells are formed by division of the entoderm-cells as above described, while the majority are in no way distinguishable from the cells which remain as the permaneut entoderm. The stage where only one, two, or three mesenchyme-cells are present is quickly followed by the appearance of others which arise from any part whatever of the entoderm al area (figs. 6, 7, and 8), and I am led to believe that the condition in Asterias vulgaris is the same as that found by Metschnikoff, and by Korschelt (21) in other Echinoderms, i. e. the absence of two bilaterally symmetrical “Urmesenchymezellen”—a view in opposition to that of Hatschek, Selenka, and Fleischmann.

My observations on Asterias vulgaris in regard to the time of the beginning of mesenchyme formation relatively to the process of gastrulation differ from those of Metschnikoff in Astropecten, inasmuch as I find that the mesenchyme formation precedes and continues throughout the progress of the invagination. No traces were found of the bilaterally symmetrical rows of cells comparable to the mesoblastic bands of Annelids, as described in other Echinoderms by Hatschek, Selenka, and Fleischmann. As the invagination progresses the spherical form of the larva changes to ovoid, the long axis corresponding with the antero-posterior axis of the future Bipinnaria. The gastrula travels through the water with two motions, one of translation in the line of the long axis of the body, the blastopore directed forwards; the other of rotation around this axis. At the completion of invagination the gastrula is much elongated (fig. 9). The archenteron, extending backwards about two thirds of the length of the embryo, is somewhat tubular in form; its blind end is bent towards that side which later proves to be the dorsal surface of the Bipinnaria. At this blind end is a considerable enlargement, where the cells by becoming flattened and losing their cilia acquire a character different from the columnar ciliated cells of the rest of the archenteron.

In section the ectoderm-cells are seen to be flatter and smaller than those of the entoderm, but the one grades insensibly into the other (fig. 9). At the pole opposite the blastopore is a point where the cells are distinctly more columnar than in any other part of the ectoderm. This point is found to become the apical pole of the Bipinnaria (fig. 9, a. p.).

FORMATION OF THE ENTEROCŒLS.—The mesoderm in Asterias as in most other Echinoderms has a twofold origin, though morphologically a sharp distinction between them in this case is not to be made: (1) mesenchyme formation; (2) enterocoel formation. The enterocœls arise as two bilaterally symmetrical diverticula of the blind end of the archenteron (fig. 10, el.). In position they are lateral and slightly dorsal. The time of complete separation shows much individual variation, in some cases being complete before, in others just after the formation of the larval oesophagus. A. Agassiz (2), working upon Asterias vulgaris, found that the stomodæal invagination united with the archenteron before the enterocoels separated from the latter. Metschnikoff (26) agrees with this, as does also Gotte (10) for Asterias glacialis. Ludwig (25) says, “In most Echinoderms the separation of the enterocoels occurs before the formation of the larval mouth. Asterids form an exception. Still this condition occurs in some Asterids, e. g. Asterina.” It would seem as if too much importance has hitherto been attached to this point, since it is subject to so much individual variation.

After separation from the archenteron the enterocoels increase slightly in size and move nearer to the dorsal wall of the larva, now appearing as ovoid vesicles with walls formed of flattened mesenchyme-like cells, which send out branching processes; these processes uniting with the branches of the mesenchyme-cells, which form an anastomosing network within the segmentation cavity, serve as supports for the enterocoels.

—Soon after the completion of the larval digestive tract by the fusion of the stomodæal ingrowth with the evaginated portion of the archenteron (fig. 11) begins the formation of the water-pores and the pore canal. On the dorsal wall of the enterocoel a diverticulum is formed. The cells of this diverticulum take on a cubical form; the cells of the rest of the entericœl wall retain the flattened branching appearance characteristic of mesenchyme (fig. 16). Above this upward projection of the enterocoel wall there appears a proliferation of the dorsal ectoderm. This as a solid plug extends downwards, meets, and fuses with the upward projection of the enterocoel; a cavity becomes formed in this ectodermal portion, and through it the cavity of the enterocœl is put in communication with the exterior. In this manner a right and a left water-pore and pore canal are formed. The walls of the pore canal are formed of columnar cells, which become ciliated. The pore canal is thus found to be made up of mesodermal and ectodermal elements. These observations are very unlike those described by Bury in his account of the mode of formation of the pore canal and water-pore in another Echinoderm. He says, “Examination of the living animal under a very high power shows that this pore is formed by a single elongated cell, perforated throughout its length and lined with cilia” (5, p. 411). The condition above described, having two bilaterally symmetrical water-pores, is found in larvae three and a half to four and a half days old (Pl. XIV, figs. 14 and 23). In a short time (eight to twelve hours after formation of the pores) the ectoderm pushes together, closing the external opening of the right water-tube; but the rest of the tube persists for some time, retaining its characteristic appearance. Fig. 19 is one of a series of sections of the larva at the stage after the closure of the right water-pore, but with the pore canal still present. In the sections succeeding the one here figured the water-tube of the right side (w. t.) was seen to end blindly. These observations were all made upon the living larvæ and confirmed by sections.

The presence of a right and a left water-pore has been noticed by several Continental investigators, but has been by them set aside as pathological. There is reason to believe, however, that this is a true ontogenetic character, and of very considerable phylogenetic significance. In figs. 21 and 22 I have drawn two sections of a series made by Professor Brooks, showing the bilaterally symmetrical water-pores and pore canals found by him in several larvæe of the same stage as that figured in side view in fig. 14, and in dorsal view in fig. 23. The specimens shown in section (figs. 21 and 22) were taken in the surface collections at Wood’s Hall. The sections are cut in a plane nearly parallel with the dorsal surface of the larva. In fig. 22 the section is seen to pass through the postoral and the preoral regions. In the postoral part are seen the two water-tubes cut transversely (w. t.). In fig. 21 only the postoral part is figured. The section passes through the waterpores (w.p.). Figs. 14 and 23 are drawn from living specimens obtained by artificial fertilisation. Specimens with two water-pores can be found in considerable numbers among larvæe three and a half to four and a half days old; but normally two water-pores are not present in larvæ after that age. In examining a number of larvæ of that age we could not expect to find a very large percentage with two waterpores, firstly on account of the individual variation in rate of development, and secondly on account of the briefness of the time in which both pores are functional. However, a number having two water-pores were isolated, and upon subsequent examination of these the right pore was found to have closed, while in two instances the process was observed.

But these facts, taken with the finding in the surface collections of several undoubtedly normal individuals of a similar age with two bilaterally symmetrical water-pores and watertubes (in one specimen the water-pore on the right side was found to be obliterated, though the pore canal still persisted), and the exceeding rarity of older larvæ with two water-pores, lead to the belief that this stage is a definite one in the ontogeny, and not a pathological condition, as has hitherto been assumed by the Continental students.

Circumoral Band

—From the time of the completion of the segmentation until the formation of the larval digestive tract, all the cells of the surface, of the oesophagus, and intestine are ciliated. These cilia serve for locomotion, and for propelling water through the digestive tract. But very early this condition of general ciliation gives way to the restriction of the cilia to definite band-like areas. Often even before the completion of the oesophagus the ectodermal cells of the ventral surface become flattened, and lose their cilia. There are left, however, two narrow transversely extending ciliated areas projecting slightly above the surface, one upon the bulging portion of the body anterior to the mouth, the other posterior, between the mouth and anus (see fig. 13, c. o. b.). As the area where the cilia have disappeared extends dorsally these two bands lengthen, the postoral one being of greater length. The original body cilia disappear last at the apex of the preoral lobe. Fig. 11, Pl. XIII, shows a stage where the end of the two parts of the ciliated bands pass into this remnant of the original general ciliation. Later, by the further disappearance of the original cilia the two ends of the preoral portion unite with the corresponding ends of the postoral portion, thus forming a single continuous bilaterally symmetrical ciliated band, by Semon named the circum-oral band. The course of this band is shown in figs. 12 and 14, c. o. b. Whether the two legs of the band touch, fuse, or remain separated at the apex of the preoral lobe is somewhat difficult to determine. Semon in Asterias rubens finds that they touch; in Asterias vulgaris I am inclined to believe that they are separated (see figs. 23 and 25).

At the apex of the preoral lobe (fig. 15) the ectoderm-cells are elongated over a small area. The outer part of the cell is clearer with fine granules, while at the base of the cells are what appear to be the cut ends of fibres, suggesting nervefibres. This may possibly be regarded as a very simple apical plate. It is the same as that referred to above in case of the gastrula (fig. 9). The direction of locomotion has become reversed. At first the blastopore was directed forwards; now the blastopore has become the larval anus, and is near the posterior end. The reversal of direction in locomotion takes place at the time of the formation of the mouth.

At first the entire surface of the larval œsophagus is ciliated. The first trace of that ciliated band immediately surrounding the mouth and sending a loop into the oesophagus, called by Semon the adoral band, is a thickened circular ring of ectoderm closely surrounding the mouth opening (fig. 12, a. o. b.). Later, with the formation of the depression of the body-wall around the mouth the circumference of this ring increases, and is seen to surround the rim of this oral depression (fig. 17, a. o. b.). It is to be noticed that up to this time there is no connection whatever between the circumoral and this adoral ciliated band. But later the transverse preoral portion of the circumoral band becomes pushed towards the hinder end of the larva, and finally vaults over the oral depression, and almost covers the mouth (figs. 24 and 30, c. o. b.). In this way the anterior part of the adoral band comes to touch and fuse with the posterior transverse preoral portion of the circumoral band. Thus the connection between these two bands is secondary, and not primary as was at first supposed by Semon in his earlier work (36), but corrected in his recent paper (38).

The cilia of the oesophagus are found to disappear except over a certain area, forming a loop-shaped ciliated band extending posteriorly and ventrally into the oesophagus, with its anterior end in connection with the above-described ciliated band bounding the rim of the oral depression, and with this circular, band constituting the adoral ciliated band.

Now to return to a further consideration of the circumoral band. How does this band, at first single and continuous (fig. 14), reach the condition shown in figs. 24, 30, and 29, the condition characteristic of the Bipinnaria ? The change occurs in the six days old larva. The original single bilaterally symmetrical circumoral ciliated band (fig. 14, c. o. b.) by a transverse division at the apex of the preoral lobe of the shaped portion, and by fusion of the divided ends takes a form; so that a plane passing between the links of the band, after the division and subsequent fusion of the broken ends, lies at right angles to its first position. At first perpendicular to the dorsal and ventral surfaces of the larva, it is after the division parallel with them (figs. 25 and 26). There are thus formed from this single band two complete bands: the upper (see fig. 26) bounds the dorsal and postoral area while the lower (p. v. a.) comes to lie entirely upon the ventral surface, bounding the preoral ventral area (fig. 30, p. v. a.). The whole history of the ciliated bands can be followed in figs. 13, 11, 12, 14, 23, 25 and 26, 24, 30, and 29.

By this division of the original single circumoral band into the two ciliated bands characteristic of the Bipinnaria ontogenetic proof is given, as pointed out by Semon (38), of the correctness of Gegenbaur’s hypothesis that the two ciliated bands of the Bipinnaria are equivalent to the single band of the Auricularia (see Balfour’s ‘Embryology,’ vol. i, pp. 554 and 557). Semon calls the stage with the single bilaterally symmetrical ciliated band the Auricularia stage of the Bipinnaria.

The ciliated bands in cross-section (fig. 34, c.o.b.) show a distinctly marked difference from the rest of the ectoderm. In the young larva all the ectodermal cells are cubical and ciliated; with the loss of the general ciliation the ectodermal cells become flattened and more irregular in outline, except in the region where the cilia persist as the ciliated bands. Here a great change has taken place. The cells become very much crowded together. The nuclei appear to be restricted to the deeper part, while the external part is formed of granular or finely fibrillated substance. The external surface is thickly ciliated, with many cilia on the free surface of each cell.

—The appearance of a larva six days old, seen from the dorsal side, is shown in fig. 20. The anus no longer has a terminal position (fig. 11), but by the bending of the intestine it has come to lie some distance forwards upon the ventral surface. The oesophagus, formed by the union of an entodermal evagination of the archenteron with the stomodieal ingrowth of the ectoderm (fig. 11), has elongated somewhat. The oral depression has become more pronounced, and the circumoral ciliated band has become divided at the apical pole, in the manner described above, into the two bands which characterise the Bipinnaria. Before this time the enteroccels have elongated somewhat. The right water-pore has disappeared, and only the left enterocœl has an opening to the exterior. At a considerably later stage (fig. 30) the enterocoels have extended anteriorly as two cylindrical tubes, nearly parallel and slightly dorsal to the œsophagus. The anterior end of the tube is solid, and its tip is formed of branching mesenchyme-like cells. Posterior to this the walls are thin, and formed of flattened cells and mesenchymatous muscle-fibres (fig. 32). In a stage a little older than fig. 30 (fig. 28) these two cylindrical tubes have extended further forwards and into the preoral lobe. The right and left enterocoels have united just in front and dorsal to the mouth. At the point of union the tubes are still solid (fig. 33). At a little later stage the cavities have united, and the two enterocœls stand in open communication with each other by their union in the preoral lobe. They later grow forwards and increase in size until they almost completely fill the cavity of the preoral lobe (figs. 24 and 29, el.). The appearance of the enterocoels in cross-section is shown in fig. 34, el.

Meantime the posterior ends of these cylindrical tubes have extended backwards and also dorsally and ventrally, so that they come to overlie the stomach and intestine. Ventrally and posteriorly the right and left enterocœls fuse together. On the left side, just posterior to the pore canal, there early appears a constriction (fig. 18, x) which finally narrows and separates the left enterocœl into two parts; the anterior, opening to the exterior at its posterior end by the pore canal and water-pore, extends forwards into the preoral lobe, where its cavity communicates with the right enterocœl, and the posterior part, the left posterior enterocœl, whose cavity is ventrally in connection with the posterior part of the right enterocœl. In the case of the right enterocœl I have often noticed a constriction, but have never found it divided into an anterior and a posterior portion as is the left enterocœl. In studying the living animal it should be noticed that the walls of the enterocoel are contractile, and that a temporary constriction may occur at almost any point. Fig. 18 is a diagram made from a reconstruction of a series of transverse sections of a larva. The outline of the left enterocœl (el), previous to its division into the anterior and posterior enterocoels, is marked by the dotted line. It does not show the posterior ventral fusion between the right and left enterocœls. The hydrocoel has not yet formed.

The amoeboid cells arising from the entodermic area press into the segmentation cavity. This cavity is filled with a transparent, jelly-like substance, and in this the mesenchymecells, by their long, delicate, anastomosing processes, form a network which serves for supporting the archenteron and the enterocoels. Mesenchyme-cells apply themselves to the walls of the body, of the digestive tract, and of the enterocœls, and there flatten and form a discontinuous covering for these organs. These mesenchyme-cells, in the later history of the larva, become differentiated into small fibres, which function as muscle-fibres. This differentiation takes place very early upon the walls of the entodermic portion of the oesophagus, where they form a circular and a longitudinal layer. The gulping movements brought about through the agency of these oesophageal muscles are very violent, and take place at intervals of fifteen to twenty seconds. A contraction starting behind the mouth travels towards the stomach, accompanied by a simultaneous longitudinal contraction of the œsophagus. As the final act the end of the oesophagus at its union with the stomach is violently pushed into the latter, and the contents of the oesophagus, driven down in front of the circular constriction, are suddenly belched into the stomach.

Differentiation of the mesenchyme-cells into muscle-fibres also takes place in the walls of the enterocoels, but no very definite layers were made out.

On the inner surface of the dorsal wall of the young Bipinnaria mesenchymatous muscle-cells and fibres are seen extending from the dorso-lateral portion above the stomach forwards along the median line (figs. 14 and 23, m. m.). These fibres have probably to do with the very considerable motion of which the preoral lobe is capable; the motion is in a dorso-ventral direction, and is accompanied by the formation of two or more wrinkles of the dorsal surface at the narrow part of the body of the larva. These fibres I judge to be the same as those described by Semon (38) as bilaterally arranged masses connected by a single commissure, and which he seems a little inclined to consider as a part or the whole of the larval nervous system, though he speaks of the possibility of their being muscular in function. However, the fact that they arise from mesenchyme-cells makes for the view that they are muscular tissue.

In the older larva (fig. 24) there is a small aggregation of mesenchyme-cells, in the main posterior to and just to the right of the pore canal. Its position is shown in fig. 18 (s. v.).

At the earliest stage in which I have yet found this it is close to the mesenchymatous lining of the body-wall; whether it arises from the lining of the wall or not I cannot yet positively say. Later a lumen is formed in the midst, and by the increase of the lumen the cells of the wall become flattened. The earliest stage which I have yet seen is shown in fig. 34, s. v. Successive stages in the growth are shown In figs. 27 and 31. In a Bipinnaria like fig. 29 the schizocœl vesicle, figured in fig. 31, measured ·04 mm. in its antero-posterior diameter. Of the ultimate fate of this schizocoel, or of its identification with the schizocoels hitherto described, I am not certain.

Morphological opinions upon the significance of the larval form of the Echinoderms fall into two diametrically opposite classes: (1) that the larval form has been coenogenetically acquired, or (2) that it is ancestral in character.

The arguments which make for the former are chiefly based. upon the necessity for better means of distribution, and in the free-swimming larval form is found a means secondarily acquired for the purpose of effecting this distribution. This view necessarily presupposes that the ancestor of the Echinoderms was sedentary; but this fundamental supposition does not seem to be well grounded. Even though the ancestors of the Crinoids appearing in the Cambrian seem geologically to be oldest, it by no means follows that we find in palaeontology a correct phylogenic record. From the nature of the case, too, we could gain from palaeontology little knowledge of the phylogenetic stages previous to the time when the hard calcareous skeletal parts appeared, and these skeletal parts are seen to be structures which have undergone exceedingly great cœnogenetic modifications. Palœontology, therefore, far from giving us a record of the phylogenetic series of ancestral forms, furnishes little more than a history of the skeletal parts of some of the later descendants. No transitional adult forms uniting the Echinoderms with the other animal groups are known; as a group they stand widely isolated. Even the numerous attempts to unite them with the other classes through the agency of larval forms have been more or less unsatisfactory.

We are asked to believe that in the life-history a form has been secondarily interpolated for the purpose of securing a wider distribution through a prolongation of the free-swimming. condition. There is no doubt but that many characters of the larva are cœnogenetic modifications, but these are of little importance when compared with those which appear to be ancestral. (1) The cleavage, total and very nearly equal, and the ciliated cœloblastula, offer simple ancestral conditions, and furnish means at the same time for wide distribution. (2) The mode of mesenchyme formation found in the Echinoderms is probably more primitive than the formation of the third germinal layer in the form of mesoblastic bands; and the derivation of this middle layer from any part whatever of the entoderm is antecedent to that condition where it is restricted to two special cells, the mesoblasts. The nature of the mesenchyme, too, filling up the cleavage cavity with its network of branching cells, is evidence of its primitive character. (3).The formation of enterocœls by archenteric diverticula is characteristic of ancestral forms. And in this larva we find this simplest condition of complete separation of enterocœls and archenteron, passing directly into corresponding parts of the adult. (4) The enterocœls open to the exterior very early by definite pores. The presence of two bilaterally symmetrical water-pores at a certain stage in the ontogeny, and the subsequent disappearance of the right pore, seems to point distinctly to the ancestral significance of the larval form ; for we are not justified in supposing that such a character is newly acquired, but that it is ancestral, as Professor Brooks has pointed out (4). It can only be in course of elimination from the ontogeny. The cause of this disappearance of the right pore may be traced to the subsequent connection between the two enterocœls by fusion in the preoral lobe. (5) The formation of the pore canals from ectodermal and mesodermal elements is similar to the condition described for the nephridia of certain worms, e. g. by Bergh for Criodrilus. From the simple conditions in the formation of the mesodermal elements above described is justified the belief that the condition here found is phylogenetically antecedent to that of the Annelids.

The function of the water-pore is difficult to determine. That its function in the adult as the stone canal is excretory is claimed by Hartog (19), but opposed by Cuenot and H. Ludwig. Bury saw exhalent currents but never inhalent, though, as he says, this does not prove that there are not inhalent currents. He found that the movements of the cilia are in such a direction as to cause an exhalent current. From Bury’s observations, and by elimination, one is almost forced to ascribe to it, at least in the larva, the excretory function, though in the adult this may be obscured by other functions; and the pore canal and water-pore of the Echinoderm larva seem in many ways to be comparable to the nephridia of Annelids, and to be ancestral in character.

The Echinoderm larva is a form which has developed along the phylogenetic line, and is in many ways differentiated and capable of free existence—an animal with a well-differentiated digestive tract, and having locomotor apparatus, enterocœls, excretory system, and well provided for respiration; to these have been coenogenetically added transparency as a protective adaptation, and the formation of long arms for protection, but primarily as a means of increased locomotor powers. The great length of the arms has probably been acquired since the time when the metamorphosis began to be accelerated by its earlier beginning ; i. e. originally metamorphosis did not begin until after the larva had become fixed to some support, but secondarily the beginning has been pushed forward, so that it now occurs long before fixation. The long arm-like projections of the larva are to be explained through the necessity for increased powers of locomotion on account of the weight of the adult starfish developing in the hinder end of the larva. But the greatest of the cœnogenetic modifications is that whereby the typical larva acquires the different forms characteristic of the various groups—the larval form distinguished as Auricularia, Bipinnaria, and Pluteus. The fact that all these forms are modifications of a single typical form was long ago pointed out by Johannes Müller. The recent work by Semon (38) has completed the confirmation.

It seems pretty certain that the radial symmetry of the Echinoderms has been derived from bilateral symmetry through the influence of a sedentary mode of life. May we not be justified in supposing that such an animal as the typical Echinoderm larva above described may upon becoming sedentary have been modified in adaptation to its mode of life, even so much as now appears between such a larva and an adult Echinoderm, and that the process of change as now shown in the metamorphosis is in its general character an expression of the course of phylogeny, but subjected to exceedingly great distortion by the constant tendency towards abbreviation, by the dropping out of details from the ontogeny, and by greater or less shifting of the relative times of formation of the various organs, particularly in the time of appearance of radial symmetry, which has been constantly carried forward to earlier appearance in the ontogeny. After assuming the sedentary condition there came in as further adaptive modifications changes in the function of organs. The greatest of these changes concerns the enterocoels; to their earlier function, probably excretory, has been secondarily added that of locomotion, of relation (feelers, tentacles), and also to some extent of respiration.

It seems more probable that the ancestral Echinoderm arose by the adaptive modification of a more primitive free-swimming form than that a larval form has been acquired for the purpose of distribution. The Echinoderm ancestor was probably a free-swimming animal, in general characters not far removed from the ancestors of the Turbellarians; a creature with a well-differentiated digestive tract, ciliary locomotor apparatus, excretory system, respiratory surface not localised ; coenogenetically modified by the acquirement of transparency, long arms, and particularly by modification of the external form by changes in the direction of the ciliated bands, as pointed out by Johannes Müller, into the forms characteristic for the various Echinoderm groups. In their ontogeny the Auricularia and the Bipinnaria have travelled together for some distance, as shown by the fact lately pointed out by Semon (38) that the Bipinnaria passes through an Auricularia stage. We may suppose that the bilateral form after a period of free-swimming life became sedentary, and after this the bilateral symmetry became more or less disguised by a radial symmetry. From this sedentary form, the Pentactea of Semon, the ancestors of the present Echinoderm groups have been derived. The earliest arising were the Synaptidae. through some archaic form; next came off the ancestors of the Holothurians, and later the ancestors of the Crinoids, and latest the ancestor of the Echinids, Ophiurids, and Asterids. We are certainly justified in applying to the tentative theory of Echinoderm phylogeny the principles which are accepted in attempts to trace the phylogeny of other classes, namely, that it is not to be expected that many of the ancestral forms connecting the groups are to-day accessible for study, either alive or as fossils > and in view of the failure of the numerous and carefully formulated theories of Echinoderm descent it seems necessary to believe that the various groups were derived from one another only through intermediate forms between each group, which forms, however, probably persisted but a comparatively short time, and palæontological evidence of them is very scanty, or in the vast majority of cases entirely wanting. The corresponding stages, too, have been for the most part eliminated from the ontogeny.

The groups of the Echinids, Ophiurids, and Asterids, and a part of the Holothurids have been coenogenetically modified for a creeping life, the original excretory system assuming the locomotor in addition to the earlier acquired sensory and respiratory functions. The early appearance of radial symmetry in the free-swimming larva shown in the radial out-pushings of the hydrocoel wall at that stage of the ontogeny generally spoken of as the beginning of metamorphosis may be regarded as coenogenetic precocious formation for the purpose of shortening the metamorphosis. Examples of further abbreviation of the metamorphosis are found in the case of the so-called viviparous Echinoderms, where it is carried to an extreme degree.

JOHNS HOPKINS UNIVERSITY;

May 1,1891.

1.
Agassiz
,
L.
‘Boston Evening Traveller,’ December 22nd, 1848; also ‘Müller’s Archiv f. Anat. u. Physio],,’
1851
, pp,
122
125
.
2.
Agassiz
,
A.
—“
North American Starfishes
,”
‘Mem. of Mus. of Comp. Zoology,’ Cambridge, Mass
.,
1873
.
3.
Balfour
,
F. M.
‘Comparative Embryology,’
vol.
i
.
4.
Brooks
,
W. K.
‘Johns Hopkins University Circular,’
No.
88
,
1891
.
5.
Bury
,
H.t
—“
Studies on the Embryology of Echinoderms
,”
‘Quart. Journ. Mier. Sei.,’
vol.
xxix
,
1889
.
6.
Desor
,
E.
—“
On the Development of Asterids
,”
‘Proc, of Boston Soc. of Nat. History,’ 1848 ; also ‘Müller’s Archiv,’
1849
, pp.
79
83
.
7.
Field
,
G. W.
‘Johns Hopkins University Circular,’
No.
88
,
1891
.
8.
Fleischmann
,
A.
—“
Die Entwickelung des Eier vonEchinocardium cordatum
,”
‘Zeit. f. wiss. Zool.,’
46
,
1888
.
9.
Götte
. —“
Bemerkungen zur Entges. der Echinodermen
,”
‘Zool. Anz.,’
1880
, pp.
324
326
.
10.
Greete
,
R.
—“
Ueber den Bau und die Entwickelung des Asteracanthion rubens vom Ei bis zur Bipinnaria und Brachiolaria
,”
‘Sitz-ber. der Markinger Naturf. Gesellsch.,’
1876
.
11.
Haeckel
,
E.
—“
Die Kometenform und die Generationswechsel der Echinodermen
,”
‘Zeit. f. wiss. Zool.,’
xxx
,
Suppl
., pp.
424
445
.
12.
Hatschek
. —“
Ueber Ent. von Teredo
,”
‘Arb.a. den Zool. Inst, der Univ. Wien,’
1880
, Bd.
iii
, pp.
1
44
.
13.
Hensen
,
V.
—“
Ueber ein Brachiolaria des Kieler Hafens
,”
‘Arch, für Naturgesch.,’
1863
, pp.
242
246
.
14.
Hertwig
,
O.
—“
Beiträge zur. Kenntniss der Bildung, Befruchtung, und Theilung des thierischen Eier
,”
‘Morph. Jahrbuch,’
i
,
1875
, pp.
347
434
.
15.
Hertwig
,
O.
—“
Vergleich der Eiund Samen-bildung bei Nematoden, Eine Grundlage für Celluläre Streitfragen
,”
1 Arch. für mikr. Anat.,’
Bd.
xxxvi
, Heft 1, 1890.
16.
Huxley
,
T. H.
—“
Report upon the Researches of Professor Müller into the Anatomy and Development of Echinoderms
,”
‘Annals and Magazine of Nat. Hist.,’
2nd series,
viii
,
1851
, pp,
1
19
.
17.
Hamann
,
O.
‘Beiträge zur Histologie der Echinodermen,’
Heft 2, Jena,
1884–7
.
18.
Hamann
,
O.
—“
Die wandernden Urkeimzellen und ihre Reifungsstätten bei den Echinodermen
,”
‘Zeit. f. w. Zool.,’
Bd.
xlvi
,
1887
.
19.
Hartog
,
M. M.
—“
True Nature of the Madreporic System of the Echinoderms, with Remarks on Nephridia
,”
‘Annals and Mag. of Nat. History,” 5th series
, vol.
xx
,
1887
, p.
321
.
20.
Krohn
,
A.
‘Beiträge zur Entges. der See-eigel,’
Heidelberg
,
1849
.
21.
Korschelt
,
E.
—“
Zur Bildung des Mittleren Keimblattes bei den Echi nodermen
,”
‘Zool. Jahrbuch,’
iv
,
1889
.
22.
Korschelt and Heideb
. —
‘Lehrbuch der Vergleichenden Entwicke lungsgeschichte der wirbellosen Thiere,’
Erster Abtheilung
,
1890
.
23.
Koben and Danielssen
. —“
Observations sur la Bipinnaria asterigera
,”
‘Annales des Sei. Nat.,’
sär. 3, t.
vii
,
1847
.
24.
Lodwig
,
H.
—“
Beiträge zur Anat. der Asteriden
,”
‘Zeit. f. Wiss Zool.,’
xxx
,
1877
.
25.
Lodwig
,
H.
—“
Entwickelungsgeschichte der Asterina gibbosa
,”
‘Zeit. f. wiss. Zool.,’
xxxviii
,
1882
.
26.
Metschnikoff
,
E.
—“
Studien über die Ent. der Echinodermen und Nemertenen
,”
‘M6m. de l’Acad. Imper. de St. Pätersbourg,’
7 sär., t.
xiv
, No.
8
,
1869
.
27.
Metschnikoff
,
E.
—“
Embryologische Mittheilungen über Echino dermen
,”
‘Zool. Anzeiger,’ Jahrgang vü
, pp.
43
and 62, y1884.
28.
Metschnikoff
,
E.
—“
Studien ü. Ent. der Medusen und Siphonophoren
,”
‘Zeit. f. wiss. Zool.,’
xxiv
,
1874
.
29.
Mölleb
,
Johannes
. —“
Ueber die Larven und die Metamorph, der Echinodermen
,”
‘Abhandl. d. Akad. d. Wissenschaften zu Berlin,’ Sieben Abteilungen
,
1848
1855
.
30.
Nedmayb
,
M.
‘Die Stämme des Thierreiches,’
Wien und Prag
,
1889
.
31.
Sans
,
M.
—“
Ueber die Ent. der Seesterne
,”
‘Archiv f. Naturgesch.,’
1844
, Bd.
i
.
32.
Schneider
,
A.
—“
Ueber die Ent. der Echinodermen; Sitzungsber. Der Gesellsch. Naturforsch
.,”
‘Freunde zu Berlin im Jahre 1869,’
Berlin
,
1870
.
33.
Selenka
. —“
Zoologische Studien :—I. Befruchtungen des Eier von Toxopneustes variegatus
,”
‘Ein Beitrag zur Lehre von der Befruchtung und Eifurchung,’
Leipzig
,
1878
.
34.
Selenka
. —“
Keimblätter der Echinodermen
,”
‘Stud. ü. Ent. d. Thieren,’
Wiesbaden
,
1883
.
35.
Sarasin
,
P.
and F
. —“
Ueber die Anat. der Echinothuriden und die Phylogenie der Echinodermen
,”
‘Ergebn. naturw. Forsch, auf Ceylon, 1884-6,’
Bd.
i
, Wiesbaden, 1888; abstract in ‘Journ. Roy. Micr. Soc.,’
London
,
1885
, pl. vi, pp.
956
958
.
36.
Semon
,
R.
—“
Die Entwickelung der Synapta digitata, und die Stammesgeschichte der Echinodermen
,”
‘Jenaische Zeitschrift für Naturwissenschaft,’
xxii
, New Series 15,
1888
.
37.
SeMon
,
R.
—“
Die Holologien innerhalb des Echinodermenstammes
,”
‘Morph. Jahrbuch,’
xv
,
1889
.
38.
Semon
,
R.
—“
Zur Morphologie der bilateralen Wimperschnure der Echinodermen-Larven
,”
‘Zeit. f. wiss. Zool.,’ N. F
., Bd.
xviii
, part 1,
1890
.
39.
Vogt
,
C.
, and
Young
,
E.
‘Lehrbuch der prak. vergl. Anatomie, Echinodermen,’ 9–11
Lieferungen
,
1886-7
, Braunschweig.

Illustrating Mr. George W. Field’s paper on “The Larva of Asterias vulgaris.”

List of Reference Letters

a. Anterior, an. Anus. ap. Apical plate, ar. Archenteron. a. o. b. Adoral ciliated band. b. e. Branched cells, bl. Blastopore, e. t. Connective tissue, c. o. b. Circumoral ciliated band. d. Dorsal. E1. Median anal-paired arm. E2. Dorsal anal-paired arm. E3. Ventral anal-paired arm. E*. Dorsal oral-paired arm. E*. Ventral oral-paired arm. E3. Unpaired anterior arm. Ec. Ectoderm. El. Enterocoel, ge. Germinal epithelium. int. Intestine, tn. Muscle. mes. Mesenchyme. m. tn. Mesenchymatous muscle-fibres, mo. Mouth. n. Nucleus. O. Ovum. o. d. Oral depression. ce. (Esophagus, p. Posterior, p. b. Polar bodies, p. c. Pore canal, pe. Peritoneum, p. I. Preoral lobe. p. v. a. Preoral ventral area. sb. Stomach. s. m. c. Sperm mother-cell. sp. Spermatozoa, st. Stomodffiurn. s. v. Schizo-coel vesicle, e. Ventral. c. d. a. Ventro-dorsal area. to. p. Water-pore. w. t. Water-tube.

All the figures are camera drawings except Fig. 18, which is a diagram of a reconstruction from serial transverse sections ; and Figs. 25 and 26, which are not drawn to scale.

FIG. 1.—Cross-section of an alveolus of the ovary; the eggs are nearly ready to be discharged. Only one half of the section is figured. Perenyi; Mayer’s cochineal preparation. × 145.

FIG. 2.—Cross-section of an alveolus of the ovary after discharge of the eggs. Perenyi; alcoholic borax-carmine preparation, × 145.

FIG. 3.—Cross-section of an alveolus of the testis. Only a small segment is drawn. The oval cells near the centre (sp.), directly without farther division, become changed into the shape characteristic of the spermatozoa. Perenyi; Kleiuenberg’s hematoxylin preparation, × 1000.

FIG. 4.—Egg in eight-celled stage, showing the bilaterally symmetrical division into a right and a left half. × 400.

FIG. 5.—An egg in sixteen-celled stage, showing the relative size of ectoderm and entoderm cells at this time. Perenyi; Kleinenberg’s hematoxylin preparation. × 400.

FIG. 6.—Optical section of living blastula soon after its escape from the egg-membrane, showing first appearance of mesenchyme-cells. × 240.

FIG. 7.—Same at beginning of invagination.

FIG. 8.—Mesenchyme formation during the progressing invagination. Optical section of living animal, × 240.

FIG. 9.—Longitudinal section of completed gastrula. Perenyi; Kleinenberg’s heematoxylin preparation, × 240.

FIG. 10.—Optical section of the living gastrula, looking down somewhat obliquely upon the blind end of the archenteron, to show the formation and relation of the enterocoels. × 400.

FIG. 11.—Living specimen, seen from the right side, to show the mode of formation of digestive tract and of the circumoral ciliated band. × 70.

FIG. 12.—Young larva, seen in ventral view, showing the original relation of the adoral (a. o. b.) and the circumoral (c. o. 6.) ciliated bands. Picric salt; Delafeld’s haematoxylin preparation, × 110.

FIG. 13.—Showing manner of formation of the ciliated band by the disappearance of the general ciliation. The dotted portions represent the ciliated areas. Living specimen. × 240,

FIG. 14.—Larva four days old, seen from the right side, showing the right water-tube and pore. Kleinenberg’s picric salt; Delafeld’s hæmatoxylin preparation, × 145.

FIG. 15.—Longitudinal section through the apex of the preoral lobe, to show the ectodermal thickening, ap. From a specimen of about the same age as Fig. 11. Perenyi ; Kleinenberg’s bæmatoxylin preparation, × 600.

FIG. 16.—Section showing mode of formation of the water-pore and pore canal. Similar conditions were observed in the living specimens. Perenyi; Kleinenberg’s hæmatoxylin preparation. × 600.

FIG. 17.—Larva of five days, seen in ventral view to further show the history of the relation of the adoral band to the preoral portion of the circum-oral band, and the formation of the oral depression. Kleinenberg’s picric salt; Delafeld’s hæmatoxylin preparation, × 70.

FIG. 18.—Diagram of a reconstruction from serial transverse sections, to show thé form and position of the left enterocoel soon after its union in the preoral lobe with the right enterocoel. As seen from the left side. x. The point where the constriction will appear which divides the left enterocoel into an anterior and a posterior.

FIG. 19.—Longitudinal section, parallel with the dorsal and ventral surfaces, of a larva in same stage as Fig. 14, but just after the closure of the right water-pore. The pore canal persists, but in the sections following the one here figured it is found to end blindly. Perenyi ; Kleinenberg’s hæmatoxylin preparation. × 240.

FIG. 20.—Larva of six days, at the stage when the division of the circum. oral band takes place (see Figs. 25 and 26), showing the dorsal surface. Only the left water-pore is now present. Kleinenberg’s picric salt; Delafeld’s hæmatoxylin preparation. × 70.

FiG. 21.—A longitudinal section, parallel with the dorsal surface, through the two Water-pores. The part of the section through the preoral lobe is not here figured, though it is in Fig. 22. × 350.

FIG. 22.—The second section, ventralwards from that shown in Fig. 21. It shows the bilaterally symmetrical water-tubes, which in Fig. 21 are seen to open on the dorsal surface. × 350.

FIG. 23.—From the living animal, four days old, showing the dorsal mesen-chymatous muscle-fibres which move the preoral lobe. It also shows the bilaterally symmetrical water-pores, × 145.

FIG. 24.—A Bipinnaria, a little older than Fig. 28,. seen from the right side. Kleinenberg’s picro-sulphuric cedar ; oil preparation. × 70.

FIG. 25.—Surface view of the tip of the preoral lobe of a four days old larva, before the division of the circutnoral ciliated band into the two bands characteristic of the Bipinnaria.

FIG. 26.—The same view of a larva six days old, after the division of the circumoral band.

FIG. 27.—Part of a transverse section of a Bipinnaria, a little older than that shown in Fig. 24, showing a later stage of the schizoccel (s. ".) and its relative position. Perenyi; Delafeld’s hsematoxylin preparation, × 600.

FIG. 28.—A stage a little older than Fig. 30. The characteristic arms have begun to form. The enterocoels have united in the preoral lobes, but the cavities have not yet become continuous. From the living specimen, viewed from the dorsal surface, × 70.

FIG. 29.—A Bipinnaria about five weeks old, seen from the ventral surface. Kleinenberg’s picro-sulphuric; alcoholic borax-carmine preparation. × 70.

FIG. 30.—Bipinnaria about eighteen days old, seen in ventral view. Shows the formation of the preoral ventral area (p. t>. a.), and the position and form of the enterocoels. From the living specimen. × 70.

FIG. 31.—Portion of a longitudinal section parallel to the dorsal surface, from a Bipinnaria of about the same stage as Fig. 29, to show a later stage of the schizoccel. Kleinenberg’s picro-sulphuric, Delafeld’s haematoxylin preparation. × 600.

FIG. 32.—The growing anterior tips of the enterocoels, drawn from the living specimen. Same stage as Fig. 30. × 240.

FIG. 33.—The union of the enterocoels in the preoral lobe. Same stage as Fig. 28. From the living specimen. × 240.

FIG. 34.—A transverse section of a Bipinnaria at about the same stage as Fig. 24, at a point just anterior to the water-pore (a tangential piece is cut from the pore canal,’ w. t.), to show the origin of the schizoccel, s. e. Kleinenberg’s picro-sulphuric, Kleinenberg’s haematoxylin preparation. × 240.