Works, Volume 3, of Francis Maitland Balfour

The developmental history of the Chordata has been studied far more completely than that of any of the groups so far considered; and the results which have been arrived at are of striking interest and importance. Three main subdivisions of this group can be recognized: (1) the Cephalochorda containing the single genus Amphioxus; (2) the Urochorda or Tunicata; and (3) the Vertebrata[1]. The members of the second and probably of the first of these groups have undergone degeneration, but at the same time the members of the first group especially undergo a less modified development than that of other Chordata.

Cephalochorda.

Our knowledge of the development of Amphioxus is mainly due to Kowalevsky (Nos. 1 and 2). The ripe eggs appear to be dehisced into the branchial or atrial cavity, and to be transported thence through the branchial clefts into the pharynx, and so through the mouth to the exterior. (Kowalevsky, No. 1, and Marshall, No. 5.)

When laid the egg is about 0.105 mm. in diameter. It is invested by a delicate membrane, and is somewhat opaque owing to the presence of yolk granules, which are however uniformly distributed through it, and proportionately less numerous than in the ova of most Chordata. Impregnation is external and the segmentation is nearly regular (fig. 1). A small segmentation cavity is visible at the stage with four segments, and increases during the remainder of the segmentation; till at the close (fig. 1 E) the embryo consists of a blastosphere formed of a single layer of cells enclosing a large segmentation cavity. One side of the blastosphere next becomes invaginated, and during the process the embryo becomes ciliated, and commences to rotate. The cells forming the invaginated layer become gradually more columnar than the remaining cells, and constitute the hypoblast; and a structural distinction between the epiblast and hypoblast is thus established. In the course of the invagination the segmentation cavity becomes gradually obliterated, and the embryo first assumes a cup-shaped form with a wide blastopore, but soon becomes elongated, while the communication of the archenteron, or cavity of invagination, with the exterior is reduced to a small blastopore (fig. 2 A), placed at the pole of the long axis which the subsequent development shews to be the hinder end of the embryo. The blastopore is often known in other Chordata as the anus of Rusconi. Before the invagination is completed the larva throws off the egg-membrane, and commences to lead a free existence.

Up to this stage the larva, although it has acquired a cylindrical elongated form, has only the structure of a simple two-layered gastrula; but the changes which next take place give rise on the one hand to the formation of the central nervous system, and on the other to the formation of the notochord and mesoblastic somites[2]. The former structure is developed from the epiblast and the two latter from the hypoblast.

The formation of the central nervous system commences with the flattening of the dorsal surface of the embryo. The flattened area forms a plate (fig. 2 B and fig. 3 A, np), extending backwards to the blastopore, which has in the meantime passed round to the dorsal surface. The sides of the plate become raised as two folds, which are most prominent posteriorly, and meet behind the blastopore, but shade off in front. The two folds next unite dorsally, so as to convert the previous groove into a canal[3]—the neural or medullary canal. They unite first of all over the blastopore, and their line of junction extends from this point forwards (fig. 2 C, D, E). There is in this way formed a tube on the floor of which the blastopore opens behind, and which is itself open in front. Finally the medullary canal is formed for the whole length of the embryo. The anterior opening persists however for some time. The communication between the neural and alimentary tracts becomes interrupted when the caudal fin appears and the anus is formed. The neural canal then extends round the end of the notochord to the ventral side, but subsequently retreats to the dorsal side and terminates in a slight dilatation.

In the formation of the medullary canal there are two points deserving notice—viz. (1) the connection with the blastopore; (2) the relation of the walls of the canal to the adjoining epiblast. With reference to the first of these points it is clear that the fact of the blastopore opening on the floor of the neural canal causes a free communication to exist between the archenteron or gastrula cavity and the neural canal; and that, so long as the anterior pore of the neural canal remains open, the archenteron communicates indirectly with the exterior (vide fig. 2 E). It must not however be supposed (as has been done by some embryologists) that the pore at the front end of the neural canal represents the blastopore carried forwards. It is even probable that what Kowalevsky describes as the carrying of the blastopore to the dorsal side is really the commencement of the formation of the neural canal, the walls of which are continuous with the lips of the blastopore. This interpretation receives support from the fact that at a later stage, when the neural and alimentary canals become separated, the neural canal extends round the posterior end of the notochord to the ventral side. The embryonic communication between the neural and alimentary canals is common to most Chordata; and the tube connecting them will be called the neurenteric canal. It is always formed in fundamentally the same manner as in Amphioxus. With reference to the second point it is to be noted that Amphioxus is exceptional amongst the Chordata in the fact that, before the closure of the neural groove, the layer of cells which will form the neural tube becomes completely separated from the adjoining epiblast (fig. 3 A), and forms a structure which may be spoken of as the medullary plate; and that in the closure of the neural canal the lateral epiblast forms a complete layer above this plate before the plate itself is folded over into a closed canal. This peculiarity will be easily understood from an examination of fig. 3 A, B and C.

The formation of the mesoblastic somites commences, at about the same time as that of the neural canal, as a pair of hollow outgrowths of the walls of the archenteron. These outgrowths, which are shewn in surface view in fig. 2 B and D, so, and in section in fig. 3 B and C, so, arise near the front end of the body and gradually extend backwards as wing-like diverticula of the archenteric cavity. As they grow backwards their dorsal part becomes divided by transverse constrictions into cubical bodies (fig. 2 D and E), which, with the exception of the foremost, soon cease to open into what may now be called the mesenteron, and form the mesoblastic somites. Each mesoblastic somite, after its separation from the mesenteron, is constituted of two layers, an inner one—the splanchnic—and an outer—the somatic, and a cavity between the two which was originally continuous with the cavity of the mesenteron. Eventually the dorsal parts of the outgrowths become separated from the ventral, and form the muscle-plates, while their cavities atrophy. The cavity of the ventral part, which is not divided into separate sections by the above described constrictions, remains as the true body cavity. The ventral part of the inner layer of the mesoblastic outgrowths gives rise to the muscular and connective tissue layers of the alimentary tract, and the dorsal part to a section of the voluntary muscular system. The ventral part of the outer layer gives rise to the somatic mesoblast, and the dorsal to a section of the voluntary muscular system. The anterior mesoblastic somite long retains its communication with the mesenteron, and was described by Max Schultze, and also at first by Kowalevsky, as a glandular organ. While the mesoblastic somites are becoming formed the dorsal wall of the mesenteron develops a median longitudinal fold (fig. 3 B, ch), which is gradually separated off from before backwards as a rod (fig. 3 C, ch), underlying the central nervous system. This rod is the notochord. After the separation of those parts the remainder of the hypoblast forms the wall of the mesenteron.

With the formation of the central nervous system, the mesoblastic somites, the notochord, and the alimentary tract the main systems of organs are established, and it merely remains briefly to describe the general changes of form which accompany the growth of the larva into the adult. By the time the larva is but twenty-four hours old there are formed about seventeen mesoblastic somites. The body, during the period in which these are being formed, remains cylindrical, but shortly afterwards it becomes pointed at both ends, and the caudal fin appears. The fine cilia covering the larva also become replaced by long cilia, one to each cell. The mesenteron is still completely closed, but on the right side of the body, at the level of the front end of the mesenteron, the hypoblast and epiblast now grow together, and a perforation becomes formed through their point of contact, which becomes the mouth. The anus is probably formed about the same time if not somewhat earlier[4].

Of the subsequent changes the two most important are (1) the formation of the gill slits or clefts; (2) the formation of the peribranchial or atrial cavity.

The formation of the gill slits is, according to Kowalevsky’s description, so peculiar that one is almost tempted to suppose that his observations were made on pathological specimens. The following is his account of the process. Shortly after the formation of the mouth there appears on the ventral line a coalescence between the epiblast and hypoblast. Here an opening is formed, and a visceral cleft is thus established, which passes to the left side, viz. the side opposite the mouth. A second and apparently a third slit are formed in the same way. The stages immediately following were not observed, but in the next stage twelve slits were present, no longer however on the left side, but in the median ventral line. There now appears on the side opposite the mouth, and the same therefore as that originally occupied by the first three clefts, a series of fresh clefts, which in their growth push the original clefts over to the same side as the mouth. Each of the fresh clefts becomes divided into two, which form the permanent clefts of their side.

The gill slits at first open freely to the exterior, but during their formation two lateral folds of the body wall, containing a prolongation of the body cavity, make their appearance (fig. 4 A), and grow downwards over the gill clefts, and finally meet and coalesce along the ventral line, leaving a widish cavity between themselves and the body wall. Into this cavity, which is lined by epiblast, the gill clefts open (fig. 4 B, br.c). This cavity—which forms a true peribranchial cavity—is completely closed in front, but owing to the folds not uniting completely behind it remains in communication with the exterior by an opening known as the atrial or abdominal pore.

The vascular system of Amphioxus appears at about the same time as the first visceral clefts.

Bibliography.

(1) A. Kowalevsky. “Entwicklungsgeschichte des Amphioxus lanceolatus.” Mém. Acad. Impér. des Sciences de St Pétersbourg, Series VII. Tom. XI. 1867.
(2) A. Kowalevsky. “Weitere Studien über die Entwicklungsgeschichte des Amphioxus lanceolatus.” Archiv f. mikr. Anat., Vol. XIII. 1877.
(3) Leuckart u. Pagenstecher. “Untersuchungen über niedere Seethiere.” Müller’s Archiv, 1858.
(4) Max Schultze. “Beobachtung junger Exemplare von Amphioxus.” Zeit. f. wiss. Zool., Bd. III. 1851.
(5) A. M. Marshall. “On the mode of Oviposition of Amphioxus.” Jour. of Anat. and Phys., Vol. X. 1876.

In the Solitaria, except Cynthia, the eggs are generally laid, and impregnation is effected sometimes before and sometimes after the eggs have left the atrial cavity. In Cynthia and most Caducichordata development takes place within the body of the parent, and in the Salpidæ a vascular connection is established between the parent and the single fœtus, forming a structure physiologically comparable with the Mammalian placenta.

Solitaria. The development of the Solitary Ascidians has been more fully studied than that of the other groups, and appears moreover to be the least modified. It has been to a great extent elucidated by the splendid researches of Kowalevsky (Nos. 18 and 20), whose statements have been in the main followed in the account below. Their truth seems to me to be established, in spite of the scepticism they have met with in some quarters, by the closeness of their correspondence with the developmental phenomena in Amphioxus.

The type most fully investigated by Kowalevsky is Ascidia (Phallusia) mammillata; and the following description must be taken as more especially applying to this type.

The segmentation is complete and regular. A small segmentation cavity appears fairly early, and is surrounded, according to Kowalevsky, by a single layer of cells, though on this point Kupffer (No. 27) and Giard (No. 11) are at variance with him.

The segmentation is followed by an invagination of nearly the same character as in Amphioxus. The blastosphere resulting from the segmentation first becomes flattened on one side, and the cells on the flatter side become more columnar (fig. 8 I.). Very shortly a cup-shaped form is assumed, the concavity of which is lined by the more columnar cells. The mouth of the cup or blastopore next becomes narrowed; while at the same time the embryo becomes oval. The blastopore is situated not quite at a pole of the oval but in a position which subsequent development shews to be on the dorsal side close to the posterior end of the embryo. The long axis of the oval corresponds with the long axis of the embryo. At this stage the embryo consists of two layers; a columnar hypoblast lining the central cavity or archenteron, and a thinner epiblastic layer. The dorsal side of the embryo next becomes flattened (fig. 8 II.), and the epiblast covering it is shortly afterwards marked by an axial groove continued forwards from the blastopore to near the front end of the body (fig. 5, mg). This is the medullary groove, and it soon becomes converted into a closed canal—the medullary or neural canal—below the external skin (fig. 6, n.c). The closure is effected by the folds on each side of the furrow meeting and coalescing dorsally. The original medullary folds fall into one another behind the blastopore so that the blastopore is situated on the floor of the groove, and, on the conversion of the groove into a canal, the blastopore connects the canal with the archenteric cavity, and forms a short neurenteric canal. The closure of the medullary canal commences at the blastopore and is thence continued forwards, the anterior end of the canal remaining open. The above processes are represented in longitudinal section in fig. 8 III, n. When the neural canal is completed for its whole length, it still communicates by a terminal pore with the exterior. In the relation of the medullary canal to the blastopore, as well as in the closure of the medullary groove from behind forwards, the Solitary Ascidians agree closely with Amphioxus.

The cells of the dorsal wall of the archenteron immediately adjoining the front and sides of the blastopore have in the meantime assumed a somewhat different character from the remaining cells of the archenteron, and give rise to a body which, when viewed from the dorsal surface, has somewhat the form of a horseshoe. This body was first observed by Metschnikoff. On the elongation of the embryo and the narrowing of the blastopore the cells forming this body arrange themselves as a broad linear cord, two cells wide, underlying about the posterior half of the neural canal (fig. 7, ch). They form the rudiment of the notochord, which, as in Amphioxus, is derived from the dorsal wall of the archenteron. They are seen in longitudinal section in fig. 8 II. and III. ch.

With the formation of the notochord the body of the embryo becomes divided into two distinct regions—a posterior region where the notochord is present, and an anterior region into which it is not prolonged. These two regions correspond with the tail and the trunk of the embryo at a slightly later stage. The section of the archenteric cavity in the trunk dilates and constitutes the permanent mesenteron (figs. 7, al, and 8 III. and IV. dd). It soon becomes shut off from the slit-like posterior part of the archenteron. The nervous system in this part also dilates and forms what may be called the cephalic swelling (fig. 8 IV.), and the pore at its anterior extremity gradually narrows and finally disappears. In the region of the tail we have seen that the dorsal wall of the archenteron becomes converted into the notochord, which immediately underlies the posterior part of the medullary canal, and soon becomes an elongated cord formed of a single or double row of flattened cells. The lateral walls of the archenteron (fig. 7, me) in the tail become converted into elongated cells arranged longitudinally, which form powerful lateral muscles (fig. 8 IV. m). After the formation of the notochord and of the lateral muscles there remains of the archenteron in the tail only the ventral wall, which according to Kowalevsky forms a simple cord of cells (fig. 6, al). It is however not always present, or else has escaped the attention of other observers. It is stated by Kowalevsky to be eventually transformed into blood corpuscles. The neurenteric canal leads at first into the narrow space between the above structures, which is the remnant of the posterior part of the lumen of the archenteron. Soon both the neurenteric canal and the caudal remnant of the archenteron become obliterated.

During the above changes the tail becomes considerably elongated and, owing to the larva being still in the egg-shell, is bent over to the ventral side of the trunk.

The larva at this stage is represented in a side view in fig. 8 IV. The epidermis is formed throughout of a single layer of cells. In the trunk the mesenteron is shewn at dd and the dilated part of the nervous system, no longer communicating with the exterior, at n. In the tail the notochord is shewn at ch, the muscles at m, and the solid remnant of the ventral wall of the archenteron at dd´. The delicate continuation of the neural canal in the tail is seen above the notochord at n. An optical section of the tail is shewn in fig. 6. It is worthy of notice that the notochord and muscles are formed in the same manner as in Amphioxus, except that the process is somewhat simplified. The mode of disappearance of the archenteric cavity in the tail, by the employment of the whole of its walls in the formation of various organs, is so peculiar, that I feel some hesitation in accepting Kowalevsky’s statements on this head[6].

The larva continues to grow in length, and the tail becomes further curled round the ventral side of the body within the egg-membrane. Before the tail has nearly reached its full length the test becomes formed as a cuticular deposit of the epiblast cells (O. Hertwig, No. 13, Semper, No. 37). It appears first in the tail and gradually extends till it forms a complete investment round both tail and trunk, and is at first totally devoid of cells. Shortly after the establishment of the test there grow out from the anterior end of the body three peculiar papillæ, developed as simple thickenings of the epidermis. At a later stage, after the hatching of the larva, these papillæ develop glands at their extremities, secreting a kind of glutinous fluid[7]. After these papillæ have become formed cells first make their appearance in the test; and there is simultaneously formed a fresh inner cuticular layer of the test, to which at first the cells are confined, though subsequently they are found in the outer layer also. On the appearance of cells in the test the latter must be regarded as a form, though a very abnormal one, of connective tissue. When the tail of the larva has reached a very considerable length the egg-membrane bursts, and the larva becomes free. The hatching takes place in Asc. canina about 48-60 hours after impregnation. The free larva (fig. 8 V.) has a swollen trunk, and a very long tail, which soon becomes straightened out. It has a striking resemblance to a tadpole (vide fig. 10).

In the free larval condition the Ascidians have in many respects a higher organization than in the adult state. It is accordingly convenient to divide the subsequent development into two periods, the first embracing the stages from the condition represented in fig. 8 V. up to the full development of the free larva, and the second the period from the full development of the larva to the attainment of the fixed adult condition.

Growth and Structure of the free larva.

The nervous system. The nervous system was left as a closed tube consisting of a dilated anterior division, and a narrow posterior one. The former may be spoken of as the brain, and the latter as the spinal cord; although the homologies of these two parts are quite uncertain. The anterior part of the spinal cord lying within the trunk dilates somewhat (fig. 8 V. and VI. Rg) and there may thus be distinguished a trunk and a caudal section of the spinal cord.

The original single vesicle of the brain becomes divided by the time the larva is hatched into two sections (fig. 9)—(1) an anterior vesicle with, for the most part, thin walls, in which unpaired auditory and optic organs make their appearance, and (2) a posterior nearly solid cephalic ganglion, through which there passes a narrow continuation of the central canal of the nervous system. This ganglion consists of a dorsal section formed of distinct cells, and a ventral section formed of a punctated material with nuclei. The auditory organ[8] consists of a ‘crista acustica’ (fig. 9), in the form of a slight prominence of columnar cells on the ventral side of the anterior cerebral vesicle; to the summit of which a spherical otolith is attached by fine hairs. In the crista is a cavity containing clear fluid. The dorsal half of the otolith is pigmented: the ventral half is without pigment. The crista is developed in situ, but the otolith is formed from a single cell on the dorsal side of the cerebral vesicle, which forms a projection into the cavity of the vesicle, and then travels (in a manner not clearly made out) round the right side of the vesicle till it comes to the crista; to which it is at first attached by a narrow pedicle. The fully developed eye (figs. 8 VI. and 9, O) consists of a cup-shaped retina, which forms a prominence slightly on the right side of the posterior part of the dorsal wall of the anterior cerebral vesicle, and of refractive media. The retina is formed of columnar cells, the inner ends of which are imbedded in pigment. The refractive media of the eye are directed towards the cavity of the cerebral vesicle, and consist of a biconvex lens and a meniscus. Half the lens is imbedded in the cavity of the retina and surrounded by the pigment, and the other half is turned toward a concavo-convex meniscus which corresponds in position with the cornea. The development of the meniscus and lens is unknown, but the retina is formed (fig. 8 V. a) as an outgrowth of the wall of the brain. At the inner ends of the cells of this outgrowth a deposit of pigment appears.

The trunk section of the spinal cord (fig. 9, N) is separated by a sharp constriction from the brain. It is formed of a superficial layer of longitudinal nervous fibres, and a central core of ganglion cells. The layer of fibres diminishes in thickness towards the tail, and finally ceases to be visible. Kupffer detected three pairs of nerves passing off from the spinal cord to the muscles of the tail. The foremost of these arises at the boundary between the trunk and the tail, and the two others at regular intervals behind this point.

The mesoblast and muscular system. It has already been stated that the lateral walls of the archenteron in the tail give rise to muscular cells. These cells lie about three abreast, and appear not to increase in number; so that with the growth of the tail they grow enormously in length, and eventually become imperfectly striated. The mesoblast cells at the hinder end of the trunk, close to its junction with the tail, do not become converted into muscle cells, but give rise to blood corpuscles; and the axial remnant of the archenteron undergoes a similar fate. According to Kowalevsky the heart is formed during larval life as an elongated closed sack on the right side of the endostyle.

The notochord. The notochord was left as a rod formed of a single row of cells, or in As. canina and some other forms of two rows, extending from just within the border of the trunk to the end of the tail.

According to Kowalevsky, Kupffer, Giard, etc. the notochord undergoes a further development which finds its only complete parallel amongst Chordata in the doubtful case of Amphioxus.

There appear between the cells peculiar, highly refractive discs (fig. 8 V. Chs). These become larger and larger, and finally, after pushing the remnants of the cells with their nuclei to the sides, coalesce together to form a continuous axis of hyaline substance. The remnants of the cells with their nuclei form a sheath round the hyaline axis (fig. 8 VI. ch.). Whether the axis is to be regarded as formed of an intercellular substance, or of a differentiation of parts of the cells is still doubtful. Kupffer inclines to the latter view: the analogy of the notochord of higher types appears to me to tell in favour of the former one.

The alimentary tract. The anterior part of the primitive archenteron alone retains a lumen, and from this part the whole of the permanent alimentary tract (mesenteron) becomes developed. The anterior part of it grows upwards, and before hatching an involution of the epiblast on the dorsal side, just in front of the anterior extremity of the nervous system, meets and opens into this upgrowth, and gives rise to the permanent mouth (fig. 8 V. o).

Kowalevsky states that a pore is formed at the front end of the nervous tube leading into the mouth (fig. 8 V. and VI. f) which eventually gives rise to the ciliated sack, which lies in the adult at the junction between the mouth and the branchial sack. Kupffer however was unable to find this opening; but Kowalevsky’s observations are confirmed by those of Salensky on Salpa.

From the hinder end of the alimentary sack an outgrowth directed dorsalwards makes its appearance (figs. 8 V. and 9, d), from which the œsophagus, stomach and intestine become developed. It at first ends blindly. The remainder of the primitive alimentary sack gives rise to the branchial sack of the adult. Just after the larva has become hatched, the outgrowth to form the stomach and œsophagus, etc. bends ventralwards and to the right, and then turns again in a dorsal and left direction till it comes close to the dorsal surface, somewhat to the left of and close to the hinder end of the trunk. The first ventral loop of this part gives rise to the œsophagus, which opens into the stomach; from this again the dorsally directed intestine passes off.

On the ventral wall of the branchial sack there is formed a narrow fold with thickened walls, which forms the endostyle. It ends anteriorly at the stomodæum and posteriorly at the point where the solid remnant of the archenteron in the tail was primitively continuous with the branchial sack. The whole of the alimentary wall is formed of a single layer of hypoblast cells.

A most important organ connected with the alimentary system still remains to be dealt with, viz. the atrial or peribranchial cavity. The first rudiments of it appear at about the time of hatching, in the form of a pair of dorsal epiblastic involutions (fig. 8 V. kl), at the level of the junction between the brain and the spinal cord. These involutions grow inwards, and meet corresponding outgrowths of the branchial sack, with which they fuse. At the junction between them is formed an elongated ciliated slit, leading from the branchial sack into the atrial cavity of each side. The slits so formed are the first pair of branchial clefts. Behind the first pair of branchial clefts a second pair is formed during larval life by a second outgrowth of the branchial sack meeting the epiblastic atrial involutions (fig. 8 VI. 1ks and 2ks). The intestine at first ends blindly close to the left atrial involution, but the anus becomes eventually formed by an opening being established between the left atrial involution and the intestine.

During the above described processes the test remains quite intact, and is not perforated at the oral or the atrial openings.

The retrogressive metamorphosis of the larva.

The development of the adult from the larva is, as has already been stated, in the main a retrogressive metamorphosis. The stages in this metamorphosis are diagrammatically shewn in figs. 10 and 11. It commences with the attachment of the larva (fig. 10 A) which takes place by one of the three papillæ. Simultaneously with the attachment the larval tail undergoes a complete atrophy (fig. 10 B), so that nothing is left of it but a mass of fatty cells situated close to the point of the previous insertion of the tail in the trunk.

The nervous system also undergoes a very rapid retrogressive metamorphosis; and the only part of it which persists would seem to be the dilated portion of the spinal cord in the trunk (Kupffer, No. 28).

The three papillæ, including that serving for attachment, early disappear, and the larva becomes fixed by a growth of the test to foreign objects.

An opening appears in the test some time after the larva is fixed, leading into the mouth, which then becomes functional. The branchial sack at the same time undergoes important changes. In the larva it is provided with only two ciliated slits, which open into the, at this stage, paired atrial cavity (fig. 10).

The openings of the atrial cavity at first are shut off from communication with the exterior by the test, but not long after the larva becomes fixed, two perforations are formed in the test, which lead into the openings of the two atrial cavities. At the same time the atrial cavities dilate so as gradually to embrace the whole branchial sack to which their inner walls attach themselves. Shortly after this the branchial clefts rapidly increase in number[9].

The increase of the branchial clefts is somewhat complicated. Between the two primitive clefts two new ones appear, and then a third appears behind the last cleft. In the interval between each branchial cleft is placed a vascular branchial vessel (fig. 8 VI. bb). Soon a great number of clefts become added in a row on each side of the branchial sack. These clefts are small ciliated openings placed transversely with reference to the long axis of the branchial sack, but only occupying a small part of the breadth of each side. The intervals dorsal and ventral to them are soon filled by series of fresh rows of slits, separated from each other by longitudinal bars. Each side of the branchial sack becomes in this way perforated by a number of small openings arranged in rows, and separated by transverse and longitudinal bars. The whole structure forms the commencement of the branchial basketwork of the adult; the arrangement of which differs considerably in structure and origin from the simple system of branchial clefts of normal vertebrate types. At the junction of the transverse and longitudinal bars papillæ are formed projecting into the lumen of the branchial sack.

After the above changes are far advanced towards completion, the openings of the two atrial sacks gradually approximate in the dorsal line, and finally coalesce to form the single atrial opening of the adult. The two atrial cavities at the same time coalesce dorsally to form a single cavity, which is continuous round the branchial sack, except along the ventral line where the endostyle is present. The atrial cavity, from its mode of origin as a pair of epiblastic involutions[10], is clearly a structure of the same nature as the branchial or atrial cavity of Amphioxus; and has nothing whatever to do with the true body cavity.

It has already been stated that the anus opens into the original left atrial cavity; when the two cavities coalesce the anus opens into the atrial cavity in the median dorsal line.

Two of the most obscure points in the development are the origin of the mesoblast in the trunk, and of the body cavity. Of the former subject we know next to nothing, though it seems that the cells resulting from the atrophy of the tail are employed in the nutrition of the mesoblastic structures of the trunk.

The body cavity in the adult is well developed in the region of the intestine, where it forms a wide cavity lined by an epithelioid mesoblastic layer. In the region of the branchial sack it is reduced to the vascular channels in the walls of the sack.

Kowalevsky believes the body cavity to be the original segmentation cavity, but this view can hardly be regarded as admissible in the present state of our knowledge. In some other Ascidian types a few more facts about the mesoblast will be alluded to.

With the above changes the retrogressive metamorphosis is completed; and it only remains to notice the change in position undergone in the attainment of the adult state. The region by which the larva is attached grows into a long process (fig. 10 B), and at the same time the part carrying the mouth is bent upwards so as to be removed nearly as far as possible from the point of attachment. By this means the condition in the adult (fig. 11) is gradually brought about; the original dorsal surface with the oral and atrial openings becoming the termination of the long axis of the body, and the nervous system being placed between the two openings.

The genus Molgula presents a remarkable exception amongst the simple Ascidians in that, in some if not all the species belonging to it, development takes place (Lacaze Duthiers 29 and 30, Kupffer 28) quite directly and without larval metamorphosis.

The ova are laid either singly or adhering together, and are very opaque. The segmentation (Lacaze Duthiers) commences by the formation of four equal spheres, after which a number of small clear spheres are formed which envelope the large spheres. The latter give rise to a closed enteric sack, and probably also to a mass of cells situated on the ventral side, which appear to be mesoblastic. The epiblast is constituted of a single layer of cells which completely envelopes the enteric sack and the mesoblast.

While the ovum is still within the chorion five peculiar processes of epiblast grow out; four of which usually lie in the same sectional plane of the embryo. They are contractile and contain prolongations of the body cavity. Their relative size is very variable.

The nervous system is formed on the dorsal side of the embryo before the above projections make their appearance, but, though it seems probable that it originates in the same manner as in the more normal forms, its development has not been worked out. As soon as it is formed it consists of a nervous ganglion similar to that usually found in the adult. The history of the mass of mesoblast cells has been inadequately followed, but it continuously disappears as the heart, excretory organs, muscles, etc. become formed. So far as can be determined from Kupffer’s descriptions the body cavity is primitively parenchymatous—an indication of an abbreviated development—and does not arise as a definite split in the mesoblast.

The primitive enteric cavity becomes converted into the branchial sack, and from its dorsal and posterior corner the œsophagus, stomach and intestine grow out as in the normal forms. The mouth is formed by the invagination of a disc-like thickening of the epidermis in front of the nervous system on the dorsal side of the body; and the atrial cavity arises behind the nervous system by a similar process at a slightly later period. The gill clefts opening into the atrial cavity are formed as in the type of simple Ascidians described by Krohn.

The embryo becomes hatched not long after the formation of the oral and atrial openings, and the five epiblastic processes undergo atrophy. They are not employed in the attachment of the adult.

The larva when hatched agrees in most important points with the adult; and is without the characteristic provisional larval organs of ordinary forms; neither organs of special sense nor a tail becoming developed. It has been suggested by Kupffer that the ventrally situated mesoblastic mass is the same structure as the mass of elements which results in ordinary types from the degeneration of the tail. If this suggestion is true it is difficult to believe that this mass has any other than a nutritive function.

The larva of Ascidia ampulloides described by P. van Beneden is regarded by Kupffer as intermediate between the Molgula larva and the normal type, in that the larval tail and notochord and a pigment spot are first developed, while after the atrophy of these organs peculiar processes like those of Molgula make their appearance.

Sedentaria. The development of the fixed composite Ascidians is, so far as we know, in the main similar to that of the simple Ascidians. The larvæ of Botryllus sometimes attain, while still in the free state, a higher stage of development with reference to the number of gill slits, etc. than that reached by the simple Ascidians, and in some instances (Botryllus auratus Metschnikoff) eight conical processes are found springing in a ring-like fashion around the trunk. The presence of these processes has led to somewhat remarkable views about the morphology of the group; in that they were regarded by Kölliker, Sars, etc. as separate individuals, and it was supposed that the product of each ovum was not a single individual, but a whole system of individuals with a common cloaca.

The researches of Metschnikoff (No. 32), Krohn (No. 25), and Giard (No. 12), etc. demonstrate that this paradoxical view is untenable, and that each ovum only gives rise to a single embryo, while the stellate systems are subsequently formed by budding.

Natantia. Our knowledge of the development of Pyrosoma is mainly due to Huxley (No. 16) and Kowalevsky (No. 22). In each individual of a colony of Pyrosoma only a single egg comes to maturity at one time. This egg is contained in a capsule formed of a structureless wall lined by a flattened epithelioid layer. From this capsule a duct passes to the atrial cavity, which, though called the oviduct, functions as an afferent duct for the spermatozoa.

The segmentation is meroblastic, and the germinal disc adjoins the opening of the oviduct. The segmentation is very similar to that which occurs in Teleostei, and at its close the germinal disc has the form of a cap of cells, without a trace of stratification or of a segmentation cavity, resting upon the surface of the yolk, which forms the main mass of the ovum.

After segmentation the blastoderm, as we may call the layer of cells derived from the germinal disc, rapidly spreads over the surface of the yolk, and becomes divided into two layers, the epiblast and the hypoblast. At the same time it exhibits a distinction into a central clearer and a peripheral more opaque region. At one end of the blastoderm, which for convenience sake may be spoken of as the posterior end, a disc of epiblast appears, which is the first rudiment of the nervous system, and on each side of the middle of the blastoderm there arises an epiblastic involution. The epiblastic involutions give rise to the atrial cavity.

These involutions rapidly grow in length, and soon form longish tubes, opening at the surface by pores situated not far from the posterior end of the blastoderm.

The blastoderm at this stage, as seen on the surface of the yolk, is shewn in fig. 12 A. It is somewhat broader than long. The nervous system is shewn at n, and at points to an atrial tube. A transverse section, through about the middle of this blastoderm, is represented in fig. 12 B. The epiblast is seen above. On each side is the section of an atrial tube (at). Below is the hypoblast which is separated from the yolk especially in the middle line; at each side it is beginning to grow in below, on the surface of the yolk. The space below the hypoblast is the alimentary cavity, the ventral wall of which is formed by the cells growing in at the sides. Between the epiblast and hypoblast are placed scattered mesoblast cells, the origin of which has not been clearly made out.

In a later stage the openings of the two atrial tubes gradually travel backwards, and at the same time approximate, till finally they meet and coalesce at the posterior end of the blastoderm behind the nervous disc (fig. 13, cl). The tubes themselves at the same time become slightly constricted not far from their hinder extremities, and so divided into a posterior region nearly coterminous with the nervous system (fig. 13), and an anterior region. These two regions have very different histories in the subsequent development.

The nervous disc has during these changes become marked by a median furrow (fig. 13, ng), which is soon converted into a canal by the same process as in the simple Ascidians. The closure of the groove commences posteriorly and travels forwards. These processes are clearly of the same nature as those which take place in Chordata generally in the formation of the central nervous system.

In the region of the germinal disc which contains the anterior part of the atrial tubes, the alimentary cavity becomes, by the growth of the layer of cells described in the last stage, a complete canal, on the outer wall of which the endostyle is formed as a median fold. The whole anterior part of the blastoderm becomes at the same time gradually constricted off from the yolk.

The fate of the anterior and posterior parts of the blastoderm is very different. The anterior part becomes segmented into four zooids or individuals, called by Huxley Ascidiozooids, which give rise to a fresh colony of Pyrosoma. The posterior part forms a rudimentary zooid, called by Huxley Cyathozooid, which eventually atrophies. These five zooids are formed by a process of embryonic fission. This fission commences by the appearance of four transverse constrictions in the anterior part of the blastoderm; by which the whole blastoderm becomes imperfectly divided into five regions, fig. 14 A.

The hindermost constriction (uppermost in my figure) lies just in front of the pericardial cavity; and separates the Cyathozooid from the four ascidiozooids. The three other constrictions mark off the four Ascidiozooids. The Cyathozooid remains for its whole length attached to the blastoderm, which has now nearly enveloped the yolk. It contains the whole of the nervous system (ng), which is covered behind by the opening of the atrial tubes (cl). The alimentary tract in the Cyathozooid forms a tube with very delicate walls. The pericardial cavity is completely contained within the Cyathozooid, and the heart itself (ht) has become formed by an involution of the walls of the cavity.

The Ascidiozooids are now completely separated from the yolk. They have individually the same structure as the undivided rudiment from which they originated; so that the organs they possess are simply two atrial tubes, an alimentary tract with an endostyle, and undifferentiated mesoblast cells.

In the following stages the Ascidiozooids grow with great rapidity. They soon cease to lie in a straight line, and eventually form a ring round the Cyathozooid and attached yolk sack.

While these changes are being accomplished in the external form of the colony, both the Cyathozooids and the Ascidiozooids progress considerably in development. In the Cyathozooid the atrial spaces gradually atrophy, with the exception of the external opening, which becomes larger and more conspicuous. The heart at the same time comes into full activity and drives the blood through the whole colony. The yolk becomes more and more enveloped by the Cyathozooid, and is rapidly absorbed; while the nutriment derived from it is transported to the Ascidiozooids by means of the vascular connection. The nervous system retains its previous condition; and round the Cyathozooid is formed the test into which cells migrate, and arrange themselves in very conspicuous hexagonal areas. The delicate alimentary tract of the Cyathozooid is still continuous with that of the first Ascidiozooid. After the Cyathozooid has reached the development just described it commences to atrophy.

The changes in the Ascidiozooids are even more considerable than those in the Cyathozooid. A nervous system appears as a fresh formation close to the end of each Ascidiozooid turned towards the Cyathozooid. It forms a tube of which the open front end eventually develops into the ciliated pit of the mouth, and the remainder into the actual nervous ganglion. Between the nervous system and the endostyle an involution appears, which gives rise to the mouth. On each side of the primitive alimentary cavity of each Ascidiozooid branchial slits make their appearance, leading into the atrial tubes; so that the primitive alimentary tract becomes converted into the branchial sacks of the Ascidiozooids. The remainder of the alimentary tract of each zooid is formed as a bud from the hind end of the branchial sack in the usual way. The alimentary tracts of the four Ascidiozooids are at first in free communication by tubes opening from the hinder extremity of one zooid into the dorsal side of the branchial sack of the next zooid. At the hinder end of each Ascidiozooid is developed a mass of fatty cells known as the elæoblast, which probably represents a rudiment of the larval tail of simple Ascidians. (Cf. pp. 30-32.)

The further changes consist in the gradual atrophy of the Cyathozooid, which becomes more and more enclosed within the four Ascidiozooids. These latter become completely enveloped in a common test, and form a ring round the remains of the yolk and of the Cyathozooid, the heart of which continues however to beat vigorously. The cloacal opening of the Cyathozooid persists through all these changes, and, after the Cyathozooid itself has become completely enveloped in the Ascidiozooids and finally absorbed, deepens to form the common cloacal cavity of the Pyrosoma colony.

The main parts of the Ascidiozooids were already formed during the last stage. The zooids long remain connected together, and united by a vascular tube with the Cyathozooid, and these connections are not severed till the latter completely atrophies. Finally, after the absorption of the Cyathozooid, the Ascidiozooids form a rudimentary colony of four individuals enveloped in a common test. The two atrial tubes of each zooid remain separate in front but unite posteriorly. An anus is formed leading from the rectum into the common posterior part of the atrial cavity; and an opening is established between the posterior end of the atrial cavity of each Ascidiozooid and the common axial cloacal cavity of the whole colony. The atrial cavities in Pyrosoma are clearly lined by epiblast, just as in simple Ascidians.

When the young colony is ready to become free, it escapes from the atrial cavity of the parent, and increases in size by budding.

Doliolidæ. The sexually developed embryos of Doliolum have been observed by Krohn (No. 23), Gegenbaur (No. 10), and Keferstein and Ehlers (No. 17); but the details of the development have been very imperfectly investigated.

The youngest embryo observed was enveloped in a large oval transparent covering, the exact nature of which is not clear. It is perhaps a larval rudiment of the test which would seem to be absent in the adult. Within this covering is the larva, the main organs of which are already developed; and which primarily differs from the adult in the possession of a larval tail similar to that of simple Ascidians.

In the body both oral and atrial openings are present, the latter on the dorsal surface; and the alimentary tract is fully established. The endostyle is already formed on the ventral wall of the branchial sack, but the branchial slits are not present. Nine muscular rings are already visible. The tail, though not so developed as in the simple Ascidians, contains an axial notochord of the usual structure, and lateral muscles. It is inserted on the ventral side, and by its slow movements the larva progresses.

In succeeding stages the tail gradually atrophies, and the gill slits, four in number, develop; at the same time a process or stolon, destined to give rise by budding to a second non-sexual generation, makes its appearance on the dorsal side in the seventh intermuscular space. This stolon is comparable with that which appears in the embryo of Salpa. When the tail completely atrophies the larva leaves its transparent covering, and becomes an asexual Doliolum with a dorsal stolon.

Salpidæ. As is well known the chains of Salpa alone are sexual, and from each individual of the chain only a single embryo is produced. The ovum from which this embryo takes its origin is visible long before the separate Salps of the chain have become completely developed. It is enveloped in a capsule continuous with a duct, which opens into the atrial cavity, and is usually spoken of as the oviduct. The capsule with the ovum is enveloped in a maternal blood sinus. Embryonic development commences after the chain has become broken up, and the spermatozoa derived from another individual would seem to be introduced to the ovum through the oviduct.

At the commencement of embryonic development the oviduct and ovicapsule undergo peculiar changes; and in part at least give rise to a structure subservient to the nutrition of the embryo, known as the placenta. These changes commence with the shortening of the oviduct, and the disappearance of a distinction between oviduct and ovicapsule. The cells lining the innermost end of the capsule, i.e. that at the side of the ovum turned away from the atrial cavity, become at the same time very columnar. The part of the oviduct between the ovum and the atrial cavity dilates into a sack, communicating on the one hand with the atrial cavity, and on the other by a very narrow opening with the chamber in which the egg is contained. This sack next becomes a prominence in the atrial cavity, and eventually constitutes a brood-pouch. The prominence it forms is covered by the lining of the atrial cavity, immediately within which is the true wall of the sack. The external opening of the sack becomes gradually narrowed, and finally disappears. In the meantime the chamber in which the embryo is at first placed acquires a larger and larger opening into the sack; till finally the two chambers unite, and a single brood-pouch containing the embryo is thus produced. The inner wall of the chamber is formed by the columnar cells already spoken of. They form the rudiment of the placenta. The double wall of the outer part of the brood-pouch becomes stretched by the growth of the embryo; the inner of its two layers then atrophies. The outer layer subsequently gives way, and becomes rolled back so as to lie at the inner end of the embryo, leaving the latter projecting freely into the atrial cavity.

While these changes are taking place the placenta becomes fully developed. The first rudiment of it consists, according to Salensky, of the thickened cells of the ovicapsule only, though this view is dissented from by Brooks, Todaro, etc. Its cells soon divide to form a largish mass, which becomes attached to a part of the epiblast of the embryo.

On the formation of the body cavity of the embryo a central axial portion of the placenta becomes separated from a peripheral layer; and a channel is left between them which leads from a maternal blood sinus into the embryonic body cavity. The peripheral layer of the placenta is formed of cells continuous with the epiblast of the embryo; while the axial portion is constituted of a disc of cells adjoining the embryo, with a column of fibres attached to the maternal side. The fibres of this column are believed by Salensky to be products of the original rudiment of the placenta. The placenta now assumes a more spherical form, and its cavity becomes shut off from the embryonic body cavity. The fibrous column breaks up into a number of strands perforating the lumen of the organ, and the cells of the wall become stalked bodies projecting into the lumen.

When the larva is nearly ready to become free the placenta atrophies.

The placenta functions in the nutrition of the embryo in the following way. It projects from its first formation into a maternal blood sinus, and, on the appearance of a cavity in it continuous with the body cavity of the embryo, the blood of the mother fully intermingles with that of the embryo. At a later period the communication with the body cavity of the embryo is shut off, but the cavity of the placenta is supplied with a continuous stream of maternal blood, which is only separated from the fœtal blood by a thin partition.

It is now necessary to turn to the embryonic development about which it is unfortunately not as yet possible to give a completely satisfactory account. The statements of the different investigators contradict each other on most fundamental points. I have followed in the main Salensky (No. 34), but have also called attention to some points where his observations diverge most from those of other writers, or where they seem unsatisfactory.

The development commences at about the period when the brood-pouch is becoming formed; and the ovum passes entirely into the brood-pouch before the segmentation is completed. The segmentation is regular, and the existence of a segmentation cavity is denied by Salensky, though affirmed by Kowalevsky and Todaro[11].

At a certain stage in the segmentation the cells of the ovum become divided into two layers, an epiblast investing the whole of the ovum with the exception of a small area adjoining the placenta, where the inner layer or hypoblast, which forms the main mass of the ovum, projects at the surface. The epiblast soon covers the whole of the hypoblast, so that there would seem (according to Salensky’s observations) to be a kind of epibolic invagination: a conclusion supported by Todaro’s figures.

At a later stage, on one side of the free apex of the embryo, a mesoblastic layer makes its appearance between the epiblast and hypoblast. This layer is derived by Salensky, as it appears to me on insufficient grounds, from the epiblast. Nearly at the same time there arises not far from the same point of the embryo, but on the opposite side, a solid thickening of epiblast which forms the rudiment of the nervous system. The nervous system is placed close to the front end of the body; and nearly at the opposite pole, and therefore at the hind end, there appears immediately below the epiblast a mass of cells forming a provisional organ known as the elæoblast. Todaro regards this organ as mesoblastic in origin, and Salensky as hypoblastic. The organ is situated in the position which would be occupied by the larval tail were it developed. It may probably be regarded (Salensky) as a disappearing rudiment of the tail, and be compared in this respect with the more or less similar mass of cells described by Kupffer in Molgula, and with the elæoblast in Pyrosoma.

After the differentiation of these organs a cavity makes its appearance between the epiblast and hypoblast, which is regarded by Salensky as the body cavity. It appears to be equivalent to the segmentation cavity of Todaro. According to Todaro’s statements, it is replaced by a second cavity, which appears between the splanchnic and somatic layers of mesoblast, and constitutes the true body cavity. The embryo now begins to elongate, and at the same time a cavity makes its appearance in the centre of the hypoblast cells. This cavity is the rudiment of the branchial and alimentary cavities: on its dorsal wall is a median projection, the rudiment of the so-called gill of Salpa.

At two points this cavity comes into close contact with the external skin. At one of these, situated immediately ventral to the nervous system, the mouth becomes formed at a later period. At the other, placed on the dorsal surface between the nervous system and the elæoblast, is formed the cloacal aperture.

By the stage under consideration the more important systems of organs are established, and the remaining embryonic history may be very briefly narrated.

The embryo at this stage is no longer covered by the walls of the brood-pouch but projects freely into the atrial cavity, and is only attached to its parent by means of the placenta. The epiblast cells soon give rise to a deposit which forms the mantle. The deposit appears however to be formed not only on the outer side of the epiblast but also on the inner side; so that the epiblast becomes cemented to the subjacent parts, branchial sack, etc., by an intercellular layer, which would seem to fill up the primitive body cavity with the exception of the vascular channels (Salensky).

The nervous system, after its separation from the epiblast, acquires a central cavity, and subsequently becomes divided into three lobes, each with an internal protuberance. At its anterior extremity it opens into the branchial sack; and from this part is developed the ciliated pit of the adult. The nervous ganglion at a later period becomes solid, and a median eye is subsequently formed as an outgrowth from it.

According to Todaro there are further formed two small auditory (? olfactory) sacks on the ventral surface of the brain, each of them placed in communication with the branchial cavity by a narrow canal.

The mesoblast gives rise to the muscles of the branchial sack, to the heart, and to the pericardium. The two latter are situated on the ventral side of the posterior extremity of the branchial cavity.

Branchial sack and alimentary tract. The first development of the enteric cavity has already been described. The true alimentary tract is formed as a bud from the hinder end of the primitive cavity. The remainder of the primitive cavity gives rise to the branchial sack. The so-called gill has at first the form of a lamella attached dorsally to the walls of the branchial sack; but its attachment becomes severed except at the two ends, and it then forms a band stretching obliquely across the branchial cavity, which subsequently becomes hollow and filled with blood corpuscles. The whole structure is probably homologous with the peculiar fold, usually prolonged into numerous processes, which normally projects from the dorsal wall of the Ascidian branchial sack.

On the completion of the gill the branchial sack becomes divided into a region dorsal to the gill, and a region ventral to it. Into the former the single atrial invagination opens. No gill slits are formed comparable with those in simple Ascidians, and the only representative of these structures is the simple communication which becomes established between the dorsal division of the branchial sack and the atrial opening. The whole branchial sack of Salpa, including both the dorsal and ventral divisions, corresponds with the branchial sack of simple Ascidians. On its ventral side the endostyle is formed in the normal way. The mouth arises at the point already indicated near the front end of the nervous system[12].

Development of the chain of sexual Salps. My description of the embryonic development of Salpa would not be complete without some reference to the development of the stolon of the Solitary generation of Salps by the segmentation of which a chain of sexual Salps originates.

The asexual Salp, the embryonic development of which has just been described, may be compared to the Cyathozooid of Pyrosoma, from which it mainly differs in being fully developed. While still in an embryonic condition it gives rise to a process or stolon, which becomes divided into a number of zooids by transverse constrictions, in the same manner that part of the germ of the ovum of Pyrosoma is divided by transverse constrictions into four Ascidiozooids.

The stolon arises as a projection on the right side of the body of the embryo close to the heart. It is formed (Salensky, No. 35) of an outgrowth of the body wall, into which there grow the following structures:
(1) A central hollow process from the end of the respiratory sack.
(2) A right and left lateral prolongation of the pericardial cavity.
(3) A solid process of cells on the ventral side derived from the same mass of the cells as the elæoblast.
(4) A ventral and a dorsal blood sinus.

Besides these parts there appears on the dorsal side a hollow tube, the origin of which is unknown, which gives rise to the nervous system.

The hollow process of the respiratory sack is purely provisional, and disappears without giving rise to any permanent structure. The right and left prolongations of the pericardial cavity become solid and eventually give origin to the mesoblast. The ventral process of cells is the most important structure in the stolon in that it gives rise both to the alimentary and respiratory sacks, and to the generative organs of the sexual Salps. The stolon containing the organs just enumerated becomes divided by transverse constrictions into a number of rings. These rings do not long remain complete, but become interrupted dorsally and ventrally. The imperfect rings so formed soon overlap, and each of them eventually gives rise to a sexual Salp. Although the stolon arises while the asexual Salp is still in an embryonic condition, it does not become fully developed till long after the asexual Salp has attained maturity.

Appendicularia. Our only knowledge of the development of Appendicularia is derived from Fol’s memoir on the group (No. 8). He simply states that it develops, as far as he was able to follow, like other Ascidians; and that the extremely minute size of the egg prevented him from pursuing the subject. He also states that the pair of pores leading from the branchial cavity to the exterior is developed from epiblastic involutions meeting outgrowths of the wall of the branchial sack.

Metagenesis.

One of the most remarkable phenomena in connection with the life history of many Ascidians is the occurrence of an alternation of sexual and gemmiparous generations. This alternation appears to have originated from a complication of the process of reproduction by budding, which is so common in this group. The mode in which this very probably took place will be best understood by tracing a series of transitional cases between simple budding and complete alternations of generations.

In the simpler cases, which occur in some Composita Sedentaria, the process of budding commences with an outgrowth of the body wall into the common test, containing a prolongation of part of the alimentary tract[13].

Between the epiblastic and hypoblastic layers of the bud so formed, a mesoblastic and sometimes a generative outgrowth of the parent also appears.

The systems of organs of the bud are developed from the corresponding layers to those in the embryo[14]. The bud eventually becomes detached, and in its turn gives rise to fresh buds. Both the bud and its parent reproduce sexually as well as by budding: the new colonies being derived from sexually produced embryos.

The next stage of complication is that found in Botryllus (Krohn, Nos. 25 and 26). The larva produced sexually gives rise to a bud from the right side of the body close to the heart. On the bud becoming detached the parent dies away without developing sexual organs. The bud of the second generation gives rise to two buds, a right one and a left one, and like the larva dies without reaching sexual maturity. The buds of the third generation each produce two buds and then suffer the same fate as their parent.

The buds of the third generation arrange themselves with their cloacal extremities in contact, and in the fourth generation a common cloaca is formed, and so a true radial system of zooids is established; the zooids of which are not however sexual.

The buds of the fourth generation in their turn produce two or three buds and then die away.

Fresh systems become formed by a continuation of the process of budding, but the zooids of the secondary systems so formed are sexual. The ova come to maturity before the spermatozoa, so that cross fertilization takes place.

In Botryllus we have clearly a rudimentary form of alternations of generations, in that the sexually produced larva is asexual, and, after a series of asexual generations, produced gemmiparously, there appear sexual generations, which however continue to reproduce themselves by budding.

The type of alternations of generations observable in Botryllus becomes, as pointed out by Huxley, still more marked in Pyrosoma.

The true product of the ovum is here (vide p. 25) a rudimentary individual called by Huxley the Cyathozooid. This gives rise, while still an embryo, by a process equivalent to budding to four fully developed zooids (Ascidiozooids) similar to the parent form, and itself dies away. The four Ascidiozooids form a fresh colony, and reproduce (1) sexually, whereby fresh colonies are formed, and (2) by ordinary budding, whereby the size of the colony is increased. All the individuals of the colony are sexual.

The alternation of generations in Pyrosoma widely differs from that in Botryllus in the fact of the Cyathozooid differing so markedly in its anatomical characters from the ordinary zooids.

In Salpa the process is slightly different[15]. The sexual forms are now incapable of budding, and, although at first a series of sexual individuals are united together in the form of a chain, so as to form a colony like Pyrosoma or Botryllus, yet they are so loosely connected that they separate in the adult state. As in Botryllus, the ova are ripe before the spermatozoa. Each sexual individual gives rise to a single offspring, which, while still in the embryonic condition, buds out a ‘stolon’ from its right ventral side. This stolon is divided into a series of lateral buds after the solitary asexual Salp has begun to lead an independent existence. The solitary asexual Salp clearly corresponds with the Cyathozooid of Pyrosoma, though it has not, like the Cyathozooid, undergone a retrogressive metamorphosis.

By far the most complicated form of alternation of generations known amongst the Ascidians is that in Doliolum. The discovery of this metamorphosis was made by Gegenbaur (No. 10). The sexual form of Doliolum is somewhat cask-shaped, with ring-like muscular bands, and the oral and atrial apertures placed at opposite ends of the cask. The number of gill slits varies according to the species. The ovum gives rise, as already described, to a tailed embryo which subsequently develops into a cask-shaped asexual form. On attaining its full size it loses its branchial sack and alimentary tract. While still in the embryonic condition, a stolon grows out from its dorsal side in the seventh intermuscular space. The stolon, like that in Salpa, contains a prolongation of the branchial sack[16].

On this stolon there develop two entirely different types of buds, (1) lateral buds, (2) dorsal median buds.

The lateral buds are developed in regular order on the two sides of the stolon, and the most advanced buds are those furthest removed from the base. They give rise to forms with a very different organization to that of the parent. They are compared by Gegenbaur to a spoon, the bowl of which is formed by the branchial sack, and the handle by the stalk attaching the bud to the stolon. The oral opening into the branchial sack is directed upwards: an atrial opening is remarkably enough not present. The branchial sack is perforated by numerous openings. It leads into an alimentary tract which opens directly to the exterior by an anus opposite the mouth.

The stalks attaching the more mature buds to the stolon are provided with ventrally directed scales, which completely hide the stolon in a view from the ventral surface.

These buds have, even after their detachment, no trace of generative organs, and shew no signs of reproducing themselves by budding. Their eventual fate is unknown.

The median dorsal buds have no such regular arrangement as the lateral buds, but arise in irregular bunches, those furthest removed from the base of the stolon being however the oldest. These buds are almost exactly similar to the original sexual form; they do not acquire sexual organs, but are provided with a stolon attached on the ventral side, in the sixth intermuscular space.

This stolon is simply the stalk by which each median bud was primitively attached to the stolon of the first asexual form.

From the stolon of the median buds of the second generation buds are developed which grow into the sexual forms.

The generations of Doliolum may be tabulated in the following way.

Sexual generation,
|
1st asexual form with dorsal stolon,
|
____________________|____________________
spoon-like forms developed as lateral buds (eventual history unknown).		2nd asexual forms developed as median buds with vental stolon,
		|
		sexual generation.

Bibliography.

(6) P. J. van Beneden. “Recherches s. l'Embryogénie, l'Anat. et la Physiol. des Ascidies simples.” Mém. Acad. Roy. de Belgique, Tom. XX.
(7) W. K. Brooks. “On the development of Salpa.” Bull. of the Museum of Comp. Anat. at Harvard College, Cambridge, Mass.
(8) H. Fol. Etudes sur les Appendiculaires du détroit de Messine. Genève et Bâle, 1872.
(9) Ganin. “Neue Thatsachen a. d. Entwicklungsgeschichte d. Ascidien.” Zeit. f. wiss. Zool., Vol. XX. 1870.
(10) C. Gegenbaur. “Ueber den Entwicklungscyclus von Doliolum nebst Bemerkungen über die Larven dieser Thiere.” Zeit. f. wiss. Zool., Bd. VII. 1856.
(11) A. Giard. “Etudes critiques des travaux d'embryogénie relatifs à la parenté des Vertebrés et des Tuniciers.” Archiv Zool. expériment., Vol. I. 1872.
(12) A. Giard. “Recherches sur les Synascidies.” Archiv Zool. expér., Vol. I. 1872.
(13) O. Hertwig. “Untersuchungen üb. d. Bau u. d. Entwicklung des Cellulose-Mantels d. Tunicaten.” Jenaische Zeitschrift, Bd. VII. 1873.
(14) Th. H. Huxley. “Remarks upon Appendicularia and Doliolum.” Phil. Trans., 1851.
(15) Th. H. Huxley. “Observations on the anatomy and physiology of Salpa and Pyrosoma.” Phil. Trans., 1851.
(16) Th. H. Huxley. “Anatomy and development of Pyrosoma.” Linnean Trans., 1860, Vol. XXIII.
(17) Keferstein u. Ehlers. Zoologische Beiträge, 1861. Doliolum.
(18) A. Kowalevsky. “Entwicklungsgeschichte d. einfachen Ascidien.” Mém. Acad. Pétersbourg, VII. série, T. X. 1866.
(19) A. Kowalevsky. “Beitrag z. Entwick. d. Tunicaten.” Nachrichten d. königl. Gesell. zu Göttingen. 1868.
(20) A. Kowalevsky. “Weitere Studien üb. d. Entwicklung d. einfachen Ascidien.” Archiv f. mikr. Anat., Vol. VII. 1871.
(21) A. Kowalevsky. “Ueber Knospung d. Ascidien.” Archiv f. mikr. Anat., Vol. X. 1874.
(22) A. Kowalevsky. “Ueber die Entwicklungsgeschichte d. Pyrosoma.” Archiv f. mikr. Anat., Vol. XI. 1875.
(23) A. Krohn. “Ueber die Gattung Doliolum u. ihre Arten.” Archiv f. Naturgeschichte, Bd. XVIII. 1852.
(24) A. Krohn. “Ueber die Entwicklung d. Ascidien.” Müller’s Archiv, 1852.
(25) A. Krohn. “Ueber die Fortpflanzungsverhältnisse d. Botrylliden.” Archiv f. Naturgeschichte, Vol. XXXV. 1869.
(26) A. Krohn. “Ueber die früheste Bildung d. Botryllenstöcke.” Archiv f. Naturgeschichte, Vol. XXXV. 1869.
(27) C. Kupffer. “Die Stammverwandschaft zwischen Ascidien u. Wirbelthieren.” Archiv f. mikr. Anat., Vol. VI. 1870.
(28) C. Kupffer. “Zur Entwicklung d. einfachen Ascidien.” Archiv f. mikr. Anat., Vol. VIII. 1872.
(29) H. Lacaze Duthiers. “Recherches sur l'organisation et l'Embryogénie des Ascidies (Molgula tubulosa).” Comptes rendus, May 30, 1870, p. 1154.
(30) H. Lacaze Duthiers. “Les Ascidies simples des Côtes de France"” (Development of Molgula). Archiv Zool. expér., Vol. III. 1874.
(31) R. Leuckart. “Salpa u. Verwandte.” Zoologische Untersuchungen, Heft II.
(32) E. Metschnikoff. “Observations sur le développement de quelques animaux (Botryllus and Simple Ascidians).” Bull. d. l'Acad. Pétersbourg, Vol. XIII. 1869.
(33) H. Milne-Edwards. “Observations s. l. Ascidies composées des côtes de la Manche.” Mémoires d. l'Institut, T. XVIII. 1842.
(34) W. Salensky. “Ueber d. embryonale Entwicklungsgeschichte der Salpen.” Zeit. f. wiss. Zool., B. XXVII. 1877.
(35) W. Salensky. “Ueber die Knospung d. Salpen.” Morphol. Jahrbuch, Bd. III. 1877.
(36) W. Salensky. “Ueber die Entwicklung d. Hoden u. über den Generationswechsel d. Salpen.” Zeit. f. wiss. Zool., Bd. XXX. Suppl. 1878.
(37) C. Semper. “Ueber die Entstehung d. geschichteten Cellulose-Epidermis d. Ascidien.” Arbeit. a. d. zool.-zoot. Instit. Würzburg, Vol. II. 1875.
(38) Fr. Todaro. Sopra lo sviluppo e l'anatomia delle Salpe. Roma, 1875.
(39) Fr. Todaro. “Sui primi fenomeni dello sviluppo delle Salpe.” Reale Accademia dei Lincei, Vol. IV. 1880.

The impregnation of the ovum is effected in the oviduct. In most forms the whole of the subsequent development, till the time when the embryo is capable of leading a free existence, takes place in the uterus; but in other cases the egg becomes enveloped, during its passage down the oviduct, first in a layer of fluid albumen, and finally in a dense horny layer, which usually takes the form of a quadrilateral capsule with characters varying according to the species. After the formation of this capsule the egg is laid, and the whole of the development, with the exception of the very first stages, takes place externally.

In many of the viviparous forms (Mustelus, Galeus, Carcharias, Sphyrna) the egg is enclosed, during the early stages of development at any rate, in a very delicate shell homologous with that of the oviparous forms; there is usually also a scanty albuminous layer. Both of these are stated by Gerbe (No. 42) to be absent in Squalus spinax.

The following are examples of viviparous genera: Hexanchus, Notidanus, Acanthias, Scymnus, Galeus, Squalus, Mustelus, Carcharias, Sphyrna, Squatina, Torpedo; and the following of oviparous genera: Scyllium, Pristiurus, Cestracion, Raja[17].

The ovum at the time of impregnation has the form of a large spherical mass, similar to the yolk of a bird’s egg, but without a vitelline membrane[18]. The greater part of it is formed of peculiar oval spherules of food-yolk, held together by a protoplasmic network. The protoplasm is especially concentrated in a small lens-shaped area, known as the germinal disc, which is not separated by a sharp line from the remainder of the ovum. Yolk spherules are present in this disc as elsewhere, but are much smaller and of a different character. The segmentation has the normal meroblastic character (fig. 15) and is confined to the germinal disc. Before it commences the germinal disc exhibits amœboid movements. During the segmentation nuclei make their appearance spontaneously (?) in the yolk adjoining the germinal disc (fig. 15, nx´), and around them portions of the yolk with its protoplasmic network become segmented off. Cells are thus formed which are added to those resulting from the segmentation proper. Even after the segmentation numerous nuclei are present in the granular matter below the blastoderm (fig. 16 A, n´); and around these cells are being continually formed, which enter the blastoderm, and are more especially destined to give rise to the hypoblast. The special destination of many of these cells is spoken of in detail below.

At the close of segmentation the blastoderm forms a somewhat lens-shaped disc, thicker at one end than at the other; the thicker end being the embryonic end. It is divided into two strata—an upper one, the epiblast—formed of a single row of columnar cells; and a lower one, the primitive hypoblast, consisting of the remaining cells of the blastoderm, and forming a mass several strata deep. These cells will be spoken of as the lower layer cells, to distinguish them from the true hypoblast which is one of their products.

A cavity very soon appears in the lower layer cells, near the non-embryonic end of the blastoderm, but the cells afterwards disappear from the floor of this cavity, which then lies between the yolk and the lower layer cells (fig. 16 A, sc). This cavity is the segmentation cavity equivalent to that present in Amphioxus, Amphibia, etc. The chief peculiarity about it is the relatively late period at which it makes its appearance, and the fact that its roof is formed both by the epiblast and by the lower layer cells. Owing to the large size of the segmentation cavity the blastoderm forms a thin layer above the cavity and a thickened ridge round its edge.

The epiblast in the next stage is inflected for a small arc at the embryonic end of the blastoderm, where it becomes continuous with the lower layer cells; at the same time some of the lower layer cells of the embryonic end of the blastoderm assume a columnar form, and constitute the true hypoblast. The portion of the blastoderm, where epiblast and hypoblast are continuous, forms a projecting structure which will be called the embryonic rim (fig. 16 B, er).

This rim is a very important structure, since it represents the dorsal portion of the lip of the blastopore of Amphioxus. The space between it and the yolk represents the commencing mesenteron, of which the hypoblast on the under side of the lip is the dorsal wall. The ventral wall of the mesenteron is at first formed solely of yolk held together by a protoplasmic network with numerous nuclei. The cavity under the lip becomes rapidly larger (fig. 17, al), owing to the continuous conversion of lower layer cells into columnar hypoblast along an axial line passing from the middle of the embryonic rim towards the centre of the blastoderm. The continuous differentiation of the hypoblast towards the centre of the blastoderm corresponds with the invagination in Amphioxus. During the formation of the embryonic rim the blastoderm grows considerably larger, but, with the exception of the formation of the embryonic rim, retains its primitive constitution.

The segmentation cavity undergoes however important changes. There is formed below it a floor of lower layer cells, derived partly from an ingrowth from the two sides, but mainly from the formation of cells around the nuclei of the yolk (fig. 16). Shortly after the floor of cells has appeared, the whole segmentation cavity becomes obliterated (fig. 17).

The disappearance of the segmentation cavity corresponds in point of time with the formation of the hypoblast by the pseudo-invagination above described; and is probably due to this pseudo-invagination, in the same way that the disappearance of the segmentation cavity in Amphioxus is due to the true invagination of the hypoblast.

When the embryonic rim first appears there are no external indications of the embryo as distinguished from the blastoderm, but when it has attained to some importance the position of the embryo becomes marked out by the appearance of a shield-like area extending inwards from the edge of the embryonic rim, and formed of two folds with a groove between them (fig. 28 B, mg), which is deepest at the edge of the blastoderm, and shallows out as it extends inwards. This groove is the medullary groove; and its termination at the edge of the blastoderm is placed at the hind end of the embryo.

At about the time of its appearance the mesoblast becomes first definitely established.

At the edge of the embryonic rim the epiblast and lower layer cells are continuous. Immediately underneath the medullary groove, as is best seen in transverse section (fig. 18), the whole of the lower layer cells become converted into hypoblast, and along this line the columnar hypoblast is in contact with the epiblast above. At the sides however this is not the case; but at the junction of the epiblast and lower layer cells the latter remain undifferentiated. A short way from the edge the lower layer cells become divided into two distinct layers, a lower one continuous with the hypoblast in the middle line, and an upper one between this and the epiblast (fig. 18 B). The upper layer is the commencement of the mesoblast (m). The mesoblast thus arises as two independent lateral plates, one on each side of the medullary groove, which are continuous behind with the undifferentiated lower layer cells at the edge of the embryonic rim. The mesoblast plates are at first very short, and do not extend to the front end of the embryo. They soon however grow forwards as two lateral ridges, attached to the hypoblast, one on each side of the medullary groove (fig. 18 A, m). These ridges become separate from the hypoblast, and form two plates, thinner in front than behind; but still continuous at the edge of the blastoderm with the undifferentiated cells of the lip of the blastopore, and laterally with the lower layer cells of the non-embryonic part of the blastoderm. It results from the above mode of development of the mesoblast, that it may be described as arising in the form of a pair of solid outgrowths of the wall of the alimentary tract; which differ from the mesoblastic outgrowths of the wall of the archenteron in Amphioxus in not containing a prolongation of the alimentary cavity.

A general idea of the structure of the blastoderm at this stage may be gathered from the diagram representing a longitudinal section through the embryo (fig. 19 B). In this figure the epiblast is represented in white and is seen to be continuous at the lip of the blastopore (x) with the shaded hypoblast. Between the epiblast and hypoblast is seen one of the lateral plates of mesoblast, represented by black cells with clear outlines. The non-embryonic lower layer cells of the blastoderm are represented in the same manner as the mesoblast of the body. The alimentary cavity is shewn at al, and below it is seen the yolk with nuclei (n). The segmentation cavity is represented as still persisting, though by this stage it would have disappeared.

As to the growth of the blastoderm it may be noted that it has greatly extended itself over the yolk. Its edge in the meantime forms a marked ridge, which is due not so much to a thickening as to an arching of the epiblast. This ridge is continuous with the embryonic rim, which gradually concentrates itself into two prominences, one on each side of the tail of the embryo, mainly formed of masses of undifferentiated lower layer cells. These prominences will be called the caudal swellings.

By this stage the three layers of the body, the epiblast, mesoblast, and hypoblast, have become definitely established. The further history of these layers may now be briefly traced.

Epiblast. While the greater part of the epiblast becomes converted into the external epidermis, from which involutions give rise to the olfactory and auditory pits, the lens of the eye, the mouth cavity, and anus, the part of it lining the medullary groove becomes converted into the central nervous system and optic cup. The medullary groove is at first continued to the front end of the medullary plate; but the anterior part of this plate soon enlarges, and the whole plate assumes a spatula form (fig. 28 C, h, and fig. 20 A and B). The enlarged part becomes converted into the brain, and may be called the cephalic plate.

The posterior part of the canal deepens much more rapidly than the rest (fig. 20 C), and the medullary folds unite dorsally and convert the posterior end of the medullary groove into a closed canal, while the groove is still widely open elsewhere. The medullary canal does not end blindly behind, but simply forms a tube not closed at either extremity. The importance of this fact will appear later.

Shortly after the medullary folds have met behind the whole canal becomes closed in. This occurs in the usual way by the junction and coalescence of the medullary folds. In the course of the closing of the medullary groove the edges of the cephalic plate, which have at first a ventral curvature, become bent up in the normal manner, and enclose the dilated cephalic portion of the medullary canal. The closing of the medullary canal takes place earlier in the head and neck than in the back.

An anterior pore at the front end of the canal, like that in Amphioxus and the Ascidians, is not found. The further differentiation of the central nervous system is described in a special chapter: it may however here be stated that the walls of the medullary canal give rise not only to the central nervous system but to the peripheral also.

Mesoblast. The mesoblast was left as two lateral plates continuous behind with the undifferentiated cells of the caudal swellings.

The cells composing them become arranged in two layers (fig. 20 C, lp), a splanchnic layer adjoining the hypoblast, and a somatic layer adjoining the epiblast. Between these two layers there is soon developed in the region of the head a well-marked cavity (fig. 20 A, pp) which is subsequently continued into the region of the trunk, and forms the primitive body cavity, equivalent to the cavity originating as an outgrowth of the archenteron in Amphioxus. The body cavities of the two sides are at first quite independent.

Coincidentally with the appearance of differentiation into somatic and splanchnic layers the mesoblast plates become in the region of the trunk partially split by a series of transverse lines of division into mesoblastic somites. Only the dorsal parts of the plates become split in this way, their ventral parts remaining quite intact. As a result of this each plate becomes divided into a dorsal portion adjoining the medullary canal, which is divided into somites, and may be called the vertebral plate, and a ventral portion not so divided, which may be called the lateral plate. These two parts are at this stage quite continuous with each other; and the body cavity originally extends uninterruptedly to the summit of the vertebral plates (fig. 21).

The next change results in the complete separation of the vertebral portion of the plate from the lateral portion; thereby the upper segmented part of the body cavity becomes isolated, and separated from the lower and unsegmented part. As a consequence of this change the vertebral plate comes to consist of a series of rectangular bodies, the mesoblastic somites, each composed of two layers, a somatic and a splanchnic, between which is the cavity originally continuous with the body cavity (fig. 23, mp). The splanchnic layer of the plates buds off cells to form the rudiments of the vertebral bodies which are at first segmented in the same planes as the mesoblastic somites (fig. 22, Vr). The plates themselves remain as the muscle-plates (mp), and give rise to the whole of the voluntary muscular system of the body. Between the vertebral and lateral plates there is left a connecting isthmus, with a narrow prolongation of the body cavity (fig. 23 B, st), which gives rise (as described in a special chapter) to the segmental tubes and to other parts of the excretory system.

In the meantime the lateral plates of the two sides unite ventrally throughout the intestinal and cardiac regions of the body, and the two primitively isolated cavities contained in them coalesce. In the tail however the plates do not unite ventrally till somewhat later, and their contained cavities remain distinct till eventually obliterated.

At first the pericardial cavity is quite continuous with the body cavity; but it eventually becomes separated from the body cavity by the attachment of the liver to the abdominal wall, and by a horizontal septum in which run the two ductus Cuvieri (fig. 23 A, sv). Two perforations in this septum (fig. 23 A) leave the cavities in permanent communication.

The parts derived from the two layers of the mesoblast (not including special organs or the vascular system) are as follows:—

From the somatic layer are formed
(1) A considerable part of the voluntary muscular system of the body.
(2) The dermis.
(3) A large part of the intermuscular connective tissue.
(4) Part of the peritoneal epithelium.

From the splanchnic layer are formed
(1) A great part of the voluntary muscular system.
(2) Part of the intermuscular connective tissue.
(3) The axial skeleton and surrounding connective tissue.
(4) The muscular and connective-tissue wall of the alimentary tract.
(5) Part of the peritoneal epithelium.

In the region of the head the mesoblast does not at first become divided into somites; but on the formation of the gill clefts a division takes place, which is apparently equivalent to the segmentation of the mesoblast in the trunk. This division causes the body cavity of the head to be divided up into a series of separate segments, one of which is shewn in fig. 24, pp. The walls of the segments eventually give rise to the main muscles of the branchial clefts, and probably also to the muscles of the mandibular arch, of the eye, and of other parts. The cephalic sections of the body cavity will be spoken of as head cavities.

In addition to the parts already mentioned the mesoblast gives rise to the whole of the vascular system, and to the generative system. The heart is formed from part of the splanchnic mesoblast, and the generative system from a portion of the mesoblast of the dorsal part of the body cavity.

The hypoblast. Very shortly after the formation of the mesoblastic plates as lateral differentiations of the lower layer cells, an axial differentiation of the hypoblast appears, which gives rise to the notochord very much in the same way as in Amphioxus.

At first the hypoblast along the axial line forms a single layer in contact with the epiblast. Along this line a rod-like thickening of the hypoblast very soon appears (fig. 25, B and C, Ch´) at the head end of the embryo, and gradually extends backwards. This is the rudiment of the notochord; it remains attached for some time to the hypoblast, and becomes separated from it first at the head end of the embryo (fig. 25 A, ch): the separation is then carried backwards.

A series of sections taken through an embryo shortly after the first differentiation of the notochord presents the following characters.

In the hindermost sections the hypoblast retains a perfectly normal structure and uniform thickness throughout. In the next few sections (fig. 25 C, Ch´) a slight thickening is to be observed in it, immediately below the medullary groove. The layer, which elsewhere is composed of a single row of cells, here becomes two cells deep, but no sign of a division into two layers is exhibited.

In the next few sections the thickening of the hypoblast becomes much more pronounced; we have, in fact, a ridge projecting from the hypoblast towards the epiblast (fig. 25 B, Ch´). This ridge is pressed firmly against the epiblast, and causes in it a slight indentation. The hypoblast in the region of the ridge is formed of two layers of cells, the ridge being entirely due to the uppermost of the two.

In sections in front of this a cylindrical rod, which can at once be recognized as the notochord, and is continuous with the ridge just described, begins to be split off from the hypoblast (fig. 25 A, Ch). It is difficult to say at what point the separation of this rod from the hypoblast is completed, since all intermediate gradations between complete separation and complete attachment are to be seen.

Shortly after the separation takes place, a fairly thick bridge is found connecting the two lateral halves of the hypoblast, but this bridge is anteriorly excessively delicate and thin, and in some cases is barely visible except with high powers. In some sections I have observed possible indications of the process like that described by Calberla for Petronyzon, by which the lateral parts of the hypoblast grow in underneath the axial part, and so isolate it bodily as the notochord.

It is not absolutely clear whether the notochord is to be regarded as an axial differentiation of the hypoblast, or as an axial differentiation of the lower layer cells.

The facts of development both in Amphioxus and Elasmobranchii tend towards the former view; but the nearly simultaneous differentiation of the notochord and the mesoblastic plates lends some support to the supposition that the notochord may be merely a median plate of mesoblast developed slightly later than the two lateral plates.

The alimentary canal or mesenteron was left as a space between the hypoblast and the yolk, ending blindly in front, but opening behind by a widish aperture, the blastopore or anus of Rusconi (vide fig. 19 B).

The conversion of this irregular cavity into a closed canal commences first of all at the anterior extremity. In this conversion two distinct processes are concerned. One of these is a process of folding off of the embryo from the blastoderm. The other is a simple growth of cells independent of any fold. To the first of these processes the depth and narrowness of the alimentary cavity is due; the second is concerned in forming its ventral wall. The process of the folding off of the embryo from the blastoderm resembles exactly the similar process in the embryo bird. The fold is a perfectly continuous one round the front end of the embryo, but may be conveniently spoken of as composed of a head-fold and two lateral folds.

Of far greater interest than the nature of these folds is the formation of the ventral wall of the alimentary canal. This originates in a growth of cells from the two sides to the middle line (fig. 26). The cells for it are not however mainly derived from pre-existing hypoblast cells, but are formed de novo around the nuclei of the yolk which have already been spoken of (fig. 26, na). The ventral wall of the mesenteron is in fact, to a large extent at any rate, formed as a differentiation of the primitive yolk floor.

The folding off and closing of the alimentary canal in the anterior part of the body proceeds rapidly, and not only is a considerable tract of the alimentary canal formed, but a great part of the head is completely folded off from the yolk before the medullary groove is closed.

The posterior part of the alimentary canal retains for a longer time its primitive condition. Finally however it also becomes closed in, by the lips of the blastopore at the hind end of the embryo meeting and uniting. The peculiarity of the closing in of the posterior part of the alimentary canal consists in the fact that a similar continuity to that in Amphioxus obtains between the neural and alimentary canals. This is due to the medullary folds being continuous at the end of the tail with the lips of the blastopore, which close in the hind end of the alimentary canal; so that, when the medullary folds unite to form a canal, this canal becomes continuous with the alimentary canal, which is closed in at the same time. In other words, the medullary folds assist in enveloping the blastopore which does not therefore become absolutely closed, but opens into the floor of the neural canal. It will afterwards be shewn that it is only the posterior part of the blastopore that becomes closed during the above process, and that the anterior and ventral part long remains open. The general arrangement of the parts, at the time when the hind end of the mesenteron is first closed, is shewn in fig. 27. The same points may be seen in the diagrammatic longitudinal section fig. 19 C.

The middle portion of the alimentary tract is the last to be closed in since it remains till late in embryonic life as the umbilical or vitelline canal, connecting the yolk-sack with the alimentary cavity. The umbilical canal falls into the alimentary tract immediately behind the entrance of the hepatic duct.

At a fairly early stage of development a rod is constricted off from the dorsal wall of the alimentary canal (figs. 27* and 23 x), which is known as the subnotochordal rod. It is placed immediately below the notochord, and disappears during embryonic life.

General features of the Elasmobranch embryo at successive stages.

Shortly after the three germinal layers become definitely established, the rudiment of the embryo, as visible from the surface, consists of an oblong plate, which extends inwards from the periphery of the blastoderm, and is bounded on its inner side by a head-fold and two lateral folds (fig. 28 B). This plate is the medullary plate; along its axial line is a shallow groove—the medullary groove (mg). The rudiment of the embryo rapidly increases in length, and takes a spatula-like form (fig. 28 C). The front part of it, turned away from the edge of the blastoderm, soon becomes dilated into a broad plate,—the cephalic plate (h)—while the tail end at the edge of the blastoderm is also enlarged, being formed of a pair of swellings—the tail swellings (ts)—derived from the lateral parts of the original embryonic rim. By this stage a certain number of mesoblastic somites have become formed but are not shewn in my figure. They are the foremost somites of the trunk, and those behind them continue to be added, like the segments in Chætopods, between the last formed somite and the end of the body. The increase in length of the body mainly takes place by growth in the region between the last mesoblastic somite and the end of the tail. The anterior part of the body is now completely folded off from the blastoderm, and the medullary groove of the earlier stage has become converted into a closed canal.

By the next stage (fig. 28 D) the embryo has become so much folded off from the yolk both in front and behind that the separate parts of it begin to be easily recognizable.

The embryo is attached to the yolk by a distinct stalk or cord, which in the succeeding stages gradually narrows and elongates, and is known as the umbilical cord (so. s.). The medullary canal has now become completely closed. The anterior region constitutes the brain; and in this part slight constrictions, not perceptible in views of the embryo as a transparent object, mark off three vesicles. These vesicles are known as the fore, mid, and hind brain. From the fore-brain there is an outgrowth on each side, the first rudiment of the optic vesicles (op). The tail swellings are still conspicuous.

The tissues of the body have now become fairly transparent, and there may be seen at the sides of the body seventeen mesoblastic somites. The notochord, which was formed long before the stage represented in figure 28 D, is now also distinctly visible. It extends from almost the extreme posterior to the anterior end of the embryo, and lies between the ventral wall of the spinal canal and the dorsal wall of the intestine. Round its posterior end the neural and alimentary tracts become continuous with each other. Anteriorly the termination of the notochord cannot be seen, it can only be traced into a mass of mesoblast at the base of the brain, which there separates the epiblast from the hypoblast. The alimentary canal (al) is completely closed anteriorly and posteriorly, though still widely open to the yolk-sack in the middle part of its course. In the region of the head it exhibits on each side a slight bulging outwards, the rudiment of the first visceral cleft. This is represented in the figure by two lines (1. v.c.).

The embryo represented in fig. 28 E is far larger than the one just described, but it has not been convenient to represent this increase of size in the figure. Accompanying this increase in size, the folding off from the yolk has considerably progressed, and the stalk which unites the embryo with the yolk is proportionately narrower and longer than before.

The brain is now very distinctly divided into the three lobes, the rudiments of which appeared during the last stage. From the foremost of these the optic vesicles now present themselves as well-marked lateral outgrowths, towards which there has appeared an involution from the external skin (op) to form the lens.

A fresh organ of sense, the auditory sack, now for the first time becomes visible as a shallow pit in the external skin on each side of the hind-brain (au.v). The epiblast which is involuted to form this pit becomes much thickened, and thereby the opacity, indicated in the figure, is produced.

The mesoblastic somites have greatly increased in number by the formation of fresh somites in the tail. Thirty-eight of them were present in the embryo figured. The mesoblast at the base of the brain is more bulky, and there is still a mass of unsegmented mesoblast which forms the tail swellings. The first rudiment of the heart (ht) becomes visible during this stage as a cavity between the mesoblast of the splanchnopleure and the hypoblast.

The fore and hind guts are now longer than they were. An invagination from the exterior to form the mouth has appeared (m) on the ventral side of the head close to the base of the thalamencephalon. The upper end of this eventually becomes constricted off as the pituitary body, and an indication of the future position of the anus is afforded by a slight diverticulum of the hind gut towards the exterior, some little distance from the posterior end of the embryo (an). The portion of the alimentary canal behind this point, though at this stage large, and even dilated into a vesicle at its posterior end (al.v), becomes eventually completely atrophied. It is known as the postanal gut. In the region of the throat the rudiment of a second visceral cleft has appeared behind the first; neither of them is as yet open to the exterior.

In a somewhat older embryo the first spontaneous movements take place, and consist in somewhat rapid excursions of the embryo from side to side, produced by a serpentine motion of the body.

A ventral flexure of the præoral part of the head, known as the cranial flexure, which commenced in earlier stages (fig. 28 D and E), has now become very evident, and the mid-brain[19] begins to project in the same manner as in the embryo fowl on the third day, and will soon form the anterior termination of the long axis of the embryo. The fore-brain has increased in size and distinctness, and the anterior part of it may now be looked on as the unpaired rudiment of the cerebral hemispheres.

Further changes have taken place in the organs of sense, especially in the eye, in which the involution for the lens has made considerable progress. The number of the muscle-plates has again increased, but there is still a region of unsegmented mesoblast in the tail. The thickened portions of mesoblast, which caused the tail swellings, are still to be seen, and would seem to act as the reserve from which is drawn the matter for the rapid growth of the tail, which occurs soon after this. The mass of the mesoblast at the base of the brain has again increased. No fresh features of interest are to be seen in the notochord. The heart is very much more conspicuous than before, and its commencing flexure is very apparent. It now beats actively. The postanal gut is much longer than during the last stage; and the point where the anus will appear is very easily detected by a bulging out of the gut towards the external skin. The alimentary vesicle at the end of the postanal gut, first observable during the last stage, is now a more conspicuous organ. There are three visceral clefts, none of which are as yet open to the exterior.

Figure 28 F represents a considerably older embryo viewed as an opaque object, and fig. 29 A is a view of the head as a transparent object. The stalk connecting it with the yolk is now, comparatively speaking, quite narrow, and is of sufficient length to permit the embryo to execute considerable movements.

The tail has grown immensely, but is still dilated terminally. The terminal dilatation is mainly due to the alimentary vesicle (fig. 28* alv), but the postanal section of the alimentary tract in front of this is now a solid cord of cells. Both the alimentary vesicle and this cord very soon disappear. Their relations are shewn in section in fig. 28*.

The two pairs of limbs have appeared as differentiations of a continuous but not very conspicuous epiblastic thickening, which is probably the rudiment of a lateral fin. The anterior pair is situated just at the front end of the umbilical stalk; and the posterior pair, which is the later developed and less conspicuous of the two, is situated some little distance behind the stalk.

The cranial flexure has greatly increased, and the angle between the long axis of the front part of the head and of the body is less than a right angle. The conspicuous mid-brain (29 A, mb) forms the anterior termination of the long axis of the body. The thin roof of the fourth ventricle (hb) may be noticed in the figure behind the mid-brain. The auditory sack (au.V) is nearly closed, and its opening is not shewn in the figure. In the eye (op) the lens is completely formed. The olfactory pit (ol) is seen a little in front of the eye.

Owing to the opacity of the embryo, the muscle-plates are only indistinctly indicated in fig. 28 F, and no other features of the mesoblast are to be seen.

The mouth is now a deep pit, the hind borders of which are almost completely formed by a thickening in front of the first branchial or visceral cleft, which may be called the first branchial arch or mandibular arch.

Four branchial clefts are now visible, all of which are open to the exterior, but in the embryo, viewed as a transparent object, two more, not open to the exterior, are visible behind the last of these.

Between each of these and behind the last one there is a thickening of the mesoblast which gives rise to a branchial arch. The arch between the first and second cleft is known as the hyoid arch.

Fig. 29 B is a representation of the head of a slightly older embryo in which papillæ may be seen in the front wall of the second, third, and fourth branchial clefts; these papillæ are the commencements of filiform processes which grow out from the gill-clefts and form external gills. The peculiar ventral curvature of the anterior end of the notochord (ch) both in this and in the preceding figure deserves notice.

A peculiar feature in the anatomy makes its appearance at this period, viz. the replacement of the original hollow œsophagus by a solid cord of cells (fig. 23 A, œs) in which a lumen does not reappear till very much later. I have found that in some Teleostei (the Salmon) long after they are hatched a similar solidity in the œsophagus is present. It appears not impossible that this feature in the œsophagus may be connected with the fact that in the ancestors of the present types the œsophagus was perforated by gill slits; and that in the process of embryonic abbreviation the stage with the perforated œsophagus became replaced by a stage with a cord of indifferent cells (the œsophagus being in the embryo quite functionless) out of which the non-perforated œsophagus was directly formed. In the higher types the process of development appears to have become quite direct.

By this stage all the parts of the embryo have become established, and in the succeeding stages the features characteristic of the genus and species are gradually acquired.

Two embryos of Scyllium are represented in fig. 28 G and H, the head and anterior part of the trunk being represented in fig. G, and the whole embryo at a much later stage in fig. H.

In both of these, and especially in the second, an apparent diminution of the cranial flexure is very marked. This diminution is due to the increase in the size of the cerebral hemispheres, which grow upwards and forwards, and press the original fore-brain against the mid-brain behind.

In fig. G the rudiments of the nasal sacks are clearly visible as small open pits.

The first cleft is no longer similar to the rest, but by the closure of the lower part has commenced to be metamorphosed into the spiracle.

Accompanying the change in position of the first cleft, the mandibular arch has begun to bend round so as to enclose the front as well as the sides of the mouth. By this change in the mandibular arch the mouth becomes narrowed in an antero-posterior direction.

In fig. H are seen the long filiform external gills which now project out from all the visceral clefts, including the spiracle. They are attached to the front wall of the spiracle, to both walls of the next four clefts, and to the front wall of the last cleft. They have very possibly become specially developed to facilitate respiration within the egg; and they disappear before the close of larval life.

When the young of Scyllium and other Sharks are hatched they have all the external characters of the adult. In Raja and Torpedo the early stages, up to the acquirement of a shark-like form, are similar to those in the Selachoidei, but during the later embryonic stages the body gradually flattens out, and assumes the adult form, which is thus clearly shewn to be a secondary acquirement.

An embryonic gill cleft behind the last present in the adult is found (Wyman, No. 54) in the embryo of Raja batis.

The unpaired fins are developed in Elasmobranchs as a fold of skin on the dorsal side, which is continued round the end of the tail along the ventral side to the anus. Local developments of this give rise to the dorsal and anal fins. The caudal fin is at first symmetrical, but a special lower lobe grows out and gives to it a heterocercal character.

Enclosure of the yolk-sack and its relation to the embryo.

The blastoderm at the stage represented in fig. 28 A and B forms a small and nearly circular patch on the surface of the yolk, composed of epiblast and lower layer cells. While the body of the embryo is gradually being moulded this patch grows till it envelopes the yolk; the growth is not uniform, but is less rapid in the immediate neighbourhood of the embryonic part of the blastoderm than elsewhere. As a consequence of this, that part of the edge, to which the embryo is attached, forms a bay in the otherwise regular outline of the edge of the blastoderm, and by the time that about two-thirds of the yolk is enclosed this bay is very conspicuous. It is shewn in fig. 30 A, where bl points to the blastoderm, and yk to the part of the yolk not yet covered by the blastoderm. The embryo at this time is only connected with the yolk-sack by a narrow umbilical cord; but, as shewn in the figure, is still attached to the edge of the blastoderm.

Shortly subsequent to this the bay in the blastoderm, at the head of which the embryo is attached, becomes obliterated by its two sides coming together and coalescing. The embryo then ceases to be attached at the edge of the blastoderm. But a linear streak formed by the coalesced edges of the blastoderm is left connecting the embryo with the edge of the blastoderm. This streak is probably analogous to (though not genetically related with) the primitive streak in the Amniota.

This stage is represented in fig. 30 B. In this figure there is only a small patch of yolk (yk) not yet enclosed, which is situated at some little distance behind the embryo. Throughout all this period the edge of the blastoderm has remained thickened: a feature which persists till the complete investment of the yolk, which takes place shortly after the stage last described. In this thickened edge a circular vein arises which brings back the blood from the yolk-sack to the embryo. The opening in the blastoderm, exposing the portion of the yolk not yet covered, may be conveniently called the yolk blastopore. It is interesting to notice that, owing to the large size of the yolk in Elasmobranchs, the posterior part of the primitive blastopore becomes encircled by the medullary folds and tail swellings, and is so closed long before the anterior and more ventral part, which is represented by the uncovered portion of the yolk. It is also worth remarking that, owing to the embryo becoming removed from the edge of the blastoderm, the final closure of the yolk blastopore takes place at some little distance from the embryo.

The blastoderm enclosing the yolk is formed of an external layer of epiblast, a layer of mesoblast below in which the blood-vessels are developed, and within this a layer of hypoblast, which is especially well marked and ciliated (Leydig, No. 46) in the umbilical stalk, where it lines the canal leading from the yolk-sack to the intestine. In the region of the yolk-sack proper the blastoderm is so thin that it is not easy to be quite sure that a layer of hypoblast is throughout distinct. Both the hypoblast and mesoblast of the yolk-sack are formed by a differentiation of the primitive lower layer cells.

Nutriment from the yolk-sack is brought to the embryo partly through the umbilical canal and so into the intestine, and partly by means of blood-vessels in the mesoblast of the sack. The blood-vessels arise before the blastoderm has completely covered the yolk.

Fig. 30 A represents the earliest stage of the circulation of the yolk-sack. At this stage there is visible a single arterial trunk (a) passing forwards from the embryo and dividing into two branches. No venous trunk could be detected with the simple microscope, but probably venous channels were present in the thickened edge of the blastoderm.

In fig. 30 B the circulation is greatly advanced. The blastoderm has now nearly completely enveloped the yolk, and there remains only a small circular space (yk) not enclosed by it. The arterial trunk is present as before, and divides in front of the embryo into two branches which turn backwards and form a nearly complete ring round the embryo. In general appearance this ring resembles the sinus terminalis of the area vasculosa of the Bird, but in reality bears quite a different relation to the circulation. It gives off branches on its inner side only.

A venous system of returning vessels is now fully developed, and its relations are very remarkable. There is a main venous ring in the thickened edge of the blastoderm, which is connected with the embryo by a single stem running along the seam where the edges of the blastoderm have coalesced. Since the venous trunks are only developed behind the embryo, it is only the posterior part of the arterial ring that gives off branches.

The succeeding stage (fig. 30 C) is also one of considerable interest. The arterial ring has greatly extended, and now embraces nearly half the yolk, and sends off trunks on its inner side along its whole circumference. More important changes have taken place in the venous system. The blastoderm has now completely enveloped the yolk, and the venous ring is therefore reduced to a point. The small veins which originally started from it may be observed diverging in a brush-like fashion from the termination of the unpaired trunk, which originally connected the venous ring with the heart.

At a still later stage the arterial ring embraces the whole yolk, and, as a result of this, vanishes in its turn, as did the venous ring before it. There is then present a single arterial and a single venous trunk. The arterial trunk is a branch of the dorsal aorta, and the venous trunk originally falls into the heart together with the subintestinal or splanchnic vein. On the formation of the liver the proximal end of the subintestinal vein becomes the portal vein, and it is joined just as it enters the liver by the venous trunk from the yolk-sack. The venous trunk leaves the body on the right side, and the arterial on the left.

The yolk-sack persists during the whole of embryonic life, and in the majority of Elasmobranch embryos there arises within the body walls an outgrowth from the umbilical canal into which a large amount of the yolk passes. This outgrowth forms an internal yolk-sack. In Mustelus vulgaris the internal yolk-sack is very small, and in Mustelus lævis it is absent. The latter species, which is one of those in which development takes place within the uterus, presents a remarkable peculiarity in that the vascular surface of the yolk-sack becomes raised into a number of folds, which fit into corresponding depressions in the vascular walls of the uterus. The yolk-sack becomes in this way firmly attached to the walls of the uterus, and the two together constitute a kind of placenta. A similar placenta is found in Carcharias.

After the embryo is hatched or born, as the case may be, the yolk-sack becomes rapidly absorbed.

Bibliography.

(40) F. M. Balfour. “A preliminary account of the development of the Elasmobranch Fishes.” Quart. J. of Micr. Science, Vol. XIV. 1876.
(41) F. M. Balfour. “A Monograph on the development of Elasmobranch Fishes.” London, 1878. Reprinted from the Journal of Anat. and Physiol. for 1876, 1877, and 1878.
(42) Z. Gerbe. Recherches sur la segmentation de la cicatrule et la formation des produits adventifs de l'œuf des Plagiostomes et particulièrement des Raies. Vide also Journal de l'Anatomie et de la Physiologie, 1872.
(43) W. His. “Ueb. d. Bildung v. Haifischenembryonen.” Zeit. für Anat. u. Entwick., Vol. II. 1877.
(44) A. Kowalevsky. “Development of Acanthias vulgaris and Mustelus lævis.” (Russian.) Transactions of the Kiew Society of Naturalists, Vol. I. 1870.
(45) R. Leuckart. “Ueber die allmählige Bildung d. Körpergestalt bei d. Rochen.” Zeit. f. wiss. Zool., Bd. II., p. 258.
(46) Fr. Leydig. Rochen u. Haie. Leipzig, 1852.
(47) A. W. Malm. “Bidrag till kännedom om utvecklingen af Rajæ.” Kongl. vetenskaps akademiens förhandlingar. Stockholm, 1876.
(48) Joh. Müller. Glatter Haie des Aristoteles und über die Verschiedenheiten unter den Haifischen und Rochen in der Entwicklung des Eies. Berlin, 1840.
(49) S. L. Schenk. “Die Eier von Raja quadrimaculata innerhalb der Eileiter.” Sitz. der k. Akad. Wien, Vol. LXXIII. 1873.
(50) Alex. Schultz. “Zur Entwicklungsgeschichte des Selachiereies.” Archiv für micro. Anat., Vol. XI.? 1875.
(51) Alex. Schultz. “Beitrag zur Entwicklungsgeschichte d. Knorpelfische.” Archiv für micro. Anat., Vol. XIII. 1877.
(52) C. Semper. “Die Stammesverwandschaft d. Wirbelthiere u. Wirbellosen.” Arbeit. a. d. zool.-zoot. Instit. Würzburg, Vol. II. 1875.
(53) C. Semper. “Das Urogenitalsystem d. Plagiostomen, etc.” Arbeit. a. d. zool.-zoot. Instit. Würzburg, Vol. II. 1875.
(54) Wyman. “Observations on the Development of Raja batis.” Memoirs of the American Academy of Arts and Sciences, Vol. IX. 1864.

The majority of the Teleostei deposit their eggs before impregnation, but some forms are viviparous, e.g. Blennius viviparus. Not a few carry their eggs about; but this operation is with a few exceptions performed by the male. In Syngnathus the eggs are carried in a brood-pouch of the male situated behind the anus. Amongst the Siluroids the male sometimes carries the eggs in the throat above the gill clefts. Ostegeniosus militaris, Arius falcarius, and Arius fissus have this peculiar habit.

The ovum when laid is usually invested in the zona radiata only, though a vitelline membrane is sometimes present in addition, e.g. in the Herring. It is in most cases formed of a central yolk mass, which may either be composed of a single large vitelline sphere, or of distinct yolk spherules. The yolk mass is usually invested by a granular protoplasmic layer, which is especially thickened at one pole to form the germinal disc.

In the Herring’s ovum the germinal disc is formed, as in many Crustacea, at impregnation; the protoplasm which was previously diffused through the egg becoming aggregated at the germinal pole and round the periphery.

Impregnation is external, and on its occurrence a contraction of the vitellus takes place, so that a space is formed between the vitellus and the zona radiata, which becomes filled with fluid.

The peculiarities in the development of the Teleostean ovum can best be understood by regarding it as an Elasmobranch ovum very much reduced in size. It seems in fact very probable that the Teleostei are in reality derived from a type of Fish with a much larger ovum. The occurrence of a meroblastic segmentation, in spite of the ovum being usually smaller than that of Amphibia and Acipenser, etc., in which the segmentation is complete, as well as the solid origin of many of the organs, receives its most plausible explanation on this hypothesis.

The proportion of the germinal disc to the whole ovum varies considerably. In very small eggs, such as those of the Herring, the disc may form as much as a fifth of the whole.

The segmentation, which is preceded by active movements of the germinal disc, is meroblastic. There is nothing very special to note with reference to its general features, but while in large ova like those of the Salmon the first furrows only penetrate for a certain depth through the germinal disc, in small ova like those of the Herring they extend through the whole thickness of the disc. During the segmentation a great increase in the bulk of the blastoderm takes place.

In hardened specimens a small cavity amongst the segmentation spheres may be present at any early stage; but it is probably an artificial product, and in any case has nothing to do with the true segmentation cavity, which does not appear till near the close of segmentation. The peripheral layer of granular matter, continuous with the germinal disc, does not undergo division, but it becomes during the segmentation specially thickened and then spreads itself under the edge of the blastoderm; and, while remaining thicker in this region, gradually grows inwards so as to form a continuous sub-blastodermic layer. In this layer nuclei appear, which are equivalent to those in the Elasmobranch ovum. A considerable number of these nuclei often become visible simultaneously (van Beneden, No. 60) and they are usually believed to arise spontaneously, though this is still doubtful[20]. Around these nuclei portions of protoplasm are segmented off, and cells are thus formed, which enter the blastoderm, and have nearly the same destination as the homologous cells of the Elasmobranch ovum.

During the later stages of segmentation one end of the blastoderm becomes thickened and forms the embryonic swelling; and a cavity appears between the blastoderm and the yolk which is excentrically situated near the non-embryonic part of the blastoderm. This cavity is the true segmentation cavity. Both the cavity and the embryonic swelling are seen in section in fig. 31 A and B.

In Leuciscus rutilus Bambeke describes a cavity as appearing in the middle of the blastoderm during the later stages of segmentation. From his figures it might be supposed that this cavity was equivalent to the segmentation cavity of Elasmobranchs in its earliest condition, but Bambeke states that it disappears and that it has no connection with the true segmentation cavity. Bambeke and other investigators have failed to recognize the homology of the segmentation cavity in Teleostei with that in Elasmobranchii, Amphibia, etc.

With the appearance of the segmentation cavity the portion of the blastoderm which forms its roof becomes thinned out, so that the whole blastoderm consists of (1) a thickened edge especially prominent at one point where it forms the embryonic swelling, and (2) a thinner central portion. The changes which now take place result in the differentiation of the embryonic layers, and in the rapid extension of the blastoderm round the yolk, accompanied by a diminution in its thickness.

The first differentiation of the layers consists in a single row of cells on the surface of the blastoderm becoming distinctly marked off as a special layer (fig. 31 A); which however does not constitute the whole epiblast but only a small part of it, which will be spoken of as the epidermic layer. The complete differentiation of the epiblast is effected by the cells of the thickened edge of the blastoderm becoming divided into two strata (fig. 31 B). The upper stratum constitutes the epiblast. It is divided into two layers, viz., the external epidermic layer already mentioned, and an internal layer known as the nervous layer, formed of several rows of vertically arranged cells. According to the unanimous testimony of investigators the roof of the segmentation cavity is formed of epiblast cells only. The lower stratum in the thickened rim of the blastoderm is several rows of cells deep, and corresponds with the lower layer cells or primitive hypoblast in Elasmobranchii. It is continuous at the edge of the blastoderm with the nervous layer of the epiblast.

In smaller Teleostean eggs there is formed, before the blastoderm becomes differentiated into epiblast and lower layer cells, a complete stratum of cells around the nuclei in the granular layer underneath the blastoderm. This layer is the hypoblast; and in these forms the lower layer cells of the blastoderm are stated to become converted into mesoblast only. In the larger Teleostean eggs, such as those of the Salmonidæ, the hypoblast, as in Elasmobranchs, appears to be only partially formed from the nuclei of the granular layer. In these forms however, as in the smaller Teleostean ova and in Elasmobranchii, the cells derived from the granular stratum give rise to a more or less complete cellular floor for the segmentation cavity. The segmentation cavity thus becomes enclosed between an hypoblastic floor and an epiblastic roof several cells deep. It becomes obliterated shortly after the appearance of the medullary plate.

At about the time when the three layers become established the embryonic swelling takes a somewhat shield-like form (fig. 33 A). Posteriorly it terminates in a caudal prominence (ts) homologous with the pair of caudal swellings in Elasmobranchs. The homologue of the medullary groove very soon appears as a shallow groove along the axial line of the shield. After these changes there takes place in the embryonic layers a series of differentiations leading to the establishment of the definite organs. These changes are much more difficult to follow in the Teleostei than in the Elasmobranchii, owing partly to the similarity of the cells of the various layers, and partly to the primitive solidity of all the organs.

The first changes in the epiblast give rise to the central nervous system. The epiblast, consisting of the nervous and epidermic strata already indicated, becomes thickened along the axis of the embryo and forms a keel projecting towards the yolk below: so great is the size of this keel in the front part of the embryo that it influences the form of the whole body and causes the outline of the surface adjoining the yolk to form a strong ridge moulded on the keel of the epiblast (fig. 32 A and B). Along the dorsal line of the epiblast keel is placed the shallow medullary groove; and according to Calberla (No. 61) the keel is formed by the folding together of the two sides of the primitively uniform epiblastic layer. The keel becomes gradually constricted off from the external epiblast and then forms a solid cord below it. Subsequently there appears in this cord a median slit-like canal, which forms the permanent central canal of the cerebrospinal cord. The peculiarity in the formation of the central nervous system of Teleostei consists in the fact that it is not formed by the folding over of the sides of the medullary groove into a canal, but by the separation, below the medullary groove, of a solid cord of epiblast in which the central canal is subsequently formed. Various views have been put forward to explain the apparently startling difference between Teleostei, with which Lepidosteus and Petromyzon agree, and other vertebrate forms. The explanations of Götte and Calberla appear to me to contain between them the truth in this matter. The groove above in part represents the medullary groove; but the closure of the groove is represented by the folding together of the lateral parts of the epiblast plate to form the medullary keel.

According to Götte this is the whole explanation, but Calberla states for Syngnathus and Salmo that the epidermic layer of the epiblast is carried down into the keel as a double layer just as if it had been really folded in. This ingrowth of the epidermic layer is shewn in fig. 32 A where it is just commencing to pass into the keel; and at a later stage in fig. 32 B where the keel has reached its greatest depth.

Götte maintains that Calberla’s statements are not to be trusted, and I have myself been unable to confirm them for Teleostei or Lepidosteus; but if they could be accepted the difference in the formation of the medullary canal in Teleostei and in other Vertebrata would become altogether unimportant and consist simply in the fact that the ordinary open medullary groove is in Teleostei obliterated in its inner part by the two sides of the groove coming together. Both layers of epiblast would thus have a share in the formation of the central nervous system; the epidermic layer giving rise to the lining epithelial cells of the central canal, and the nervous layer to the true nervous tissue.

The separation of the solid nervous system from the epiblast takes place relatively very late; and, before it has been completed, the first traces of the auditory pits, of the optic vesicles, and of the olfactory pits are visible. The auditory pit arises as a solid thickening of the nervous layer of the epiblast at its point of junction with the medullary keel; and the optic vesicles spring as solid outgrowths from part of the keel itself. The olfactory pits are barely indicated as thickenings of the nervous layer of the epiblast.

At this early stage all the organs of special sense are attached to a layer continuous with or forming part of the central nervous system; and this fact has led Götte (No. 63) to speak of a special-sense plate, belonging to the central nervous system and not to the skin, from which all the organs of special sense are developed; and to conclude that a serial homology exists between these organs in their development. A comparison between Teleostei and other forms shews that this view cannot be upheld; even in Teleostei the auditory and olfactory rudiments arise rather from the epiblast at the sides of the brain than from the brain itself, while the optic vesicles spring from the first directly from the medullary keel, and are therefore connected with the central nervous system rather than with the external epiblast. In a slightly later stage the different connections of the two sets of sense organs is conclusively shewn by the fact that, on the separation of the central nervous system from the epiblast, the optic vesicles remain attached to the former, while the auditory and olfactory vesicles are continuous with the latter.

After its separation from the central nervous system the remainder of the epiblast gives rise to the skin, etc., and most probably the epidermic stratum develops into the outer layer of the epidermis and the nervous stratum into the mucous layer. The parts of the organs of special sense, which arise from the epiblast, are developed from the nervous layer. In the Trout (Oellacher, No. 72) both layers are continued over the yolk-sack; but in Clupeus and Gasterosteus only the epidermic has this extension. According to Götte the distinction between the two layers becomes lost in the later embryonic stages.

Although it is thoroughly established that the mesoblast originates from the lower of the two layers of the thickened embryonic rim, it is nevertheless not quite certain whether it is a continuous layer between the epiblast and hypoblast, or whether it forms two lateral masses as in Elasmobranchs. The majority of observers take the former view, while Calberla is inclined to adopt the latter. In the median line of the embryo underneath the medullary groove there are undoubtedly from the first certain cells which eventually give rise to the notochord; and it is these cells the nature of which is in doubt. They are certainly at first very indistinctly separated from the mesoblast on the two sides, and Calberla also finds that there is no sharp line separating them from the secondary hypoblast (fig. 32 A). Whatever may be the origin of the notochord the mesoblast very soon forms two lateral plates, one on each side of the body, and between them is placed the notochord (fig. 32 B). The general fate of the two mesoblast plates is the same as in Elasmobranchs. They are at first quite solid and exhibit relatively late a division into splanchnic and somatic layers, between which is placed the primitive body cavity. The dorsal part of the plates becomes transversely segmented in the region of the trunk; and thus gives rise to the mesoblastic somites, from which the muscle plates and the perichordal parts of the vertebral column are developed. The ventral or outer part remains unsegmented. The cavity of the ventral section becomes the permanent body cavity. It is continued forward into the head (Oellacher), and part of it becomes separated off from the remainder as the pericardial cavity.

The hypoblast forms a continuous layer below the mesoblast, and, in harmony with the generally confined character of the development of the organs in Teleostei, there is no space left between it and the yolk to represent the primitive alimentary cavity. The details of the formation of the true alimentary tube have not been made out; it is not however formed by a folding in of the lateral parts of the hypoblast, but arises as a solid or nearly solid cord in the axial line, between the notochord and the yolk, in which a lumen is gradually established.

In the just hatched larva of an undetermined freshwater fish with a very small yolk-sack I found that the yolk extended along the ventral side of the embryo from almost the mouth to the end of the gut. The gut had, except in the hinder part, the form of a solid cord resting in a concavity of the yolk. In the hinder part of the gut a lumen was present, and below this part the amount of yolk was small and the yolk nuclei numerous. Near the limit of its posterior extension the yolk broke up into a mass of cells, and the gut ended behind by falling into this mass. These incomplete observations appear to shew that the solid gut owes its origin in a large measure to nuclei derived from the yolk.

When the yolk has become completely enveloped a postanal section of gut undoubtedly becomes formed; and although, owing to the solid condition of the central nervous system, a communication between the neural and alimentary canals cannot at first take place, yet the terminal vesicle of the postanal gut of Elasmobranchii is represented by a large vesicle, originally discovered by Kupffer (No. 68), which can easily be seen in the embryos of most Teleostei, but the relations of which have not been satisfactorily worked out (vide fig. 34, hyv). As the tail end of the embryo becomes separated off from the yolk the postanal vesicle atrophies.

General development of the Embryo. Attention has already been called to the fact that the embryo first appears as a thickening of the edge of the blastoderm which soon assumes a somewhat shield-like form (fig. 33 A). The hinder end of the embryo, which is placed at the edge of the blastoderm, is somewhat prominent, and forms the caudal swelling (ts). The axis of the embryo is marked by a shallow groove.

The body now rapidly elongates, and at the same time becomes considerably narrower, while the groove along the axis becomes shallower and disappears. The anterior, and at first proportionately a very large part, soon becomes distinguished as the cephalic region (fig. 33 B). The medullary cord in this region becomes very early divided into three indistinctly separated lobes, representing the fore, the mid, and the hind brains: of these the anterior is the smallest. With it are connected the optic vesicles (oc)—solid at first—which are pushed back into the region of the mid-brain.

The trunk grows in the usual way by the addition of fresh somites behind.

After the yolk has become completely enveloped by the blastoderm the tail becomes folded off, and the same process takes place at the front end of the embryo. The free tail end of the embryo continues to grow, remaining however closely applied to the yolk-sack, round which it curls itself to an extent varying with the species (vide fig. 34).

The general growth of the embryo during the later stages presents a few special features of interest. The head is remarkable for the small apparent amount of the cranial flexure. This is probably due to the late development of the cerebral hemispheres. The flexure of the floor of the brain is however quite as considerable in the Teleostei as in other types. The gill clefts develop from before backwards. The first cleft is the hyomandibular, and behind this there are the hyobranchial and four branchial clefts. Simultaneously with the clefts there are developed the branchial arches. The postoral arches formed are the mandibular, hyoid and five branchial arches. In the case of the Salmon all of these appear before hatching.

The first cleft closes up very early (about the time of hatching in the Salmon); and about the same time there springs a membranous fold from the hyoid arch, which gradually grows backwards over the arches following, and gives rise to the operculum. There appear in the Salmon shortly before hatching double rows of papillæ on the four anterior arches behind the hyoid. They are the rudiments of the branchiæ. They reach a considerable length before they are covered in by the opercular membrane. In Cobitis (Götte, No. 64) they appear in young larvæ as filiform processes equivalent to the external gills of Elasmobranchs. The extremities of these processes atrophy; while the basal portions became the permanent gill lamellæ. The general relation of the clefts, after the closure of the hyomandibular, is shewn in fig. 35.

The air-bladder is formed as a dorsal outgrowth of the alimentary tract very slightly in front of the liver. It grows in between the two limbs of the mesentery, in which it extends itself backwards. It appears in the Salmon, Carp, and other types to originate rather on the right side of the median dorsal line, but whether this fact has any special significance is rather doubtful. In the Salmon and Trout it is formed considerably later than the liver, but the two are stated by Von Baer to arise in the Carp nearly at the same time. The absence of a pneumatic duct in the Physoclisti is due to a post-larval atrophy. The region of the stomach is reduced almost to nothing in the larva.

The œsophagus becomes solid, like that of Elasmobranchs, and remains so for a considerable period after hatching.

The liver, in the earliest stage in which I have met with it in the Trout (27 days after impregnation), is a solid ventral diverticulum of the intestine, which in the region of the liver is itself without a lumen.

The excretory system commences with the formation of a segmental duct, formed by a constriction of the parietal wall of the peritoneal cavity. The anterior end remains open to the body cavity, and forms a pronephros (head kidney). On the inner side of and opposite this opening a glomerulus is developed, and the part of the body cavity containing both the glomerulus and the opening of the pronephros becomes shut off from the remainder of the body cavity, and forms a completely closed Malpighian capsule.

The mesonephros (Wolffian body) is late in developing.

The unpaired fins arise as simple folds of the skin along the dorsal and ventral edges, continuous with each other round the end of the tail. The ventral fold ends anteriorly at the anus.

The dorsal and anal fins are developed from this fold by local hypertrophy. The caudal fin[21], however, undergoes a more complicated metamorphosis. It is at first symmetrical or nearly so on the dorsal and ventral sides of the hinder end of the notochord. This symmetry is not long retained, but very soon the ventral part of the fin with its fin rays becomes much more developed than the dorsal part, and at the same time the posterior part of the notochord bends up towards the dorsal side.

In some few cases, e.g. Gadus, Salmo, owing to the simultaneous appearance of a number of fin rays on the dorsal and ventral side of the notochord the external symmetry of the tail is not interfered with in the above processes. In most instances this is far from being the case.

In the Flounder, which may serve as a type, the primitive symmetry is very soon destroyed by the appearance of fin rays on the ventral side. The region where they are present soon forms a lobe; and an externally heterocercal tail is produced (fig. 36 A). The ventral lobe with its rays continues to grow more prominent and causes the tail fin to become bilobed (fig. 36 B); there being a dorsal embryonic lobe without fin rays (c), which contains the notochord, and a ventral lobe with fin rays, which will form the permanent caudal fin. In this condition the tail fin resembles the usual Elasmobranch form or still more that of some Ganoids, e.g. the Sturgeon. The ventral lobe continues to develop; and soon projects beyond the dorsal, which gradually atrophies together with the notochord contained in it, and finally disappears, leaving hardly a trace on the dorsal side of the tail (fig. 36 C, c). In the meantime the fin rays of the ventral lobe gradually become parallel to the axis of the body; and this lobe, together with a few accessory dorsal and ventral fin rays supported by neural and hæmal processes, forms the permanent tail fin, which though internally unsymmetrical, assumes an externally symmetrical form. The upturned end of the notochord which was originally continued into the primitive dorsal lobe becomes ensheathed in a bone without a division into separate vertebræ. This bone forms the urostyle (u). The hæmal processes belonging to it are represented by two cartilaginous masses, which subsequently ossify, forming the hypural bones, and supporting the primary fin rays of the tail (fig. 36 C). The ultimate changes of the notochord and urostyle vary very considerably in the different types of Teleostei. Teleostei may fairly be described as passing through an Elasmobranch stage or a stage like that of most pre-jurassic Ganoids or the Sturgeon as far as concerns their caudal fin.

The anterior paired fins arise before the posterior; and there do not appear to be any such indications as in Elasmobranchii of the paired fins arising as parts of a continuous lateral fin.

Most osseous fishes pass through more or less considerable post-embryonic changes, the most remarkable of which are those undergone by the Pleuronectidæ[22]. These fishes, which in the adult state have the eyes unsymmetrically placed on one side of the head, leave the egg like normal Teleostei. In the majority of cases as they become older the eye on the side, which in the adult is without an eye, travels a little forward and then gradually rotates over the dorsal side of the head, till finally it comes to lie on the same side as the other eye. During this process the rotating eye always remains at the surface and continues functional; and on the two eyes coming to the same side of the head the side of the body without an organ of vision loses its pigment cells, and becomes colourless.

The dorsal fin, after the rotation of the eye, grows forward beyond the level of the eyes. In the genus Plagusia (Steenstrup, Agassiz, No. 56) the dorsal fin grows forward before the rotation of the eye (the right eye in this form), and causes some modifications in the process. The eye in travelling round gradually sinks into the tissues of the head, at the base of the fin above the frontal bone; and in this process the original large opening of the orbit becomes much reduced. Soon a fresh opening on the opposite and left side of the dorsal fin is formed; so that the orbit has two external openings, one on the left and one on the right side. The original one on the right soon atrophies, and the eye passes through the tissues at the base of the dorsal fin completely to the left side.

The rotating eye may be either the right or the left according to the species.

The most remarkable feature in which the young of a large number of Teleostei differ from the adults is the possession of provisional spines, very often formed as osseous spinous projections the spaces between which become filled up in the adult. These processes are probably, as suggested by Günther, secondary developments acquired, like the Zoœa spines of larval Crustaceans, for purposes of defence.

The yolk-sack varies greatly in size in the different types of Teleostei.

According as it is enclosed within the body-wall, or forms a distinct ventral appendage, it is spoken of by Von Baer as an internal or external yolk-sack. By Von Baer the yolk-sack is stated to remain in communication with the intestine immediately behind the liver, while Lereboullet states that there is a vitelline pedicle opening between the stomach and the liver which persists till the absorption of the yolk-sack. My own observations do not fully confirm either of these statements for the Salmon and Trout. So far as I have been able to make out, all communication between the yolk-sack and the alimentary tract is completely obliterated very early. In the Trout the communication between the two is shut off before hatching, and in the just-hatched Salmon I can find no trace of any vitelline pedicle. The absorption of the yolk would seem therefore to be effected entirely by blood-vessels.

The yolk-sack persists long after hatching, and is gradually absorbed. There is during the stages either just before hatching or shortly subsequent to hatching (Cyprinus) a rich vascular development in the mesoblast of the yolk-sack. The blood is at first contained in lacunar spaces, but subsequently it becomes confined to definite channels. As to its exact relations to the vascular system of the embryo more observations seem to be required.

The following account is given by Rathke (No. 72*) and Lereboullet (No. 71). At first a subintestinal vein (vide chapter on Circulation) falls into the lacunæ of the yolk-sack, and the blood from these is brought back direct to the heart. At a later period, when the liver is developed, the subintestinal vessel breaks up into capillaries in the liver, thence passes into the yolk-sack, and from this to the heart. An artery arising from the aorta penetrates the liver, and there breaks up into capillaries continuous with those of the yolk-sack. This vessel is perhaps the equivalent of the artery which supplies the yolk-sack in Elasmobranchii, but it seems possible that there is some error in the above description.

Bibliography.

(55) Al. Agassiz. “On the young Stages of some Osseous Fishes. I. Development of the Tail.” Proceedings of the American Academy of Arts and Sciences, Vol. XIII. Presented Oct. 11, 1877.
(56) Al. Agassiz. “II. Development of the Flounders.” Proceedings of the American Acad. of Arts and Sciences, Vol. XIV. Presented June, 1878.
(57) K. E. v. Baer. Untersuchungen über die Entwicklungsgeschichte der Fische. Leipzig, 1835.
(58) Ch. van Bambeke. “Premiers effets de la fécondation sur les œufs de Poissons: sur l'origine et la signification du feuillet muqueux ou glandulaire chez les Poissons Osseux.” Comptes Rendus des Séances de l'Académie des Sciences, Tome LXXIV. 1872.
(59) Ch. van Bambeke. “Recherches sur l'Embryologie des Poissons Osseux.” Mém. couronnés et Mém. de savants étrangers de l'Académie roy. Belgique, Vol. XL. 1875.
(60) E. v. Beneden. “A contribution to the history of the Embryonic development of the Teleosteans.” Quart. J. of Micr. Sci., Vol. XVIII. 1878.
(61) E. Calberla. “Zur Entwicklung des Medullarrohres u. d. Chorda dorsalis d. Teleostier u. d. Petromyzonten.” Morphologisches Jahrbuch, Vol. III. 1877.
(62) A. Götte. “Beiträge zur Entwicklungsgeschichte der Wirbelthiere.” Archiv f. mikr. Anat., Vol. IX. 1873.
(63) A. Götte. “Ueber d. Entwicklung d. Central-Nervensystems der Teleostier.” Archiv f. mikr. Anat., Vol. XV. 1878.
(64) A. Götte. “Entwick. d. Teleostierkeime.” Zoologischer Anzeiger, No. 3. 1878.
(65) W. His.. “Untersuchungen über die Entwicklung von Knochenfischen, etc.” Zeit. f. Anat. u. Entwicklungsgeschichte, Vol. I. 1876.
(66) W. His. “Untersuchungen über die Bildung des Knochenfischembryo (Salmen).” Archiv f. Anat. u. Physiol., 1878.
(67) E. Klein. “Observations on the early Development of the Common Trout.” Quart. J. of Micr. Science, Vol. XVI. 1876.
(68) C. Kupffer. “Beobachtungen über die Entwicklung der Knochenfische.” Archiv f. mikr. Anat., Bd. IV. 1868.
(69) C. Kupffer. Ueber Laichen u. Entwicklung des Ostsee-Herings. Berlin, 1878.
(70) M. Lereboullet. “Recherches sur le développement du brochet de la perche et de l'écrevisse.” Annales des Sciences Nat., Vol. I., Series IV. 1854.
(71) M. Lereboullet. “Recherches d'Embryologie comparée sur le développement de la Truite.” An. Sci. Nat., quatrième série, Vol. XVI. 1861.
(72) T. Oellacher. “Beiträge zur Entwicklungsgeschichte der Knochenfische nach Beobachtungen am Bachforellenei.” Zeit. f. wiss. Zool., Vol. XXII., 1872, and Vol. XXIII., 1873.
(72*) H. Rathke. Abh. z. Bildung u. Entwick. d. Menschen u. Thiere. Leipzig, 1832-3. Part II. Blennius.
(73) Reineck. “Ueber die Schichtung des Forellenkeims.” Archiv f. mikr. Anat., Bd. V. 1869.
(74) S. Stricker. “Untersuchungen über die Entwicklung der Bachforelle.” Sitzungsberichte der Wiener k. Akad. d. Wiss., 1865. Vol. LI. Abth. 2.
(75) Carl Vogt. “Embryologie des Salmones.” Histoire Naturelle des Poissons de l'Europe Centrale. L. Agassiz. 1842.
(76) C. Weil. “Beiträge zur Kenntniss der Knochenfische.” Sitzungsber. der Wiener kais. Akad. der Wiss., Bd. LXVI. 1872.

Petromyzon is the only type of this degenerated but primitive group of Fishes the development of which has been as yet studied[24].

The development does not however throw any light on the relationships of the group. The similarity of the mouth and other parts of Petromyzon to those of the Tadpole probably indicates that there existed a common ancestral form for the Cyclostomata and Amphibia. Embryology does not however add anything to the anatomical evidence on this subject. The fact of the segmentation being complete was at one time supposed to indicate an affinity between the two groups; but the discovery that the segmentation is also complete in the Ganoids deprives this feature in the development of any special weight. In the formation of the layers and in most other developmental characters there is nothing to imply a special relationship with the Amphibia, and in the mode of formation of the nervous system Petromyzon exhibits a peculiar modification, otherwise only known to occur in Teleostei and Lepidosteus.

Dohrn[25] was the first to bring into prominence the degenerate character of the Cyclostomata. I cannot however assent to his view that they are descended from a relatively highly-organized type of Fish. It appears to me almost certain that they belong to a group of fishes in which a true skeleton of branchial bars had not become developed, the branchial skeleton they possess being simply an extra-branchial system; while I see no reason to suppose that a true branchial skeleton has disappeared. If the primitive Cyclostomata had not true branchial bars, they could not have had jaws, because jaws are essentially developed from the mandibular branchial bar. These considerations, which are supported by numerous other features of their anatomy, such as the character of the axial skeleton, the straightness of the intestinal tube, the presence of a subintestinal vein etc., all tend to prove that these fishes are remnants of a primitive and prægnathostomatous group. The few surviving members of the group have probably owed their preservation to their parasitic or semiparasitic habits, while the group as a whole probably disappeared on the appearance of gnathostomatous Vertebrata.

The ripe ovum of Petromyzon Planeri is a slightly oval body of about 1 mm. in diameter. It is mainly formed of an opaque nearly white yolk, invested by a membrane composed of an inner perforated layer, and an outer structureless layer. There appears to be a pore perforating the inner layer at the formative pole, which may be called a micropyle (Kupffer and Benecke, No. 79). Enclosing the egg-membranes there is present a mucous envelope, which causes the egg, when laid, to adhere to stones or other objects.

Impregnation is effected by the male attaching itself by its suctorial mouth to the female. The attached couple then shake together; and, as they do so, they respectively emit from their abdominal pores ova and spermatozoa which pass into a hole previously made[26].

The segmentation is total and unequal, and closely resembles that in the Frog’s egg (Vol. II. p. 96). The upper pole is very slightly whiter than the lower. A segmentation cavity is formed very early, and is placed between the small cells of the upper pole and the large cells of the lower pole. It is proportionately larger than in the Frog; and the roof eventually thins out so as to be formed of a single row of small cells. At the sides of the segmentation cavity there are always several rows of small cells, which gradually merge into the larger cells of the lower pole of the egg. The segmentation is completed in about fifty hours.

The segmentation is followed by an asymmetrical invagination (fig. 37) which leads to a mode of formation of the hypoblast fundamentally similar to that in the Frog. The process has been in the main correctly described by M. Schultze (No. 81).

On the border between the large and small cells of the embryo, at a point slightly below the segmentation cavity, a small circular pit appears; the roof of which is formed by an infolding of the small cells, while the floor is formed of the large cells. This pit is the commencing mesenteron. It soon grows deeper (fig. 37, al) and extends as a well-defined tube (shewn in transverse section in fig. 38, al) in the direction of the segmentation cavity. In the course of the formation of the mesenteron the segmentation cavity gradually becomes smaller, and is finally (about the 200th hour) obliterated. The roof of the mesenteron is formed by the continued invagination of small cells, and its floor is composed of large yolk-cells. The wide external opening is arched over dorsally by a somewhat prominent lip—the homologue of the embryonic rim. The opening persists till nearly the time of hatching; but eventually becomes closed, and is not converted into the permanent anus. On the formation of the mesenteron the hypoblast is composed of two groups of cells, (1) the yolk-cells, and (2) the cells forming the roof of the mesenteron.

While the above changes are taking place, the small cells, or as they may now be called the epiblast cells, gradually spread over the large yolk-cells, as in normal types of epibolic invagination. The growth over the yolk-cells is not symmetrical, but is most rapid in the meridian opposite the opening of the alimentary cavity, so that the latter is left in a bay (cf. Elasmobranchii, p. 63). The epibolic invagination takes place as in Molluscs and many other forms, not simply by the division of pre-existing epiblast cells, but by the formation of fresh epiblast cells from the yolk-cells (fig. 37); and till after the complete enclosure of the yolk-cells there is never present a sharp line of demarcation between the two groups of cells. By the time that the segmentation cavity is obliterated the whole yolk is enclosed by the epiblast. The yolk-cells adjoining the opening of the mesenteron are the latest to be covered in, and on their enclosure this opening constitutes the whole of the blastopore. The epiblast is composed of a single row of columnar cells.

Mesoblast and notochord. During the above changes the mesoblast becomes established. It arises, as in Elasmobranchs, in the form of two plates derived from the primitive hypoblast. During the invagination to form the mesenteron some of the hypoblast cells on each side of the invaginated layer become smaller, and marked off as two imperfect plates (fig. 38, ms). It is difficult to say whether these plates are entirely derived from invaginated cells, or are in part directly formed from the pre-existing yolk-cells, but I am inclined to adopt the latter view; the ventral extension of the mesoblast plates undoubtedly takes place at the expense of the yolk-cells. The mesoblast plates soon become more definite, and form (fig. 39, ms) well-defined structures, triangular in section, on the two sides of the middle line.

At the time the mesoblast is first formed the hypoblast cells, which roof the mesenteron, are often imperfectly two layers thick (fig. 38). They soon however become constituted of a single layer only. When the mesoblast is fairly established, the lateral parts of the hypoblast grow inwards underneath the axial part, so that the latter (fig. 39, ch) first becomes isolated as an axial cord, and is next inclosed between the medullary cord (nc) (which has by this time been formed) and a continuous sheet of hypoblast below (fig. 40). Here its cells divide and it becomes the notochord. The notochord is thus bodily formed out of the axial portion of the primitive hypoblast. Its mode of origin may be compared with that in Amphioxus, in which an axial fold of the archenteric wall is constricted off as the notochord. The above features in the development of the notochord were first established by Calberla[27] (No. 78).

General history of the development. Up to about the time when the enclosure of the hypoblast by the epiblast is completed, no external traces are visible of any of the organs of the embryo; but about this time, i.e. about 180 hours after impregnation, the rudiment of the medullary plate becomes established, as a linear streak extending forwards from the blastopore over fully one half the circumference of the embryo. The medullary plate first contains a shallow median groove, but it is converted into the medullary cord, not in the usual vertebrate fashion, but, as first shewn by Calberla, in a manner much more closely resembling the formation of the medullary cord in Teleostei. Along the line of the median groove the epiblast becomes thickened and forms a kind of keel projecting inwards towards the hypoblast (fig. 39, nc). This keel is the rudiment of the medullary cord. It soon becomes more prominent, the median groove in it disappears, and it becomes separated from the epiblast as a solid cord (fig. 40, mc).

By this time the whole embryo has become more elongated, and on the dorsal surface is placed a ridge formed by the projection of the medullary cord. At the lip of the blastopore the medullary cord is continuous with the hypoblast, thus forming the rudiment of a neurenteric canal.

Calberla gives a similar account of the formation of the neural canal to that which he gives for the Teleostei (vide p. 72).

He states that the epiblast becomes divided into two layers, of which the outer is involuted into the neural cord, a median slit in the involution representing the neural groove. The eventual neural canal is stated to be lined by the involuted cells. Scott (No. 87) fully confirms Calberla on this point, and, although my own sections do not clearly shew an involution of the outer layer of epiblast cells, the testimony of these two observers must no doubt be accepted on this point.

Shortly after the complete establishment of the neural cord the elongation of the embryo proceeds with great rapidity. The processes in this growth are shewn in fig. 41, A, B, and C. The cephalic portion (A, c) first becomes distinct, forming an anterior protuberance free from yolk. About the time it is formed the mesoblastic plates begin to be divided into somites, but the embryo is so opaque that this process can only be studied in sections. Shortly afterwards an axial lumen appears in the centre of the neural cord, in the same manner as in Teleostei.

The general elongation of the embryo continues rapidly, and, as shewn in my figures, the anterior end is applied to the ventral surface of the yolk (B). With the growth of the embryo the yolk becomes entirely confined to the posterior part. This part is accordingly greatly dilated, and might easily be mistaken for the head. The position of the yolk gives to the embryo a very peculiar appearance. The apparent difference between it and the embryos of other Fishes in the position of the yolk is due in the main to the fact that the postanal portion of the tail is late in developing, and always small. As the embryo grows longer it becomes spirally coiled within the egg-shell. Before hatching the mesoblastic somites become distinctly marked (C).

The hatching takes place at between 13-21 days after impregnation; the period varying according to the temperature.

During the above changes in the external form of the embryo, the development of the various organs makes great progress. This is especially the case in the head. The brain becomes distinct from the spinal cord, and the auditory sacks and the optic vesicles of the eye become formed. The branchial region of the mesenteron becomes established, and causes a dilatation of the anterior part of the body, and the branchial pouches grow out from the throat. The anus becomes formed, and a neurenteric canal is also established (Scott). The nature of these and other changes will best be understood by a description of the structure of the just-hatched larva. The general appearance of the larva immediately after hatching is shewn in fig. 41, D. The body is somewhat curved; the posterior extremity being much dilated with yolk, while the anterior is very thin. All the cells still contain yolk particles, which render the embryo very opaque. The larva only exhibits slow movements, and is not capable of swimming about.

The structure of the head is shewn in figs. 42 and fig. 43. Fig. 42 is a section through a very young larva, while fig. 43 is taken from a larva three days after hatching, and shews the parts with considerably greater detail.

On the ventral side of the head is placed the oral opening (fig. 43, m) leading into a large stomodæum which is still without a communication with the mesenteron. Ventrally the stomodæum is prolonged for a considerable distance under the anterior part of the mesenteron. Immediately behind the stomodæum is placed the branchial region of the mesenteron. Laterally it is produced on each side into seven or perhaps eight branchial pouches (fig. 43, br.c), which extend outwards nearly to the skin but are not yet open. Between the successive pouches are placed mesoblastic segments, of the same nature and structure as the walls of the head cavities in the embryos of Elasmobranchs, and like them enclosing a central cavity. A similar structure is placed behind the last, and two similar structures in front of the first persistent pouch. This pouch is situated in the same vertical line as the auditory sack (au.v), and would appear therefore to be the hyobranchial cleft; and this identification is confirmed by the fact of two head cavities being present in front of it. At the front end of the branchial region of the mesenteron is placed a thickened ridge of tissue, which, on the opening of the passage between the stomodæum and the mesenteron, forms a partial septum between the two, and is known as the velum (fig. 43, tv).

According to Scott (No. 87) a hyomandibular pouch forming the eighth pouch is formed in front of the pouch already defined as the hyobranchial. It disappears early and does not acquire gill folds[28]. The tissue forming the line of insertion of the velum appears to me to represent the mandibular arch. The grounds for this view are the following:
(1) The structure in question has exactly the position usually occupied by the mandibular arch.
(2) There is present in late larvæ (about 20 days after hatching) an arterial vessel, continued from the ventral prolongation of the bulbus arteriosus along the insertion of the velum towards the dorsal aorta, which has the relations of a true branchial artery.

On the ventral aspect of the branchial region is placed a sack (figs. 42, h, and 43, th), which extends from the front end of the branchial region to the fourth cleft. At first it constitutes a groove opening into the throat above (fig. 44), but soon the opening becomes narrowed to a pore placed between the second and third of the permanent branchial pouches (fig. 43, th). In Ammocœtes[29] the simple tube becomes divided, and assumes a very complicated form, though still retaining its opening into the branchial region of the throat. In the adult it forms a glandular mass underneath the branchial region of the throat equivalent to the thyroid gland of higher Vertebrates.

On the ventral aspect of the head, and immediately in front of the mouth, is placed the olfactory pit (fig. 43, ol). It is from the first unpaired, and in just-hatched larvæ simply forms a shallow groove of thickened epiblast at the base of the front of the brain. By the stage represented in fig. 43 the ventral part of the original groove is prolonged into a pit, extending backwards beneath the brain nearly up to the infundibulum.

On the side of the head, nearly on a level with the front end of the notochord, is placed the eye (fig. 43, op). It is constituted (figs. 45 and 46) of a very shallow optic cup with a thick outer (retinal) layer, and a thin inner choroid layer. In contact with the retinal layer is placed the lens. The latter is formed as an invagination of the skin; to which it is still attached in the just-hatched larva (fig. 45). The eye only differs at this stage from that of other Vertebrata in its extraordinarily small size, and the rudimentary character of its constituent parts.

The auditory sack is a large vesicle (fig. 43, au.v.), placed at the side of the brain opposite the first persistent branchial pouch.

The brain is formed of the usual vertebrate parts[30], but is characterized by the very slight cranial flexure. The fore-brain consists (fig. 43) of a thalamencephalon (th) and an undivided cerebral rudiment (ch). To the roof of the thalamencephalon is attached a flattened sack (pn) which is probably the pineal gland. The floor is prolonged into an infundibulum (in) which contains a prolongation of the third ventricle. The lateral walls of the cerebral rudiment are much thickened.

Behind the thalamencephalon follows the mid-brain (mb), the sides of which form the optic lobes, and behind this again the hind-brain (md); the front border of the roof of which is thickened to form the cerebellum (cb). The medulla passes without any marked line of demarcation into the spinal cord.

The histological differentiation of the brain has already proceeded to some extent; and it has in the main the same character as the spinal cord. Before the larva has been hatched very long a lateral investment of white matter is present throughout. The notochord (ch) is continued forwards in the head to the hinder border of the infundibulum. It is slightly flexed anteriorly.

From the hinder border of the auditory region to the end of the branchial region the mesoblast is dorsally divided into myotomes, which nearly, though not quite, correspond in number with the branchial pouches.

The growth of the myotomes would seem, as might be anticipated from their independent innervation, not to be related to that of the branchial pouches, so that there is a want of correspondence between these parts, the extent of which varies at different periods of life. The relation between the two in an old larva is shewn in fig. 47.

The head of the larva of Petromyzon differs very strikingly in general appearance from that of the normal Vertebrata. This is at once shewn by a comparison of fig. 43 with fig. 29. The most important difference between the two is due to the absence of a pronounced cranial flexure in Petromyzon; an absence which is in its turn probably caused by the small development of the fore-brain.

The stomodæum of Petromyzon is surprisingly large, and its size and structure in this type militate against the view of some embryologists that the stomodæum originated from the coalescence of a pair of branchial pouches.

In the region of the trunk there is present an uninterrupted dorsal fin continuous with a ventral fin round the end of the tail.

There is a well-developed body cavity, which is especially dilated in front, in the part which afterwards becomes the pericardium. In this region is placed the nearly straight heart, divided into an auricle and ventricle (figs. 42 and 43), the latter continued forwards into a bulbus arteriosus.

The myotomes are now very numerous (about 57, including those of the head, in a three days’ larva). They are separated by septa, but do not fill up the whole space between the septa, and have a peculiar wavy outline. The notochord is provided with a distinct sheath, and below it is placed a subnotochordal rod.

The alimentary canal consists of a narrow anterior section free from yolk, and a posterior region, the walls of which are largely swollen with yolk. The anterior section corresponds to the region of the œsophagus and stomach, but exhibits no distinction of parts. Immediately behind this point the alimentary canal dilates considerably, and on the ventral side is placed the opening of a single large sack, which forms the commencement of the liver. The walls of the hepatic sack are posteriorly united to the yolk-cells. At the region where the hepatic sack opens into the alimentary tract the latter dilates considerably.

The posterior part of the alimentary tract still constitutes a kind of yolk-sack, the ventral wall being enormously thick and formed of several layers of yolk-cells. The dorsal wall is very thin.

The excretory system is composed of two segmental ducts, each connected in front with a well-developed pronephros (head-kidney), with about five ciliated funnels opening into the pericardial region of the body cavity. The segmental ducts in the larvæ open behind into the cloacal section of the alimentary tract.

The development of the larva takes place with considerable rapidity. The yolk becomes absorbed and the larva becomes accordingly more transparent. It generally lies upon its side, and resembles in general appearance and habit a minute Amphioxus. It is soon able to swim with vigour, but usually, unless disturbed, is during the day quite quiescent, and chooses by preference the darkest situations. It soon straightens out, and, with the disappearance of the yolk, the tail becomes narrower than the head. A large caudal fin becomes developed.

When the larva is about twenty days old, it bears in most anatomical features a close resemblance to an Ammocœtes; though the histological differences between my oldest larva (29 days) and even very young Ammocœtes are considerable.

The mouth undergoes important changes. The upper lip becomes much more prominent, forming of itself the anterior end of the body (fig. 47, ul). The opening of the nasal pit is in this way relatively thrown back, and at the same time is caused to assume a dorsal position. This will be at once understood by a comparison of fig. 43 with fig. 47. On the inner side of the oral cavity a ring of papillæ is formed (fig. 47, or.p). Dorsally these papillæ are continued forward as a linear streak on the under side of the upper lip. A communication between the oral cavity and the branchial sack is very soon established.

The gill pouches gradually become enlarged; but it is some time before their small external openings are established. Their walls, which are entirely lined by hypoblast, become raised in folds, forming the branchial lamellæ. The walls of the head cavities between them become resolved into the contractors and dilators of the branchial sacks. The extra-branchial basketwork becomes established very early (it is present in the larva of 6 millimetres, about 9 days after hatching) and is shewn in an older larva in fig. 47, br.s. It is not so complicated in these young larvæ as in the Ammocœtes, but in Max Schultze’s figure, which I have reproduced, the dorsal elements of the system are omitted. On the dorsal wall of the branchial region a ciliated ridge is formed, which may be homologous with the ridge on the dorsal wall of the branchial sack of Ascidians. It has been described by Schneider in Ammocœtes.

With reference to the remainder of the alimentary canal there is but little to notice. The primitive hepatic diverticulum rapidly sprouts out and forms a tubular gland. The opening into the duodenum changes from a ventral to a lateral or even dorsal position. The duct leads into a gall-bladder imbedded in the substance of the liver. Ventrally the liver is united with the abdominal wall, but laterally passages are left by which the pericardial and body cavities continue to communicate.

The greater part of the yolk becomes employed in the formation of the intestinal wall. This part of the intestine in a nine days’ larva (67 mm.) has the form of a cylindrical tube with very thick columnar cells entirely filled with yolk particles. The dorsal wall is no longer appreciably thinner than the ventral. In the later stages the cells of this part of the intestine become gradually less columnar as the yolk is absorbed.

The fate of the yolk-cells in the Lamprey is different from that in most other Vertebrata with an equally large amount of yolk. They no doubt supply nutriment for the growth of the embryo, and although in the anterior part of the intestine they become to some extent enclosed in the alimentary tract and break up, yet in the posterior part they become wholly transformed into the regular epithelium of the intestine.

On the ninth day a slight fold filled with mesoblastic tissue is visible on the dorsal wall of the intestine. This fold appears to travel towards the ventral side; at any rate a similar but better-marked fold is visible in a ventro-lateral position at a slightly later period. This fold is the commencement of the fold which in the adult makes a half spiral, and is no doubt equivalent to the spiral valve of Elasmobranchs and Ganoids. It contains a prolongation of the cœliac artery, which constitutes at first the vitelline artery.

The nervous system does not undergo during the early larval period changes which require a description.

The opening of the olfactory sack becomes narrowed and ciliated (fig. 47, ol). It is carried by the process already mentioned to the dorsal surface of the head. The lumen of the sack is well developed; and lies in contact with the base of the fore part of the brain.

The vascular system presents no very remarkable features. The heart is two-chambered and straight. The ventricle is continued forwards as a bulbus arteriosus, which divides into two arteries at the thyroid body. From the bulbus and its continuations eight branches are given off to the gills; and, as mentioned above, a vessel, probably of the same nature, is given off in the region of the velum. The blood from the branchial sacks is collected into the dorsal aorta. Some of it is transmitted to the head, but the greater part flows backwards under the notochord.

The venous system consists of the usual anterior and posterior cardinal veins which unite on each side into a ductus Cuvieri, and of a great subintestinal vessel of the same nature as that in embryo Elasmobranchs, which persists however in the adult. It breaks up into capillaries in the liver, and constitutes therefore the portal vein. From the liver the blood is brought by the hepatic vein into the sinus venosus. In addition to these vessels there is a remarkable unpaired sub-branchial vein, which brings back the blood directly to the heart from the ventral part of the branchial region.

Metamorphosis. The larva just described does not grow directly into the adult, but first becomes a larval form, known as Ammocœtes, which was supposed to be a distinct species till Aug. Müller (No. 80) made the brilliant discovery of its nature.

The Ammocœtes does not differ to any marked extent from the larva just described. The histological elements become more differentiated, and a few organs reach a fuller development.

The branchial skeleton becomes more developed, and capsules for the olfactory sack and auditory sacks are established.

The olfactory sack is nearly divided into two by a ventral septum. The eye (fig. 48) is much more fully developed, but lies a long way below the surface. The optic cup forms a deep pit, in the mouth of which is placed the lens. The retinal layers are well developed (cf. Langerhans), and the outer layer of the optic cup or layer of retinal pigment (rp) contains numerous pigment granules, especially on its dorsal side. At the edge of the optic cup the two layers fall into each other. They constitute the commencement of the pigment layer of the iris; but at this stage they are not pigmented. The mesoblast of the iris is hardly differentiated. The lens (l) has the normal structure of the embryonic lens of Vertebrata. The inner wall is thick and doubly convex, while the outer wall, which will form the anterior epithelium, is very thin. There is a large space between the lens and the retina containing the vitreous humour (v.h). There is no aqueous humour, and the tissues in front of the lens bear but little resemblance to those in higher Vertebrata. The cornea is represented by (1) the epidermis (ep); (2) the dermis (d.c); (3) the subdermal connective tissue (s.d.c) which passes without any sharp line of demarcation into the dermis; (4) a thick membrane continuous with the choroid which represents Descemet’s membrane. The subdermal connective tissue is continued as an investment round the whole eye. There is no specially differentiated sclerotic, and a choroid is only imperfectly indicated[31]. The peculiar features of the eye of the young larva of the Ammocœtes are probably due to degeneration.

In the brain the two cerebral hemispheres lie one on each side of the anterior end of the thalamencephalon. There are well-defined olfactory lobes, and two distinct olfactory nerves are present.

The excretory system has undergone great changes. A series of segmental tubes, which first appear in a larva of about 9 mm., becomes established behind the pronephros, and in an Ammocœtes of 65 mm. the pronephros has begun to atrophy. The generative organs are formed in a larva of about 35 mm. Shortly before the metamorphosis the portion of the cloaca into which the segmental tubes open becomes separated off as a distinct urinogenital sinus, the walls of which become perforated by the two abdominal pores.

The Ammocœtes of Petromyzon Planeri lives in the mud in streams. Without undergoing any marked changes in structure it gradually grows larger, and after three or four years undergoes a metamorphosis. The full-grown larva may be as large or even larger than the adult. The metamorphosis takes place from August till January. The breeding season sets in during the second half of April; and shortly after depositing its generative products the Lamprey dies. The changes which take place in the metamorphosis are of a most striking kind.

The dome-shaped mouth of the larva is replaced (fig. 47) by a more definitely suctorial mouth with horny cuticular teeth (fig. 49). The eyes appear on the surface; and the dorsal fin becomes more prominent, and is divided into two parts.

Besides these obvious external changes very great modifications are effected in almost all the organs, which may be very briefly enumerated.

1. Very profound changes take place in the skeleton. An elaborate system of cartilages is developed in connection with the mouth; the cranium itself undergoes important modifications; and neural arches become formed.

2. Considerable changes are effected in the gill pouches, and, according to Schneider, whose statements must however be received with some caution, the branchial sack becomes detached posteriorly from the œsophagus, the œsophagus then sends forwards a prolongation above the branchial sack which is at first solid. This prolongation forms the anterior part of the œsophagus of the adult, and joins the primitive oral cavity at the velum. The so-called bronchus of the adult is thus the whole branchial region of the Ammocœtes, and the anterior part of the œsophagus of the adult is an entirely new formation.

3. The posterior part of the alimentary tract of the Ammocœtes undergoes partial atrophy. The gall-bladder of the liver is absorbed; and the liver itself ceases to communicate with the intestine.

4. The eye undergoes important changes in that it travels to the surface, and acquires all the characters of the normal vertebrate eye.

5. The brain becomes relatively larger but more compact, and the optic lobes (corpora bigemina) become more distinct.

6. The pericardial cavity becomes completely separated from the body cavity, and a distinct pericardium is formed.

7. The mesonephros of the larva disappears, and a fresh posterior part is formed.

Myxine. The ovum of Myxine when ready to be laid is inclosed, as shewn by Allen Thomson[32], in an oval horny shell in many respects similar to that of Elasmobranchii; from its ends there project a number of trumpet-shaped tubular processes, which no doubt serve to attach it to marine objects. No observations have been made on the development.

Bibliography.

(77) E. Calberla. “Der Befruchtungsvorgang beim Petromyzon Planeri.” Zeit. f. wiss. Zool., Vol. XXX. 1877.
(78) E. Calberla. “Ueb. d. Entwicklung d. Medullarrohres u. d. Chorda dorsalis d. Teleostier u. d. Petromyzonten.” Morpholog. Jahrbuch, Vol. III. 1877.
(79) C. Kupffer u. B. Benecke. Der Vorgang d. Befruchtung am Ei d. Neunaugen. Königsberg, 1878.
(80) Aug. Müller. “Ueber die Entwicklung d. Neunaugen.” Müller’s Archiv, 1856.
(81) Aug. Müller. Beobachtungen üb. d. Befruchtungserscheinungen im Ei d. Neunaugen. Königsberg, 1864.
(82) W. Müller. “Das Urogenitalsystem d. Amphioxus u. d. Cyclostomen.” Jenaische Zeitschrift, Vol. IX. 1875.
(83) Ph. Owsjannikoff. “Die Entwick. von d. Flussneunaugen.” Vorläuf. Mittheilung. Mélanges Biologiques tirés du Bulletin de l'Acad. Imp. St Pétersbourg, Vol. VII. 1870.
(84) Ph. Owsjannikoff. On the development of Petromyzon fluviatilis (Russian).
(85) Anton Schneider. Beiträge z. vergleich. Anat. u. Entwick. d. Wirbelthiere. Quarto. Berlin, 1879.
(86) M. S. Schultze. “Die Entwickl. v. Petromyzon Planeri.” Gekrönte Preisschrift. Haarlem, 1856.
(87) W. B. Scott. “Vorläufige Mittheilung üb. d. Entwicklungsgeschichte d. Petromyzonten.” Zoologischer Anzeiger, Nos. 63 and 64. III. Jahrg. 1880.

It is only within quite recent times that any investigations have been made on the embryology of this heterogeneous, but primitive group of fishes. Much still remains to be done, but we now know the main outlines of the development of Acipenser and Lepidosteus, which are representatives of the two important subdivisions of the Ganoids. Both types have a complete segmentation, but Lepidosteus presents in its development some striking approximations to the Teleostei. I have placed at the end of the chapter a few remarks with reference to the affinities indicated by the embryology.

Acipenser[34].

The freshly laid ovum is 2 mm. in diameter and is invested by a two-layered shell, covered by a cellular layer derived from the follicle[35]. The segmentation, though complete, approaches the meroblastic type more nearly than the segmentation of the frog’s egg. The first furrow appears at the formative pole, at which the germinal vesicle was situated. The earlier phases of the segmentation are like those of meroblastic ova, in that the furrows only penetrate for a certain distance into the egg. Eight vertical furrows appear before the first equatorial furrow; which is somewhat irregular, and situated close to the formative pole.

In the later stages the vertical furrows extend through the whole egg, and a segmentation cavity appears between the small and the large spheres. The segmentation is thus in the main similar to that of a frog, from which it diverges in the fact that there is a greater difference in size between the small and the large segments.

In the final stages of the segmentation the cells become distinctly divided into two layers. A layer of small cells is placed at the formative pole, and constitutes the epiblast. The cells composing it are divided, like those of Teleostei, etc., into a superficial epidermic and a deeper nervous layer. The remaining cells constitute the primitive hypoblast (the eventual hypoblast and mesoblast); they form a great mass of yolk-cells at the lower pole, and also spread along the roof of the segmentation cavity, on the inner side of the epiblast.

A process of unsymmetrical invagination now takes place, which is in its essential features exactly similar to that in the frog or the lamprey, and I must refer the reader for the details of the process to the chapter on the Amphibia. The edge of the cap of epiblast forms an equatorial line. For the greater extent of this line the epiblast cells grow over the hypoblast, as in an epibolic gastrula, but for a small arc they are inflected. At the inflected edge an invagination of cells takes place, underneath the epiblast, towards the segmentation cavity, and gives rise to the dorsal wall of the mesenteron and the main part of the dorsal mesoblast. The slit below the invaginated layer gradually dilates to form the alimentary cavity; the ventral wall of which is at first formed of yolk-cells. The epiblast along the line of the invaginated cells soon becomes thickened, and forms a medullary plate, which is not very distinct in surface views. The cephalic extremity of this plate, which is furthest removed from the edge, dilates, and the medullary plate then assumes a spatula form (fig. 50 A, Mg).

By the continued extension of the epiblast the uncovered part of the hypoblast has in the meantime become reduced to a small circular pore—the blastopore—and in surface views of the embryo has the form represented in fig. 50 A, bl.p. The invagination of the mesenteron has in the meantime extended very far forwards, and the segmentation cavity has become obliterated. The lip of the blastopore has moreover become inflected for its whole circumference.

The invaginated cells forming the dorsal wall of the mesenteron soon become divided into a pigmented hypoblastic epithelium adjoining the lumen of the mesenteron (fig. 51, En) and a mesoblastic layer (Sgp), between the hypoblast and the epiblast. The mesoblast is divided into two plates, between which is placed the notochord[36] (Ch).

With the completion of the medullary plate and the germinal layers, the first embryonic period may be considered to come to a close. The second period ends with the hatching of the embryo. During it the rudiments of the greater number of organs make their appearance. The general form of the embryo during this period is shewn in figs. 50 B and 52 A and B.

One of the first changes to take place is the conversion of the medullary plate into the medullary canal. This, as shewn in fig. 51, is effected in the usual vertebrate fashion, by the establishment of a medullary groove which is then converted into a closed canal by the folding over of the sides.

The uncovered patch of yolk in the blastoporic area soon becomes closed over; and on the formation of the medullary canal the usual neurenteric canal becomes established.

The further changes which take place are in the main similar to those in other Ichthyopsida, but in some ways the appearance of the embryo is, as may be gathered from fig. 52, rather strange. This is mainly due to the fact that the embryo does not become folded off from the yolk in the manner usual in Vertebrates; and as will be shewn in the sequel, the relation of the yolk to the embryo is unlike that in any other known Vertebrate. The appearance of the embryo is something like that of an ordinary embryo slit open along the ventral side and then flattened out. Organs which properly belong to the ventral side appear on the lateral parts of the dorsal surface. Owing to the great forward extension of the yolk the heart (fig. 52 B) appears to be placed directly in front of the head.

Even before the formation of the medullary canal the cephalic portion of the nervous system becomes marked out. This part, after the closure of the medullary groove, becomes divided into two (fig. 50 B), and then three lobes—the fore-, the mid-, and the hind-brain (fig. 52, A and B). From the lateral parts of the at first undivided fore-brain the optic vesicles (fig. 52 B, Op) soon sprout out; and in the hind-brain a dilatation to form the fourth ventricle appears in the usual fashion.

The epiblast at the sides of the brain constitutes a more or less well-defined structure, which may be spoken of as a cephalic plate (fig. 52 A, cp). From this plate are formed the essential parts of the organs of special sense. Anteriorly the olfactory pits arise (fig. 52 B, Olp) as invaginations of both layers of the epiblast. The lens of the eye is formed as an ingrowth of the nervous layer only, and opposite the hind-brain the auditory sack (fig. 52 A and B, Auv) is similarly formed from the nervous layer of the epiblast. At the sides of the cephalic plate the visceral arches make their appearance; and in fig. 52 A and B there are shewn the mandibular (Md), hyoid (Ha) and first branchial (Br´) arches, with the hyomandibular (spiracle) and hyobranchial clefts between them. They constitute peculiar concentric circles round the cephalic plate; their shape being due to the flattened form of the embryo, already alluded to.

While the above structures are being formed in the head the changes in the trunk have also been considerable. The mesoblastic plates at the junction of the head and trunk become very early segmented, the segments being formed from before backwards (fig. 50 B). With their formation the trunk rapidly increases in length. At their outer border the segmental duct (fig. 50 B, and fig. 52 A, Sd) is very early established. It is formed, as in Elasmobranchs, as a solid outgrowth of the mesoblast (fig. 51, Wg); but its anterior extremity becomes converted into a pronephros (fig. 57, pr.n).

Before hatching, the embryo has to a small extent become folded off from the yolk both anteriorly and posteriorly; and has also become to some extent vertically compressed. As a result of these changes, the general form of its body becomes much more like that of an ordinary Teleostean embryo.

The general features of the larva after hatching are illustrated by figs. 53, 54 and 55. Fig. 53 represents a larva of about 7 mm. and fig. 54 a lateral and fig. 55 a ventral view of the head of a larva of about 11 mm.

There are only a few points which call for special attention in the general form of the body. In the youngest larva figured the ventral part of the hyomandibular cleft is already closed: the dorsal part of the cleft is destined to form the spiracle (sp). The arch behind is the hyoid: on its posterior border is a membranous outgrowth, which will develop into the operculum. In older larvæ, a very rudimentary gill appears to be developed on the front walls of the spiracular cleft (Parker), but I have not succeeded in satisfying myself about its presence; and rows of gill papillæ appear on the hyoid and the true branchial arches (figs. 54 and 55, g). The biserially-arranged gill papillæ of the true branchial arches are of considerable length, and are not at first covered by the operculum; but they do not form elongated thread-like external gills similar to those of the Elasmobranchii.

The oral cavity is placed on the ventral side of the head; it has at first a more or less rhomboidal form. It soon however (fig. 55) becomes narrowed to a slit with projecting lips, and eventually becomes converted into the suctorial mouth of the adult. The most remarkable feature connected with the mouth is the development of provisional teeth (fig. 55) on both jaws.

These teeth were first discovered by Knock (No. 88). They do not appear to be calcified, and might be supposed to be of the same nature as the horny teeth of the Lamprey. They are however developed like true teeth, as a deposit between a papilla of subepidermic tissue and an epidermic cap. The substance of which they are formed corresponds morphologically to the enamel of ordinary teeth. As they grow they pierce the epidermis, and form hollow spine-like structures with a central axis filled with subepidermic (mesoblastic) cells. They disappear after the third month of larval life.

In front of the mouth two pairs of papillæ grow out, which appear to be of the same nature as the papillæ on the suctorial disc in the embryo of Lepidosteus (vide p. 115). They are very short in the embryo represented in fig. 53; soon however they grow in length (figs. 54 and 55, st); and it is probable that they become the barbels, since these occupy a precisely similar position[37].

The openings of the nasal pits are at first single; but the opening of each becomes gradually divided into two by the growth of a flap on the outer side (fig. 54, ol). It is probable that this flap is equivalent to the fold of the superior maxillary process of the Amniota, which by its growth roofs over the open groove which originally leads from the external to the internal nares; so that the two openings of each nasal sack, so established in these and in other fishes, correspond to the external and internal nares of higher Vertebrata.

At the time of hatching there is a continuous dorso-ventral fin, which, by atrophy in some parts, and hypertrophy in other parts, gives rise to all the unpaired fins of the adult, except the first dorsal and the abdominal. The caudal part of the fin is at first symmetrical, and the heterocercal tail is produced by the special growth of the ventral part of the fin.

Of the internal features of development in the Sturgeon the most important concern the relation of the yolk to the alimentary tract. In most Vertebrata the yolk-cells form a protuberance of the part of the alimentary canal, immediately behind the duodenum. The yolk may either, as in the lamprey or frog, form a simple thickening of the alimentary wall in this region, or it may constitute a well-developed yolk-sack as in Elasmobranchii and the Amniota. In either case the liver is placed in front of the yolk. In the Sturgeon on the contrary the yolk is placed almost entirely in front of the liver, and the Sturgeon appears to be also peculiar in that the yolk, instead of constituting an appendage of the alimentary tract, is completely enclosed in a dilated portion of the tract which becomes the stomach (figs. 56 and 57). It dilates this portion to such extent that it might be supposed to form a true external yolk-sack. In the stages before hatching the glandular hypoblast, which was established on the dorsal side of the primitive mesenteron, envelops the yolk-cells, which fuse together into a yolk-mass, and lose all trace of their original cellular structure.

The peculiar flattening out of the embryo over the yolk (vide p. 105) is no doubt connected with the mode in which the yolk becomes enveloped by the hypoblast.

As the posterior part of the trunk, containing the intestine, becomes formed, the yolk is gradually confined to the anterior part of the alimentary tract, which, as before stated, becomes the stomach. The epithelial cells of the stomach, as well as those of the intestine, are enormously dilated with food-yolk (fig. 57, st). Behind the stomach is formed the liver. The subintestinal vein bringing back the blood to the liver appears to have the same course as in Teleostei, in that the blood, after passing through the liver, is distributed to the walls of the stomach and is again collected into a venous trunk which falls into the sinus venosus. As the yolk becomes absorbed, the liver grows forwards underneath the stomach till it comes in close contact with the heart. The relative position of the parts at this stage is shewn diagrammatically in fig. 56. At the commencement of the intestine there arises in the larva of about 14 mm. a great number of diverticula, which are destined to form the compact glandular organ, which opens at this spot in the adult. At this stage there is also a fairly well developed pancreas opening into the duodenum at the same level as the liver.

No trace of the air-bladder was present at the stage in question.

The spiral valve is formed, as in Elasmobranchii, as a simple fold in the wall of the intestine.

There is a well developed subnotochordal rod (fig. 57) which, according to Salensky, becomes the subvertebral ligament of the adult; a statement which confirms an earlier suggestion of Bridge. The pronephros (head-kidney) resembles in the main that of Teleostei (fig. 57); while the front end of the mesonephros, which is developed considerably later than the pronephros, is placed some way behind it. In my oldest larva (14 mm.) the mesonephros did not extend backwards into the posterior part of the abdominal cavity.

Bibliography.

(88) Knock. “Die Beschr. d. Reise z. Wolga Behufs d. Sterlettbefruchtung.” Bull. Soc. Nat. Moscow, 1871.
(89) A. Kowalevsky, Ph. Owsjannikoff, and N. Wagner. “Die Entwick. d. Störe.” Vorläuf. Mittheilung. Mélanges Biologiques tirés du Bulletin d. l'Acad. Imp. St Pétersbourg, Vol. VII. 1870.
(90) W. Salensky. “Development of the Sterlet (Acipenser ruthenus).” 2 Parts. Proceedings of the Society of Naturalists in the imperial University of Kasan. 1878 and 9 (Russian). Part I., abstracted in Hoffmann and Schwalbe’s Jahresbericht for 1878.
(91) W. Salensky. “Zur Embryologie d. Ganoiden (Acipenser).” Zoologischer Anzeiger, Vol. I., Nos. 11, 12, 13.

Lepidosteus[38].

The ova of Lepidosteus are spherical bodies of about 3 mm. in diameter. They are invested by a tough double membrane, composed of (1) an outer layer of somewhat pyriform bodies, radiately arranged, which appear to be the remains of the follicular cells; and (2) of an inner zona radiata, the outer part of which is radiately striated, while the inner part is homogeneous.

The segmentation, as in the Sturgeon, is complete, but approaches closely the meroblastic type. It commences with a vertical furrow at the animal pole, extending through about one-fifth of the circumference. Before this furrow has proceeded further a second furrow is formed at right angles to it. The next stages have not been observed, but on the third day after impregnation (fig. 58), the animal pole is completely divided into small segments, which form a disc similar to the blastoderm of meroblastic ova; while the vegetative pole, which subsequently forms a large yolk-sack, is divided by a few vertical furrows, four of which nearly meet at the pole opposite the blastoderm. The majority of the vertical furrows extend only a short way from the edge of the small spheres, and are partially intercepted by imperfect equatorial furrows.

The stages immediately following the segmentation are still unknown, and in the next stage satisfactorily observed, on the fifth day after impregnation, the body of the embryo is distinctly differentiated. The lower pole of the ovum is then formed of a mass in which no traces of segments or segmentation furrows can be detected.

The embryo (fig. 59) has a dumbbell-shaped outline, and is composed of (1) an outer area, with some resemblance to the area pellucida of an avian embryo, forming the lateral part of the body; and (2) a central portion consisting of the vertebral plates and medullary plate. The medullary plate is dilated in front to form the brain (br). Two lateral swellings in the brain are the commencing optic vesicles. The caudal extremity of the embryo is somewhat swollen.

Sections of this stage (fig. 60) are interesting as shewing a remarkable resemblance between Lepidosteus and Teleostei.

The three layers are fully established. The epiblast (ep) is formed of a thicker inner nervous stratum, and an outer flattened epidermic stratum. Along the axial line there is a solid keel-like thickening of the nervous layer of the epidermis, which projects towards the hypoblast. This thickening (MC) is the medullary cord; and there is no evidence of the epidermic layer being at this or any subsequent period concerned in its formation (vide chapter on Teleostei, p. 72). In the region of the brain the medullary cord is so thick that it gives rise, as in Teleostei, to a projection of the whole body of the embryo towards the yolk. Posteriorly it is flatter. The mesoblast (Me) in the trunk has the form of two plates, which thin out laterally. The hypoblast (hy) is a single layer of cells, and is nowhere folded in to form a closed alimentary canal. The hypoblast is separated from the neural cord by the notochord (Ch), which throughout the greater part of the embryo is a distinct structure.

In the region of the tail, the axial part of the hypoblast, the notochord, and the neural cord fuse together, the fused part so formed is the homologue of the neurenteric canal of other types. Quite at the hinder end of the embryo the mesoblastic plates cease to be separable from the axial structures between them.

In a somewhat later stage the embryo is considerably more elongated, embracing half the circumference of the ovum. The brain is divided into three distinct vesicles.

Anteriorly the neural cord has now become separated from the epidermis. The whole of the thickened nervous layer of the epiblast appears to remain united with the cerebrospinal cord, so that the latter organ is covered dorsally by the epidermic layer of the epiblast only. The nervous layer soon however grows in again from the two sides.

Where the neural cord is separated from the epidermis, it is already provided with a well-developed lumen. Posteriorly it remains in its earlier condition.

In the region of the hind-brain traces of the auditory vesicles are present in the form of slightly involuted thickenings of the nervous layer of the epidermis.

The mesoblast of the trunk is divided anteriorly into splanchnic and somatic layers.

In the next stage, on the sixth day after impregnation (fig. 61), there is a great advance in development. The embryo is considerably longer, and a great number of mesoblastic somites are visible. The body is now laterally compressed and raised from the yolk.

The region of the head is more distinct, and laterally two streaks are visible (br.c), which, by comparison with the Sturgeon, would seem to be the two first visceral clefts[39]: they are not yet perforated. In the lateral regions of the trunk the two segmental ducts are visible in surface views (fig. 61, sd) occupying the same situation as in the Sturgeon. Their position in section is shewn in fig. 62, sg.

With reference to the features in development, visible in sections, a few points may be alluded to.

The optic vesicles are very prominent outgrowths of the brain, but are still solid, though the anterior cerebral vesicle has a well-developed lumen. The auditory vesicles are now deep pits of the nervous layer of the epiblast, the openings of which are covered by the epidermic layer. They are shewn for a slightly later stage in fig. 63 (au.v).

There is now present a subnotochordal rod, which develops as in other types from a thickening of the hypoblast (fig. 62, x).

In an embryo of the seventh day after impregnation, the features of the preceding stage become generally more pronounced.

The optic vesicles are now provided with a lumen (fig. 64), and have approached close to the epidermis. Adjoining them a thickening (l) of the nervous layer of the epidermis has appeared, which will form the lens. The cephalic extremity of the segmental duct, which, as shewn in fig. 61, is bent inwards towards the middle line, has now become slightly convoluted, and forms the rudiment of a pronephros (head-kidney).

During the next few days the folding off of the embryo from the yolk commences, and proceeds till the embryo acquires the form represented in fig. 65.

Both the head and tail are quite free from the yolk; and the embryo presents a general resemblance to that of a Teleostean.

On the ventral surface of the front of the head there is a disc (figs. 65, 66, sd), which is beset with a number of processes, formed as thickenings of the epiblast. As shewn by Agassiz, these eventually become short suctorial papillæ[40]. Immediately behind this disc is placed a narrow depression which forms the rudiment of the mouth.

The olfactory pits are now developed, and are placed near the front of the head.

A great advance has taken place in the development of the visceral clefts and arches. The oral region is bounded behind by a well-marked mandibular arch, which is separated by a shallow depression from a still more prominent hyoid arch (fig. 65, hy). Between the hyoid and mandibular arches a double lamella of hypoblast, which represents the hyomandibular cleft, is continued from the throat to the external skin, but does not, at this stage at any rate, contain a lumen.

The hyoid arch is prolonged backwards into a considerable opercular fold, which to a great extent overshadows the branchial clefts behind. The hyobranchial cleft is widely open.

Behind the hyobranchial cleft are four pouches of the throat on each side, not yet open to the exterior. They are the rudiments of the four branchial clefts of the adult.

The trunk has the usual compressed piscine form, and there is a well-developed dorsal fin continuous round the end of the tail, with a ventral fin. There is no trace of the paired fins.

The anterior and posterior portions of the alimentary tract are closed in, but the middle region is still open to the yolk. The circulation is now fully established, and the vessels present the usual vertebrate arrangement. There is a large subintestinal vein.

The first of Agassiz’ embryos was hatched about ten days after impregnation. The young fish on hatching immediately used its suctorial disc to attach itself to the sides of the vessel in which it was placed.

The general form of Lepidosteus shortly after hatching is shewn in fig. 67. On the ventral part of the front of the head is placed the large suctorial disc. At the side of the head are seen the olfactory pit, the eye and the auditory vesicle; while the projecting vesicle of the mid-brain is very prominent above. Behind the mouth follow the visceral arches. The mandibular arch (md) is placed on the hinder border of the mouth, and is separated by a deep groove from the hyoid arch (hy). This groove is connected with the hyomandibular cleft, but I have not determined whether it is now perforated. The posterior border of the hyoid arch is prolonged into an opercular fold. Behind the hyoid arch are seen the true branchial arches.

There is still a continuous dorso-ventral fin, in which there are as yet no fin-rays, and the anterior paired fins are present.

The yolk-sack is very large, but its communication with the alimentary canal is confined to a narrow vitelline duct, which opens into the commencement of the intestine immediately behind the duct of the liver, which is now a compact gland. The yolk in Lepidosteus thus behaves very differently from that in the Sturgeon. In the first place it forms a special external yolk-sack, instead of an internal dilatation of part of the alimentary tract; and in the second place it is placed behind instead of in front of the liver.

I failed to find any trace of a pancreas. There is however, opening on the dorsal side of the throat, a well-developed appendage continued backwards beyond the level of the commencement of the intestine. This appendage is no doubt the air-bladder.

In the course of the further growth of the young Lepidosteus, the yolk-sack is rapidly absorbed, and has all but disappeared after three weeks. A rich development of pigment early takes place; and the pigment is specially deposited on the parts of the embryonic fin which will develop into the permanent fins.

The notochord in the tail bends slightly upwards, and by the special development of a caudal lobe an externally heterocercal tail like that of Acipenser is established. The ventral paired fins are first visible after about the end of the third week, and by this time the operculum has grown considerably, and the gills have become well developed.

The most remarkable changes in the later periods are those of the mouth.

The upper and lower jaws become gradually prolonged, till they eventually form a snout; while at the end of the upper jaw is placed the suctorial disc, which is now considerably reduced in size (fig. 68, sd). The “fleshy globular termination of the upper jaw of the adult Lepidosteus is the remnant of this embryonic sucking disc.” (Agassiz, No. 92.)

The fin-rays become formed as in Teleostei, and parts of the continuous embryonic fin gradually undergo atrophy. The dorsal limb of the embryonic tail, as has been shewn by Wilder, is absorbed in precisely the same manner as in Teleostei, leaving the ventral lobe to form the whole of the permanent tail-fin.

Bibliography.

(92) Al. Agassiz. “The development of Lepidosteus.” Proc. Amer. Acad. of Arts and Sciences, Vol. XIII. 1878.

General observations on the Embryology of the Ganoids.

The very heterogeneous character of the Ganoid group is clearly shewn both in its embryology and its anatomy. The two known types of formation of the central nervous system are exemplified in the two species which have been studied, and these two species, though in accord in having a holoblastic segmentation, yet differ in other important features of development, such as the position of the yolk etc. Both types exhibit Teleostean affinities in the character of the pronephros; but as might have been anticipated Lepidosteus presents in the origin of the nervous system, the relations of the hypoblast, and other characters, closer approximations to the Teleostei than does Acipenser. There are no very prominent Amphibian characters in the development of either type, other than a general similarity in the segmentation and formation of the layers. In the young of Polypterus an interesting amphibian and dipnoid character is found in the presence of a pair of true external gills covered by epiblast. These gills are attached at the hinder end of the operculum, and receive their blood from the hyoid arterial arch[41]. In the peculiar suctorial disc of Lepidosteus, and in the more or less similar structure in the Sturgeon, these fishes retain, I believe, a very primitive vertebrate organ, which has disappeared in the adult state of almost all the Vertebrata; but it is probable that further investigations will shew that the Teleostei, and especially the Siluroids, are not without traces of a similar structure.

The eggs of most Amphibia[43] are laid in water. They are smallish nearly spherical bodies, and in the majority of known Anura (all the European species), and in many Urodela (Amblystoma, Axolotl, though not in the common Newt) part of the surface is dark or black, owing to the presence of a superficial layer of pigment, while the remainder is unpigmented. The pigmented part is at the upper pole of the egg, and contains the germinal vesicle till the time of its atrophy; and the yolk-granules in it are smaller than those in the unpigmented part. The ovum is closely surrounded by a vitelline membrane[44], and receives, in its passage down the oviduct, a gelatinous investment of varying structure.

In the Anura the eggs are fertilized as they leave the oviduct. In some of the Urodela the mode of fertilization is still imperfectly understood. In Salamanders and probably Newts it is internal[45]; but in Amblystoma punctatum (Clark, No. 98), the male deposits the semen in the water. The eggs are laid by the Anura in masses or strings. By Newts they are deposited singly in the angle of a bent blade of grass or leaf of a water-plant, and by Amblystoma punctatum in masses containing from four eggs to two hundred. Salamandra atra and Salamandra maculosa are viviparous. The period of gestation for the latter species lasts a whole year.

A good many exceptions to the above general statements have been recorded[46].

In Notodelphis ovipara the eggs are transported (by the male?) into a peculiar dorsal pouch of the skin of the female, which has an anterior opening, but is continued backwards into a pair of diverticula. The eggs are very large, and in this pouch, which they enormously distend, they undergo their development. A more or less similar pouch is found in Nototrema marsupiatum.

In the Surinam toad (Pipa dorsigera) the eggs are placed by the male on the back of the female. A peculiar pocket of skin becomes developed round each egg, the open end of which is covered by a gelatinous operculum. The larvæ are hatched, and actually undergo their metamorphosis, in these pockets. The female during this period lives in water. Pipa Americana (if specifically distinct from P. dorsigera) presents nearly the same peculiarities. The female of a tree frog of Ceylon (Polypedates reticulatus) carries the eggs attached to the abdomen.

Rhinoderma Darwinii[47] behaves like some of the Siluroid fishes, in that the male carries the eggs during their development in an enormously developed laryngeal pouch.

Some Anura do not lay their eggs in water. Chiromantis Guineensis attaches them to the leaves of trees; and Cystignathus mystacius lays them in holes near ponds, which may become filled with water after heavy rains.

The eggs of Hylodes Martinicensis are laid under dead leaves in moist situations.

Formation of the layers.

Anura. The formation of the germinal layers has so far only been studied in some Anura and in the Newt. The following description applies to the Anura, and I have called attention, at the end of the section, to the points in which the Newt is peculiar.

The segmentation of the Frog’s ovum has already been described (Vol. II. pp. 95-7), but I may remind the reader that the segmentation (fig. 69) results in the formation of a vesicle, the cavity of which is situated excentrically; the roof of the cavity being much thinner than the floor. The cavity is the segmentation cavity. The roof is formed of two or three layers of smallish pigmented cells, and the floor of large cells, which form the greater part of the ovum. These large cells, which are part of the primitive hypoblast, will be spoken of in the sequel as yolk-cells: they are equivalent to the food-yolk of the majority of vertebrate ova.

The cells forming the roof of the cavity pass without any sharp boundary into the yolk-cells, there being at the junction of the two a number of cells of an intermediate character. The cells both of the roof and the floor continue to increase in number, and those of the roof become divided into two distinct strata (fig. 70, ep).

The upper of these is formed of a single row of somewhat cubical cells, and the lower of several rows of more rounded cells. Both of these strata eventually become the epiblast, of which they form the epidermic and nervous layers. The roof of the segmentation cavity appears therefore to be entirely constituted of epiblast.

The next changes which take place lead (1) to the formation of the mesenteron[48], and (2) to the enclosure of the yolk-cells by the epiblast.

The mesenteron is formed as in Petromyzon and Lepidosteus by an unsymmetrical form of invagination. The invagination first commences by an inflection of the epiblast-cells for a small arc on the equatorial line which marks the junction between the epiblastic cells and the yolk-cells (fig. 70, x).

The inflected cells become continuous with the adjoining cells; and the region where the inflection is formed constitutes a kind of lip, below which a slit-like cavity is soon established. This lip is equivalent to the embryonic rim of the Elasmobranch blastoderm, and the cavity beneath it is the rudiment of the mesenteron.

The mesenteron now rapidly extends by the invagination of the cells on its dorsal side. These cells grow inwards towards the segmentation cavity as a layer of cells several rows deep. At its inner end, this layer is continuous with the yolk-cells; and is divided into two strata (fig. 71 A), viz. (1) a stratum of several rows of cells adjoining the epiblast, which becomes the mesoblast (m), and (2) a stratum of a single row of more columnar cells lining the cavity of the mesenteron, which forms the hypoblast (hy). The growth inwards of the dorsal wall of the mesenteron is no doubt in part a true invagination, but it seems probable that it is also due in a large measure to an actual differentiation of yolk-cells along the line of growth. The mesenteron is at first a simple slit between the yolk and the hypoblast (fig. 71 A), but as the involution of the hypoblast and mesoblast extends further inwards, this slit enlarges, especially at its inner end, into a considerable cavity; the blind end of which is separated by a narrow layer of yolk-cells from the segmentation cavity (fig. 71 B).

In the course of the involution, the segmentation cavity becomes gradually pushed to one side and finally obliterated. Before obliteration, it appears in some forms (Pelobates fuscus) to become completely enclosed in the yolk-cells.

While the invagination to form the mesenteron takes place as above described, the enclosure of the yolk has been rapidly proceeding. It is effected by the epiblast growing over the yolk at all points of its circumference. The nature of the growth is however very different at the embryonic rim and elsewhere. At the embryonic rim it takes place by the simple growth of the rim, so that the point x in figs. 70 and 71 is carried further and further over the surface of the yolk. Elsewhere the epiblast at first extends over the yolk as in a typical epibolic gastrula, without being inflected to form a definite lip. While a considerable patch of yolk is still left uncovered, the whole of the edge of the epiblast becomes however inflected, as at the embryonic rim (fig. 71 A); and a circular blastopore is established, round the whole edge of which the epiblast and intermediate cells are continuous.

From the ventral lip of the blastopore the mesoblast (fig. 71, m´), derived from the small intermediate cells, grows inwards till it comes to the segmentation cavity; the growth being not so much due to an actual invagination of cells at the lip of the blastopore, as to a differentiation of yolk-cells in situ. Shortly after the stage represented in fig. 71 B, the plug of yolk, which fills up the opening of the blastopore, disappears, and the mesenteron communicates freely with the exterior by a small circular blastopore (fig. 73). The position of the blastopore is the same as in other types, viz. at the hinder end of the embryo.

By this stage the three layers of the embryo are definitely established. The epiblast, consisting from the first of two strata, arises from the small cells forming the roof of the segmentation-cavity. It becomes continuous at the lip of the blastopore with cells intermediate in size between the cells of which it is formed and the yolk-cells. These latter, increasing in number by additions from the yolk-cells, give rise to the mesoblast and to part of the hypoblast; while to the latter layer the yolk-cells, as mentioned above, must also be considered as appertaining. Their history will be dealt with in treating of the general fate of the hypoblast.

Urodela. The early stages of the development of the Newt have been adequately investigated by Scott and Osborn (No. 114). The segmentation and formation of the layers is in the main the same as in the Frog. The ovum is without black pigment. There is a typical unsymmetrical invagination, but the dorsal lip of the blastopore is somewhat thickened. The most striking feature in which the Newt differs from the Frog is the fact that the epiblast is at first constituted of a single layer of cells (fig. 75, ep). The roof of the segmentation cavity is constituted, during the later stages of segmentation, of several rows of cells (Bambeke, No. 95), but subsequently it would appear to be formed of a single row of cells only (Scott and Osborn, No. 114).

General history of the layers.

Epiblast: Anura. At the completion of the invagination the epiblast forms a continuous layer enclosing the whole ovum, and constituted throughout of two strata. The formation of the medullary canal commences by the nervous layer along the axial dorsal line becoming thickened, and giving rise to a somewhat pyriform medullary plate, the sides of which form the projecting medullary folds (fig. 77 A). The medullary plate is thickened at the two sides, and is grooved in the median line by a delicate furrow (fig. 72, r). The dilated extremity of the medullary plate, situated at the end of the embryo opposite the blastopore, is the cerebral part of the plate, and the remainder the spinal. The medullary folds bend upwards, and finally meet above, enclosing a central cerebrospinal canal (fig. 74). The point at which they first meet is nearly at the junction of the brain and spinal cord, and from this point their junction extends backwards and forwards; but the whole process is so rapid that the closure of the medullary canal for its whole length is effected nearly simultaneously. In front the medullary canal ends blindly, but behind it opens freely into the still persisting blastopore, with the lips of which the medullary folds become, as in other types, continuous. Fig. 73 represents a longitudinal section through an embryo, shortly after the closure of the medullary canal (nc); the opening of which into the blastopore (x) is clearly seen.

On the closure of the medullary canal, its walls become separated from the external epiblast, which extends above it as a continuous layer. In the formation of the central nervous system both strata of the epiblast have a share, though the main mass is derived from the nervous layer. After the central nervous tube has become separated from the external skin, the two layers forming it fuse together; but there can be but little doubt that at a later period the epidermic layer separates itself again as the central epithelium of the nervous system.

Both the nervous and epidermic strata have a share in forming the general epiblast; and though eventually they partially fuse together yet the horny layer of the adult epidermis, where such can be distinguished, is probably derived from the epidermic layer of the embryo, and the mucous layer of the epidermis from the embryonic nervous layer.

In the formation of the organs of sense the nervous layer shews itself throughout as the active layer. The lens of the eye and the auditory sack are derived exclusively from it, the latter having no external opening. The nervous layer also plays the more important part in the formation of the olfactory sack.

The outer layer of epiblast-cells becomes ciliated after the close of the segmentation, but the cilia gradually disappear on the formation of the internal gills. The cilia cause a slow rotatory movement of the embryo within the egg, and probably assist in the respiration after it is hatched. They are especially developed on the external gills.

Urodela. In the Newt (Scott and Osborn, No. 114) the medullary plate becomes established, while the epiblast is still formed of a single row of cells; and it is not till after the closure of the neural groove that any distinction is observable between the epithelium of the central canal, and the remaining cells of the cerebrospinal cord (fig. 75).

Before the closure of the medullary folds the lateral epiblast becomes divided into the two strata present from the first in the Frog; and in the subsequent development the inner layer behaves as the active layer, precisely as in the Anura.

The mesoblast and notochord: Anura. After the disappearance of the segmentation cavity, the mesoblast is described by most observers, including Götte, as forming a continuous sheet round the ovum, underneath the epiblast. The first important differentiations in it take place, as in the case of the epiblast, in the axial dorsal line. Along this line a central cord of the mesoblast becomes separated from the two lateral sheets to form the notochord. Calberla states, however, that when the mesoblast is distinctly separated from the hypoblast it does not form a continuous sheet, but two sheets one on each side, between which is placed a ridge of cells continuous with the hypoblastic sheet. This ridge subsequently becomes separated from the hypoblast as the notochord. Against this view Götte has recently strongly protested, and given a series of careful representations of his sections which certainly support his original account.

My own observations are in favour of Calberla’s statement, and so far as I can determine from my sections the mesoblast never appears as a perfectly continuous sheet, but is always deficient in the dorsal median line. My observations are unfortunately not founded on a sufficient series of sections to settle the point definitely.

After the formation of the notochord (fig. 72), the mesoblast may be regarded as consisting of two lateral plates, continuous ventrally, but separated in the median dorsal line. By the division of the dorsal parts of these plates into segments, which commences in the region of the neck and thence extends backwards, the mesoblast of the trunk becomes divided into a vertebral portion, cleft into separate somites, and a lateral unsegmented portion (fig. 74).

The history of these two parts and of the mesoblast is generally the same as in Elasmobranchs.

The mesoblast in the head becomes, according to Götte, divided into four segments, equivalent to the trunk somites. Owing to a confusion into which Götte has fallen from not recognizing the epiblastic origin of the cranial nerves, his statements on this head must, I think, be accepted with considerable reserve; but some part of his segments appears to correspond with the head-cavities of Elasmobranchii.

Urodela. Scott and Osborn (No. 114) have shewn that in the Newt the mesoblast (fig. 75) is formed of two lateral plates, split off from the hypoblast, and that the ventral growth of these plates is largely effected by the conversion of yolk-cells into mesoblast-cells. They have further shewn that the notochord is formed of an axial portion of the hypoblast, as in the types already considered (fig. 75). The body cavity is continued into the region of the head; and the mesoblast lining the cephalic section of the body cavity is divided into the same number of head cavities as in Elasmobranchii, viz. one in front of the mouth, and one in the mandibular and one in each of the following arches.

The hypoblast. There are no important points of difference in the relations of the hypoblast between the Anura and Urodela. The mesenteron, at the stage represented in fig. 73, forms a wide cavity lined dorsally by a layer of invaginated hypoblast, and ventrally by the yolk-cells. The hypoblast is continuous laterally and in front with the yolk-cells (figs. 72, 74 and 75). At an earlier stage, when the mesenteron has a less definite form, such a continuity between the true hypoblast and the yolk-cells does not exist at the sides of the cavity.

The definite closing in of the mesenteron by the true hypoblast-cells commences in front and behind, and takes place last of all in the middle (fig. 76). In front this process takes place with the greatest rapidity. The cells of the yolk-floor become continuously differentiated into hypoblast-cells, and very soon the whole of the front end becomes completely lined by true hypoblastic cells, while the yolk-cells become confined to the floor of the middle part.

The front portion of the mesenteron gives rise to the œsophagus, stomach and duodenum. Close to its hinder boundary there appears a ventral outgrowth, which is the commencement of the hepatic diverticulum (fig. 76, l). The yolk is thus post-hepatic, as in Vertebrates generally.

The stomodæum is formed comparatively late by an epiblastic invagination (fig. 76, m).

It should be noticed that the conversion of the yolk-cells into hypoblast-cells to form the ventral wall of the anterior region of the alimentary tract is a closely similar occurrence to the formation of cells in the yolk-floor of the anterior part of the alimentary tract in Elasmobranchii. This conversion is apparently denied by Götte, but since I find cells in all stages of transition between yolk-cells and hypoblast-cells I cannot doubt the fact of its occurrence.

At first, the mesenteron freely communicates with the exterior by the opening of the blastopore. The lips of the blastopore gradually approximate, and form a narrow passage on the dorsal side of which the neural tube opens, as has already been described (fig. 73). The external opening of this passage finally becomes obliterated, and the passage itself is left as a narrow diverticulum leading from the hind end of the mesenteron into the neural canal (fig. 76). It forms the postanal gut, and gradually narrows and finally atrophies. At its front border, on the ventral side, there may be seen a slight ventrally directed diverticulum of the alimentary tract, which first becomes visible at a somewhat earlier stage (fig. 73). This diverticulum becomes longer and meets an invagination of the skin (fig. 76, an), which arises in Rana temporaria at a somewhat earlier period than represented by Götte in Bombinator. This epiblastic invagination is the proctodæum, and an anal perforation eventually appears at its upper extremity.

The differentiation of the hinder end of the præanal gut proceeds in the same fashion as that of the front end, though somewhat later. It gives rise to the cloacal and intestinal part of the alimentary tract. From the ventral wall of the cloacal section, there grows out the bifid allantoic bladder, which is probably homologous with the allantois of the higher Vertebrata. After the differentiation of the ventral wall of the fore and hind ends of the alimentary tract has proceeded for a certain distance, the yolk only forms a floor for a restricted median region of the alimentary cavity, which corresponds to the umbilical canal of the Amniota. The true hypoblastic epithelium then grows over the outer side of the yolk, which thus constitutes a true, though small, and internal yolk-sack. The yolk-cells enclosed in this sack become gradually absorbed, and the walls of the sack form part of the intestine.

General growth of the Embryo.

Anura. The pyriform medullary plate, already described, is the first external indication of the embryo. This plate appears about the stage represented in longitudinal section in fig. 71 B. The feature most conspicuous in it at first is the axial groove. It soon becomes more prominent (fig. 77 A), and ends behind at the blastopore (bl), the lips of which are continuous with the two medullary folds. As the sides of this plate bend upwards to form the closed medullary canal, the embryo elongates itself and assumes a somewhat oval form. At the same time the cranial flexure becomes apparent (fig. 73), and the blastopore shortly afterwards becomes shut off from the exterior. The embryo now continues to grow in length (fig. 77 B), and the mesoblast becomes segmented. The somites are first formed in the neck, and are added successively behind in the unsegmented posterior region of the embryo. The hind end of the embryo grows out into a rounded prominence, which rapidly elongates, and becomes a well-marked tail entirely formed by the elongation of the postanal section of the body. The whole body has a very decided dorsal flexure, the ventral surface being convex. Fig. 78 represents an embryo of Bombinator in side view, with the tail commencing to project. The longitudinal section (fig. 76) is taken through an embryo of about the same age. In the cephalic region important changes have taken place. The cranial flexure has become more marked, but is not so conspicuous a feature in the Amphibia as in most other types, owing to the small size of the cerebral rudiment. The mid-brain is shewn at fig. 78 a forming the termination of the long axis of the body, and the optic vesicles (a´) are seen at its sides.

The rudiments of the mandibular (d), hyoid (e), and first branchial (e´) arches project as folds at the side of the head, but the visceral clefts are not yet open. Rudiments of the proctodæum and stomodæum have appeared, but neither of them as yet communicates with the mesenteron. Below the hyoid arch is seen a peculiar disc (k) which is an embryonic suctorial organ, formed of a plate of thickened epiblast. There is a pair of these discs, one on each side, but only one of them is shewn in the figure. At a later period they meet each other in the middle line, though they separate again before their final atrophy. They are found in the majority of the Anura, but are absent according to Parker in the Aglossa (Pipa and Dactylethra (fig. 83)). They are probably remnants of the same primitive organs as the suctorial disc of Lepidosteus.

The embryo continues to grow in length, while the tail becomes more and more prominent, and becomes bent round to the side owing to the confinement of the larva within the egg-membrane. At the front of the head the olfactory pits become distinct. The stomodæum deepens, though still remaining blind, and three fresh branchial arches become formed; the last two being very imperfectly differentiated, and not visible from the exterior. There are thus six arches in all, viz. the mandibular, the hyoid and four branchial arches. Between the mandibular and the hyoid, and between each of the following arches, pouches of the mesenteron push their way towards the external skin. Of these pouches there are five, there being no pouch behind the last branchial arch. The first of these will form the hyomandibular cleft, the second the hyobranchial, and the third, fourth and fifth the three branchial clefts.

Although the pouches of the throat meet the external skin, an external opening is not formed in them till after the larva is hatched. Before this takes place there grow, in the majority of forms, from the outer side of the first and second branchial arches small processes, each forming the rudiment of an external gill; a similar rudiment is formed, either before or after hatching, on the third arch; but the fourth arch is without it (figs. 80 and 82).

These external gills, which differ fundamentally from the external gills of Elasmobranchii in being covered by epiblast, soon elongate and form branched ciliated processes floating freely in the medium around the embryo (fig. 80).

Before hatching the excretory system begins to develop. The segmental duct is formed as a fold of the somatic wall at the dorsal side of the body cavity (fig. 79, u). Its anterior end alone remains open to the body cavity, and gives rise to a pronephros with two or three peritoneal openings, opposite to which a glomerulus is formed.

The mesonephros (permanent kidney of Amphibia) is formed as a series of segmental tubes much later than the pronephros, during late larval life. Its anterior end is situated some distance behind the pronephros, and during its formation the pronephros atrophies.

The period of hatching varies in different larvæ, but in most cases, at the time of its occurrence, the mouth has not yet become perforated. The larva, familiarly known as a tadpole, is at first enclosed in the detritus of the gelatinous egg envelopes. The tail, by the development of a dorsal and ventral fin, very soon becomes a powerful swimming organ. Growth, during the period before the larva begins to feed, is no doubt carried on at the expense of the yolk, which is at this time enclosed within the mesenteron.

The mouth and anal perforations are not long in making their appearance, and the tadpole is then able to feed. The gill slits also become perforated, but the hyomandibular diverticulum in most species never actually opens to the exterior, and in all cases becomes very soon closed.

There can be but little doubt that the hyomandibular diverticulum gives rise, as in the Amniota, to the Eustachian tube and tympanic cavity, except when these are absent (i.e. Bombinatoridæ). Götte holds however that these parts are derived from the hyobranchial cleft, but his statements on this head, which would involve us in great morphological difficulties, stand in direct contradiction to the careful researches of Parker.

Shortly after hatching, there grows out from the hyoid arch on each side an opercular fold of skin, which gradually covers over the posterior branchial arches and the external gills (fig. 80 d). It fuses with the skin at the upper part of the gill arches, and also with that of the pericardial wall below them; but is free in the middle, and so assists in forming a cavity, known as the branchial cavity, in which the gills are placed. Each branchial cavity at first opens by a separate widish pore behind (fig. 80), and in Dactylethra both branchial apertures are preserved (Huxley). In the larva of Bombinator, and it would seem also that of Alytes and Pelodytes, the original widish openings of the two branchial chambers meet together in the ventral line, and form a single branchial opening or spiracle. In most other forms, i.e. Rana, Bufo, Pelobates, etc., the two branchial chambers become united by a transverse canal, and the opening of the right sack then vanishes, while that of the left remains as the single unsymmetrical spiracle. In breathing the water is taken in at the mouth, passes through the branchial clefts into the branchial cavities, and is thence carried out by the spiracle.

Immediately after the formation of the branchial cavities, the original external gills atrophy, but in their place fresh gills, usually called internal gills, appear on the outer side of the middle region of the four branchial arches.

There is a single row of these on the first and fourth branchial arches, and two rows on the second and third. In addition to these gills, which are vascular processes of the mesoblast, covered, according to Götte, with an epiblastic (?) epithelium, branchial processes appear on the hypoblastic walls of the three branchial clefts. The last-named branchial processes would appear to be homologous with the gills of Lampreys. In Dactylethra no other gills but these are formed (Parker).

The mouth, even before the tadpole begins to feed, acquires a transversely oval form (fig. 81), and becomes armed with provisional structures in the form of a horny beak and teeth, which are in use during larval life.

The beak is formed of a pair of horny plates moulded on the upper and lower pairs of labial cartilages. The upper valve of the beak is the larger of the two, and covers the lower. The beak is surrounded by a projecting lip formed of a circular fold of skin, the free edge of which is covered by papillæ. Between the papillæ and the beak rows of horny teeth are placed on the inner surface of the lip. There are usually two rows of these on the upper side, the inner one not continuous across the middle line, and three or four rows on the lower side, the inner one or two divided into two lateral parts.

As the tadpole attains its full development, the suctorial organs behind the mouth gradually atrophy. The alimentary canal, which is (fig. 81) at first short, rapidly elongates, and fills up with its numerous coils the large body cavity. In the meantime, the lungs develop as outgrowths from the œsophagus.

Various features in the anatomy of the Tadpole point to its being a repetition of a primitive vertebrate type. The nearest living representative of this type appears to be the Lamprey.

The resemblance between the mouths of the Tadpole and Lamprey is very striking, and many of the peculiarities of the larval skull of the Anura, especially the position of the Meckelian cartilages and the subocular arch, perhaps find their parallel in the skull of the Lamprey[49]. The internal hypoblastic gill-sacks of the Frog, with their branchial processes, are probably equivalent to the gill-sacks of the Lamprey[50]; and it is not impossible that the common posterior openings of the gill-pouches in Myxine are equivalent to the originally paired openings of the branchial sack of the Tadpole.

The resemblances between the Lamprey and the Tadpole appear to me to be sufficiently striking not to be merely the results of more or less similar habits; but at the same time there are no grounds for supposing that the Lamprey itself is closely related to an ancestral form of the Amphibia. In dealing with the Ganoids and other types arguments have been adduced to shew that there was a primitive vertebrate stock provided with a perioral suctorial disc; and of this stock the Cyclostomata are the degraded, but at the same time the nearest living representatives. The resemblances between the Tadpole and the Lamprey are probably due to both of them being descended from this stock. The Ganoids, as we have seen, also shew traces of a similar descent; and the resemblance between the larva of Dactylethra (fig. 83), the Old Red Sandstone Ganoids[51] and Chimæra, probably indicates that an extension of our knowledge will bring to light further affinities between the primitive Ganoid and Holocephalous stocks and the Amphibia.

Metamorphosis. The change undergone by the Tadpole in its passage into the Frog is so considerable as to deserve the name of a metamorphosis. This metamorphosis essentially consists in the reduction and atrophy of a series of provisional embryonic organs, and the appearance of adult organs in their place. The stages of this metamorphosis are shewn in fig. 82, 5, 6, 7, 8.

The two pairs of limbs appear nearly simultaneously as small buds; the hinder pair at the junction of the tail and body (fig. 82, 5), and the anterior pair concealed under the opercular membrane. The lungs acquire a greater and greater importance, and both branchial and pulmonary respirations go on together for some time.

When the adult organs are sufficiently developed an ecdysis takes place, in which the gills are completely lost, the provisional horny beak is thrown off, and the mouth loses its suctorial form. The eyes, hitherto concealed under the skin, become exposed on the surface, and the front limbs appear (fig. 82, 6). With these external changes important internal modifications of the mouth, the vascular system, and the visceral arches take place. A gradual atrophy of the tail, commencing at the apex, next sets in, and results in the complete absorption of this organ.

The long alimentary canal becomes shortened, and the, in the main, herbivorous Tadpole gradually becomes converted into the carnivorous Frog (fig. 82, 6, 7, 8).

The above description of the metamorphosis of the Frog applies fairly to the majority of the Anura, but it is necessary to notice a few of the more instructive divergences from the general type.

In the first place, several forms are known, which are hatched in the condition of the adult. The exact amount of metamorphosis which these forms pass through in the egg is still a matter of some doubt. Hylodes Martinicensis is one of these forms. The larva no doubt acquires within the egg a long tail; but while Bavay[52] states that it is provided with external gills, which however are not covered by an operculum, Peters[53] was unable to see any traces of such structures.

In Pipa Americana, and apparently in Pipa dorsigera also if a distinct species, the larva leaves the cells on the back of the mother in a condition closely resembling the adult. The embryos of both species develop a long tail in the egg, which is absorbed before hatching, and according to Wyman[54] P. Americana is also temporarily provided with gills, which atrophy early.

The larva of Rhinoderma Darwinii is stated by Jiminez de la Espada to be without external gills, and it appears to be hatched while still in the laryngeal pouch of the male. In Nototrema marsupiatum the larvæ are also stated to be without external gills.

Amongst the forms with remarkable developments Pseudis paradoxa deserves especial mention, in that the tadpole of this form attains an immensely greater bulk than the adult; a peculiarity which may be simply a question of nutrition, or may perhaps be explained by supposing that the larva resembles a real ancestral form, which was much larger than the existing Frog.

Another form of perhaps still greater morphological interest is the larva of Dactylethra. The chief peculiarities of this larva (fig. 83) have been summarized by Parker (No. 107, p. 626), from whom I quote the following passage:

a. “The mouth is not inferior in position, suctorial and small, but is very wide like that of the ‘Siluroids and Lophius;’ has an underhung lower jaw, an immensely long tentacle from each upper lip, and possesses no trace of the primordial horny jaws of the ordinary kind.

b. “In conformity with these characters the head is extremely flat or depressed, instead of being high and thick.

c. “There are no claspers beneath the chin.

d. “The branchial orifice is not confined to the left side, but exists on the right side also.

e. “The tail, like the skull, is remarkably chimæroid; it terminates in a long thin pointed lash, and the whole caudal region is narrow and elongated as compared with that of our ordinary Batrachian larvæ.

f. “The fore-limbs are not hidden beneath the opercular fold.”

Although most Anurous embryos are not provided with a sufficient amount of yolk to give rise to a yolk-sack as an external appendage of the embryo, yet in some forms a yolk-sack, nearly as large as that of Teleostei, is developed. One of these forms, Alytes obstetricans, belongs to a well-known European genus allied to Pelobates. The embryos of Pipa dorsigera (Parker) are also provided with a very large yolk-sack, round which they are coiled like a Teleostean embryo. A large yolk-sack is also developed in the embryo of Pseudophryne australis.

The actual complexity of the organization of different tadpoles, and their relative size, as compared with the adult, vary considerably. The tadpoles of Toads are the smallest, Pseudophryne australis excelling in this respect; those of Pseudis are the largest known.

The external gills reach in certain forms, which are hatched in late larval stages, a very great development. It seems however that this development is due to these gills being especially required in the stages before hatching. Thus in Alytes, in which the larva leaves the egg in a stage after the loss of the external gills, these structures reach in the egg a very great development. In Notodelphis ovipara, in which the eggs are carried in a dorsal pouch of the mother, the embryos are provided with long vesicular gills attached to the neck by delicate threads. The fact (if confirmed) that some of the forms which are not hatched till post-larval stages are without external gills, probably indicates that there may be various contrivances for embryonic respiration[55]; and that the external gills only attain a great development in those instances in which respiration is mainly carried on by their means. The external gills of Elasmobranchii are probably, as stated in a previous chapter, examples of secondarily developed structures, which have been produced by the same causes as the enlarged gills of Alytes, Notodelphis, etc.

Urodela. Up to the present time complete observations on the development of the Urodela are confined to the Myctodera[56].

The early stages are in the main similar to those of the Anura. The body of the embryo is, as pointed out by Scott and Osborn, ventrally instead of dorsally flexed. The metamorphosis is much less complete than in the Anura. The larva of Triton may be taken as typical. At hatching, it is provided with a powerful swimming tail bearing a well-developed fin: there are three pairs of gills placed on the three anterior of the true branchial arches.

Between the hyoid and first branchial arch, and between the other branchial arches, slits are developed, there being four slits in all. At the period just before hatching, only three of these have made their appearance. The hyomandibular cleft is not perforated. Stalked suckers, of the same nature as the suckers of the Anura, are formed on the ventral surface behind the mouth. A small opercular fold, developed from the lower part of the hyoid arch, covers over the bases of the gills. The suctorial mouth and the provisional horny beak of the Anura have no counterpart in these larvæ. The skin is ciliated, and the cilia cause a rotation in the egg. Even before hatching, a small rudiment of the anterior pair of limbs is formed, but the hind-limbs are not developed till a later stage, and the limbs do not attain to any size till the larva is well advanced. In the course of the subsequent metamorphosis lungs become developed, and a pulmonary respiration takes the place of the branchial one. The branchial slits at the same time close and the branchiæ atrophy.

The other types of Myctodera, so far investigated, agree fairly with the Newt.

The larva of Amblystoma punctatum (fig. 84) is provided with two very long processes (s), like the suctorial processes in Triton, placed on the throat in front of the external gills. They are used to support the larva when it sinks to the bottom, and have been called by Clarke (No. 98) balancers. On the development of the limbs, these processes drop off. The external gills atrophy about one hundred days after hatching.

It might have been anticipated that the Axolotl, being a larval form of Amblystoma, would agree in development with Amblystoma punctatum. The conspicuous suctorial processes of the latter form are however represented by the merest rudiments in the Axolotl.

The young of Salamandra maculata leave the uterus with external gills, but those of the Alpine Salamander (Salamandra atra) are born in the fully developed condition without gills. In the uterus they pass through a metamorphosis, and are provided (in accordance with the principle already laid down) with very long gill-filaments[57].

Salamandra atra has only two embryos, but there are originally a larger number of eggs (Von Siebold), of which all but two fail to develop, while their remains are used as pabulum by the two which survive. Both species of Salamander have a sufficient quantity of food-yolk to give rise to a yolk-sack.

Spelerpes only develops three post-hyoid arches, between which slits are formed as in ordinary types. Menobranchus and Proteus agree with Spelerpes in the number of post-hyoid arches.

One of the most remarkable recent discoveries with reference to the metamorphosis of the Urodela was made by Dumeril[58]. He found that some of the larvæ of the Axolotl, bred in the Jardin des Plantes, left the water, and in the course of about a fortnight underwent a similar metamorphosis to that of the Newt, and became converted into a form agreeing in every particular with the American genus Amblystoma. During this metamorphosis a pulmonary respiration takes the place of a branchial one, the gills are lost, and the gill slits close. The tail loses its fin and becomes rounded, the colour changes, and alterations take place in the gums, teeth, and lower jaw.

Madame von Chauvin[59] was able, by gradually accustoming Axolotl larvæ to breathe, artificially to cause them to undergo the above metamorphosis.

It seems very possible, as suggested by Weismann[60], that the existing Axolotls are really descendants of Amblystoma forms, which have reverted to a lower stage. In favour of this possibility a very interesting discovery of Filippi’s[61] may be cited. He found in a pond in a marsh near Andermat some examples of Triton alpestris, which, though they had become sexually mature, still retained the external gills and the other larval characters. Similar sexually mature larval forms of Triton tæniatus have been described by Jullien. These discoveries would seem to indicate that it might be possible artificially to cause the Newt to revert to a perennibranchiate condition.

Gymnophiona. The development of the Gymnophiona is almost unknown, but it is certain that some larval forms are provided with a single gill-cleft, while others have external gills.

A gill-cleft has been noticed in Epicrium glutinosum (Müller), and in Cœcilia oxyura. In Cœcilia compressicauda, Peters (No. 108) was unable to find any trace of a gill-cleft, but he observed in the larvæ within the uterus two elongated vesicular gills.

Bibliography.

Amphibia.

(93) Ch. van Bambeke. “Recherches sur le développement du Pélobate brun.” Mémoires couronnés, etc. de l'Acad. roy. de Belgique, 1868.
(94) Ch. van Bambeke. “Recherches sur l'embryologie des Batraciens.” Bulletin de l'Acad. roy. de Belgique, 1875.
(95) Ch. van Bambeke. “Nouvelles recherches sur l'embryologie des Batraciens.” Archives de Biologie, Vol. I. 1880.
(96) K. E. von Baer. “Die Metamorphose des Eies der Batrachier.” Müller’s Archiv, 1834.
(97) B. Benecke. “Ueber die Entwicklung des Erdsalamanders.” Zoologischer Anzeiger, 1880.
(98) S. F. Clarke. “Development of Amblystoma punctatum,” Part I., External. Studies from the Biological Laboratory of the Johns Hopkins University, No. II. 1880.
(99) H. Cramer. “Bemerkungen üb. d. Zellenleben in d. Entwick. d. Froscheies.” Müller’s Archiv, 1848.
(100) A. Ecker. Icones Physiolog. 1851-1859.
(101) A. Götte. Die Entwicklungsgeschichte der Unke. Leipzig, 1875.
(102) C. K. Hoffmann. “Amphibia.” Klassen u. Ordnungen d. Thierreichs, 1873-1879.
(103) T. H. Huxley. Article “Amphibia” in the Encyclopædia Britannica.
(104) A. Moquin-Tandon. “Développement des Batraciens anures.” Annales des Sciences Naturelles, III. 1875.
(105) G. Newport. “On the impregnation of the Ovum in Amphibia” (three memoirs). Phil. Trans. 1851, 1853, and 1854.
(106) W. K. Parker. “On the structure and development of the Skull of the common Frog.” Phil. Trans., CLXI. 1871.
(107) W. K. Parker. “On the structure and development of the Skull of the Batrachia.” Phil. Trans., Vol. CXLVI., Part 2. 1876.
(108) W. C. H. Peters. “Ueber die Entwicklung der Coecilien und besonders von Cœcilia compressicauda.” Berlin Monatsbericht, p. 40, 1874.
(109) W. C. H. Peters. “Ueber die Entwicklung der Coecilien.” Berl. Monatsbericht, p. 483, 1875.
(110) J. L. Prevost and J. B. Dumas. “Deuxième Mém. s. l. génération. Développement de l'œuf d. Batraciens.” Ann. Sci. Nat. II. 1824.
(111) R. Remak. Untersuchungen über die Entwicklung der Wirbelthiere, 1850-1858.
(112) M. Rusconi. Développement de la grenouille commune depuis le moment de sa naissance jusqu'à son état parfait, 1826.
(113) M. Rusconi. Histoire naturelle, développement et métamorphose de la Salamandre terrestre, 1854.
(114) W. B. Scott and H. F. Osborn. “On the early development of the common Newt.” Quart. J. of Micr. Science, Vol. XXIX. 1879.
(115) S. Stricker. “Entwicklungsgeschichte von Bufo cinereus.” Sitzb. der kaiserl. Acad. zu Wien, 1860.
(116) S. Stricker. “Untersuchungen über die ersten Anlagen in Batrachier-Eiern.” Zeitschrift f. wiss. Zoologie, Bd. XI. 1861.

Introduction.

The variations in the character of the embryonic development of the Amniota are far less important than in the case of the Ichthyopsida. There are, it is true, some very special features in the early developmental history of the Mammalia, but apart from these there is such a striking uniformity in the embryos of all the groups that it would, in many cases, be difficult to assign a young embryo to its proper class.

Amongst the Sauropsida the Aves have for obvious reasons received a far fuller share of attention than any other group; and an account of their embryology forms a suitable introduction to this part of our subject. For the convenience of the student many parts of their developmental history will be dealt with at greater length than in the case of the previous groups.

The development of the Aves.

Comparatively few types of Birds have been studied embryologically. The common Fowl has received a disproportionately large share of attention; although within quite recent times the Duck, the Goose, the Pigeon, the Starling, and a Parrot (Melopsittacus undulatus) have also been studied. The result of these investigations has been to shew that the variations in the early development of different Birds are comparatively unimportant. In the sequel the common Fowl will be employed as type, attention being called when necessary to the development of the other forms.

The ovum of the Fowl, at the time when it is clasped by the expanded extremity of the oviduct, is a large yellow body enclosed in a vitelline membrane. It is mainly formed of spherules of food-yolk. Of these there are two varieties; one known as yellow yolk, and the other as white. The white yolk spherules form a small mass at the centre of the ovum, which is continued to the surface by a narrow stalk, and there expands into a somewhat funnel-shaped disc, the edges of which are continued over the surface of the ovum as a delicate layer. The major part of the ovum is formed of yellow yolk. The yellow yolk consists of large delicate spheres, filled with small granules (fig. 85 A); while the white yolk is formed of vesicles of a smaller size than the yellow yolk spheres, in which are a variable number of highly refractive bodies (fig. 85 B).

In addition to the yolk there is present in the ovum a small protoplasmic region, containing the remains of the germinal vesicle, which forms the germinal disc (fig. 86). It overlies the funnel-shaped disc of white yolk, into which it is continued without any marked line of demarcation. It contains numerous minute spherules of the same nature as the smallest white yolk spherules.

Impregnation takes place at the upper extremity of the oviduct.

In its passage outwards the ovum gradually receives its accessory coverings in the form of albumen, shell-membrane, and shell (fig. 87).

The segmentation commences in the lower part of the oviduct, shortly before the shell has begun to be formed. It is meroblastic, being confined to the germinal disc, through the full depth of which however the earlier furrows do not extend. It is mainly remarkable for being constantly somewhat unsymmetrical (Kölliker)—a feature which is not represented in fig. 88, copied from Coste. Owing to the absence of symmetry the cells at one side of the germinal disc are larger than those at the other, but the relations between the disc and the axis of the embryo are not known. During the later stages the segmentation is irregular, and not confined to the surface; and towards its close the germinal disc becomes somewhat lenticular in shape; and is formed of segments, which are smallest in the centre and increase in size towards the periphery (figs. 89 and 90). The superficial segments in the centre of the germinal disc are moreover smaller than those below, and more or less separated as a distinct layer (fig. 90). As development proceeds the segmentation reaches its limits in the centre, but continues at the periphery; and thus eventually the masses at the periphery become of the same size as those at the centre. At the time when the ovum is laid (fig. 91) the uppermost layer of segments has given rise to a distinct membrane, the epiblast, formed of a single row of columnar cells (ep). The lower or hypoblast segments are larger, in some cases very much larger, than those of the epiblast, and are so granular that their nuclei can only with difficulty be seen. They form a somewhat irregular mass, several layers deep, and thicker at the periphery than at the centre: they rest on a bed of white yolk, from which they are in parts separated by a more or less developed cavity, which is probably filled with fluid yolk matter about to be absorbed. In the bed of white yolk nuclei are present, which are of the same character, and have the same general fate, as those in Elasmobranchii. They are generally more numerous in the neighbourhood of the thickened periphery of the blastoderm than elsewhere. Peculiar large spherical bodies are to be found amongst the lower layer cells, which superficially resemble the larger cells around them, and have been called formative cells [vide Foster and Balfour (No. 126)]. Their real nature is still very doubtful, and though some are no doubt true cells, others are perhaps only nutritive masses of yolk. In a surface view the blastoderm, as the segmented germinal disc may now be called, appears as a circular disc; the central part of which is distinguished from the peripheral by its greater transparency, and forms what is known in the later stages as the area pellucida. The narrow darker ring of blastoderm, outside the area pellucida, is the commencing area opaca.

As a result of incubation the blastoderm undergoes a series of changes, which end in the definite formation of three germinal layers, and in the establishment of the chief systems of organs of the embryo. The more important of these changes are accomplished in the case of the common Fowl during the first day and the early part of the second day of incubation.

There is hardly any question in development which has been the subject of so much controversy as the mode of formation of the germinal layers in the common Fowl. The differences in the views of authors have been caused to a large extent by the difficulties of the investigation, but perhaps still more by the fact that many of the observations were made at a time when the methods of making sections were very inferior to those of the present day. The subject itself is by no means of an importance commensurate with the attention it has received. The characters which belong to the formation of the layers in the Sauropsida are secondarily derived from those in the Ichthyopsida, and are of but little importance for the general questions which concern the nature and origin of the germinal layers. In the account in the sequel I have avoided as much as possible discussion of controverted points. My statements are founded in the main on my own observations, more especially on a recent investigation carried on in conjunction with my pupil, Mr Deighton. It is to Kölliker (No. 135), and to Gasser (No. 127) that the most important of the more recent advances in our knowledge are due. Kölliker, in his great work on Embryology, definitely established the essential connection between the primitive streak and the formation of the mesoblast; but while confirming his statement on this head, I am obliged to differ from him with reference to some other points.

Gasser’s work, especially that part of it which relates to the passages leading from the neural to the alimentary canal, which he was the first to discover, is very valuable.

The blastoderm gradually grows in size, and extends itself over the yolk; the growth over the yolk being very largely effected by an increase in the size of the area opaca, which during this process becomes more distinctly marked off from the area pellucida. The area pellucida gradually assumes an oval form, and at the same time becomes divided into a posterior opaque region and an anterior transparent region. The posterior opacity is named by some authors the embryonic shield.

During these changes the epiblast (fig. 92) becomes two layers deep over the greater part of the area pellucida, though still only one cell deep in the area opaca. The irregular hypoblast spheres of the unincubated blastoderm flatten themselves out, and unite into a definite hypoblastic membrane (fig. 92). Between this membrane and the epiblast there remain a number of scattered cells (fig. 92) which cannot however be said to form a definite layer altogether distinct from the hypoblast. They are almost entirely confined to the posterior part of the area pellucida, and give rise to the opacity of that part.

At the edge of the area pellucida the hypoblast becomes continuous with a thickened rim of material, underlying the epiblast, and derived from the original thickened edge of the blastoderm and the subjacent yolk. It is mainly formed of yolk granules, with a varying number of cells and nuclei imbedded in it. It is known as the germinal wall, and is spoken of more in detail on pp. 160 and 161.

The changes which next take place result in the complete differentiation of the embryonic layers, a process which is intimately connected with the formation of the structure known as the primitive streak. The meaning of the latter structure, and its relation to the embryo, can only be understood by comparison with the development of the forms already considered. The most striking peculiarity in the first formation of the embryo Bird, as also in that of the embryos of all Amniota, consists in the fact that they do not occupy a position at the edge of the blastoderm, but are placed near its centre. Behind the embryo there is however a peculiar structure—the primitive streak above mentioned—which is a linear body placed in the posterior region of the blastoderm. This body, the nature of which will be more fully explained in the chapter on the comparative development of Vertebrates, is really a rudimentary part of the blastopore, of the same nature as the linear streak behind the embryo in Elasmobranchii formed by the concrescence of the edges of the blastoderm (vide p. 64); although there is no ontogenetic process in the Amniota, like the concrescence in Elasmobranchii. The relations of the blastopore in Elasmobranchii and Aves is shewn in figs. B and C of the diagram (fig. 93).

In describing in detail the succeeding changes we may at first confine our attention to the area pellucida. As this gradually assumes an oval form the posterior opacity becomes replaced by a very dark median streak, which extends forwards some distance from the posterior border of the area (fig. 94). This is the first rudiment of the primitive streak. In the region in front of it the blastoderm is still formed of two layers only, but in the region of the streak itself the structure of the blastoderm is greatly altered. The most important features in it are represented in fig. 95. This figure shews that the median portion of the blastoderm has become very much thickened (thus producing the opacity of the primitive streak), and that this thickening is caused by a proliferation of rounded cells from the epiblast. In the very young primitive streak, of which fig. 95 is a section, the rounded cells are still continuous throughout with the epiblast, but they form nevertheless the rudiment of the greater part of a sheet of mesoblast, which will soon arise in this region.

In addition to the cells clearly derived from the epiblast, there are certain other cells (vide fig. 95), closely adjoining the hypoblast, which appear to me to be the derivatives of the cells interposed between the epiblast and hypoblast, which gave rise to the posterior opacity in the blastoderm during the previous stage. In my opinion these cells also have a share in forming the future mesoblast.

The number and distribution of these cells is subject to not inconsiderable variations. In a fair number of cases they are entirely congregated along the line of the primitive streak, leaving the sides of the blastoderm quite free. They then form a layer, which can only with difficulty be distinguished from the cells derived from the epiblast by slight peculiarities of staining, and by the presence of a considerable proportion of large granular cells. It is, I believe, by the study of such blastoderms that Kölliker has been led to deny to the intermediate cells of the previous stage any share in the formation of the mesoblast. In other instances, of which fig. 95 is a fairly typical example, they are more widely scattered. To follow with absolute certainty the history of these cells, and to prove that they join the mesoblast is not, I believe, possible by means of sections, and I must leave the reader to judge how far the evidence given in the sequel is sufficient to justify my opinions on this subject.

In the course of further growth the area pellucida soon becomes pyriform, the narrower extremity being the posterior. The primitive streak (fig. 96) elongates considerably, so as to occupy about two-thirds of the length of the area pellucida; but its hinder end in many instances does not extend to the posterior border of the area pellucida. The median line of the primitive streak becomes marked by a shallow groove, known as the primitive groove.

During these changes in external appearance there grow from the sides of the primitive streak two lateral wings of mesoblast cells, which gradually extend till they reach the sides of the area pellucida (fig. 97). The mesoblast still remains attached to the epiblast along the line of the primitive streak. During this extension many sections through the primitive streak give an impression of the mesoblast being involuted at the lips of a fold, and so support the view above propounded, that the primitive streak is the rudiment of the coalesced lips of the blastopore. The hypoblast below the primitive streak is always quite independent of the mesoblast above, though much more closely attached to it in the median line than at the sides. The part of the mesoblast, which I believe to be derived from the primitive hypoblast, can generally be distinctly traced. In many cases, especially at the front end of the primitive streak, it forms, as in fig. 97, a distinct layer of stellate cells, quite unlike the rounded cells of the mesoblastic involution of the primitive streak.

In the region in front of the primitive streak, where the first trace of the embryo will shortly appear, the layers at first undergo no important changes, except that the hypoblast becomes somewhat thicker. Soon, however, as shewn in longitudinal section in fig. 98, the hypoblast along the axial line becomes continuous behind with the front end of the primitive streak. Thus at this point, which is the future hind end of the embryo, the mesoblast, the epiblast, and the hypoblast all unite together; just as they do in all the types of Ichthyopsida.

Shortly afterwards, at a slightly later stage than that represented in fig. 96, an important change takes place in the constitution of the hypoblast in front of the primitive streak. The rounded cells, of which it is at first composed (fig. 98), break up into (1) a layer formed of a single row of more or less flattened elements below—the hypoblast—and (2) into a layer formed of several rows of stellate elements, between the hypoblast and the epiblast—the mesoblast (fig. 99). A separation between these two layers is at first hardly apparent, and before it has become at all well marked, especially in the median line, an axial opaque line makes its appearance in surface views, continued forwards from the front end of the primitive streak, but stopping short at a semicircular fold—the future head-fold—near the front end of the area pellucida. In section (fig. 100) this opaque line is seen to be due to a special concentration of cells in the form of a cord. This cord is the commencement of the notochord (ch). In some instances the commencing notochord remains attached to the hypoblast, while the mesoblast is laterally quite distinct (vide fig. 100), and is therefore formed in the same manner as in most Ichthyopsida; while in other instances, and always apparently in the Goose (Gasser, No. 127), the notochord appears to become differentiated in the already separated layer of mesoblast. In all cases the notochord and the hypoblast below it unite with the front end of the primitive streak; with which also the two lateral plates of mesoblast become continuous.

From what has just been said it is clear that in the region of the embryo the mesoblast originates as two lateral plates split off from the hypoblast, and that the notochord originates as a median plate, simultaneously with the mesoblast, with which it may sometimes be at first continuous.

Kölliker holds that the mesoblast of the region of the embryo is derived from a forward growth from the primitive streak. There is no theoretical objection to this view, and I think it would be impossible to shew for certain by sections whether or not there is a growth such as he describes; but such sections as that represented in fig. 99 (and I have series of similar sections from several embryos) appear to me to be conclusive in favour of the view that the mesoblast of the region of the embryo is to a large extent derived from a differentiation of the primitive hypoblast. I am however inclined to believe that some of the mesoblast cells of the embryonic region have the derivation which Kölliker ascribes to all of them.

As regards the mesoblast of the primitive streak, in a purely objective description like that given above, the greater part of it may fairly be described as being derived from the epiblast. But if it is granted that the primitive streak corresponds with the blastopore, it is obvious to the comparative embryologist that the mesoblast derived from it really originates from the lips of the blastopore, as in so many other cases; and that to describe it, without explanation, as arising from the epiblast, would give an erroneous impression of the real nature of the process.

The differentiation of the embryo may be said to commence with the formation of the notochord and the lateral plates of mesoblast. Very shortly after the formation of these structures the axial part of the epiblast, above the notochord and in front of the primitive streak, which is somewhat thicker than the lateral parts, becomes differentiated into a distinct medullary plate, the sides of which form two folds—the medullary folds—enclosing between them a medullary groove (fig. 101).

In front the two medullary folds meet, while posteriorly they thin out and envelop between them the front end of the primitive streak. On the formation of the medullary folds the embryo assumes a form not unlike that of the embryos of many Ichthyopsida at a corresponding stage. The appearance of the embryo, and its relation to the surrounding parts is somewhat diagrammatically represented in fig. 102. The primitive streak now ends with an anterior swelling (not represented in the figure), and is usually somewhat unsymmetrical. In most cases its axis is more nearly continuous with the left, or sometimes the right, medullary fold than with the medullary groove. In sections its front end appears as a ridge on one side or on the middle of the floor of the widened end of the medullary groove.

The mesoblast and hypoblast, within the area pellucida, do not give rise to the whole of these two layers in the surrounding area opaca; but the whole of the hypoblast of the area opaca, and a large portion of the mesoblast, and possibly even some of the epiblast, take their origin from the peculiar material already spoken of, which forms the germinal wall, and is continuous with the hypoblast at the edge of the area opaca (vide figs. 91, 94, 97, 98, 99, 100).

The exact nature of this material has been the subject of many controversies. Into these controversies it is not my purpose to enter, but subjoined are the results of my own examination. The germinal wall first consists, as already mentioned, of the lower cells of the thickened edge of the blastoderm, and of the subjacent yolk material with nuclei. During the period before the formation of the primitive streak the epiblast extends itself over the yolk, partly, it appears, at the expense of the cells of the germinal wall, and possibly even of cells formed around the nuclei in this part. This mode of growth of the epiblast is very similar to that in the epibolic gastrulas of many Invertebrata, of the Lamprey, etc.; but how far this process is continued in the subsequent extension of the epiblast I am unable to say. The cells of the germinal wall, which are at first well separated from the yolk below, become gradually absorbed in the growth of the hypoblast, and the remaining cells and yolk then become mingled together, and constitute a compound structure, continuous at its inner border with the hypoblast. This structure is the germinal wall usually so described. It is mainly formed of yolk granules with numerous nuclei, and a somewhat variable number of largish cells imbedded amongst them. The nuclei typically form a special layer immediately below the epiblast, some of which are probably enclosed by a definite cell-body. A special mass of nuclei (vide figs. 98 and 100, n) is usually present at the junction of the hypoblast with the germinal wall.

The germinal wall at this stage corresponds in many respects with the granular material, forming a ring below the edge of the blastoderm in Teleostei.

It retains the characters above enumerated till near the close of the first day of incubation, i.e. till several mesoblastic somites have become established. It then becomes more distinctly separated from the subjacent yolk, and its component parts change very considerably in character. The whole wall becomes much less granular. It is then mainly formed of large vesicles, which often assume a palisade-like arrangement, and contain granular balls, spherules of white yolk, and in an early stage a good deal of granular matter (vide fig. 115). These bodies have some resemblance to cells, and have been regarded as such by Kölliker (No. 135) and Virchow (No. 150): they contain however nothing which can be considered as a nucleus. Between them however nuclei[62] may easily be seen in specimens hardened in picric acid, and stained with hæmatoxylin (these nuclei are not shewn in fig. 115). These nuclei are about the same size as those of the hypoblast cells, and are surrounded by a thin layer of granular protoplasm, which is continuous with a mesh-work of granular protoplasm enveloping the above described vesicles. The germinal wall is still continuous with the hypoblast at its edge; and close to the junction of the two the hypoblast at first forms a layer of moderately columnar cells, one or two deep and directly continuous with the germinal wall, and at a later period usually consists of a mass of rounder cells lying above the somewhat abrupt inner edge of the germinal wall.

The germinal wall certainly gives rise to the hypoblast cells, which mainly grow at its expense. They arise at the edge of the area pellucida, and when first formed are markedly columnar, and enclose in their protoplasm one of the smaller vesicles of the germinal wall.

In the later stages (fourth day and onwards) the whole germinal wall is stated to break up into columnar hypoblast cells, each of them mainly formed of one of the vesicles just spoken of. After the commencing formation of the embryo the mesoblast becomes established at the inner edge of the area opaca, between the germinal wall and the epiblast; and gives rise to the tissue which eventually forms the area vasculosa. It seems probable that the mesoblast in this situation is mainly derived from cells formed around the nuclei of the germinal wall, which are usually specially aggregated close below the epiblast. Disse (No. 122) has especially brought evidence in favour of this view, and my own observations also support it.

The mesoblastic somites begin to be formed in the lateral plates of the mesoblast before the closure of the medullary folds. The first somite arises close to the foremost extremity of the primitive streak, but the next is stated to arise in front of this, so that the first formed somite corresponds to the second permanent vertebra[63]. The region of the embryo in front of the second formed somite—at first the largest part of the embryo—is the cephalic region. The somites following the second are formed in the regular manner, from before backwards, out of the unsegmented posterior part of the embryo, which rapidly grows in length to supply the necessary material (fig. 103). As the somites retain during the early stages of development an approximately constant breadth, their number is a fair test of the length of the trunk. With the growth of the embryo the primitive streak is continually carried back, the lengthening of the embryo always taking place between the front end of the primitive streak and the last somite; and during this process the primitive streak undergoes important changes both in itself and in its relation to the embryo. Its anterior thicker part, which is enveloped in the diverging medullary folds, soon becomes distinguished in structure from the part behind this, and placed symmetrically in relation to the axis of the embryo (fig. 103, a.pr), and at the same time the medullary folds, which at first simply diverge on each side of the primitive streak, bend in again and meet behind so as completely to enclose the front part of the primitive streak. The region of the embryo bird, where the medullary folds diverge, is known as the sinus rhomboidalis, though it has no connection with the similarly named structure in the adult. By the time that ten somites are formed the sinus rhomboidalis is completely established, and the medullary groove has become converted into a tube till close up to the front end of the sinus. In the following stages the closure of the medullary canal extends to the sinus rhomboidalis, and the folding off of the hind end of the embryo from the yolk commences. Coincidently with the last-named changes the sides of the front part of the primitive streak become thickened, and give rise to conspicuous caudal swellings; in which the layers of the embryo are indistinguishably fused. The apparently hinder part of the primitive streak becomes, as more particularly explained in the sequel, folded downwards and forwards on the ventral side.

This is a convenient place to notice remarkable appearances which present themselves close to the junction of the neural plate and the primitive streak. These are temporary passages leading from the hinder end of the neural tube into the alimentary canal. They vary somewhat in different species of birds, and it appears that in the same species there may be several openings of the kind, which appear one after the other and then close again. They were first discovered by Gasser (No. 127). In all cases[64] they lead round the posterior end of the notochord, or through the point where the notochord falls into the primitive streak.

If the primitive streak is, as I believe, formed of the lips of the blastopore, there can be but little doubt that these structures are disappearing, and functionless rudiments of the opening of the blastopore, and they thus lend support to my view as to the nature of the primitive streak. That, in part, they correspond with the neurenteric canal of the Ichthyopsida is clear from the detailed statements below. Till their relations have been more fully worked out it is not possible to give a more definite explanation of them.

According to Braun (No. 120) three independent communications are to be distinguished in Birds. These are best developed in the Duck. The first of these is a small funnel-shaped diverticulum leading from the neural groove through the hypoblast. It is visible when eight mesoblastic somites are present, and soon disappears. The second, which is the only one I have myself investigated, is present in the embryo duck with twenty-six mesoblastic somites, and is represented in the series of sections (fig. 104). The passage leads obliquely backwards and ventralwards from the hind end of the neural tube into the notochord, where the latter joins the primitive streak (B). A narrow diverticulum from this passage is continued forwards for a short distance along the axis of the notochord (A, ch). After traversing the notochord, the passage is continued into a hypoblastic diverticulum, which opens ventrally into the future lumen of the alimentary tract (C). Shortly behind the point where the neurenteric passage communicates with the neural tube the latter structure opens dorsally, and a groove on the surface of the primitive streak is continued backwards from it for a short distance (C). The first part of this passage to appear is the hypoblastic diverticulum above mentioned.

This passage does not long remain open, but after its closure, when the tail-end of the embryo has become folded off from the yolk, a third passage is established, and leads round the end of the notochord from the closed medullary canal into the postanal gut. It is shewn diagrammatically in fig. 106, ne, and, as may be gathered from that figure, has the same relations as the neurenteric canal of the Ichthyopsida.

In the goose a passage has been described by Gasser, which appears when about fourteen or fifteen somites are present, and lasts till twenty-three are formed. Behind its opening the medullary canal is continued back as a small diverticulum, which follows the course of the primitive groove and is apparently formed by the conversion of this groove into a canal. It is at first open to the exterior, but soon becomes closed, and then atrophies.

In the chick there is a perforation on the floor of the neural canal, which is not so marked as those in the goose or duck, and never results in a complete contin164uity between the neural and alimentary tracts; but simply leads from the floor of the neural canal into the tissues of the tail swelling, and thence into a cavity in the posterior part of the notochord. The hinder diverticulum of the neural canal along the line of the primitive groove is, moreover, very considerable in the chick, and is not so soon obliterated as in the goose. The incomplete passage in the chick arises when about twelve somites are present. It is regarded by Braun as equivalent to the first formed passage in the duck, but I very much doubt whether there is a very exact equivalence between the openings in different types, and think it more probable that they are variable remnants of a primitive neurenteric canal, which in the ancestors of those forms persisted through the whole period of the early development. The third passage is formed in the chick (Kupffer) during the third day of incubation. In Melopsittacus undulatus the two first communications are stated by Braun (No. 120) to be present at the same time, the one in front of the other.

It is probable, from the above description, that the front portion of the primitive streak in the bird corresponds with that part of the lips of the blastopore in Elasmobranchii which becomes converted into the tail swelling and the lining of the neurentic canal; while the original groove of the front part of the primitive streak appears to be converted into the posterior diverticulum of the neural canal. The hinder part of the primitive streak of the bird corresponds, in a very general way, with the part of the blastopore in Elasmobranchii, which shuts off the embryo from the edge of the blastoderm (vide p. 64), though there is of course no genetic relation between the two structures. When the anterior part of the streak is becoming converted into the tail swelling, the groove of the posterior part gradually shallows and finally disappears. The hinder part itself atrophies from behind forwards, and in the course of the folding off of the embryo from the yolk the part of the blastoderm where it was placed becomes folded in, so as to form part of the ventral wall of the embryo. The apparent hinder part of the primitive streak is therefore in reality the ventral and anterior part[65].

It has generally been maintained that the primitive streak and groove become wholly converted into the dorsal portion of the trunk of the embryo, i.e. into the posterior part of the medullary plate and subjacent structures. This view appears to me untenable in itself, and quite incompatible with the interpretation of the primitive streak given above. To shew how improbable it is, apart from any theoretical considerations, I have compiled two tables of the relative lengths of the primitive streak and the body of the embryo, measured by the number of sections made through them, in a series of examples from the data in Gasser’s important memoir (No. 127). In these tables each horizontal line relates to a single embryo. The first column shews the number of somites, and the second the number of sections through the primitive streak. Where the primitive streak becomes divided into two parts the sections through the two parts are given separately: the left column (A) referring to the anterior part of the streak; the right column (P) to the posterior part. The third column gives the number of sections through the embryo. The first table is for fowl embryos, the second for goose embryos.

No. of Somites.	No. of sections through the Primitive Streak.	No. of sections through the Embryo.
0	29	7
0	45	10
0	39	23
2	30	30
4	30	30
	A P
5 or 6	10 + 17 = 27
8	12 + 20 = 32	48
12	13 + 10 = 23
14	9 + 12 = 21
18	10 +  7 = 17	70
	8 +  4 = 12
	8 +  3 = 11

No. of Somites.	No. of sections through the Primitive Streak.	No. of sections through the Embryo.
0	10	4
0	28	5
0	44	12
2	36	32
4	24	42
	A P
9	10 + 10 = 20	61
14	8 + 10 = 18	68
17	8 +  5 = 13
22	9 +  6 = 15
26	6 +  5 = 11

An inspection of these two tables shews that an actual diminution in the length of the primitive streak takes place just about the time when the first somites are being formed, but there is no ground for thinking that the primitive streak becomes then converted into the medullary plate. Subsequently the primitive streak does not for a considerable time become markedly shorter, and certainly its curtailment is not really sufficient to account for the increased length of the embryo—an increase in length, which (with the exception of the head) takes place entirely by additions at the hind end. At the stage with fourteen somites the primitive streak is still pretty long. In the later stages, as is clearly demonstrated by the tables, the diminution in the length of the primitive streak mainly concerns the posterior part and not that adjoining the embryo.

General history of the germinal layers.

The epiblast. The epiblast of the body of the embryo, though several rows of cells deep, does not become divided into two strata till late in embryonic life; so that the organs of sense formed from the epiblast, which are the same as in the types already described, are not specially formed from an inner nervous stratum. The medullary canal is closed in the same manner as in Elasmobranchii, the Frog, etc., by the simple conversion of an open groove into a closed canal. The closure commences first of all in the region of the mid-brain, and extends rapidly backwards and more slowly forwards. It is completed in the Fowl by about the time that twelve mesoblastic somites are formed.

The mesoblast. The general changes of this layer do not exhibit any features of special interest—the division into lateral and vertebral plates, etc., being nearly the same as in the lower forms.

The hypoblast. The closure of the alimentary canal is entirely effected by a process of tucking in or folding off of the embryo from the yolk-sack. The general nature of the process is seen in the diagrams figs. 105 and 121. The folds by which it is effected are usually distinguished as the head-, the tail- and the lateral folds. The head-fold (fig. 105) is the first to appear; and in combination with the lateral folds gives rise to the anterior part of the mesenteron (D) (including the œsophagus, stomach and duodenum), which by its mode of formation clearly ends blindly in front. The tail-fold, in combination with the two lateral folds, gives rise to the hinder part of the alimentary tract, including the cloaca, which is a true part of the mesenteron. At the junction between the two folds there is present a circular opening leading into the yolk-sack, which becomes gradually narrowed as development proceeds. The opening is completely closed long before the embryo is hatched. Certain peculiarities in reference to the structure of the tail-fold are caused by the formation of the allantois, and are described with the embryonic appendages. The stomodæum and proctodæum are formed by epiblastic invaginations. The communication between the stomodæum and the mesenteron is effected comparatively early (on the 4th day in the chick), while that between the proctodæum and mesenteron does not take place till very late (15th day in the chick). The proctodæum gives rise to the bursa Fabricii, as well as to the anus. Although the opening of the anus is so late in being formed, the proctodæum itself is very early apparent. Soon after the hinder part of the primitive streak becomes tucked in on the ventral side of the embryo, an invagination may be noticed where the tail of the embryo is folded off. This gradually becomes deeper, and finally comes into contact with the hypoblast at the front (primitively the apparent hind) border of the posterior section of the primitive streak. An early stage in the invagination is shewn in the diagram (fig. 106, an). It deserves to be noted that the anus lies some way in front of the blind end of the mesenteron, so that there is in fact a well-developed postanal section of the gut (fig. 106, p.a.g), which corresponds with that in the Ichthyopsida. For a short period, as mentioned above (p. 163), a neurenteric canal is present connecting the postanal gut with the medullary tube in the duck, fowl, and other birds. On the ventral wall of the postanal gut there are at first two prominences. The posterior of these is formed of part of the tail swelling, and is therefore derived from the apparent anterior part of the primitive streak. The anterior is formed from what was originally the apparent posterior part of the primitive streak. The postanal gut becomes gradually less and less prominent, and finally atrophies.

General development of the Embryo.

It will be convenient to take the Fowl as a type for the general development of the Sauropsida.

The embryo occupies a fairly constant position with reference to the egg-shell. Its long axis is placed at right angles to that of the egg, and the broad end of the egg is on the left side of the embryo. The general history of the embryo has already been traced up to the formation of the first formed mesoblastic somites (fig. 107). This stage is usually reached at about the close of the first day. After this stage the embryo rapidly grows in length, and becomes, especially in front and to the sides, more and more definitely folded off from the yolk-sack.

The general appearance of the embryo between the 30th and 40th hours of incubation is shewn in fig. 108 from the upper surface, and in fig. 109 from the lower. The outlines of the embryo are far bolder than during the earlier stages. Fig. 109 shews the nature of the folding, by which the embryo is constricted off from the yolk-sack. The folds are complicated by the fact that the mesoblast has already become split into two layers—a splanchnic layer adjoining the hypoblast and a somatic layer adjoining the epiblast—and that the body cavity between these two layers has already become pretty wide in the lateral parts of the body of the embryo and the area pellucida. The fold by which the embryo is constricted off from the yolk-sack is in consequence a double one, formed of two limbs or laminæ, an inner limb constituted by the splanchnopleure, and an outer limb by the somatopleure. The relation of these two limbs is shewn in the diagrammatic longitudinal section (fig. 105), and in the surface view (fig. 109) the splanchnic limb being shewn at sf and the somatic at so. Between the two limbs, and closely adjoining the splanchnopleure, is seen the heart (ht). At the stage figured the head is well marked off from the trunk, but the first separation between the two regions was effected at an earlier period, on the appearance of the foremost somite (fig. 107). Very shortly after the cephalic region is established, and before the closure of the medullary folds, the anterior part of the neural canal becomes enlarged to form the first cerebral vesicle, from which two lateral diverticula—rudiments of the optic lobes—are almost at once given off (fig. 108, op.v). By the stage figured the cephalic part of the neural canal has become distinctly differentiated into a fore- (f.b), a mid- (m.b) and a hind-brain (h.b); and the hind-brain is often subdivided into successive lobes. In the region of the hind-brain two shallow epiblastic invaginations form the rudiments of the auditory pits (au.p).

A section through the posterior part of the head of an embryo of 30 hours is represented in fig. 110. The enlarged part of the neural tube, forming the hind-brain, is shewn at (hb). It is still connected with the epidermis, and at its dorsal border an outgrowth on each side forming the root of the vagus nerve is present (vg). The notochord (ch) is seen below the brain, and below this again the crescentic foregut (al). The commencing heart (ht), formed at this stage of two distinct tubes, is attached to the ventral side of the foregut.

On the dorsal side of the foregut immediately below the notochord is seen a small body (x) formed as a thickening of the hypoblast. This may possibly be a rudiment of the subnotochordal rod of the Ichthyopsida.

In the trunk (fig. 108) the chief point to be noticed is the complete closure of the neural canal, though in the posterior part, where the open sinus rhomboidalis was situated at an earlier stage, there may still be seen a dilatation of the canal (fig. 108, s.r), on each side of which are the tail swellings; while the mesoblastic somites stop short somewhat in front of it. Underneath the neural canal may be seen the notochord (fig. 109, ch) extending into the head, as far as the base of the mid-brain. At the sides of the trunk are seen the mesoblastic somites (p.v), the outer edges of which mark the boundary between the vertebral and lateral plates. A fainter line can be seen marking off the part of the lateral plates which will become part of the body-wall, from that which pertains to the yolk-sack.

During the latter half of the second day, and during the third day, great progress is made in the folding off of the embryo. Both the head- and tail-ends of the embryo become quite distinct, and the side-folds make such considerable progress that the embryo is only connected with the yolk by a broad stalk. This stalk is double, and consists of an inner splanchnic stalk, continuous with the walls of the alimentary canal, and an outer somatic stalk, continuous with the body-walls of the embryo. The somatic stalk is very much wider than the splanchnic. (Compare fig. 121 E and F, which may be taken as diagrammatic longitudinal and transverse sections of the embryo on the third day.) A change also takes place in the position of the embryo. Up to the third day it is placed symmetrically, on the yolk, with its ventral face downwards. During this day it turns so as partially to lie on its left side. This rotation affects first the head (fig. 111), but in the course of the fourth day gradually extends to the rest of the body (fig. 118). Coincidently with this change in position the whole embryo undergoes a ventral and somewhat spiral flexure.

During the latter part of the second day and during the third day important changes take place in the head. One of these is the cranial flexure. This, which must not be confounded with the curvature of the body just referred to, commences by the bending downwards of the front part of the head round a point which may be considered as the extreme end either of the notochord or of the alimentary canal.

The cranial flexure progresses rapidly, the front-brain being more and more folded down till, at the end of the third day, it is no longer the first vesicle or fore-brain, but the second cerebral vesicle or mid-brain, which occupies the extreme front of the long axis of the embryo. In fact a straight line through the long axis of the embryo would now pass through the mid-brain instead of, as at the beginning of the second day, through the fore-brain, so completely has the front end of the neural canal been folded over the end of the notochord. The commencement of this cranial flexure gives the body of an embryo of the third day somewhat the appearance of a chemist’s retort, the head of the embryo corresponding to the bulb. On the fourth day the flexure is still greater than on the third, but on the fifth and succeeding days it becomes less obvious.

The anterior part of the fore-brain has now become greatly dilated, and may be distinguished from the posterior part as the unpaired rudiment of the cerebral hemispheres. It soon bulges out laterally into two lobes, which do not however become separated by a median partition till a much later period.

Owing to the development of the cerebral rudiment the posterior part of the fore-brain no longer occupies the front position (fig. 111, and 112 FB), and ceases to be the conspicuous object that it was. Inasmuch as its walls will hereafter be developed into the parts surrounding the so-called third ventricle of the brain, it is known as the vesicle of the third ventricle, or the thalamencephalon.

On the summit of the thalamencephalon there may now be seen a small conical projection, the rudiment of the pineal gland, while the centre of the floor is produced into a funnel-shaped process, the infundibulum, which, stretching towards the extreme end of the alimentary canal, joins the pituitary body.

Beyond an increase in size, which it shares with nearly all parts of the embryo, and the change of position which has already been referred to, the mid-brain undergoes no great alterations during the third day. Its sides will ultimately become developed into the corpora bigemina or optic lobes, its floor will form the crura cerebri, and its cavity will be reduced to the narrow canal known as the iter a tertio ad quartum ventriculum and two diverticula leading from this into the optic lobes.

In the hind-brain, or third cerebral vesicle, the roof of the part which lies nearest to the mid-brain, becomes during the third day marked off from the rest by a slight constriction. This distinction, which becomes much more evident later on by a thickening of the walls and roof of the front portion, separates the hind-brain into the cerebellum and the medulla oblongata (fig. 112 Cb and HB). While the walls of the cerebellar portion of the hind-brain become very much thickened as well at the roof as at the sides, the roof of the posterior portion or medulla oblongata thins out into a mere membrane, forming a delicate covering to the cavity of the vesicle (fig. 114 IV), which here becoming broad and shallow with greatly thickened floor and sides, is known as the fourth ventricle, subsequently overhung by the largely-developed posterior portion of the cerebellum.

The third day, therefore, marks the distinct differentiation of the brain into five distinct parts: the cerebral hemispheres, the central masses round the third ventricle, the corpora bigemina, the cerebellum and the medulla oblongata; the original cavity of the neural canal at the same time passing from its temporary division of three single cavities into the permanent arrangement of a series of connected ventricles, viz. the lateral ventricles, the third ventricle, the iter (with a prolongation into the optic lobe on each side), and the fourth ventricle.

By the third day the lens of the eye has become formed by an invagination of the epiblast, and other changes in the eye have taken place. The external opening of the auditory pit is closed before the completion of the third day (fig. 114, RL); and the rudiments of the external parts of the organ of smell have become formed as small pits on the under surface of the fore-brain (fig. 112, N). Like the lens and the labyrinth of the ear, they are formed as invaginations of the external epiblast; unlike them they are never closed up.

During the second and third days there are formed the visceral or branchial clefts, homologous with those of the Ichthyopsida, though never developing branchial processes from their walls.

They are however real clefts or slits passing right through the walls of the throat, and are placed in series on either side across the axis of the alimentary canal, lying not quite at right angles to that axis nor parallel to each other, but converging somewhat to the middle of the throat in front (fig. 112 and fig. 113).

Four in number on either side, the anterior is the first to be formed, the other three following in succession. They originate as pouches of the hypoblast, which meet the epiblast. At the junction of the epiblast and hypoblast an absorption of the tissue is effected, placing the pouches in communication with the exterior.

No sooner has a cleft been formed than its anterior border (i.e. the border nearer the head) becomes raised into a thick lip or fold, the visceral or branchial fold. Each cleft has its own fold on its anterior border, and in addition the posterior border of the fourth or last visceral cleft is raised into a similar fold. There are thus five visceral folds to four visceral clefts (figs. 112 and 113). The last two folds however, and especially the last, are not nearly so thick and prominent as the other three, the second being the broadest and most conspicuous of all. The first fold meets, or nearly meets, its fellow in the middle line in front, but the second falls short of reaching the middle line, and the third, fourth and fifth do so in an increasing degree. Thus in front views of the neck a triangular space with its apex directed towards the head is observed between the ends of the several folds (fig. 113 A).

Into this space the pleuroperitoneal cavity extends, the somatopleure separating from the splanchnopleure along the ends of the folds; and it is here that the aorta plunges into the mesoblast of the body.

The history of these most important visceral folds and clefts will be dealt with in detail hereafter; meanwhile I may say that in the Chick and higher Vertebrates the first three pairs of folds are those which call for most notice.

The first fold on either side, increasing rapidly in size and prominence, does not, like the others, remain single, but sends off in the course of the third day a branch or bud-like process from its upper edge (fig. 113). This branch, starting from near the outer end of the fold, runs forwards and upwards in front of the stomodæum, tending to meet the corresponding branch from the fold on the other side, at a point in the middle line nearer the front of the head than the junction of the main folds (fig. 113, sm). The two branches do not quite meet, being separated by a median process, which at the same time grows down from the extreme front of the head, and against which they abut (fig. 120, k). Between the main folds, which are directed somewhat downwards and their branches which slant upwards, the somewhat lozenge-shaped stomodæum is placed, which, as the folds become more and more prominent, grows deeper and deeper (fig. 120 A). The main folds form the mandibular arch, and their branches the maxillary processes, and the descending process which helps to complete the anterior margin of the stomodæum or oral cavity is called, from the parts which will be formed out of it, the frontonasal process.

In two succeeding pairs of visceral folds, which correspond with the hyoid and first branchial arches of the Ichthyopsida, are developed the parts of the hyoid bone, which will be best considered in connection with the development of the skull. The last two disappear in the Chick without giving rise to any permanent structures. The external opening of the first visceral i.e. hyomandibular cleft becomes closed[66], but the inner part of the cleft, opening into the mouth, gives rise to the Eustachian tube and the tympanic cavity, the latter being formed as a special diverticulum.

Part of the membranous mandibular and hyoid arches form a wall round the dorsal part of the original opening of this cleft, and so give rise to the meatus auditorius externus. At the bottom of this is placed the tympanic membrane, which is probably derived from the tissue which grows over the dorsal part of the opening of the first cleft. It is formed of an external epiblast epithelium, a middle layer of mesoblast, and an internal hypoblastic epithelium.

The general nature of the changes, which take place in the trunk between the commencement of the second half of the second day and the end of the third day, is illustrated by the sections figs. 115, 116, 117.

In the earliest of these sections there is not a trace of a folding off of the embryo from the yolk, and the body walls are quite horizontal. In the second section (fig. 116), from an embryo of about two days, the body walls are already partially inclined, and the splanchnopleure is very distinctly folded inwards. There is a considerable space between the notochord and the hypoblast, which forms the rudiment of the mesentery.

In the third section (fig. 117) the body walls have become nearly vertical, the folding of the splanchnopleure is nearly completed, and it is only for a small region that the alimentary tract is open, by the vitelline duct, to the yolk-sack.

These three sections further illustrate (1) the gradual differentiation of the mesoblastic somites (fig. 115, P.v) into (a) the muscle-plates (figs. 116, ms and 117, m.p), and (b) the tissue to form the vertebral bodies and adjacent connective tissue; (2) the formation of a mass of tissue between the lateral plates and the mesoblastic somites (fig. 115), known as the intermediate cell mass, on the dorsal side of which the Wolffian duct is formed, while the intermediate cell mass itself breaks up into the segmental tubes (fig. 116, st) and connective tissue of the Wolffian body.

Various other features in the development of the vascular system, general mesoblast, etc., are also represented in these sections. It may more especially be noted that there are at first two widely separated dorsal aortæ, which gradually approach (figs. 115 and 116); and meeting first of all in front finally coalesce (figs. 117 and 119) for their whole length.

The general appearance of the embryo of the fourth day may be gathered from fig. 118.

The changes which have taken place consist for the most part in the further development of the parts already present, and do not need to be specified in detail. The most important event of the day is perhaps the formation of the limbs. They appear as outgrowths from a slightly marked lateral ridge (fig. 119, WR), which runs on the level of the lower end of the muscle-plates for nearly the whole length of the trunk. This ridge is known as the Wolffian ridge. The first trace of the limbs can be seen towards the end of the third day; and their appearance at the end of the fourth day is shown in fig. 118, W and HL.

A section through the trunk of the embryo on the fourth day is represented in fig. 119. The section passes through the region of the trunk behind the vitelline duct. The mesentery (M) is very much deeper and thinner than on the previous day. The notochord has become invested by a condensed mesoblastic tissue, which will give rise to the vertebral column. The two dorsal aortæ have now completely coalesced into the single dorsal aorta, and the Wolffian body has reached a far more complete development.

In the course of the fifth day the face begins to assume a less embryonic character, and by the sixth and succeeding days presents distinctive avian characters.

The general changes which take place between the sixth day and the time of hatching do not require to be specified in detail.

Fœtal Membranes.

The Reptilia, Aves and Mammalia are distinguished from the Ichthyopsida by the possession of certain provisional fœtal membranes, known as the amnion and allantois.

As the mode of development of these membranes may be most conveniently studied in the Chick, I have selected this type for their detailed description.

The Amnion. The amnion is a peculiar sack which envelopes and protects the embryo.

At the end of the first day of incubation, when the cleavage of the mesoblast has somewhat advanced, there appears, a little way in front of the semilunar head-fold, a second fold (fig. 102, also fig. 121 C, af and fig. 122, Am), running more or less parallel or rather concentric with the first and not unlike it in general appearance, though differing widely from it in nature. This second fold gives rise to the amnion, and is limited entirely to the somatopleure. Rising up as a semilunar fold with its concavity directed towards the embryo (fig. 121 C, af), as it increases in height it is gradually drawn backwards over the developing head of the embryo. The fold thus covering the head is in due time accompanied by similar folds of somatopleure, starting at some little distance behind the tail, and at some little distance from the side (fig. 121 C, D, E, F, and 116, am). In this way the embryo becomes surrounded by a series of folds of thin somatopleure, which form a continuous wall all round it. All are drawn gradually over the body of the embryo, and at last meet and completely coalesce (fig. 121, H, I, and 117, Am), all traces of their junction being removed. Beneath these united folds there is therefore a cavity, within which the embryo lies (fig. 121 H, ae). This cavity is the cavity of the amnion.

Each fold is necessarily formed of two limbs, both limbs consisting of epiblast and a very thin layer of mesoblast; but in one limb the epiblast looks towards the embryo, while in the other it looks away from it. The space between the two limbs of the fold, as can easily be seen in fig. 121, is really part of the space between the somatopleure and splanchnopleure; it is therefore continuous with the general space, part of which afterwards becomes the pleuroperitoneal cavity of the body, shaded with dots in the figure and marked (pp); so that it is possible to pass from the cavity between the two limbs of the amniotic folds into the cavity which surrounds the alimentary canal. When the several folds meet and coalesce together above the embryo, they unite in such a way that all their inner limbs unite to form a continuous inner membrane or sack, and all their outer limbs a similarly continuous outer membrane or sack. The inner membrane thus built up forms a completely closed sack round the body of the embryo, and is called the amniotic sack, or amnion proper (fig. 121, H, I, &c., a), and the fluid which it afterwards contains is called the amniotic fluid, or liquor amnii. The space between the inner and outer sack is, from the mode of its formation, simply a part of the general cavity found everywhere between somatopleure and splanchnopleure. The outer sack over the embryo lies close under the vitelline membrane, and the cavity between it and the true amnion is gradually extended over the whole yolk-sack.

The actual manner in which the amniotic folds meet is somewhat peculiar (His and Kölliker). The head-fold of the amnion is the earliest formed, and completely covers over the head before the end of the second day. The side and tail folds are later in developing. The side-folds finally meet in the dorsal line, and their coalescence proceeds backwards from the head-fold in a linear direction, till there is only a small opening left over the tail. This also becomes closed early on the third day.

The allantois[67] is essentially a diverticulum of the alimentary tract into which it opens immediately in front of the anus. Its walls are formed of splanchnic mesoblast with blood-vessels, within which is a lining of hypoblast. It becomes a conspicuous object on the third day of incubation, but its first development takes place at an earlier period, and is intimately connected with the formation of the posterior section of the gut.

At the time of the folding in of the hinder end of the mesenteron the splitting of the mesoblast into somatopleure and splanchnopleure has extended up to the border of the hinder division of the primitive streak. As has been already mentioned, the ventral wall of the postanal section of the alimentary tract is formed by the primitive streak. Immediately in front of this is the involution which forms the proctodæum; while the wall of the hindgut in front of the anus owes its origin to a folding in of the splanchnopleure.

The allantois first appears as a protuberance of the splanchnopleure just in front of the anus. This protuberance arises, however, before the splanchnopleure has begun to be tucked in so as to form the ventral wall of the hindgut; and it then forms a diverticulum (fig. 123 A, All) the open end of which is directed forward, while its blind end points somewhat upwards and towards the peritoneal space behind the embryo.

As the hindgut becomes folded in the allantois shifts its position, and forms (figs. 123 B and 124) a rather wide vesicle lying immediately below the hind end of the digestive canal, with which it communicates freely by a still considerable opening; its blind end projects into the pleuroperitoneal cavity below.

Still later the allantois grows forward, and becomes a large spherical vesicle, still however remaining connected with the cloaca by a narrow canal which forms its neck or stalk (fig. 121 G, al). From the first the allantois lies in the pleuroperitoneal cavity. In this cavity it grows forwards till it reaches the front limit of the hindgut, where the splanchnopleure turns back to enclose the yolk-sack. It does not during the third day project beyond this point; but on the fourth day begins to pass out beyond the body of the chick, along the as yet wide space between the splanchnic and somatic stalks of the embryo, on its way to the space between the external and internal folds of the amnion, which it will be remembered is directly continuous with the pleuroperitoneal cavity (fig. 121 K). In this space it eventually spreads out over the whole body of the chick. On the first half of the fourth day the vesicle is still very small, and its growth is not very rapid. Its mesoblast wall still remains very thick. In the latter half of the day its growth becomes very rapid, and it forms a very conspicuous object in a chick of that date (fig. 118, Al). At the same time its blood-vessels become important. It receives its supply of blood from two branches of the iliac arteries known as the allantoic arteries[68], and the blood is brought back from it by two allantoic veins which run along in the body walls (fig. 119) and after uniting into a single trunk fall into the vitelline vein close behind the liver.

Before dealing with the later history of the fœtal membranes, it will be convenient to complete the history of the yolk-sack.

Yolk-Sack. The origin of the area opaca has already been described. It rapidly extends over the yolk underneath the vitelline membrane; and is composed of epiblast and of the hypoblast of the germinal wall continuous with that of the area pellucida, which on the fourth day takes the form of a more or less complete layer of columnar cells[69]. Between the epiblast and hypoblast there is a layer of mesoblast, which does not extend as far as the two other layers. The yolk is completely surrounded by the seventh day.

Towards the end of the first day blood-vessels begin to be developed in the inner part of the mesoblast of the area opaca. Their development is completed on the second day; and the region through which they extend is known as the area vasculosa. The area vasculosa also grows round the yolk, and completely encloses it not long after the area opaca. The part of the blastoderm which thus encloses the yolk forms the yolk-sack. The splitting of the mesoblast gradually extends to the mesoblast of the yolk-sack, and eventually the somatopleure of the sack, which is continuous, it will be remembered, with the outer limb of the amnion, separates completely from the splanchnopleure; and between the two the allantois inserts itself. These features are represented in fig. 121 E, K, and L.

The circulation of the yolk-sack is most important during the third day of incubation. The arrangement of the vessels during that day is shewn in fig. 125.

The blood leaving the body of the embryo by the vitelline arteries (fig. 125, R.Of.A, L.Of.A), which are branches of the dorsal aortæ, is carried to the small vessels and capillaries of the vascular area, a small portion only being appropriated by the pellucid area.

From the vascular area part of the blood returns directly to the sinus venosus by the main lateral trunks of the vitelline veins (R.Of, L.Of), and so to the heart. During the second day these venous trunks join the body of the embryo considerably in front of, that is nearer, the head than the corresponding arterial ones. Towards the end of the third day, owing to the continued lengthening of the heart, the veins and arteries run not only parallel to each other, but almost in the same line, the points at which they respectively join and leave the body being nearly at the same distance from the head.

The rest of the blood brought by the vitelline arteries finds its way into the lateral portions of a venous trunk bounding the vascular area, which is known as the sinus terminalis, S.T., and there divides on each side into two streams. Of these, the two which, one on either side, flow backward, meet at a point about opposite to the tail of the embryo, and are conveyed along a distinct vein which, running straight forward parallel to the axis of the embryo, empties itself into the left vitelline vein. The two forward streams reaching a gap in the front part of the sinus terminalis fall into either one, or in some cases two veins, which run straight backwards parallel to the axis of the embryo, and so reach the roots of the heart. When one such vein only is present it joins the left vitelline trunk; where there are two they join the left and right vitelline trunks respectively. The left vein is always considerably larger than the right; and the latter when present rapidly gets smaller and speedily disappears. After the third day, although the vascular area goes on increasing in size until it finally all but encompasses the yolk, the prominence of the sinus terminalis becomes less and less.

The fœtal membranes and the yolk-sack may conveniently be treated of together in the description of their later changes and final fate.

On the sixth and seventh days they exhibit changes of great importance.

The amnion, at its complete closure on the fourth day, very closely invested the body of the chick: the true cavity of the amnion was then therefore very small. On the fifth day fluid begins to collect in the cavity, and raises the membrane of the amnion to some distance from the embryo. The cavity becomes still larger by the sixth day, and on the seventh day is of very considerable dimensions, the fluid increasing with it. On the sixth day Von Baer observed movements of the embryo, chiefly of the limbs; he attributes them to the stimulation of the cold air on opening the egg. By the seventh day very obvious movements begin to appear in the amnion itself; slow vermicular contractions creeping rhythmically over it. The amnion in fact begins to pulsate slowly and rhythmically, and by its pulsation the embryo is rocked to and fro in the egg. This pulsation is probably due to the contraction of involuntary muscular fibres, which seem to be present in the attenuated portion of the mesoblast, forming part of the amniotic fold. Similar movements are also seen in the allantois at a considerably later period.

The growth of the allantois has been very rapid, and it forms a flattened bag, covering the right side of the embryo, and rapidly spreading out in all directions between the primitive folds of the amnion, that is, between the amnion proper and the false amnion or serous envelope. It is filled with fluid, so that in spite of its flattened form its opposite walls are distinctly separated from each other.

The vascular area has become still further extended than on the fifth day, but with a corresponding loss in the definite character of its blood-vessels. The sinus terminalis has indeed by the end of the seventh day lost all its previous distinctness; and the vessels which brought back the blood from it to the heart are no longer to be seen.

Both the vitelline arteries and veins now pass to and from the body of the chick as single trunks, assuming more and more the appearance of being merely branches of the mesenteric vessels.

The yolk is still more fluid than on the previous day, and its bulk has (according to von Baer) increased. This can only be due to its absorbing the white of the egg, which indeed is diminishing rapidly.

During the eighth, ninth, and tenth days, the amnion does not undergo any very important changes. Its cavity is still filled with fluid, and on the eighth day its pulsations are at their height, henceforward diminishing in intensity.

The splitting of the mesoblast has now extended to the outer limit of the vascular area, i.e. over about three-quarters of the yolk-sack. The somatopleure at this point is continuous (as can be easily seen by reference to fig. 121) with the original outer fold of the amnion. It thus comes about that the further splitting of the mesoblast merely enlarges the cavity in which the allantois lies. The growth of this organ keeps pace with that of the cavity in which it is placed. Spread out over the greater part of the yolk-sack as a flattened bag filled with fluid, it now serves as the chief organ of respiration. It is indeed very vascular and a marked difference may be observed between the colour of the blood in the outgoing and the returning vessels.

The yolk now begins to diminish rapidly in bulk. The yolk-sack becomes flaccid, and on the eleventh day is thrown into a series of internal folds, abundantly supplied by large venous trunks. By this means the surface of absorption is largely increased, and the yolk is more and more rapidly taken up by the blood-vessels, and in a partially assimilated condition transferred to the body of the embryo[70].

By the eleventh day the abdominal parietes, though still much looser and less firm than the walls of the chest, may be said to be definitely established; and the loops of intestine, which have hitherto been hanging down into the somatic stalk, are henceforward confined within the cavity of the abdomen. The body of the embryo is therefore completed; but it still remains connected with its various appendages by a narrow somatic umbilicus, in which run the stalk of the allantois and the solid cord suspending the yolk-sack.

The cleavage of the mesoblast is still progressing, and the yolk is completely invested by a splanchnopleural sack.

The allantois meanwhile spreads out rapidly, and lies over the embryo close under the shell, being separated from the shell membrane by nothing more than the attenuated serous envelope, formed out of the outer primitive fold of the amnion and the remains of the vitelline membrane. With this membrane the allantois partially coalesces, and in opening an egg at the later stages of incubation, unless care be taken, the allantois is in danger of being torn in the removal of the shell-membrane. As the allantois increases in size and importance, the allantoic vessels are correspondingly developed.

On about the sixteenth day, the white having entirely disappeared, the cleavage of the mesoblast is carried right over the pole of the yolk opposite the embryo, and is thus completed (fig. 121). The yolk-sack now, like the allantois which closely wraps it all round, lies loose in a space bounded outside the body by the serous membrane, and continuous with the pleuroperitoneal cavity of the body of the embryo. Deposits of urates now become abundant in the allantoic fluid.

The loose and flaccid walls of the abdomen enclose a space which the empty intestines are far from filling, and on the nineteenth day the yolk-sack, diminished greatly in bulk but still of some considerable size, is withdrawn through the somatic stalk into the abdominal cavity, which it largely distends. Outside the embryo there now remains nothing but the highly vascular allantois and the bloodless serous membrane and amnion. The amnion, whose fluid during the later days of incubation rapidly diminishes, is continuous at the umbilicus with the body-walls of the embryo. The serous membrane (or outer primitive amniotic fold) is, by the completion of the cleavage of the mesoblast and the withdrawal of the yolk-sack, entirely separated from the embryo. The cavity of the allantois, by means of its stalk passing through the umbilicus, is of course continuous with the cloaca.

When the chick is about to be hatched it thrusts its beak through the egg-membranes and begins to breathe the air contained in the air chamber. Thereupon the pulmonary circulation becomes functionally active, and at the same time blood ceases to flow through the allantoic arteries. The allantois shrivels up, the umbilicus becomes completely closed, and the chick, piercing the shell at the broad end of the egg with repeated blows of its beak, casts off the dried remains of allantois, amnion and serous membrane, and steps out into the world.

Bibliography.

(117) K. E. von Baer. Ueb. Entwicklungsgeschichte d. Thiere. Königsberg, 1828-1837.
(118) F. M. Balfour. “The development and growth of the layers of the Blastoderm,” and “On the disappearance of the Primitive Groove in the Embryo Chick.” Quart. J. of Micros. Science, Vol. XIII. 1873.
(119) M. Braun. “Die Entwicklung d. Wellenpapagei’s.” Part I. Arbeit. d. zool.-zoot. Instit. Würzburg. Vol. V. 1879.
(120) M. Braun. “Aus d. Entwick. d. Papageien; I. Rückenmark; II. Entwicklung d. Mesoderms; III. Die Verbindungen zwischen Rückenmark u. Darm bei Vögeln.” Verh. d. phys.-med. Ges. zu Würzburg. N. F. Bd. XIV. and XV. 1879 and 1880.
(121) J. Disse. “Die Entwicklung des mittleren Keimblattes im Hühnerei.” Archiv für mikr. Anat., Vol. XV. 1878.
(122) J. Disse. “Die Entstehung d. Blutes u. d. ersten Gefässe im Hühnerei.” Archiv f. mikr. Anat., Vol. XVI. 1879.
(123) Fr. Durante. “Sulla struttura della macula germinativa delle uova di Gallina.” Ricerche nel Laboratorio di Anatomia della R. Università di Roma.
(124) E. Dursy. Der Primitivstreif des Hühnchens. 1867.
(125) M. Duval. “Etude sur la ligne primitive de l'embryon de Poulet.” Annales des Sciences Naturelles, Vol. VII. 1879.
(126) M. Foster and F. M. Balfour. Elements of Embryology. Part I. London, 1874.
(127) Gasser. “Der Primitivstreifen bei Vogelembryonen.” Schriften d. Gesell. zur Beförd. d. gesammten Naturwiss. zu Marburg, Vol. II. Supplement I. 1879.
(128) A. Götte. “Beiträge zur Entwicklungsgeschichte d. Wirbelthiere. II. Die Bildung d. Keimblatter u. d. Blutes im Hühnerei.” Archiv für mikr. Anat., Vol. X. 1874.
(129) V. Hensen. “Embryol. Mitth.” Archiv f. mikr. Anat., Vol. III. 1867.
(130) W. His. Untersuch. üb. d. erste Anlage d. Wirbelthierleibes. Leipzig, 1868.
(131) W. His. Unsere Körperform und das physiol. Problem ihrer Entstehung. Leipzig, 1875.
(132) W. His. “Der Keimwall des Hühnereies u. d. Entstehung d. parablastischen Zellen.” Zeit. f. Anat. u. Entwicklungsgeschichte. Bd. I. 1876.
(133) W. His. “Neue Untersuchungen üb. die Bildung des Hühnerembryo I.” Archiv f. Anat. u. Phys. 1877.
(134) E. Klein. “Das mittlere Keimblatt in seiner Bezieh. z. Entwick. d. ers. Blutgefässe und Blutkörp. im Hühnerembryo.” Sitzungsber. Wien. Akad., Vol. LXIII. 1871.
(135) A. Kölliker. Entwicklungsgeschichte d. Menschen u. d. höheren Thiere. Leipzig, 1879.
(136) C. Kupffer. “Die Entsteh. d. Allantois u. d. Gastrula d. Wirbelth.” Zoolog. Anzeiger, Vol. II. 1879, pp. 520, 593, 612.
(137) C. Kupffer and B. Benecke. “Photogramme z. Ontogenie d. Vögel.” Nov. Act. d. k. Leop.-Carol.-Deutschen Akad. d. Naturforscher, Vol. XLI. 1879.
(138) J. Oellacher. “Untersuchungen über die Furchung u. Blatterbildung im Hühnerei.” Stricker’s Studien. 1870.
(139) C. H. Pander. Beiträge z. Entwick. d. Hühnchens im Eie. Würzburg, 1817.
(140) A. Rauber. “Ueber die Embryonalanlage des Hühnchens.” Centralblatt für d. medic. Wissenschaften. 1874-75.
(141) A. Rauber. Ueber die Stellung des Hühnchens im Entwicklungsplan. 1876.
(142) A. Rauber. “Primitivrinne und Urmund. Beiträge zur Entwicklungsgeschichte des Hühnchens.” Morphol. Jahrbuch, B. II. 1876.
(143) A. Rauber. Primitivstreifen und Neurula der Wirbelthiere in normaler und pathologischer Beziehung. 1877.
(144) R. Remak. Untersuch. üb. d. Entwicklung d. Wirbelthiere. Berlin, 1850-55.
(145) S. L. Schenk. “Beiträge z. Lehre v. d. Organanlage im motorischen Keimblatt.” Sitz. Wien. Akad., Vol. LVII. 1860.
(146) S. L. Schenk. “Beiträge z. Lehre v. Amnion.” Archiv f. mikr. Anat., Vol. VII. 1871.
(147) S. L. Schenk. Lehrbuch d. vergleich. Embryol. d. Wirbelthiere. Wien, 1874.
(148) S. Stricker. “Mittheil. üb. d. selbstständigen Bewegungen embryonaler Zellen.” Sitz. Wien. Akad., Vol. XLIX. 1864.
(149) S. Stricker. “Beiträge zur Kenntniss des Hühnereies.” Wiener Sitzungsber., Vol. LIV. 1866.
(150) H. Virchow. Ueber d. Epithel d. Dottersackes im Hühnerei. Inaug. Diss. Berlin, 1875.
(151) W. Waldeyer. “Ueber die Keimblätter und den Primitivstreifen bei der Entwicklung des Hühnerembryo.” Zeitschrift für rationelle Medicin. 1869.
(152) C. F. Wolff. Theoria generationis. Halæ, 1759.
(153) C. F. Wolff. Ueb. d. Bildung d. Darmcanals im bebrüteten Hühnchen. Halle, 1812.

The formation of the germinal layers in the Reptilia is very imperfectly known. The Lizard has been studied in this respect more completely than other types, and there are a few scattered observations on Turtles and Snakes.

The ovum has in all Reptilia a very similar structure to that in Birds. Impregnation is effected in the upper part of the oviduct, and the early stages of development invariably take place in the oviduct. A few forms are viviparous, viz. some of the blindworms amongst Lizards (Anguis, Seps), and some of the Viperidæ and Hydrophidæ amongst the Serpents. In the majority of cases, however, the eggs are laid in moist earth, sand, &c. Around the true ovum an egg-shell (of the same general nature as that in birds, though usually soft), and a variable quantity of albumen, are deposited in the oviduct. The extent to which development has proceeded in the oviparous forms before the eggs are laid varies greatly in different species.

The general features of the development (for a knowledge of which we are mainly indebted to Rathke’s beautiful memoirs), the structure of the amnion and allantois, &c. are very much the same as in Birds.

The Lizards will be taken as type of the class, and a few noteworthy points in the development of other groups will be dealt with at the close of the Chapter. The following description, taken in the main from my own observations, applies to Lacerta muralis.

The segmentation is meroblastic, and similar to that in Birds. At its close the resulting blastoderm becomes divided into two layers, a superficial epiblast formed of a single row of cells, and a layer below this several rows deep. Below this layer fresh segments continue for some time to be added to the blastoderm from the subjacent yolk.

The blastoderm, which is thickened at its edge, spreads rapidly over the yolk. Shortly before the yolk is half enclosed a small embryonic shield (area pellucida) makes its appearance near the centre of the blastoderm. The embryonic shield is mainly distinguished from the remainder of the blastoderm by the more columnar character of its constituent epiblast cells. It is somewhat pyriform in shape, the narrower end corresponding with the future posterior end of the embryo. At the hind end of the shield a somewhat triangular primitive streak is formed, consisting of epiblast continuous below with a great mass of rounded mesoblast cells, probably mainly formed, as in the bird, by a proliferation of the epiblast. To this mass of cells the hypoblast is also partially adherent. At the front end of the streak an epiblastic involution appears, which soon becomes extended into a passage open at both extremities, leading obliquely forwards through the epiblast to the space below the hypoblast. The walls of the passage are formed of a layer of columnar cells continuous both with epiblast and hypoblast. In front of the primitive streak the body of the embryo becomes first differentiated by the formation of a medullary plate; and at the same time there grows out from the primitive streak a layer of mesoblast, which spreads out in all directions between the epiblast and hypoblast. In the region of the embryo the mesoblast plate is stated by Kupffer and Benecke to be continuous across the middle line, but this appears very improbable. In a slightly later stage the medullary plate becomes marked by a shallow groove, and the mesoblast of the embryo is then undoubtedly constituted of two lateral plates, one on each side of the median line. In the median line the notochord arises as a ridge-like thickening of the hypoblast, which is continued posteriorly into the front wall of the passage mentioned above.

The notochord does not long remain attached to the hypoblast, and the separation between the two is already effected for the greater part of the length of the embryo by the stage represented in fig. 129. Fig. 126 represents a series of sections through this embryo.

In a section (A) through the trunk of the embryo a short way in front of the primitive streak, there is a medullary plate with a shallow groove (mg), well-developed mesoblastic plates (mep), already divided into somatic and splanchnic layers, and a completely formed notochord independent of the hypoblast (hy). In the next section (B), taken just in front of the primitive streak, the notochord is attached to the hypoblast, and the medullary groove is deeper; while in the section following (C), which passes through the front border of the primitive streak, the notochord and hypoblast have become fused with the epiblast. The section behind (D) shews the neurenteric passage leading through the floor of the medullary groove and through the hypoblast (ne). On the right side the mesoblastic plate has become continuous with the walls of the passage. The last section (E) passes through the front part of the primitive streak behind the passage. The mesoblast, epiblast, and to some extent the hypoblast, are now fused together in the axial line, and in the middle of the fused mass is seen a narrow diverticulum (ne) which is probably equivalent to the posterior diverticulum of the neural canal in Birds (vide p. 164).

The general features of the stage will best be understood by an examination of the diagrammatic longitudinal section represented in fig. 127. In front is shewn the amnion (am), growing over the head of the embryo. The notochord (ch) is seen as an independent cord for the greater part of the length of the embryo, but falls into the hypoblast shortly in front of the neurenteric passage. The neurenteric passage is shewn at ne, and behind it is the front part of the primitive streak.

It is interesting to notice the remarkable relations of the notochord to the walls of the neurenteric passage. More or less similar relations are also well marked in the case of the goose and the fowl, and support the conclusion, deducible from the lower forms of Vertebrata, that the notochord is essentially hypoblastic.

The passage at the front end of the primitive streak forms the posterior boundary of the medullary plate, though the medullary groove is not at first continued back to it. The anterior wall of this passage connects together the medullary plate and the notochordal ridge of the hypoblast. In the stage represented in fig. 126 and 129 the medullary groove has become continued back to the opening of the passage, which thus becomes enclosed in the medullary folds, and forms a true neurenteric passage[71].

It will be convenient at this point to say a few words as to what is known of the further fate of the neurenteric canal, and the early development of the allantois. According to Strahl, who has worked on Lacerta vivipara, the canal gradually closes from below upwards, and is obliterated before the completion of the neural canal. The hind end of the alimentary tract appears also to become a closed canal before this stage.

In Lacerta muralis the history appears to be somewhat different, and it is more especially to be noticed that in this species the hindgut does not become closed till considerably after the completion of the neural canal. In a stage shortly after that last described, the neurenteric passage becomes narrower. The next stage which I have observed is considerably later. The neural canal has become completely closed, and the flexure of the embryo has already made its appearance. There is still a well-developed, though somewhat slit-like, neurenteric passage, but from the analogy of birds, it is not impossible that it may have in the meantime closed up and opened again. It has, in any case, the same relations as in the previous stage.

It leads from the end of the medullary canal (at the point where its walls are continuous with the cells of the primitive streak) round the end of the notochord, which here becomes continuous with the medullary cord, and so through the hypoblast. The latter layer is still a flat sheet without any lateral infolding; but it gives rise, behind the neurenteric passage, to a blind posteriorly directed diverticulum, placed in the body cavity behind the embryo, and opening at the ventral face of the apparent hind end of the primitive streak. There is very little doubt that this diverticulum is the commencing allantois.

At a somewhat later stage the arrangement of these parts has undergone some changes. Their relations are shewn in the sections represented in fig. 128.

The foremost section (A) passes through the alimentary opening of the neurenteric passage (ne). Above this opening the section passes through the primitive streak (pr) close to its junction with the walls of the medullary canal. The hypoblast is folded in laterally, but the gut is still open below. The amnion is completely established. In the next section figured (B), the fourth of my series, the gut is completely closed in; and the mesoblast has united laterally with the axial tissue of the primitive streak. Vessels to supply the allantois are shewn at v.

The three following sections are not figured, but they present the same features as B, except that the primitive streak gets rapidly smaller, and the lumen of the gut narrower. The section following (C) represents, I believe, only the stalk of the allantoic diverticulum. This diverticulum appears to be formed as usual of hypoblast (hy) enveloped by splanchnic mesoblast (me), and projects into the section of the body cavity present behind the embryo. Its position in the body cavity is the cause of its somewhat peculiar appearance in the figure. Had the whole section been represented the allantois would have been enclosed in a space between the serous membrane (se) and a layer of splanchnic mesoblast below which has also been omitted in fig. B[72]. It still points directly backwards, as it primitively does in the chick, vide fig. 123 A, and Gasser, No. 127, Pl. V. figs. 1 and 2. I do not understand the apparently double character of the lumen of the allantois. In the next section (not figured) the lumen of the allantoic stalk is larger, but still apparently double, while in the last section (D) the lumen is considerably enlarged and single. The neurenteric canal appears to close shortly after the stage last described, though its further history has not been followed in detail.

General development of the Embryo.

The formation of the embryo commences with the appearance of the medullary plate, the sides of which soon grow up to form the medullary folds. The medullary groove is developed anteriorly before any trace of it is visible behind. In a general way the closure of the groove takes place as in Birds, but the anterior part of the body is very early folded off, sinks into the yolk, and becomes covered over by the amnion as by a hood (figs. 127 and 129). All this takes place before the closure of the medullary canal; and the changes of this part are quite concealed from view.

The closure of the medullary canal commences in the neck, and extends forwards and backwards; and the whole region of the brain becomes closed in, while the groove is still largely open behind.

The later stages in the development of the Lacertilian embryo do not require a detailed description, as they present the closest analogy with those already described for Aves. The embryo soon turns on to its left side; and then, becoming continuously folded off from the yolk, passes through the series of changes of form with which the reader is already familiar. An advanced embryo is represented in fig. 130. The early development and great length of the tail, which is spirally coiled on the ventral surface, is a special feature to which the attention of the reader may be called.

Embryonic Membranes and Yolk-Sack.

The early development of the cephalic portion of the amnion has already been alluded to. The first traces of it become apparent while the medullary groove is still extremely shallow. The medullary plate in the region of the head forms an axial strip of a thickish plate of epiblast. The edge of this plate coincides with the line of the amniotic fold, and as this fold rises up the two sides of the plate become bent over the embryo and give rise to the inner limb of the amnion or amnion proper. The section (fig. 127), representing the origin of the amniotic hood of the head, shews very well how the space between the two limbs of the amnion is continuous with the body cavity. The amnion very early completely encloses the embryo (fig. 128 A and B), and its external limb or serous membrane, after separating from the true amnion, soon approaches and fuses with the vitelline membrane.

The first development of the allantois as a diverticulum of the hypoblast covered by splanchnic mesoblast, at the apparent posterior end of the primitive streak, has been described on p. 207. The allantois continues for some time to point directly backwards; but gradually assumes a more ventral direction; and, as it increases in size, extends into the space between the serous membrane and amnion, eventually to form a large, highly vascular, flattened sack immediately below the serous membrane.

The Yolk-Sack. The blastoderm spreads in the Lizard with very great rapidity over the yolk to form the yolk-sack. The early appearance of the area pellucida, or as it has been called by Kupffer and Benecke the embryonic shield, has already been noted. Outside this a vascular area, which has the same function as in the chick, is not long in making its appearance. In all Reptilia the vascular channels which arise in the vascular area, and the vessels carrying the blood to and from the vascular area, are very similar to those in the chick. In the Snake the sinus terminalis never attains so conspicuous a development and in Chelonia the stage with a pair of vitelline arteries is preceded by a stage in which the vascular area is supplied, as it permanently is in many Mammals, by numerous transverse arterial trunks, coming off from the dorsal aorta (Agassiz, No. 164). The vascular area gradually envelops the whole yolk, although it does so considerably more slowly than the general blastoderm.

Ophidia. There is, as might have been anticipated, a very close correspondence in general development between the Lacertilia and Ophidia. The embryos of all the Amniota are, during part of their development, more or less spirally coiled about their long axis. This is well marked in the chick of the third day; it is still more pronounced in the Lizard (fig. 130); but it reaches its maximum in the Snake. The whole Snake embryo has at the time when most coiled (Dutrochet, Rathke) somewhat the form of a Trochus. The base of the spiral is formed by the head, while the majority of the coils are supplied by the tail. There are in all at this stage seven coils, and the spiral is right-handed.

Another point, which deserves notice in the Snake, is the absence in the embryo of all external trace of the limbs. It might have been anticipated, on the analogy of the branchial arches, that rudiments of the limbs would be preserved in the embryo even when limbs were absent in the adult. Such, however, is not the case. It is however very possible that rudiments of the branchial arches and clefts have been preserved because these structures were functional in the larva (Amphibia) after they ceased to have any importance in the adult; and that the limbs have disappeared even in the embryo because in the course of their gradual atrophy there was no advantage to the organism in their being specially preserved at any period of life[74].

Chelonia[75]. In their early development the Chelonia resemble, so far as is known, the Lacertilia. The amnion arises early, and soon forms a great cephalic hood. Before development has proceeded very far the embryo turns over on to its left side. The tail in many species attains a very considerable development (fig. 133). The chief peculiarity in the form of the embryo (figs. 131, 132, and 133) is caused by the development of the carapace. The first rudiment of the carapace appears in the form of two longitudinal folds, extending above the line of insertion of the fore- and hind-limbs, which have already made their appearance (fig. 131). These folds are subsequently prolonged so as to mark out the area of the carapace on the dorsal surface. On the surface of this area there are formed the horny plates (tortoise shell), and in the mesoblast below the bony elements of the carapace (figs. 132 and 133).

Immediately after hatching the yolk-sack becomes withdrawn into the body; while the external part of the allantois shrivels up.

Bibliography.

General.

(154) C. Kupffer and Benecke. Die erste Entwicklung am Ei d. Reptilien. Königsberg, 1878.
(155) C. Kupffer. “Die Entstehung d. Allantois u. d. Gastrula d. Wirbelthiere.” Zoologischer Anzeiger, Vol. II. 1879, pp. 520, 593, 612.

Lacertilia.

(156) F. M. Balfour. “On the early Development of the Lacertilia, together with some observations, etc.” Quart. J. of Micr. Science, Vol. XIX. 1879.
(157) Emmert u. Hochstetter. “Untersuchung üb. d. Entwick. d. Eidechsen in ihren Eiern.” Reil’s Archiv, Vol. X. 1811.
(158) M. Lereboullet. “Développement de la Truite, du Lézard et du Limnée. II. Embryologie du Lézard.” An. Sci. Nat., Ser. IV., Vol. XXVII. 1862.
(159) W. K. Parker. “Structure and Devel. of the Skull in Lacertilia.” Phil. Trans., Vol. 170, p. 2. 1879.
(160) H. Strahl. “Ueb. d. Canalis myeloentericus d. Eidechse.” Schrift. d. Gesell. z. Beför. d. gesam. Naturwiss. Marburg. July 23, 1880.

Ophidia.

(161) H. Dutrochet. “Recherches s. l. enveloppes du fœtus.” Phil. Trans., Paris, Vol. VIII. 1816.
(162) W. K. Parker. “On the skull of the common Snake.” Phil. Trans., Vol. 169, Part II. 1878.
(163) H. Rathke. Entwick. d. Natter. Königsberg, 1839.

Chelonia.

(164) "L. Agassiz. Contributions to the Natural History of the United States, Vol. II. 1857. Embryology of the Turtle.
(165) W. K. Parker. “On the development of the skull and nerves in the green Turtle.” Proc. of the Roy. Soc., Vol. XXVIII. 1879. Vide also Nature, April 14, 1879, and Challenger Reports, Vol. I. 1880.
(166) H. Rathke. Ueb. d. Entwicklung d. Schildkröten. Braunschweig, 1848.

Crocodilia.

(167) H. Rathke. Ueber die Entwicklung d. Krokodile. Braunschweig, 1866.

The classical researches of Bischoff on the embryology of several mammalian types, as well as those of other observers, have made us acquainted with the general form of the embryos of the Placentalia, and have shewn that, except in the earliest stages of development, there is a close agreement between them. More recently Hensen, Schäfer, Kölliker, Van Beneden and Lieberkühn have shed a large amount of light on the obscurer points of the earliest developmental periods, especially in the rabbit. For the early stages the rabbit necessarily serves as type; but there are grounds for thinking that not inconsiderable variations are likely to be met with in other species, and it is not at present easy to assign to some of the developmental features their true value. We have no knowledge of the early development of the Ornithodelphia or Marsupialia.

The ovum on leaving the ovary is received by the fimbriated extremity of the Fallopian tube, down which it slowly travels. It is still invested by the zona radiata, and in the rabbit an albuminous envelope is formed around it in its passage downwards. Impregnation takes place in the upper part of the Fallopian tube, and is shortly followed by the segmentation, which is remarkable amongst the Amniota for being complete.

Although this process (the details of which have been made known by the brilliant researches of Ed. van Beneden) has already been shortly dealt with as it occurs in the rabbit (Vol. II. p. 98) it will be convenient to describe it again with somewhat greater detail.

The ovum first divides into two nearly equal spheres, of which one is slightly larger and more transparent than the other. The larger sphere and its products will be spoken of as the epiblastic spheres, and the smaller one and its products as the hypoblastic spheres, in accordance with their different destinations.

Both the spheres are soon divided into two, and each of the four so formed into two again; and thus a stage with eight spheres ensues. At the moment of their first separation these spheres are spherical, and arranged in two layers, one of them formed of the four epiblastic spheres, and the other of the four hypoblastic. This position is not long retained, but one of the hypoblastic spheres passes to the centre; and the whole ovum again takes a spherical form.

In the next phase of segmentation each of the four epiblastic spheres divides into two, and the ovum thus becomes constituted of twelve spheres, eight epiblastic and four hypoblastic. The epiblastic spheres have now become markedly smaller than the hypoblastic.

The four hypoblastic spheres next divide, giving rise, together with the eight epiblastic spheres, to sixteen spheres in all; which are nearly uniform in size. Of the eight hypoblastic spheres four soon pass to the centre, while the eight superficial epiblastic spheres form a kind of cup partially enclosing the hypoblastic spheres. The epiblastic spheres now divide in their turn, giving rise to sixteen spheres which largely enclose the hypoblastic spheres. The segmentation of both epiblastic and hypoblastic spheres continues, and in the course of it the epiblastic spheres spread further and further over the hypoblastic, so that at the close of segmentation the hypoblastic spheres constitute a central solid mass almost entirely surrounded by the epiblastic spheres. In a small circular area however the hypoblastic spheres remain for some time exposed at the surface (fig. 134 A).

The whole process of segmentation is completed in the rabbit about seventy hours after impregnation. At its close the epiblast cells, as they may now be called, are clear, and have an irregularly cubical form; while the hypoblast cells are polygonal and granular, and somewhat larger than the epiblast cells.

The opening in the epiblastic layer where the hypoblast cells are exposed on the surface may for convenience be called with Van Beneden the blastopore, though it is highly improbable that it in any way corresponds with the blastopore of other vertebrate ova[76].

After its segmentation the ovum passes into the uterus. The epiblast cells soon grow over the blastopore and thus form a complete superficial layer. A series of changes next take place which result in the formation of what has been called the blastodermic vesicle. To Ed. van Beneden we owe the fullest account of these changes; to Hensen and Kölliker however we are also indebted for valuable observations, especially on the later stages in the development of this vesicle.

The succeeding changes commence with the appearance of a narrow cavity between the epiblast and hypoblast, which extends so as completely to separate these two layers except in the region adjoining the original site of the blastopore (fig. 134 B)[77]. The cavity so formed rapidly enlarges, and with it the ovum also; which soon takes the form of a thin-walled vesicle with a large central cavity. This vesicle is the blastodermic vesicle. The greater part of its walls are formed of a single row of flattened epiblast cells; while the hypoblast cells form a small lens-shaped mass attached to the inner side of the epiblast cells (fig. 135).

In the Vespertilionidæ Van Beneden and Julin have shewn that the ovum undergoes at the close of segmentation changes of a more or less similar nature to those in the rabbit; the blastopore would however appear to be wider, and to persist even after the cavity of the blastodermic vesicle has commenced to be developed.

Although by this stage, which occurs in the rabbit between seventy and ninety hours after impregnation, the blastodermic vesicle has by no means attained its greatest dimensions, it has nevertheless grown from about 0.09 mm.—the size of the ovum at the close of segmentation—to about 0.28. It is enclosed by a membrane formed from the zona radiata and the mucous layer around it. The blastodermic vesicle continues to enlarge rapidly, and during the process the hypoblastic mass undergoes important changes. It spreads out on the inner side of the epiblast and at the same time loses its lens-like form and becomes flattened. The central part of it remains however thicker, and is constituted of two rows of cells, while the peripheral part, the outer boundary of which is irregular, is formed of an imperfect layer of amœboid cells which continually spread further and further within the epiblast. The central thickening of the hypoblast forms an opaque circular spot on the blastoderm, which constitutes the commencement of the embryonic area.

The history of the stages immediately following, from about the commencement of the fifth day to the seventh day, when a primitive streak makes its appearance, is imperfectly understood, and has been interpreted very differently by Van Beneden (No. 171) on the one hand and by Kölliker (184), Rauber (187) and Lieberkühn (186) on the other. I have myself in conjunction with my pupil, Mr Heape, also conducted some investigations on these stages, which have unfortunately not as yet led me to a completely satisfactory reconciliation of the opposing views.

Van Beneden states that about five days after impregnation the hypoblast cells in the embryonic area become divided into two distinct strata, an upper stratum of small cells adjoining the epiblast and a lower stratum of flattened cells which form the true hypoblast. At the edge of the embryonic area the hypoblast is continuous with a peripheral ring of the amœboid cells of the earlier stage, which now form, except at the edge of the ring, a continuous layer of flattened cells in contact with the epiblast. During the sixth day the flattened epiblast cells are believed by Van Beneden to become columnar. The embryonic area gradually extends itself, and as it does so becomes oval. A central lighter portion next becomes apparent, which gradually spreads, till eventually the darker part of the embryonic area forms a crescent at the posterior part of the now somewhat pyriform embryonic area. The lighter part is formed of columnar epiblast and hypoblast only, while in the darker area a layer of the mesoblast, derived from the intermediate layer of the fifth day, is also found. In this darker area the primitive streak originates early on the seventh day.

Kölliker, following the lines originally laid down by Rauber, has arrived at very different results. He starts from the three-layered condition described by Van Beneden for the fifth day, but does not give any investigations of his own as to the origin of the middle layer. He holds the outer layer to be a provisional layer of protective cells, forming part of the wall of the original vesicle, the middle layer he regards as the true epiblast and the inner layer as the hypoblast.

During the sixth day he finds that the cells of the outer layer gradually cease to form a continuous layer and finally disappear; while the cells of the middle layer become columnar, and form the columnar epiblast present in the embryonic area at the end of the sixth day. The mesoblast first takes its origin in the region and on the formation of the primitive streak.

The investigations of Heape and myself do not extend to the first formation of the intermediate layer found on the fifth day. We find on the sixth day in germinal vesicles of about 2.2-2.5 millimetres in diameter with embryonic areas of about .8 mm. that the embryonic area (fig. 136) is throughout composed of:
(1) A layer of flattened hypoblast cells;
(2) A somewhat irregular layer of more columnar elements, in some places only a single row deep and in other places two or more rows deep.
(3) Flat elements on the surface, which do not, however, form a continuous layer, and are intimately attached to the columnar cells below.

Our results as to the structure of the blastoderm at this stage closely correspond therefore with those of Kölliker, but on one important point we have arrived at a different conclusion. Kölliker states that he has never found the flattened elements in the act of becoming columnar. We believe that we have in many instances been able to trace them in the act of undergoing this change, and have attempted to shew this in our figure.

Our next oldest embryonic areas were somewhat pyriform measuring about 1.19 mm. in length and .85 in breadth. Of these we have several, some from a rabbit in which we also met with younger still nearly circular areas. All of them had a distinctly marked posterior opacity forming a commencing primitive streak, though decidedly less advanced than in the blastoderm represented in fig. 140. In the younger specimens the epiblast in front of the primitive streak was formed of a single row of columnar cells (fig. 138 A), no mesoblast was present and the hypoblast formed a layer of flattened cells. In the region immediately in front of the primitive streak, an irregular layer of mesoblast cells was interposed between the epiblast and hypoblast. In the anterior part of the primitive streak itself (fig. 138 B) there was a layer of mesoblast with a considerable lateral extension, while in the median line there was a distinct mesoblastic proliferation of epiblast cells. In the posterior sections the lateral extension of the mesoblast was less, but the mesoblast cells formed a thicker cord in the axial line.

Owing to the unsatisfactory character of our data the following attempt to fill in the history of the fifth and sixth days must be regarded as tentative[78]. At the commencement of the fifth day the central thickening, of what has been called above the primitive hypoblast, becomes divided into two layers: the lower of these is continuous with the peripheral hypoblast and is formed of flattened cells, while the upper one is formed of small rounded elements. The superficial epiblast again is formed of flattened cells.

During the fifth day remarkable changes take place in the epiblast of the embryonic area. It is probable that its constituent cells increase in number and become one by one columnar; and that in the process they press against the layer of rounded elements below them, so that the two layers cease to be distinguishable, and the whole embryonic area acquires in section the characters represented in fig. 136[79]. Towards the end of the sixth day the embryonic area becomes oval, but the changes which next take place are not understood. In the front part of the area only two layers of cells are found, (1) an hypoblast, and (2) an epiblast of columnar cells probably derived from the flattened epiblast cells of the earlier stages. In the posterior part of the blastoderm a middle layer is present (Van Beneden) in addition to the two other layers; and this layer probably originates from the middle layer which extended throughout the area at the beginning of the fifth day, and then became fused with the epiblast. The middle layer does not give rise to the whole of the eventual mesoblast, but only to part of it. From its origin it may be called the hypoblastic mesoblast, and it is probably equivalent to the hypoblastic mesoblast already described in the chick (pp. 154 and 155). The stage just described has only been met with by Van Beneden[80].

A diagrammatic view of the whole blastodermic vesicle at about the beginning of the seventh day is given in fig. 137. The embryonic area is represented in white. The line ge in B shews the extension of the hypoblast round the inner side of the vesicle. The blastodermic vesicle is therefore formed of three areas, (1) the embryonic area with three layers: this area is placed where the blastopore was originally situated. (2) The ring around the embryonic area where the walls of the vesicle are formed of epiblast and hypoblast. (3) The area beyond this again where the vesicle is formed of epiblast only[81].

The changes which next take place begin with the formation of a primitive streak, homologous with, and in most respects similar to, the primitive streak in Birds. The formation of the streak is preceded by that of a clear spot near the middle of the blastoderm, forming the nodal point of Hensen. This spot subsequently constitutes the front end of the primitive streak.

The history of the primitive streak was first worked out in a satisfactory manner by Hensen (No. 182), from whom however I differ in admitting the existence of a certain part of the mesoblast before its appearance.

Early on the seventh day the embryonic area becomes pyriform, and at its posterior and narrower end a primitive streak makes its appearance, which is due to a proliferation of rounded cells from the epiblast. At the time when this proliferation commences the layer of hypoblastic mesoblast is present, especially just in front of, and at the sides of, the anterior part of the streak; but no mesoblast is found in the anterior part of the embryonic area. These features are shewn in fig. 138 A and B. The mesoblast derived from the proliferation of the epiblast soon joins the mesoblast already present; though in many sections it seems possible to trace a separation between the two parts (fig. 139 B) of the mesoblast.

During the seventh day the primitive streak becomes a more pronounced structure, the mesoblast in its neighbourhood increases in quantity, while an axial groove—the primitive groove—is formed on its upper surface. The mesoblastic layer in front of the primitive streak becomes thicker, and, in the two-layered region in front, the epiblast becomes several rows deep (fig. 139 A).

In the part of the embryonic area in front of the primitive streak there arise during the eighth day two folds bounding a shallow median groove, which meet in front, but diverge behind, and enclose between them the foremost end of the primitive streak (fig. 141). These folds are the medullary folds and they constitute the first definite traces of the embryo. The medullary plate bounded by them rapidly grows in length, the primitive streak always remaining at its hinder end. While the lateral epiblast is formed of several rows of cells, that of the medullary plate is at first formed of but a single row (fig. 142, mg). The mesoblast, which appears to grow forward from the primitive streak, is stated to be at first a continuous sheet between the epiblast and hypoblast (Hensen). The evidence on this point does not however appear to me to be quite conclusive. In any case, as soon as ever the medullary groove is formed, the mesoblast becomes divided, exactly as in Lacerta and Elasmobranchii, into two independent lateral plates, which are not continuous across the middle line (fig. 142, me). The hypoblast cells are flattened laterally, but become columnar beneath the medullary plate (fig. 142).

In tracing the changes which take place in the relations of the layers, in passing from the region of the embryo to that of the primitive streak, it will be convenient to follow the account given by Schäfer for the guinea-pig (No. 190), which on this point is far fuller and more satisfactory than that of other observers. In doing so I shall leave out of consideration the fact (fully dealt with later in this chapter) that the layers in the guinea-pig are inverted. Fig. 143 represents a series of sections through this part in the guinea-pig. The anterior section (D) passes through the medullary groove near its hinder end. The commencement of the primitive streak is marked by a slight prominence on the floor of the medullary groove between the two diverging medullary folds (fig. 143 C, ae). Where this prominence becomes first apparent the epiblast and hypoblast are united together. The mesoblast plates at the two sides remain in the meantime quite free. Slightly further back, but before the primitive groove is reached, the epiblast and hypoblast are connected together by a cord of cells (fig. 143 B. f), which in the section next following becomes detached from the hypoblast and forms a solid keel projecting from the epiblast. In the following section the hitherto independent mesoblast plates become united with this keel (fig. 143 A); and in the posterior sections, through the part of the primitive streak with the primitive groove, the epiblast and mesoblast continue to be united in the axial line, but the hypoblast remains distinct. These peculiar relations may shortly be described by saying that in the axial line the hypoblast becomes united with the epiblast at the posterior end of the embryo; and that the cells which connect the hypoblast and epiblast are posteriorly continuous with the fused epiblast and mesoblast of the primitive streak, the hypoblast in the region of the primitive streak having become distinct from the other layers.

The peculiar relations just described, which hold also for the rabbit, receive their full explanation by a comparison of the Mammal with the Bird and the Lizard, but before entering into this comparison, it will be well to describe the next stage in the rabbit, which is in many respects very instructive. In this stage the thickened axial portion of the hypoblast in the region of the embryo becomes separated from the lateral part as the notochord. Very shortly after the formation of the notochord, the hypoblast grows in from the two sides, and becomes quite continuous across the middle line. The formation of the notochord takes place from before backwards; and at the hinder end of the embryo the notochord is continued into the mass of cells which forms the axis of the primitive streak, becoming therefore at this point continuous with the epiblast. The notochord in fact behaves exactly as did the axial hypoblast in the earlier stage.

In comparison with Lacerta (pp. 203-205) it is obvious that the axial hypoblast and the notochord derived from it have exactly the same relations in Mammalia and Lacertilia. In both they are continued at the hind end of the embryo into the epiblast; and close to where they join it, the mesoblast and epiblast fuse together to form the primitive streak. The difference between the two types consists in the fact that in Reptilia there is formed a passage connecting the neural and alimentary canals, the front wall of which is constituted by the cells which form the above junction between the notochord and epiblast; and that in Mammalia this passage—which is only a rudimentary structure in Reptilia—has either been overlooked or else is absent. In any case the axial junction of the epiblast and hypoblast in Mammalia is shewn by the above comparison with Lacertilia to represent the dorsal lip of the true vertebrate blastopore. The presence of this blastopore seems to render it clear that the blastopore discovered by Ed. van Beneden cannot have the meaning he assigned to it in comparing it with the blastopore of the frog.

Kölliker adduces the fact that the notochord is continuous with the axial cells of the primitive streak as an argument against its hypoblastic origin. The above comparison with Lacertilia altogether deprives this argument of any force.

At the stage we have now reached the three layers are definitely established. The epiblast (on the view adopted above) clearly originates from epiblastic segmentation cells. The hypoblast without doubt originates from the hypoblastic segmentation spheres which give rise to the lenticular mass within the epiblast on the appearance of the cavity of the blastodermic vesicle; while, though the history of the mesoblast is still obscure, part of it appears to originate from the hypoblastic mass, and part is undoubtedly formed from the epiblast of the primitive streak.

While these changes have been taking place the rudiments of a vascular area become formed, and it is very possible that part of the hypoblastic mesoblast passes in between the epiblast and hypoblast, immediately around the embryonic area, to give rise to the area vasculosa. From Hensen’s observation it seems at any rate clear that the mesoblast of the vascular area arises independently of the primitive streak: an observation which is borne out by the analogy of Birds.

General growth of the Embryo.

We have seen that the blastodermic vesicle becomes divided at an early stage of development into an embryonic area, and a non-embryonic portion. The embryonic area gives rise to the whole of the body of the embryo, while the non-embryonic part forms an appendage, known as the umbilical vesicle, which becomes gradually folded off from the embryo, and has precisely the relations of the yolk-sack of the Sauropsida. It is almost certain that the Placentalia are descended from ancestors, the embryos of which had large yolk-sacks, but that the yolk has become reduced in quantity owing to the nutriment received from the wall of the uterus taking the place of that originally supplied by the yolk. A rudiment of the yolk-sack being retained in the umbilical vesicle, this structure may be called indifferently umbilical vesicle or yolk-sack.

The yolk which fills the yolk-sack in Birds is replaced in Mammals by a coagulable fluid; while the gradual extension of the hypoblast round the wall of the blastodermic vesicle, which has already been described, is of the same nature as the growth of the hypoblast round the yolk-sack in Birds.

The whole embryonic area would seem to be employed in the formation of the body of the embryo. Its long axis has no very definite relation to that of the blastodermic vesicle. The first external trace of the embryo to appear is the medullary plate, bounded by the medullary folds, and occupying at first the anterior half of the embryonic area (fig. 141). The two medullary folds diverge behind and enclose the front end of the primitive streak. As the embryo elongates, the medullary folds nearly meet behind and so cut off the front portion of the primitive streak, which then appears as a projection in the hind end of the medullary groove. In an embryo rabbit, eight days after impregnation, the medullary groove is about 1.80 mm. in length. At this stage a division may be clearly seen in the lateral plates of mesoblast into a vertebral zone adjoining the embryo and a more peripheral lateral zone; and in the vertebral zone indications of two somites, about 0.37 mm. from the hinder end of the embryo, become apparent. The foremost of these somites marks the junction, or very nearly so, of the cephalic region and trunk. The small size of the latter as compared with the former is very striking, but is characteristic of Vertebrates generally. The trunk gradually elongates relatively to the head, by the addition behind of fresh somites. The embryo has not yet begun to be folded off from the yolk-sack. In a slightly older embryo of nine days there appears (Hensen, Kölliker) round the embryonic area a delicate clear ring which is narrower in front than behind (fig. 144 A, ap). This ring is regarded by these authors as representing the peripheral part of the area pellucida of Birds, which does not become converted into the body of the embryo. Outside the area pellucida, an area vasculosa has become very well defined. In the embryo itself (fig. 144 A) the disproportion between head and trunk is less marked than before; the medullary plate dilates anteriorly to form a spatula-shaped cephalic enlargement; and three or four somites are established. In the lateral parts of the mesoblast of the head there may be seen on each side a tube-like structure (hz). Each of these is part of the heart, which arises as two independent tubes. The remains of the primitive streak (pr) are still present behind the medullary groove.

In somewhat older embryos (fig. 144 B) with about eight somites, in which the trunk considerably exceeds the head in length, the first distinct traces of the folding-off of the head end of the embryo become apparent, and somewhat later a fold also appears at the hind end. In the formation of the hind end of the embryo the primitive streak gives rise to a tail swelling and to part of the ventral wall of the postanal gut. In the region of the head the rudiments of the heart (h) are far more definite. The medullary groove is still open for its whole length, but in the head it exhibits a series of well-marked dilatations. The foremost of these (vh) is the rudiment of the fore-brain, from the sides of which there project the two optic vesicles (ab); the next is the mid-brain (mh), and the last is the hind-brain (hh), which is again divided into smaller lobes by successive constrictions. The medullary groove behind the region of the somites dilates into an embryonic sinus rhomboidalis like that of the Bird. Traces of the amnion (af) are now apparent both in front of and behind the embryo.

The structure of the head and the formation of the heart at this age are illustrated in fig. 145. The widely-open medullary groove (rf) is shewn in the centre. Below it the hypoblast is thickened to form the notochord dd´; and at the sides are seen the two tubes, which, on the folding-in of the foregut, give rise to the unpaired heart. Each of these is formed of an outer muscular tube of splanchnic mesoblast (ahh), not quite closed towards the hypoblast, and an inner epithelioid layer (ihh); and is placed in a special section of the body cavity (ph), which afterwards forms the pericardial cavity.

Before the ninth day is completed great external changes are usually effected. The medullary groove becomes closed for its whole length with the exception of a small posterior portion. The closure commences, as in Birds, in the region of the mid-brain. Anteriorly the folding-off of the embryo proceeds so far that the head becomes quite free, and a considerable portion of the throat, ending blindly in front, becomes established. In the course of this folding the, at first widely separated, halves of the heart are brought together, coalesce on the ventral side of the throat, and so give rise to a median undivided heart. The fold at the tail end of the embryo progresses considerably, and during its advance the allantois is formed in the same way as in Birds. The somites increase in number to about twelve. The amniotic folds nearly meet above the embryo.

The later stages in the development proceed in the main in the same manner as in the Bird. The cranial flexure soon becomes very marked, the mid-brain forming the end of the long axis of the embryo (fig. 146). The sense organs have the usual development. Under the fore-brain appears an epiblastic involution giving rise both to the mouth and to the pituitary body. Behind the mouth are three well-marked pairs of visceral arches. The first of these is the mandibular arch (fig. 146, md), which meets its fellow in the middle line, and forms the posterior boundary of the mouth. It sends forward on each side a superior maxillary process (mx) which partially forms the anterior margin of the mouth. Behind the mandibular arch are present a well-developed hyoid (hy) and a first branchial arch (not shewn in fig. 146). There are four clefts, as in other Amniota, but the fourth is not bounded behind by a definite arch. Only the first of these clefts persists as the tympanic cavity and Eustachian tube.

At the time when the cranial flexure appears, the body also develops a sharp flexure immediately behind the head, which is thus bent forwards upon the posterior straight part of the body (fig. 146). The amount of this flexure varies somewhat in different forms. It is very marked in the dog (Bischoff). At a later period, and in some species even before the stage figured, the tail end of the body also becomes bent (fig. 146), so that the whole dorsal side assumes a convex curvature, and the head and tail become closely approximated. In most cases the embryo, on the development of the tail, assumes a more or less definite spiral curvature (fig. 146); which however never becomes nearly so marked a feature as it commonly is in Lacertilia and Ophidia. With the more complete development of the lower wall of the body the ventral flexure partially disappears, but remains more or less persistent till near the close of intra-uterine life. The limbs are formed as simple buds in the same manner as in Birds. The buds of the hind-limbs are directed somewhat forwards, and those of the fore-limb backwards.

Embryonic membranes and yolk-sack.

The early stages in the development of the embryonic membranes are nearly the same as in Aves; but during the later stages in the Placentalia the allantois enters into peculiar relations with the uterine walls, and the two, together with the interposed portion of the subzonal membrane or false amnion, give rise to a very characteristic Mammalian organ—the placenta—into the structure of which it will be necessary to enter at some length. The embryonic membranes vary so considerably in the different forms that it will be advantageous to commence with a description of their development in an ideal case.

We may commence with a blastodermic vesicle, closely invested by the delicate remnant of the zona radiata, at the stage in which the medullary groove is already established. Around the embryonic area a layer of mesoblast would have extended for a certain distance; so as to give rise to an area vasculosa, in which however the blood-vessels would not have become definitely established. Such a vesicle is represented diagrammatically in fig. 147, 1. Somewhat later the embryo begins to be folded off, first in front and then behind (fig. 147, 2). These folds result in a constriction separating the embryo and the yolk-sack (ds), or as it is known in Mammalian embryology, the umbilical vesicle. The splitting of the mesoblast into a splanchnic and a somatic layer has taken place, and at the front and hind end of the embryo a fold (ks) of the somatic mesoblast and epiblast begins to rise up and grow over the head and tail of the embryo. These two folds form the commencement of the amnion. The head and tail folds of the amnion are continued round the two sides of the embryo, till they meet and unite into a continuous fold. This fold grows gradually upwards, but before it has completely enveloped the embryo, the blood-vessels of the area vasculosa become fully developed. They are arranged in a manner not very different from that in the chick.

The following is a brief account of their arrangement in the Rabbit:—

The outer boundary of the area, which is continually extending further and further round the umbilical vesicle, is marked by a venous sinus terminalis (fig. 147, st). The area is not, as in the chick, a nearly complete circle, but is in front divided by a deep indentation extending inwards to the level of the heart. In consequence of this indentation the sinus terminalis ends in front in two branches, which bend inwards and fall directly into the main vitelline veins. The blood is brought from the dorsal aortæ by a series of lateral vitelline arteries, and not by a single pair as in the chick. These arteries break up into a more deeply situated arterial network, from which the blood is continued partly into the sinus terminalis, and partly into a superficial venous network. The hinder end of the heart is continued into two vitelline veins, each of which divides into an anterior and a posterior branch. The anterior branch is a limb of the sinus terminalis, and the posterior and smaller branch is continued towards the hind part of the sinus, near which it ends. On its way it receives, on its outer side, numerous branches from the venous network, which connect by their anastomoses the posterior branch of the vitelline vein and the sinus terminalis.

While the above changes have been taking place the whole blastodermic vesicle, still enclosed in the zona, has become attached to the walls of the uterus. In the case of the typical uterus with two tubular horns, the position of each embryo, when there are several, is marked by a swelling in the walls of the uterus, preparatory to the changes which take place on the formation of the placenta. In the region of each swelling the zona around the blastodermic vesicle is closely embraced, in a ring-like fashion, by the epithelium of the uterine wall. The whole vesicle assumes an oval form, and it lies in the uterus with its two ends free. The embryonic area is placed close to the mesometric attachment of the uterus. In many cases peculiar processes or villi grow out from the ovum (fig. 147, 4, sz), which fit into the folds of the uterine epithelium. The nature of these processes requires further elucidation, but in some instances they appear to proceed from the zona (the Rabbit) and in other instances from the subzonal membrane (the Dog). In any case the attachment between the blastodermic vesicle and the uterine wall becomes so close at the time when the body of the embryo is first formed out of the embryonic area, that it is hardly possible to separate them without laceration; and at this period—from the 8th to the 9th day in the Rabbit—it requires the greatest care to remove the ovum from the uterus without injury. It will be understood of course that the attachment above described is at first purely superficial and not vascular.

Shortly after the establishment of the circulation of the yolk-sack the folds of the amnion meet and coalesce above the embryo (fig. 147, 3 and 4, am). After this the inner or true amnion becomes severed from the outer or false amnion, though the two sometimes remain connected by a narrow stalk. Between the true and false amnion is a continuation of the body cavity. The true amnion consists of a layer of epiblastic epithelium and generally also of somatic mesoblast, while the false amnion consists, as a rule, of epiblast only; though it is possible that in some cases (the Rabbit?) the mesoblast may be continued along its inner face.

Before the two limbs of the amnion are completely severed, the epiblast of the umbilical vesicle becomes separated from the mesoblast and hypoblast of the vesicle (fig. 147, 3), and, together with the false amnion (sh), with which it is continuous, forms a complete lining for the inner face of the zona radiata. The space between this membrane and the umbilical vesicle with the attached embryo is obviously continuous with the body cavity (vide figs. 147, 4 and 147*). To this membrane Turner has given the appropriate name of subzonal membrane: by Von Baer it was called the serous envelope. It soon fuses with the zona radiata, or at any rate the zona ceases to be distinguishable.

While the above changes are taking place in the amnion, the allantois grows out from the hind gut as a vesicle lined by hypoblast, but covered externally by a layer of splanchnic mesoblast (fig. 147, 3 and 4, al)[83]. The allantois soon becomes a flat sack, projecting into the now largely developed space between the subzonal membrane and the amnion, on the dorsal side of the embryo (fig. 147*, ALC). In some cases it extends so as to cover the whole inner surface of the subzonal membrane; in other cases again its extension is much more limited. Its lumen may be retained or may become nearly or wholly aborted. A fusion takes place between the subzonal membrane and the adjoining mesoblastic wall of the allantois, and the two together give rise to a secondary membrane round the ovum, known as the chorion. Since however the allantois does not always come in contact with the whole inner surface of the subzonal membrane, the term chorion is apt to be somewhat vague; and in the rabbit, for instance, a considerable part of the so-called chorion is formed by a fusion of the wall of the yolk-sack with the subzonal membrane (fig. 148). The placental region of the chorion may in such cases be distinguished as the true chorion, from the remaining part which will be called the false chorion.

The mesoblast of the allantois, especially that part of it which assists in forming the chorion, becomes highly vascular; the blood being brought to it by two allantoic arteries continued from the terminal bifurcation of the dorsal aorta, and returned to the body by one, or rarely two, allantoic veins, which join the vitelline veins from the yolk-sack. From the outer surface of the true chorion (fig. 147, 5, d, 148) villi grow out and fit into crypts or depressions which have in the meantime made their appearance in the walls of the uterus[84]. The villi of the chorion are covered by an epithelium derived from the subzonal membrane, and are provided with a connective tissue core containing an artery and vein and a capillary plexus connecting them. In most cases they assume a more or less arborescent form, and have a distribution on the surface of the chorion varying characteristically in different species. The walls of the crypts into which the villi are fitted also become highly vascular, and a nutritive fluid passes from the maternal vessels of the placenta to the fœtal vessels by a process of diffusion; while there is probably also a secretion by the epithelial lining of the walls of the crypts, which becomes absorbed by the vessels of the fœtal villi. The above maternal and fœtal structures constitute together the organ known as the placenta. The maternal portion consists essentially of the vascular crypts in the uterine walls, and the fœtal portion of more or less arborescent villi of the true chorion fitting into these crypts.

While the placenta is being developed, the folding-off of the embryo from the yolk-sack becomes more complete; and the yolk-sack remains connected with the ileal region of the intestine by a narrow stalk, the vitelline duct (fig. 147, 4 and 5 and fig. 147*), consisting of the same tissues as the yolk-sack, viz. hypoblast and splanchnic mesoblast. While the true splanchnic stalk of the yolk-sack is becoming narrow, a somatic stalk connecting the amnion with the walls of the embryo is also formed, and closely envelops the stalk both of the allantois and the yolk-sack. The somatic stalk together with its contents is known as the umbilical cord. The mesoblast of the somatopleuric layer of the cord develops into a kind of gelatinous tissue, which cements together the whole of the contents. The allantoic arteries in the cord wind in a spiral manner round the allantoic vein. The yolk-sack in many cases atrophies completely before the close of intra-uterine life, but in other cases it is only removed with the other embryonic membranes at birth. The intra-embryonic portion of the allantoic stalk gives rise to two structures, viz. to (1) the urinary bladder formed by a dilatation of its proximal extremity, and to (2) a cord known as the urachus connecting the bladder with the wall of the body at the umbilicus. The urachus, in cases where the cavity of the allantois persists till birth, remains as an open passage connecting the intra- and extra-embryonic parts of the allantois. In other cases it gradually closes, and becomes nearly solid before birth, though a delicate but interrupted lumen would appear to persist in it. It eventually gives rise to the ligamentum vesicæ medium.

At birth the fœtal membranes, including the fœtal portion of the placenta, are shed; but in many forms the interlocking of the fœtal villi with the uterine crypts is so close that the uterine mucous membrane is carried away with the fœtal part of the placenta. It thus comes about that in some placentæ the maternal and fœtal parts simply separate from each other at birth, and in others the two remain intimately locked together, and both are shed together as the afterbirth. These two forms of placenta are distinguished as non-deciduate and deciduate, but it has been shewn by Ercolani and Turner that no sharp line can be drawn between the two types; moreover, a larger part of the uterine mucous membrane than that forming the maternal part of the placenta is often shed in the deciduate Mammalia, and in the non-deciduate Mammalia it is probable that the mucous membrane (not including vascular parts) of the maternal placenta either peels or is absorbed.

Comparative history of the Mammalian fœtal membranes.

Two groups of Mammalia—the Monotremata and the Marsupialia—are believed not to be provided with a true placenta.

The nature of the fœtal membranes in the Monotremata is not known. Ova, presumably in an early stage of development, have been found free in the uterus of Ornithorhyncus by Owen. The lining membrane of the uterus was thickened and highly vascular. The females in which these were found were killed early in October[85].

Marsupialia. Our knowledge of the fœtal membranes of the Marsupialia is almost entirely due to Owen. In Macropus major he found that birth took place thirty-eight days after impregnation. A fœtus at the twentieth day of gestation measured eight lines from the mouth to the root of the tail. The fœtus was enveloped in a large subzonal membrane, with folds fitting into uterine furrows, but not adhering to the uterus, and without villi. The embryo was enveloped in an amnion reflected over the stalk of the yolk-sack, which was attached by a filamentary pedicle to near the end of the ileum. The yolk-sack was large and vascular, and was connected with the fœtal vascular system by a vitelline artery and two veins. The yolk-sack was partially adherent, especially at one part, to the subzonal membrane. No allantois was observed. In a somewhat older fœtus of ten lines in length there was a small allantois supplied by two allantoic arteries and one vein. The allantois was quite free and not attached to the subzonal membrane. The yolk-sack was more closely attached to the subzonal membrane than in the younger embryo[86].

All Mammalia, other than the Monotremata and Marsupialia, have a true allantoic placenta. The placenta presents a great variety of forms, and it will perhaps be most convenient first to treat these varieties in succession, and then to give a general exposition of their mutual affinities[87].

Amongst the existing Mammals provided with a true placenta, the most primitive type is probably retained by those forms in which the placental part of the chorion is confined to a comparatively restricted area on the dorsal side of the embryo; while the false chorion is formed by the vascular yolk-sack fusing with the remainder of the subzonal membrane. In all the existing forms with this arrangement of fœtal membranes, the placenta is deciduate. This, however, was probably not the case in more primitive forms from which these are descended[88]. The placenta would appear from Ercolani’s description to be simpler in the mole (Talpa) than in other species. The Insectivora, Cheiroptera, and Rodentia are the groups with this type of placenta; and since the rabbit, amongst the latter, has been more fully worked out than other species, we may take it first.

The Rabbit. In the pregnant female Rabbit several ova are generally found in each horn of the uterus. The general condition of the egg-membranes at the time of their full development is shewn in fig. 148.

The embryo is surrounded by the amnion, which is comparatively small. The yolk-sack (ds) is large and attached to the embryo by a long stalk. It has the form of a flattened sack closely applied to about two-thirds of the surface of the subzonal membrane. The outer wall of this sack, adjoining the subzonal membrane, is formed of hypoblast only; but the inner wall is covered by the mesoblast of the area vasculosa, as indicated by the thick black line (fd). The vascular area is bordered by the sinus terminalis (st). In an earlier stage of development the yolk-sack had not the compressed form represented in the figure. It is, however, remarkable that the vascular area never extends over the whole yolk-sack; but the inner vascular wall of the yolk-sack fuses with the outer, and with the subzonal membrane, and so forms a false chorion, which receives its blood supply from the yolk-sack. This part of the chorion does not develop vascular villi.

The allantois (al) is a simple vascular sack with a large cavity. Part of its wall is applied to the subzonal membrane, and gives rise to the true chorion, from which there project numerous vascular villi. These fit into corresponding uterine crypts. It seems probable, from Bischoff’s and Kölliker’s observations, that the subzonal membrane in the area of the placenta becomes attached to the uterine wall, by means of villi, even before its fusion with the allantois. In the later periods of gestation the intermingling of the maternal and fœtal parts of the placenta becomes very close, and the placenta is truly deciduate. The cavity of the allantois persists till birth. Between the yolk-sack, the allantois, and the embryo, there is left a large cavity filled with an albuminous fluid.

The Hare does not materially differ in the arrangement of its fœtal membranes from the Rabbit.

In the Rat (Mus decumanus) (fig. 149) the sack of the allantois completely atrophies before the close of fœtal life[89], and there is developed, at the junction of the maternal part of the placenta and the unaltered mucous membrane of the uterus, a fold of the mucous membrane which completely encapsules the whole chorion, and forms a separate chamber for it, distinct from the general lumen of the uterus. Folds of this nature, which are specially developed in Man and Apes, are known as a decidua reflexa. The decidua reflexa of the Rat is reduced to extreme tenuity, or even vanishes before the close of gestation.

Guinea-pig. The development of the Guinea-pig is dealt with elsewhere, but, so far as its peculiarities permit a comparison with the Rabbit, the agreement between the two types appears to be fairly close.

The blastodermic vesicle of the Guinea-pig becomes completely enveloped in a capsule of the uterine wall (decidua reflexa) (fig. 150). The epithelium of the blastodermic vesicle in contact with the uterine wall is not epiblastic, but corresponds with the hypoblast of the yolk-sack of other forms, and the mesoblast of the greater part of the inner side of this becomes richly vascular (yk); the vascular area being bounded by a sinus terminalis.

The blastodermic vesicle is so situated within its uterine capsule that the embryo is attached to the part of it adjoining the free side of the uterus. From the opposite side of the uterus, viz. that to which the mesometrium is attached, there grow into the wall of the blastodermic vesicle numerous vascular processes of the uterine wall, which establish at this point an organic connection between the two (pl). The blood-vessels of the blastodermic vesicle (yolk-sack) stop short immediately around the area of attachment to the uterus; but at a late period the allantois grows towards, and fuses with this area. The blood-vessels of the allantois and of the uterus become intertwined, and a disc-like placenta more or less similar to that in the Rabbit becomes formed (pl). The cavity of the allantois, if developed, vanishes completely.

In all the Rodentia the placenta appears to be situated on the mesometric side of the uterus.

Insectivora. In the Mole (Talpa) and the Shrew (Sorex), the fœtal membranes are in the main similar to those in the rabbit, and a deciduate discoidal placenta is always present. It may be situated anywhere in the circumference of the uterine tube. The allantoic cavity persists (Owen), but the allantois only covers the placental area of the chorion. The yolk-sack is persistent, and fuses with the non-allantoic part of the subzonal membrane; which is rendered vascular by its blood-vessels. There would seem to be (Owen) a small decidua reflexa. A similar arrangement is found in the Hedgehog (Erinaceus Europæus) (Rolleston), in which the placenta occupies the typical dorsal position. It is not clear from Rolleston’s description whether the yolk-sack persists till the close of fœtal life, but it seems probable that it does so. There is a considerable reflexa which does not, however, cover the whole chorion. In the Tenrec (Centetes) the yolk-sack and non-placental part of the chorion are described by Rolleston as being absent, but it seems not impossible that this may have been owing to the bad state of preservation of the specimen. The amnion is large. In the Cheiroptera (Vespertilio and Pteropus), the yolk-sack is large, and coalesces with part of the chorion. The large yolk-sack has been observed in Pteropus by Rolleston, and in Vespertilio by Owen. The allantoic vessels supply the placenta only. The Cheiroptera are usually uniparous.

Simiadæ and Anthropidæ. The fœtal membranes of Apes and Man, though in their origin unlike those of the Rodentia and Insectivora, are in their ultimate form similar to them, and may be conveniently dealt with here. The early stages in the development of these membranes in the human embryo have not been satisfactorily observed; but it is known that the ovum, shortly after its entrance into the uterus, becomes attached to the uterine wall, which in the meantime has undergone considerable preparatory changes. A fold of the uterine wall appears to grow round the blastodermic vesicle, and to form a complete capsule for it, but the exact mode of formation of this capsule is a matter of inference and not of observation. During the first fortnight of pregnancy villi grow out, according to Allen Thomson over its whole surface, but according to Reichert in a ring-like fashion round the edge of the somewhat flattened ovum, and attach it to the uterus. The further history of the early stages is extremely obscure, and to a large extent a matter of speculation: what is known with reference to it will be found in a special section, but I shall here take up the history at about the fourth week.

At this stage a complete chorion has become formed, and is probably derived from a growth of the mesoblast of the allantois (unaccompanied by the hypoblast) round the whole inner surface of the subzonal membrane. From the whole surface of the chorion there project branched vascular processes, covered by an epithelium. The allantois is without a cavity, but a hypoblastic epithelium is present in the allantoic stalk, through which it does not, however, form a continuous tube. The blood-vessels of the chorion are derived from the usual allantoic arteries and vein. The general condition of the embryo and of its membranes at this period is shewn diagrammatically in fig. 147, 5. Around the embryo is seen the amnion, already separated by a considerable interval from the embryo. The yolk-sack is shewn at ds. Relatively to the other parts it is considerably smaller than it was at an earlier stage. The allantoic stalk is shewn at al. Both it and the stalk of the yolk-sack are enveloped by the amnion (am). The chorion with its vascular processes surrounds the whole embryo.

It may be noted that the condition of the chorion at this stage is very similar to that of the normal diffused type of placenta, described in the sequel.

While the above changes are taking place in the embryonic membranes, the blastodermic vesicle greatly increases in size, and forms a considerable projection from the upper wall of the uterus. Three regions of the uterine wall, in relation to the blastodermic vesicle, are usually distinguished; and since the superficial parts of all of these are thrown off with the afterbirth, each of them is called a decidua. They are represented at a somewhat later stage in fig. 151. There is (1) the part of the wall reflected over the blastodermic vesicle, called the decidua reflexa (dr); (2) the part of the wall forming the area round which the reflexa is inserted, called the decidua serotina (ds); (3) the general wall of the uterus, not related to the embryo, called the decidua vera (du).

The decidua reflexa and serotina together envelop the chorion, the processes of which fit into crypts in them. At this period both of them are highly and nearly uniformly vascular. The general cavity of the uterus is to a large extent obliterated by the ovum, but still persists as a space filled with mucus, between the decidua reflexa and the decidua vera.

The changes which ensue from this period onwards are fully known. The amnion continues to dilate (its cavity being intensely filled with amniotic fluid) till it comes very close to the chorion (fig. 151, am); from which, however, it remains separated by a layer of gelatinous tissue. The villi of the chorion in the region covered by the decidua reflexa, gradually cease to be vascular, and partially atrophy, but in the region in contact with the decidua serotina increase and become more vascular and more arborescent (fig. 151, z). The former region becomes known as the chorion læve, and the latter as the chorion frondosum. The chorion frondosum, together with the decidua serotina, gives rise to the placenta.

Although the vascular supply is cut off from the chorion læve, the processes on its surface do not completely abort. It becomes, as the time of birth approaches, more and more closely united with the reflexa, till the union between the two is so close that their exact boundaries cannot be made out. The umbilical vesicle (fig. 151, nb), although it becomes greatly reduced in size and flattened, persists in a recognisable form till the time of birth.

As the embryo enlarges, the space between the decidua vera and decidua reflexa becomes reduced, and finally the two parts unite together. The decidua vera is mainly characterised by the presence of peculiar roundish cells in its subepithelial tissue, and by the disappearance of a distinct lining of epithelial cells. During the whole of pregnancy it remains highly vascular. The decidua reflexa, on the disappearance of the vessels in the chorion læve, becomes non-vascular. Its tissue undergoes changes in the main similar to those of the decidua vera, and as has been already mentioned, it fuses on the one hand with the chorion, and on the other with the decidua vera. The membrane resulting from its fusion with the latter structure becomes thinner and thinner as pregnancy advances, and is reduced to a thin layer at the time of birth.

The placenta has a somewhat discoidal form, with a slightly convex uterine surface and a concave embryonic surface. At its edge it is continuous both with the decidua reflexa and decidua vera. Near the centre of the embryonic surface is implanted the umbilical cord. As has already been mentioned, the placenta is formed of the decidua serotina and the fœtal villi of the chorion frondosum. The fœtal and maternal tissues are far more closely united (fig. 152) than in the forms described above. The villi of the chorion, which were originally comparatively simple, become more and more complicated, and assume an extremely arborescent form. Each of them contains a vein and an artery, which subdivide to enter the complicated ramifications; and are connected together by a rich anastomosis. The villi are formed mainly of connective tissue, but are covered by an epithelial layer generally believed to be derived from the subzonal membrane; but, as was first stated by Goodsir, and has since been more fully shewn by Ercolani and Turner, this epithelial layer is really a part of the cellular decidua serotina of the uterine wall, which has become adherent to the villi in the development of the placenta (fig. 161, g). The placenta is divided into a number of lobes, usually called cotyledons, by septa which pass towards the chorion. These septa, which belong to the serotina, lie between the arborescent villi of the chorion. The cotyledons themselves consist of a network of tissue permeated by large vascular spaces, formed by the dilatation of the maternal blood-vessels of the serotina, into which the ramifications of the fœtal villi project. In these spaces they partly float freely, and partly are attached to delicate trabeculæ of the maternal tissue (fig. 161, G). They are, of course, separated from the maternal blood by the uterine epithelial layer before mentioned. The blood is brought to the maternal part of the placenta by spirally coiled arteries, which do not divide into capillaries, but open into the large blood-spaces already spoken of. From these spaces there pass off oblique utero-placental veins, which pierce the serotina, and form a system of large venous sinuses in the adjoining uterine wall (fig. 152, F), and eventually fall into the general uterine venous system. At birth the whole placenta, together with the fused decidua vera, and reflexa, with which it is continuous, is shed; and the blood-vessels thus ruptured are closed by the contraction of the uterine wall.

The fœtal membranes and the placenta of the Simiadæ (Turner, No. 225) are in most respects closely similar to those in Man; but the placenta is, in most cases, divided into two lobes, though in the Chimpanzee, Cynocephalus, and the Apes of the New World, it appears to be single.

The types of deciduate placenta so far described, are usually classified by anatomists as discoidal placentæ, although it must be borne in mind that they differ very widely. In the Rodentia, Insectivora, and Cheiroptera there is a (usually) dorsal placenta, which is co-extensive with the area of contact between the allantois and the subzonal membrane, while the yolk-sack adheres to a large part of the subzonal membrane. In Apes and Man the allantois spreads over the whole inner surface of the subzonal membrane; the placenta is on the ventral side of the embryo, and occupies only a small part of the surface of the allantois. The placenta of Apes and Man might be called metadiscoidal, in order to distinguish it from the primitive discoidal placenta of the Rodentia and Insectivora.

In the Armadilloes (Dasypus) the placenta is truly discoidal and deciduate (Owen and Kölliker). Alf. Milne Edwards states that in Dasypus novemcinctus the placenta is zonary, and both Kölliker and he found four embryos in the uterus, each with its own amnion, but the placenta of all four united together; and all four enclosed in a common chorion. A reflexa does not appear to be present. In the Sloths the placenta approaches the discoidal type (Turner, No. 218). It occupies in Cholæpus Hoffmanni about four-fifths of the surface of the chorion, and is composed of about thirty-four discoid lobes. It is truly deciduate, and the maternal capillaries are replaced by a system of sinuses (fig. 161). The amnion is close to the inner surface of the chorion. A dome-shaped placenta is also found amongst the Edentata in Myrmecophaga and Tamandua (Milne Edwards, No. 208).

Zonary Placenta. Another form of deciduate placenta is known as the zonary. This form of placenta occupies a broad zone of the chorion, leaving the two poles free. It is found in the Carnivora, Hyrax, Elephas, and Orycteropus.

It is easy to understand how the zonary placenta may be derived from the primitive arrangement of the membranes (vide p. 240) by the extension of a discoidal placental area to a zonary area, but it is possible that some of the types of zonary placenta may have been evolved from the concentration of a diffused placenta (vide p. 261) to a zonary area. The absence of the placenta at the extreme poles of the chorion is explained by the fact of their not being covered by a reflection of the uterine mucous membrane. In the later periods of pregnancy the placental area becomes, however, in most forms much more restricted than the area of contact between the uterus and chorion.

In the Dog[90], which may be taken as type, there is a large vascular yolk-sack formed in the usual way, which does not however fuse with the chorion. It extends at first quite to the end of the citron-shaped ovum, and persists till birth. The allantois first grows out on the dorsal side of the embryo, where it coalesces with the subzonal membrane, over a small discoidal area.

Before the fusion of the allantois with the subzonal membrane, there grow out from the whole surface of the external covering of the ovum, except the poles, numerous non-vascular villi, which fit into uterine crypts. When the allantois adheres to the subzonal membrane vascular processes grow out from it into these villi. The vascular villi so formed are of course at first confined to the disc-shaped area of adhesion between the allantois and the subzonal membrane; and there is thus formed a rudimentary discoidal placenta, closely resembling that of the Rodentia. The view previously stated, that the zonary placenta is derived from the discoidal one, receives from this fact a strong support.

The cavity of the allantois is large, and its inner part is in contact with the amnion. The area of adhesion between the outer part of the allantois and subzonal membrane gradually spreads over the whole interior of the subzonal membrane, and vascular villi are formed over the whole area of adhesion except at the two extreme poles of the egg. The last part to be covered is the ventral side where the yolk-sack adjoins the subzonal membrane.

During the extension of the allantois its cavity persists, and its inner part covers not only the amnion, but also the yolk-sack. It adheres to the amnion and supplies it with blood-vessels (Bischoff).

With the full growth of the allantois there is formed a broad placental zone, with numerous branched villi, fitting into corresponding pits which become developed in the uterine walls. The maternal and fœtal structures become closely interlocked and highly vascular; and at birth a large part of the maternal part is carried away with the placenta; some of it however still remains attached to the muscular wall of the uterus. The villi of the chorion do not fit into uterine glands. The zone of the placenta diminishes greatly in proportion to the chorion as the latter elongates, and at the full time the breadth of the zone is not more than about one-fifth of the whole length of the chorion.

At the edge of the placental zone there is a very small portion of the uterine mucous membrane reflected over the non-placental part of the chorion, which forms a small reflexa analogous with the reflexa in Man.

The Carnivora generally closely resemble the Dog, but in the Cat the whole of the maternal part of the placenta is carried away with the fœtal parts, so that the placenta is more completely deciduate than in the Dog. In the Grey Seal (Halichœrus gryphus, Turner, No. 219) the general arrangement of the fœtal membranes is the same as in the other groups of the Carnivora, but there is a considerable reflexa developed at the edge of the placenta. The fœtal part of the placenta is divided by a series of primary fissures which give off secondary and tertiary fissures. Into the fissures there pass vascular laminæ of the uterine wall. The general surface of the fœtal part of the placenta between the fissures is covered by a greyish membrane formed of the coalesced terminations of the fœtal villi.

The structure of the placenta in Hyrax is stated by Turner (No. 221) to be very similar to that in the Felidæ. The allantoic sack is large, and covers the whole surface of the subzonal membrane. The amnion is also large, but the yolk-sack would seem to disappear at an early stage, instead of persisting, as in the Carnivora, till the close of fœtal life.

The Elephant (Owen, Turner, Chapman) is provided with a zonary deciduate placenta, though a villous patch is present near each pole of the chorion.

Turner (No. 220) has shewn that in Orycteropus there is present a zonary placenta, which differs however in several particulars from the normal zonary placenta of the Carnivora; and it is even doubtful whether it is truly deciduate. There is a single embryo, which fills up the body of the uterus and also projects into only one of the horns. The placenta forms a broad median zone, leaving the two poles free. The breadth of the zone is considerably greater than is usual in Carnivora, one-half or more of the whole longitudinal diameter of the chorion being occupied by the placenta. The chorionic villi are arborescent, and diffusely scattered, and though the maternal and fœtal parts are closely interwoven, it has not been ascertained whether the adhesion between them is sufficient to cause the maternal subepithelial tissue to be carried away with the fœtal part of the placenta at birth. The allantois is adherent to the whole chorion, the non-placental parts of which are vascular. In the umbilical cord a remnant of the allantoic vesicle was present in the embryos observed by Turner, but in the absence of a large allantoic cavity the Cape Ant-eater differs greatly from the Carnivora. The amnion and allantois were in contact, but no yolk sack was observed.

Non-deciduate placenta. The remaining Mammalia are characterized by a non-deciduate placenta; or at least by a placenta in which only parts of the maternal epithelium and no vascular maternal structures are carried away at parturition. The non-deciduate placentæ are divided into two groups: (1) The polycotyledonary placenta, characteristic of the true Ruminantia (Cervidæ, Antilopidæ, Bovidæ, Camelopardalidæ); (2) the diffused placenta found in the other non-deciduate Mammalia, viz. the Perissodactyla, the Suidæ, the Hippopotamidæ, the Tylopoda, the Tragulidæ, the Sirenia, the Cetacea, Manis amongst the Edentata, and the Lemuridæ. The polycotyledonary form is the most differentiated; and is probably a modification of the diffused form. The diffused non-deciduate placenta is very easily derived from the primitive type (p. 240) by an extension of the allantoic portion of the chorion; and the exclusion of the yolk-sack from any participation in forming the chorion.

The possession in common of a diffused type of placenta is by no means to be regarded as a necessary proof of affinity between two groups, and there are often, even amongst animals possessing a diffused form of placenta, considerable differences in the general arrangement of the embryonic membranes.

Ungulata. Although the Ungulata include forms with both cotyledonary and diffused placentæ, the general arrangement of the embryonic membranes is so similar throughout the group, that it will be convenient to commence with a description of them, which will fairly apply both to the Ruminantia and to the other forms.

The blastodermic vesicle during the early stages of development lies freely in the uterus; and no non-vascular villi, similar to those of the Dog or the Rabbit, are formed before the appearance of the allantois. The blastodermic vesicle has at first the usual spherical form, but it grows out at an early period, and with prodigious rapidity, into two immensely long horns; which in cases where there is only one embryo are eventually prolonged for the whole length of the two horns of the uterus. The embryonic area is formed in the usual way, and its long axis is placed at right angles to that of the vesicle. On the formation of an amnion there is formed the usual subzonal membrane, which soon becomes separated by a considerable space from the yolk-sack (fig. 153). The yolk-sack is, however, continued into two elongated processes (yk), which pass to the two extremities of the subzonal membrane. It is supplied with the normal blood-vessels. As soon as the allantois appears (fig. 153 all), it grows out into a right and a left process, which rapidly fill the whole free space within the subzonal membrane and in many cases, e.g. the Pig (Von Baer), break through the ends of the membrane, from which they project as the diverticula allantoidis. The cavity of the allantois remains large, but the lining of hypoblast becomes separated from the mesoblast, owing to the more rapid growth of the latter. The mesoblast of the allantois applies itself externally to the subzonal membrane to form the chorion[91], and internally to the amnion, the cavity of which remains very small. The chorionic portion of the allantoic mesoblast is very vascular, and that applied to the amnion also becomes vascular in the later developmental periods.

The horns of the yolk-sack gradually atrophy, and the whole yolk-sack disappears some time before birth.

Where two or more embryos are present in the uterus, the chorions of the several embryos may unite where they are in contact.

From the chorion there grow out numerous vascular villi, which fit into corresponding pits in the uterine walls. According to the distribution of these villi, the allantois is either diffused or polycotyledonary.

The pig presents the simplest type of diffused placenta. The villi of the surface of the chorion cover a broad zone, leaving only the two poles free; their arrangement differs therefore from that in a zonary placenta in the greater breadth of the zone covered by them. The villi have the form of simple papillæ, arranged on a series of ridges, which are highly vascular as compared with the intervening valleys. If an injected chorion is examined (fig. 154), certain clear non-vascular spots are to be seen (b), from which the ridges of villi radiate. The surface of the uterus adapts itself exactly to the elevations of the chorion; and the furrows which receive the chorionic ridges are highly vascular (fig. 155). On the other hand, there are non-vascular circular depressions corresponding to the non-vascular areas on the chorion; and in these areas, and in these alone, the glands of the uterus open (fig. 155 g) (Turner). The maternal and fœtal parts of the placenta in the pig separate with very great ease.

In the mare (Turner), the fœtal villi are arranged in a less definite zonary band than in the pig, though still absent for a very small area at both poles of the chorion, and also opposite the os uteri. The filiform villi, though to the naked eye uniformly scattered, are, when magnified, found to be clustered together in minute cotyledons, which fit into corresponding uterine crypts (fig. 156). Surrounding the uterine crypts are reticulate ridges on which are placed the openings of the uterine glands. The remaining Ungulata with diffused placentæ do not differ in any important particulars from those already described.

The polycotyledonary form of placenta is found in the Ruminantia alone. Its essential character consists in the fœtal villi not being uniformly distributed, but collected into patches or cotyledons which form as it were so many small placentæ (fig. 157). The fœtal villi of these patches fit into corresponding pits in thickened patches of the wall of the uterus (figs. 158 and 159). In many cases (Turner), the interlocking of the maternal and fœtal structures is so close that large parts of the maternal epithelium are carried away when the fœtal villi are separated from the uterus. The glands of the uterus open in the intervals between the cotyledons. The character of the cotyledons differs greatly in different types. The maternal parts are cup-shaped in the sheep, and mushroom-shaped in the cow. There are from 60-100 in the cow and sheep, but only about five or six in the Roe-deer. In the Giraffe there are, in addition to larger and smaller cotyledons, rows and clusters of short villi, so that the placenta is more or less intermediate between the polycotyledonary and diffused types (Turner). A similarly intermediate type of placenta is found in Cervus mexicanus (Turner).

The groups not belonging to the Ungulata which are characterized by the possession of a diffused placenta are the Sirenia, the Cetacea, Manis, and the Lemuridæ.

Sirenia. Of the Sirenia, the placentation of the Dugong is known from some observations of Harting (No. 201).

It is provided with a diffuse and non-deciduate placenta; with the villi generally scattered except at the poles. The umbilical vesicle vanishes early.

Cetacea. In the Cetacea, if we may generalize from Turner’s observations on Orca Gladiator and the Narwhal, and those of Anderson (No. 191) on Platanista and Orcella, the blastodermic vesicle is very much elongated, and prolonged unsymmetrically into two horns. The mesoblast (fig. 160) of the allantois would appear to grow round the whole inner surface of the subzonal membrane, but the cavity of the allantois only persists as a widish sack on the ventral aspect of the embryo (al). The amnion (am) is enormous, and is dorsally in apposition with, and apparently coalesces with the chorion, and ventrally covers the inner wall of the persistent allantoic sack. The chorion, except for a small area at the two poles and opposite the os uteri, is nearly uniformly covered with villi, which are more numerous than in fig. 160. In the large size of the amnion, and small dimensions of the persistent allantoic sack, the Cetacea differ considerably from the Ungulata.

Manis. Manis amongst the Edentata presents a type of diffused placenta[92]. The villi are arranged in ridges which radiate from a non-villous longitudinal strip on the concave surface of the chorion.

Manis presents us with the third type of placenta found amongst the Edentata. On this subject, I may quote the following sentence from Turner (Journal of Anat. and Phys., vol. X., p. 706).

“The Armadilloes (Dasypus), according to Professor Owen, possess a single, thin, oblong, disc-shaped placenta; a specimen, probably Dasypus gymnurus, recently described by Kölliker[93], had a transversely oval placenta, which occupied the upper ²⁄₃rds of the uterus. In Manis, as Dr Sharpey has shewn, the placenta is diffused over the surfaces of the chorion and uterine mucosa. In Myrmecophaga and Tamandua, as MM. Milne Edwards have pointed out, the placenta is set on the chorion in a dome-like manner. In the Sloths, as I have elsewhere described, the placenta is dome-like in its general form, and consists of a number of aggregated, discoid lobes. In Orycteropus, as I have now shewn, the placenta is broadly zonular.”

Lemuridæ. The Lemurs in spite of their affinities with the Primates and Insectivora have, as has been shewn by Milne Edwards and Turner, an apparently very different form of placenta. There is only one embryo, which occupies the body and one of the cornua of the uterus. The yolk-sack disappears early, and the allantois (Turner) bulges out into a right and left lobe, which meet above the back of the embryo. The cavity of the allantois persists, and the mesoblast of the outer wall fuses with the subzonal membrane (the hypoblastic epithelium remaining distinct) to give rise to the chorion.

On the surface of the chorion are numerous vascular villi, which fit into uterine crypts. They are generally distributed, though absent at the two ends of the chorion and opposite the os uteri. Their distribution accords with Turner’s diffused type. Patches bare of villi correspond with smooth areas on the surface of the uterine mucosa in which numerous utricular glands open. There is no reflexa.

Although the Lemurian type of placenta undoubtedly differs from that of the Primates, it must be borne in mind that the placenta of the Primates may easily be conceived to be derived from a Lemurian form of placenta. It will be remembered that in Man, before the true placenta becomes developed, there is a condition with simple vascular villi scattered over the chorion. It seems very probable that this is a repetition of the condition of the placenta of the ancestors of the Primates which has probably been more or less retained by the Lemurs. It was mentioned above that the resemblance between the metadiscoidal placenta of Man and that of the Cheiroptera, Insectivora and Rodentia is rather physiological than morphological.

Comparative histology of the Placenta.

It does not fall within the province of this work to treat from a histological standpoint the changes which take place in the uterine walls during pregnancy. It will, however, be convenient to place before the reader a short statement of the relations between the maternal and fœtal tissues in the different varieties of placenta. This subject has been admirably dealt with by Turner (No. 222), from whose paper fig. 161 illustrating this subject is taken.

The simplest known condition of the placenta is that found in the pig (B). The papilla-like fœtal villi fit into the maternal crypts. The villi (v) are formed of a connective tissue cone with capillaries, and are covered by a layer of very flat epithelium (e) derived from the subzonal membrane. The maternal crypts are lined by the uterine epithelium (e´), immediately below which is a capillary flexus. The maternal and fœtal vessels are here separated by a double epithelial layer. The same general arrangement holds good in the diffused placentæ of other forms, and in the polycotyledonary placenta of the Ruminantia, but the fœtal villi (C) in the latter acquire an arborescent form. The maternal vessels retain the form of capillaries.

In the deciduate placenta a considerably more complicated arrangement is usually found. In the typical zonary placenta of the fox and cat (D and E), the maternal tissue is broken up into a complete trabecular mesh-work, and in the interior of the trabeculæ there run dilated maternal capillaries (d´). The trabeculæ are covered by a more or less columnar uterine epithelium (e´), and are in contact on every side with fœtal villi. The capillaries of the fœtal villi preserve their normal size, and the villi are covered by a flat epithelial layer (e).

In the sloth (F) the maternal capillaries become still more dilated, and the epithelium covering them is formed of very flat polygonal cells.

In the human placenta (G), as in that of Apes, the greatest modification is found in that the maternal vessels have completely lost their capillary form, and have become expanded into large freely communicating sinuses (d´). In these sinuses the fœtal villi hang for the most part freely, though occasionally attached to their walls (t). In the late stages of fœtal life there is only one epithelial layer (e´) between the maternal and fœtal vessels, which closely invests the fœtal villi, but, as shewn by Turner and Ercolani, is part of the uterine tissue. In the fœtal villi the vessels retain their capillary form.

Evolution of the Placenta.

From Owen’s observations on the Marsupials it is clear that the yolk-sack in this group plays an important, if not the most important part, in absorbing the maternal nutriment destined for the fœtus. The fact that in Marsupials both the yolk-sack and the allantois are functional in rendering the chorion vascular makes it à priori probable that this was also the case in the primitive types of the Placentalia, and this deduction is supported by the fact that in the Rodentia, Insectivora and Cheiroptera this peculiarity of the fœtal membranes is actually found. In the primitive Placentalia there was probably present a discoidal allantoic region of the chorion, from which simple fœtal villi, like those of the pig (fig. 161 B), projected into uterine crypts; but it is not certain how far the umbilical part of the chorion, which was no doubt vascular, may also have been villous. From such a primitive type of fœtal membranes divergences in various directions have given rise to the types of fœtal membranes now existing.

In a general way it may be laid down that variations in any direction which tended to increase the absorbing capacities of the chorion would be advantageous. There are two obvious ways in which this might be done, viz. (1) by increasing the complexity of the fœtal villi and maternal crypts over a limited area, (2) by increasing the area of the part of the chorion covered by placental villi. Various combinations of the two processes would also of course be advantageous.

The most fundamental change which has taken place in all the existing Placentalia is the exclusion of the umbilical vesicle from any important function in the nutrition of the fœtus.

The arrangement of the fœtal parts in the Rodentia, Insectivora and Cheiroptera may be directly derived from the primitive form by supposing the villi of the discoidal placental area to have become more complex, so as to form a deciduate discoidal placenta; while the yolk-sack still plays a part, though physiologically an unimportant part, in rendering the chorion vascular.

In the Carnivora again we have to start from the discoidal placenta, as shewn by the fact that the allantoic region of the placenta is at first discoidal (p. 248). A zonary deciduate placenta indicates an increase both in area and in complexity. The relative diminution of the breadth of the placental zone in late fœtal life in the zonary placenta of the Carnivora is probably due to its being on the whole advantageous to secure the nutrition of the fœtus by insuring a more intimate relation between the fœtal and maternal parts, than by increasing their area of contact. The reason of this is not obvious, but as mentioned below, there are other cases where it can be shewn that a diminution in the area of the placenta has taken place, accompanied by an increase in the complexity of its villi.

The second type of differentiation from the primitive form of discoidal placenta is illustrated by the Lemuridæ, the Suidæ, and Manis. In all these cases the area of the placental villi appears to have increased so as to cover nearly the whole subzonal membrane, without the villi increasing to any great extent in complexity. From the diffused placenta covering the whole surface of the chorion, differentiations appear to have taken place in various directions. The metadiscoidal placenta of Man and Apes, from its mode of ontogeny (p. 248), is clearly derived from a diffused placenta—very probably similar to that of Lemurs—by a concentration of the fœtal villi, which are originally spread over the whole chorion, to a disc-shaped area, and by an increase in their arborescence.

The polycotyledonary forms of placenta are due to similar concentrations of the fœtal villi of an originally diffused placenta.

In the Edentata we have a group with very varying types of placenta. Very probably these may all be differentiations within the group itself from a diffused placenta, such as that found in Manis. The zonary placenta of Orycteropus is capable of being easily derived from that of Manis, by the disappearance of the fœtal villi at the two poles of the ovum. The small size of the umbilical vesicle in Orycteropus indicates that its discoidal placenta is not, like that in Carnivora, directly derived from a type with both allantoic and umbilical vascularization of the chorion. The discoidal and dome-shaped placentæ of the Armadilloes, Myrmecophaga, and the Sloths may easily have been formed from a diffused placenta, just as the discoidal placenta of the Simiadæ and Anthropidæ appears to have been formed from a diffused placenta like that of the Lemuridæ.

The presence of zonary placentæ in Hyrax and Elephas does not necessarily afford any proof of affinity of these types with the Carnivora. A zonary placenta may quite easily be derived from a diffused placenta; and the presence of two villous patches at the poles of the chorion in Elephas indicates that this was very probably the case with the placenta of this form.

Although it is clear from the above considerations that the placenta is capable of being used to some extent in classification, yet at the same time the striking resemblances which can exist between such essentially different forms of placenta, as for instance those of Man and the Rodentia, are likely to prevent it being employed, except in conjunction with other characters.

Special types of development.

The Guinea-pig, Cavia cobaya. Many years ago Bischoff (No. 176) shewed that the development of the guinea-pig was strikingly different from that of other Mammalia. His statements, which were at first received with some doubt, have been in the main fully confirmed by Hensen (No. 182) and Schäfer (No. 190), but we are still as far as ever from explaining the mystery of the phenomenon.

The ovum, enclosed by the zona radiata, passes into the Fallopian tube and undergoes a segmentation which has not been studied with great detail. On the close of segmentation, about six days after impregnation, it assumes (Hensen) a vesicular form not unlike that of other Mammalia. To the inner side of one wall of this vesicle is attached a mass of granular cells similar to the hypoblastic mass in the blastodermic vesicle of the rabbit. The egg still lies freely in the uterus, and is invested by its zona radiata. The changes which next take place are in spite of Bischoff’s, Reichert’s (No. 188) and Hensen’s observations still involved in great obscurity. It is certain, however, that during the course of the seventh day a ring-like thickening of the uterine mucous membrane, on the free side of the uterus, gives rise to a kind of diverticulum of the uterine cavity, in which the ovum becomes lodged. Opposite the diverticulum the mucous membrane of the mesometric side of the uterus also becomes thickened, and this thickening very soon (shortly after the seventh day) unites with the wall of the diverticulum, and completely shuts off the ovum in a closed capsule.

The history of the ovum during the earlier period of its inclusion in the diverticulum of the uterine wall is not satisfactorily elucidated. There appears in the diverticulum during the eighth and succeeding days a cylindrical body, one end of which is attached to the uterine walls at the mouth of the diverticulum. The opposite end of the cylinder is free, and contains a solid body.

With reference to the nature of this cylinder two views have been put forward. Reichert and Hensen regard it as an outgrowth of the uterine wall, while the body within its free apex is regarded as the ovum. Bischoff and Schäfer maintain that the cylinder itself is the ovum attached to the uterine wall. The observations of the latter authors, and especially those of Schäfer, appear to me to speak for the correctness of their view[94].

The cylinder gradually elongates up to the twelfth day. Before this period it becomes attached by its base to the mesometric thickening of the uterus, and enters into vascular connection with it. During its elongation it becomes hollow, and is filled with a fluid not coagulable in alcohol, while the body within its apex remains unaltered till the tenth day.

On this day a cavity develops in the interior of this body which at the same time enlarges itself. The greater part of its wall next attaches itself to the free end of the cylinder, and becomes considerably thickened. The remainder of the wall adjoining the cavity of the cylinder becomes a comparatively thin membrane. At the free end of the cylinder there appears on the thirteenth day an embryonic area similar to that of other Mammalia. It is at first round but soon becomes pyriform, and in it there appear a primitive streak and groove; and on their appearance it becomes obvious that the outer layer of the cylinder is the hypoblast[95], instead of, as in all other Mammalia, the epiblast; and that the epiblast is formed by the wall of the inner vesicle, i.e. the original solid body placed at the end of the cylinder. Thus the dorsal surface of the embryo is turned inwards, and the ventral surface outwards, and the ordinary position of the layers is completely inverted.

The previously cylindrical egg next assumes a spherical form, and the mesoblast arises in connection with the primitive streak in the manner already described. A splanchnic layer of mesoblast attaches itself to the inner side of the outer hypoblastic wall of the egg, a somatic layer to the epiblast of the inner vesicle, and a mass of mesoblast grows out into the cavity of the larger vesicle forming the commencement of the allantois. The general structure of the ovum at this stage is represented on fig. 162, copied from Schäfer; and the condition of the whole ovum will best be understood by a description of this figure.

It is seen to consist of two vesicles, (1) an outer larger one (h)—the original egg-cylinder—united to the mesometric wall of the uterus by a vascular connection at m´m´, and (2) an inner smaller one (ev)—the originally solid body at the free end of the egg-cylinder. The outer vesicle is formed of (1) an external lining of columnar hypoblast (h) which is either pierced or invaginated at the area of vascular connection with the uterus, and (2) of an inner layer of splanchnic mesoblast (m´´) which covers without a break the vascular uterine growth. At the upper pole of the ovum is placed the smaller epiblastic vesicle, and where the two vesicles come together is situated the embryonic area with the primitive streak (f), and the medullary plate seen in longitudinal section. The thinner wall of the inner vesicle is formed of epiblast and somatic mesoblast, and covers over the dorsal face of the embryo just like the amnion. It is in fact usually spoken of as the amnion. The large cavity of the outer vesicle is continuous with the body cavity, and into it projects the solid mesoblastic allantois (all), so far without hypoblast[96].

The outer vesicle corresponds exactly with the yolk-sack, and its mesoblastic layer receives the ordinary vascular supply.

The embryo becomes folded off from the yolk-sack in the usual way, but comes to lie not outside it as in the ordinary form, but in its interior, and is connected with it by an umbilical stalk. The yolk-sack forms the substitute for part of the subzonal membrane of other Mammalia. The so-called amnion appears to me from its development and position rather to correspond with the non-embryonic part of the epiblastic wall (true subzonal membrane) of the blastodermic vesicle of the ordinary mammalian forms than with the true amnion; and a true amnion would seem not to be developed.

The allantois meets the yolk-sack on about the seventeenth day at the region of its vascular connection with the uterine wall, and gives rise to the placenta. A diagrammatic representation of the structure of the embryo at this stage is given in fig. 163.

The peculiar inversion of the layers in the Guinea-pig has naturally excited the curiosity of embryologists, but as yet no satisfactory explanation has been offered of it.

At the time when the ovum first becomes fixed it will be remembered that it resembles the early blastodermic vesicle of the Rabbit, and it is natural to suppose that the apparently hypoblastic mass attached to the inner wall of the vesicle becomes the solid body at the end of the egg-cylinder. This appears to be Bischoff’s view, but, as shewn above, the solid mass is really the epiblast! Is it conceivable that the hypoblast in one species becomes the epiblast in a closely allied species? To my mind it is not conceivable, and I am reduced to the hypothesis, put forward by Hensen, that in the course of the attachment of the ovum to the wall of the uterus a rupture of walls of the blastodermic vesicle takes place, and that they become completely turned inside out. It must be admitted, however, that in the present state of our knowledge of the development of the ovum on the seventh and eighth days it is not possible to frame a satisfactory explanation how such an inversion can take place.

The Human Embryo. Our knowledge as to the early development of the human embryo is in an unsatisfactory state. The positive facts we know are comparatively few, and it is not possible to construct from them a history of the development which is capable of satisfactory comparison with that in other forms, unless all the early embryos known are to be regarded as abnormal. The most remarkable feature in the development, which was first clearly brought to light by Allen Thomson in 1839, is the very early appearance of branched villi. In the last few years several ova, even younger than those described by Allen Thomson, have been met with, which exhibit this peculiarity.

The best-preserved of these ova is one described by Reichert (No. 237). This ovum, though probably not more than thirteen days old, was completely enclosed by a decidua reflexa. It had (fig. 164 A and B) a flattened oval form, measuring in its two diameters 5.5 mm. and 3.5 mm. The edge was covered with branched villi, while in the centre of each of the flattened surfaces there was a spot free from villi. On the surface adjoining the uterine wall was a darker area (e) formed of two layers of cells, which is interpreted by Reichert as the embryonic area, while the membrane forming the remainder of the ovum, including the branched villi, was stated by Reichert to be composed of a single row of epithelial cells.

Whether or no Reichert is correct in identifying his darker spot as the embryonic area, it is fairly certain from the later observations of Beigel and Löwe (No. 228), Ahlfeld (No. 227), and Kollmann (No. 234) on ova nearly as young as that of Reichert, that the wall of very young ova has a more complicated structure than Reichert is willing to admit. These authors do not however agree amongst themselves, but from Kollmann’s description, which appears to me the most satisfactory, it is probable that it is composed of an outer epithelial layer, and an inner layer of connective tissue, and that the connective tissue extends at a very early period into the villi; so that the latter are not hollow, as Reichert supposed them to be.

The villi, which at first leave the flattened poles free, seem soon to extend first over one of the flat sides, and finally over the whole ovum (fig. 164 C).

Unless the two-layered region of Reichert’s ovum is the embryonic area, nothing which can clearly be identified as an embryo has been detected in these early ova. In an ovum described by Breus (No. 228), and in one described long ago by Wharton-Jones a mass found in the interior of the egg may perhaps be interpreted (His) as the remains of the yolk. It is, however, very probable that all the early ova so far discovered are more or less pathological.

The youngest ovum with a distinct embryo is one described by His (No. 232). This ovum, which is diagrammatically represented in fig. 168 in longitudinal section, had the form of an oval vesicle completely covered by villi, and about 8.5 mm. and 5.5 mm. in its two diameters, and flatter on one side than on the other. An embryo with a yolk-sack was attached to the inner side of the flatter wall of the vesicle by a stalk, which must be regarded as the allantoic stalk[97], and the embryo and yolk-sack filled up but a very small part of the whole cavity of the vesicle.

The embryo, which was probably not quite normal (fig. 165 A), was very imperfectly developed; a medullary plate was hardly indicated, and, though the mesoblast was unsegmented, the head fold, separating the embryo from the yolk-sack (um), was already indicated. The amnion (am) was completely formed, and vitelline vessels had made their appearance.

Two embryos described by Allen Thomson (No. 239) are but slightly older than the above embryos of His. Both of them probably belong to the first fortnight of pregnancy. In both cases the embryo was more or less folded off from the yolk-sack, and in one of them the medullary groove was still widely open, except in the region of the neck (fig. 165 B). The allantoic stalk, if present, was not clearly made out, and the condition of the amnion was also not fully studied. The smaller of the two ova was just 6 mm. in its largest diameter, and was nearly completely covered with simple villi, more developed on one side than on the other.

In a somewhat later period, about the stage of a chick at the end of the second day, the medullary folds are completely closed, the region of the brain already marked, and the cranial flexure commencing. The mesoblast is divided up into numerous somites, and the mandibular and first two branchial arches are indicated. The embryo is still but incompletely folded off from the yolk-sack below.

In a still older stage the cranial flexure becomes still more pronounced, placing the mid-brain at the end of the long axis of the body. The body also begins to be ventrally curved (fig. 165 C).

Externally human embryos at this age are characterised by the small size of the anterior end of the head.

The flexure goes on gradually increasing, and in the third week of pregnancy in embryos of about 4 mm. the limbs make their appearance. The embryo at this stage (fig. 166), which is about equivalent to that of a chick on the fourth day, resembles in almost every respect the normal embryos of the Amniota. The cranial flexure is as pronounced as usual, and the cerebral region has now fully the normal size. The whole body soon becomes flexed ventrally, and also somewhat spirally. The yolk-sack (b) forms a small spherical appendage with a long wide stalk, and the embryo (B) is attached by an allantoic stalk with a slight swelling (all), probably indicating the presence of a small hypoblastic diverticulum, to the inner face of the chorion.

A remarkable exception to the embryos generally observed is afforded by an embryo which has been described by Krause (No. 235). In this embryo, which probably belongs to the third week of pregnancy, the limbs were just commencing to be indicated, and the embryo was completely covered by an amnion, but instead of being attached to the chorion by an allantoic cord, it was quite free, and was provided with a small spherical sack-like allantois, very similar to that of a fourth-day chick, projected from its hind end.

No details are given as to the structure of the chorion or the presence of villi upon it. The presence of such an allantois at this stage in a human embryo is so unlike what is usually found that Krause’s statements have been received with considerable scepticism. His even holds that the embryo is a chick embryo, and not a human one; while Kölliker regards Krause’s allantois as a pathological structure. The significance to be attached to this embryo is dealt with below.

A detailed history of the further development of the human embryo does not fall within the province of this work; while the later changes in the embryonic membranes have already been dealt with (pp. 244-248).

For the changes which take place on the formation of the face I may refer the reader to fig. 167.

The most obscure point connected with the early history of the human ovum concerns the first formation of the allantois, and the nature of the villi covering the surface of the ovum. The villi, if really formed of mesoblast covered by epiblast, have the true structure of chorionic villi; and can hardly be compared to the early villi of the dog which are derived from the subzonal membrane, and still less to those of the rabbit formed from the zona radiata.

Unless all the early ova so far described are pathological, it seems to follow that the mesoblast of the chorion is formed before the embryo is definitely established, and even if the pathological character of these ova is admitted, it is nevertheless probable (leaving Krause’s embryo out of account), as shewn by the early embryos of Allen Thomson and His, that it is formed before the closure of the medullary groove. In order to meet this difficulty His supposes that the embryo never separates from the blastodermic vesicle, but that the allantoic stalk of the youngest embryo (fig. 168) represents the persistent attachment between the two[98]. His’ view has a good deal to be said for it. I would venture, however, to suggest that Reichert’s embryonic area is probably not in the two-layered stage, but that a mesoblast has already become established, and that it has grown round the inner face of the blastodermic vesicle from the (apparent) posterior end of the primitive streak. This growth I regard as a precocious formation of the mesoblast of the allantois—an exaggeration of the early formation of the allantoic mesoblast which is characteristic of the Guinea-pig (vide p. 264). This mesoblast, together with the epiblast, forms a true chorion, so that in fig. 168, and probably also in fig. 164 A and B, a true chorion has already become established. The stalk connecting the embryo with the chorion in His’ earliest embryo (fig. 168) is therefore a true allantoic stalk into which the hypoblastic allantoic diverticulum grows in for some distance. How the yolk-sack (umbilical vesicle) is formed is not clear. Perhaps, as suggested by His, it arises from the conversion of a solid mass of primitive hypoblast directly into a yolk-sack. The amnion is probably formed as a fold over the head end of the embryo in the manner indicated in His’ diagram (fig. 168 Am).

These speculations have so far left Krause’s embryo out of account. How is this embryo to be treated? Krause maintains that all the other embryos shewing an allantoic stalk at an early age are pathological. This, though not impossible, appears to me, to say the least of it, improbable; especially when it is borne in mind that embryos, which have every appearance of being normal, of about the same age and younger than Krause’s, have been frequently observed, and have always been found attached to the chorion by an allantoic stalk.

We are thus provisionally reduced to suppose either that the structure figured by Krause is not the allantois, or that it is a very abnormal allantois. It is perhaps just possible that it may be an abnormally developed hypoblastic vesicle of the allantois artificially detached from the mesoblastic layer,—the latter having given rise to the chorion at an earlier date.

Bibliography.

General.

(168) K. E. von Baer. Ueb. Entwicklungsgeschichte d. Thiere. Königsberg, 1828-1837.
(169) Barry. “Researches on Embryology.” First Series. Philosophical Transactions, 1838, Part II. Second Series, Ibid. 1839, Part II. Third Series, Ibid. 1840.
(170) Ed. van Beneden. La maturation de l'œuf, la fécondation et les premières phases du développement embryonaire d. Mammifères. Bruxelles, 1875.
(171) Ed. van Beneden. “Recherches sur l'embryologie des Mammifères.” Archives de Biologie, Vol. I. 1880.
(172) Ed. v. Beneden and Ch. Julin. “Observations sur la maturation etc. de l'œuf chez les Cheiroptères.” Archives de Biologie, Vol. I. 1880.
(173) Th. L. W. Bischoff. Entwicklungsgeschichte d. Säugethiere u. des Menschen. Leipzig, 1842.
(174) Th. L. W. Bischoff. Entwicklungsgeschichte des Kanincheneies. Braunschweig, 1842.
(175) Th. L. W. Bischoff. Entwicklungsgeschichte des Hundeeies. Braunschweig, 1845.
(176) Th. L. W. Bischoff. Entwicklungsgeschichte des Meerschweinchens. Giessen, 1852.
(177) Th. L. W. Bischoff. Entwicklungsgeschichte des Rehes. Giessen, 1854.
(178) Th. L. W. Bischoff. “Neue Beobachtungen z. Entwicklungsgesch. des Meerschweinchens.” Abh. d. bayr. Akad., Cl. II. Vol. X. 1866.
(179) Th. L. W. Bischoff. Historisch-kritische Bemerkungen z. d. neuesten Mittheilungen üb. d. erste Entwick. d. Säugethiereier. München, 1877.
(180) M. Coste. Embryogénie comparée. Paris, 1837.
(181) E. Haeckel. Anthropogenie, Entwicklungsgeschichte des Menschen. Leipzig, 1874.
(182) V. Hensen. “Beobachtungen üb. d. Befrucht. u. Entwick. d. Kaninchens u. Meerschweinchens.” Zeit. f. Anat. u. Entwick., Vol. I. 1876.
(183) A. Kölliker. Entwicklungsgeschichte d. Menschen u. d. höheren Thiere. Leipzig, 1879.
(184) A. Kölliker. “Die Entwick. d. Keimblätter des Kaninchens.” Zoologischer Anzeiger, Nos. 61, 62, Vol. III. 1880.
(185) N. Lieberkühn. Ueber d. Keimblätter d. Säugethiere. Doctor-Jubelfeier d. Herrn. H. Nasse. Marburg, 1879.
(186) N. Lieberkühn. “Z. Lehre von d. Keimblättern d. Säugethiere.” Sitz. d. Gesell. z. Beförd d. gesam. Naturwiss. Marburg, No. 3. 1880.
(187) Rauber. “Die erste Entwicklung d. Kaninchens.” Sitzungsber. d. naturfor. Gesell. z. Leipzig. 1875.
(188) C. B. Reichert. “Entwicklung des Meerschweinchens.” Abh. der. Berl. Akad. 1862.
(189) E. A. Schäfer. “Description of a Mammalian ovum in an early condition of development.” Proc. Roy. Soc., No. 168. 1876.
(190) E. A. Schäfer. “A contribution to the history of development of the guinea-pig.” Journal of Anat. and Phys., Vol. X. and XI. 1876 and 1877.

Fœtal Membranes and Placenta.

(191) John Anderson. Anatomical and Zoological Researches in Western Yunnan. London, 1878.
(192) K. E. von Baer. Untersuchungen über die Gefässverbindung zwischen Mutter und Frucht, 1828.
(193) C. G. Carus. Tabulae anatomiam comparativam illustrantes. 1831, 1840.
(194) H. C. Chapman. “The placenta and generative apparatus of the Elephant.” Journ. Acad. Nat. Sc., Philadelphia. Vol. VIII. 1880.
(195) C. Creighton. “On the formation of the placenta in the guinea-pig.” Journal of Anat. and Phys., Vol. XII. 1878.
(196) Ecker. Icones Physiologicae. 1852-1859.
(197) G. B. Ercolani. The utricular glands of the uterus, etc., translated from the Italian under the direction of H. O. Marcy. Boston, 1880. Contains translations of memoirs published in the Mem. dell'Accad. d. Scienze d. Bologna, and additional matter written specially for the translation.
(198) G. B. Ercolani. Nuove ricerche sulla placenta nei pesci cartilaginosi e nei mammiferi. Bologna, 1880.
(199) Eschricht. De organis quae respirationi et nutritioni fœtus Mammalium inserviunt. Hafniae, 1837.
(200) A. H. Garrod and W. Turner. “The gravid uterus and placenta of Hyomoschus aquaticus.” Proc. Zool. Soc., London, 1878.
(201) P. Harting. Het ei en de placenta van Halicore Dugong. Inaug. diss. Utrecht. “On the ovum and placenta of the Dugong.” Abstract by Prof. Turner. Journal of Anat. and Phys., Vol. XIII.
(202) Th. H. Huxley. The Elements of Comparative Anatomy. London, 1864.
(203) A. Kölliker. “Ueber die Placenta der Gattung Tragulus.” Verh. der Würzb. phys.-med. Gesellschaft, Bd. X.
(204) C. D. Meigs. “On the reproduction of the Opossum (Didelphis Virginiana).” Amer. Phil. Soc. Trans., Vol. X. 1853.
(205) H. Milne-Edwards. “Sur la Classification Naturelle.” Ann. Sciences Nat., Sér. 3, Vol. I. 1844.
(206) Alf. Milne-Edwards. “Recherches sur la famille des Chevrotains.” Ann. des Sciences Nat., Séries V., Vol. II. 1864.
(207) Alf. Milne-Edwards. “Observations sur quelques points de l'Embryologie des Lemuriens, etc.” Ann. Sci. Nat., Sér. V., Vol. XV. 1872.
(208) Alf. Milne-Edwards. “Sur la conformation du placenta chez le Tamandua.” Ann. des Sci. Nat., XV. 1872.
(209) Alf. Milne-Edwards. “Recherches s. l. enveloppes fœtales du Tatou à neuf bandes.” Ann. Sci. Nat., Sér. VI., Vol. VIII. 1878.
(210) R. Owen. “On the generation of Marsupial animals, with a description of the impregnated uterus of the Kangaroo.” Phil. Trans., 1834.
(211) R. Owen. “Description of the membranes of the uterine fœtus of the Kangaroo.” Mag. Nat. Hist., Vol. I. 1837.
(212) R. Owen. “On the existence of an Allantois in a fœtal Kangaroo (Macropus major).” Zool. Soc. Proc., v. 1837.
(213) R. Owen. “Description of the fœtal membranes and placenta of the Elephant.” Phil. Trans., 1857.
(214) R. Owen. On the Anatomy of Vertebrates, Vol. III. London, 1868.
(215) G. Rolleston. “Placental structure of the Tenrec, etc.” Transactions of the Zoological Society, Vol. V. 1866.
(216) W. Turner. “Observations on the structure of the human placenta.” Journal of Anat. and Phys., Vol. VII. 1868.
(217) W. Turner. “On the placentation of the Cetacea.” Trans. Roy. Soc. Edinb., Vol. XXVI. 1872.
(218) W. Turner. “On the placentation of Sloths (Cholœpus Hoffmanni).” Trans. of R. Society of Edinburgh, Vol. XXVII. 1875.
(219) W. Turner. “On the placentation of Seals (Halichœrus gryphus).” Trans. of R. Society of Edinburgh, Vol. XXVII. 1875.
(220) W. Turner. “On the placentation of the Cape Ant-eater (Orycteropus capensis).” Journal of Anat. and Phys., Vol. X. 1876.
(221) W. Turner. Lectures on the Anatomy of the Placenta. First Series. Edinburgh, 1876.
(222) W. Turner. “Some general observations on the placenta, with special reference to the theory of Evolution.” Journal of Anat. and Phys., Vol. XI. 1877.
(223) W. Turner. “On the placentation of the Lemurs.” Phil. Trans., Vol. 166, p. 2. 1877.
(224) W. Turner. “On the placentation of Apes.” Phil. Trans., 1878.
(225) W. Turner. “The cotyledonary and diffused placenta of the Mexican deer (Cervus Americanus).” Journal of Anat. and Phys., Vol. XIII. 1879.

Human Embryo.

(226) Fried. Ahlfeld. “Beschreibung eines sehr kleinen menschlichen Eies.” Archiv f. Gynaekologie, Bd. XIII. 1878.
(227) Herm. Beigel und Ludwig Loewe. “Beschreibung eines menschlichen Eichens aus der zweiten bis dritten Woche der Schwangerschaft.” Archiv f. Gynaekologie, Bd. XII. 1877.
(228) K. Breus. “Ueber ein menschliches Ei aus der zweiten Woche der Gravidität.” Wiener medicinische Wochenschrift, 1877.
(229) M. Coste. Histoire générale et particulière du développement des corps organisés, 1847-59.
(230) A. Ecker. Icones Physiologicae. Leipzig, 1851-1859.
(231) V. Hensen. “Beitrag z. Morphologie d. Körperform u. d. Gehirns d. menschlichen Embryos.” Archiv f. Anat. u. Phys., 1877.
(232) W. His. Anatomie menschlicher Embryonen, Part I. Embryonen d. ersten Monats. Leipzig, 1880.
(233) J. Kollmann. “Die menschlichen Eier von 6 MM. Grösse.” Archiv f. Anat. und Phys., 1879.
(234) W. Krause. “Ueber d. Allantois d. Menschen.” Archiv f. Anat. und Phys., 1875.
(235) W. Krause. “Ueber zwei frühzeitige menschliche Embryonen.” Zeit. f. wiss. Zool., Vol. XXXV. 1880.
(236) L. Loewe. “Im Sachen der Eihäute jüngster menschlicher Eier.” Archiv für Gynaekologie, Bd. XIV. 1879.
(237) C. B. Reichert. “Beschreibung einer frühzeitigen menschlichen Frucht im bläschenformigen Bildungszustande (sackformiger Keim von Baer) nebst vergleichenden Untersuchungen über die bläschenformigen Früchte der Säugethiere und des Menschen.” Abhandlungen der königl. Akad. d. Wiss. zu Berlin, 1873.
(238) Allen Thomson. “Contributions to the history of the structure of the human ovum and embryo before the third week after conception; with a description of some early ova.” Edinburgh Med. Surg. Journal, Vol. LII. 1839.

Although the preceding chapters of this volume contain a fairly detailed account of the early developmental stages of different groups of the Chordata, it will nevertheless be advantageous to give at this place a short comparative review of the whole subject.

In this review only the most important points will be dwelt upon, and the reader is referred for the details of the processes to the sections on the development of the individual groups.

The subject may conveniently be treated under three heads.
(1) The formation of the gastrula and behaviour of the blastopore: together with the origin of the hypoblast.
(2) The mesoblast and notochord.
(3) The epiblast.

At the close of the chapter is a short summary of the organs derived from the several layers, together with some remarks on the growth in length of the vertebrate embryo, and some suggestions as to the origin of the allantois and amnion.

Formation of the gastrula. Amphioxus is the type in which the developmental phenomena are least interfered with by the presence of food-yolk.

In this form the segmentation results in a uniform, or nearly uniform, blastosphere, one wall of which soon becomes thickened and invaginated, giving rise to the hypoblast; while the larva takes the form of a gastrula, with an archenteric cavity opening by a blastopore. The blastopore rapidly narrows, while the embryo assumes an elongated cylindrical form with the blastopore at its hinder extremity (fig. 169 A). The blastopore now passes to the dorsal surface, and by the flattening of this surface a medullary plate is formed extending forwards from the blastopore (fig. 169 B). On the formation of the medullary groove and its conversion into a canal, the blastopore opens into this canal, and gives rise to a neurenteric passage, leading from the neural canal into the alimentary tract (fig. 169 C and E). At a later period this canal closes, and the neural and alimentary canals become separated.

Such is the simple history of the layers in Amphioxus. In the simplest types of Ascidians the series of phenomena is almost the same, but the blastopore assumes a more definitely dorsal position.

Here also the blastopore lies at the hinder end of the medullary groove, and on the closure of the groove becomes converted into a neurenteric passage.

In the true Vertebrates the types which most approach Amphioxus are the Amphibia, Acipenser and Petromyzon. We may take the first of these as typical (though Petromyzon is perhaps still more so) and fig. 170 A B C D represents four diagrammatic longitudinal vertical sections through a form belonging to this group (Bombinator). The food-yolk is here concentrated in what I shall call the lower pole of the egg, which becomes the ventral aspect of the future embryo. The part of the egg containing the stored-up food-yolk is, as has already been explained in the chapter on segmentation (Vol. II. pp. 94 and 95), to be regarded as equivalent to part of those eggs which do not contain food-yolk; a fact which requires to be borne in mind in any attempt to deal comparatively with the formation of the layers in the Vertebrata. It may be laid down as a general law, which holds very accurately for the Vertebrata, that in eggs in which the distribution of food-yolk is not uniform, the size of the cells resulting from segmentation is proportional to the quantity of food-material they contain. In accordance with this law the cells of the Amphibian ovum are of unequal size even at the close of segmentation. They may roughly be divided into two categories, viz. the smaller cells of the upper pole and the larger of the lower (fig. 170 A). The segmentation cavity (sg) lies between the two, but is unsymmetrically placed near the upper pole of the egg, owing to the large bulk of the ventrally placed yolk-segments. In the inequality of the cells at the close of segmentation the Amphibia stand in contrast with Amphioxus. The upper cells are mainly destined to form the epiblast, and the lower the hypoblast and mesoblast.

The next change which takes place is an invagination, the earliest traces of which are observable in fig. 170 A. The invagination is not however so simple as in Amphioxus. Owing in fact to the presence of the food-yolk it is a mixture of invagination by epibole and by embole.

At the point marked x in fig. 170 A, which corresponds with the future hind end of the embryo, and is placed on the equatorial line marking the junction of the large and small cells, there takes place a normal invagination, which gives rise solely to the hypoblast of the dorsal wall of the alimentary tract and to part of the dorsal mesoblast. The invaginated layer grows inwards from the point x along what becomes the dorsal side of the embryo; and between it and the yolk-cells below is formed a slit-like space (fig. 170 B and C). This space is the mesenteron. It is even better shewn in fig. 171 representing the process of invagination in Petromyzon. The point x in fig. 170 where epiblast, mesoblast and hypoblast are continuous, is homologous with the dorsal lip of the blastopore in Amphioxus. In the course of the invagination the segmentation cavity, as in Amphioxus, becomes obliterated.

While the above invagination has been taking place, the epiblast cells have been simply growing in an epibolic fashion round the yolk; and by the stage represented in fig. 170 C and D the exposed surface of yolk has become greatly diminished; and an obvious blastopore is thus established. Along the line of the growth a layer of mesoblast cells (m´), continuous at the sides with the invaginated mesoblast layer, has become differentiated from the small cells (fig. 170 A) intermediate between the epiblast cells and the yolk.

Owing to the nature of the above process of invagination the mesenteron is at first only provided with an epithelial wall on its dorsal side, its ventral wall being formed of yolk-cells (fig. 170). At a later period some of the yolk-cells become transformed into the epithelial cells of the ventral wall, while the remainder become enclosed in the alimentary cavity and employed as pabulum. The whole of the yolk-cells, after the separation of the mesoblast, are however morphologically part of the hypoblast.

The final fate of the blastopore is nearly the same as in Amphioxus. It gradually narrows, and the yolk-cells which at first plug it up disappear (fig. 170 C and D). The neural groove, which becomes formed on the dorsal surface of the embryo, is continued forwards from the point x in fig. 170 C. On the conversion of this groove into a canal the canal freely opens behind into the blastopore; and a condition is reached in which the blastopore still opens to the exterior and also into the neural canal fig. 170 D. In a later stage (fig. 172) the external opening of the blastopore becomes closed by the medullary folds meeting behind it, but the passage connecting the neural and alimentary canals is left. There is one small difference between the Frog and Amphioxus in the relation of the neural canal to the blastopore. In both types the medullary folds embrace and meet behind it, so that it comes to occupy a position at the hind extremity of the medullary groove. In Amphioxus the closure of the medullary folds commences behind, so that the external opening of the blastopore is obliterated simultaneously with the commencing formation of the medullary canal; but in the Frog the closure of the medullary folds commences anteriorly and proceeds backwards, so that the obliteration of the external opening of the blastopore is a late event in the formation of the medullary canal.

The anus is formed (vide fig. 172) some way in front of the blastopore, and a postanal gut, continuous with the neurenteric canal, is thus established. Both the postanal gut and the neurenteric canal eventually disappear.

The two other types classed above with the Amphibia, viz. Petromyzon and Acipenser, agree sufficiently closely with them to require no special mention; but with reference to both types it may be pointed out that the ovum contains relatively more food-yolk than that of the Amphibian type just described, and that this leads amongst other things to the lower layer cells extending up the sides of the segmentation cavity, and assisting in forming its roof.

The next type to be considered is that of Elasmobranchii. The yolk in the ovum of these forms is enormously bulky, and the segmentation is in consequence a partial one. At first sight the differences between their development and that of Amphibia would appear to be very great. In order fully to bridge over the gulf which separates them I have given three diagrammatic longitudinal sections of an ideal form intermediate between Amphibia and Elasmobranchii, which differs however mainly from the latter in the smaller amount of food-yolk; and by their aid I trust it will be made clear that the differences between the Amphibia and Elasmobranchii are of an insignificant character. In fig. 174 A B C are represented three diagrammatic longitudinal sections of Elasmobranch embryos, and in fig. 173 A B C three longitudinal sections of the ideal intermediate form. The diagrams correspond with the Amphibian diagrams already described (fig. 170). In the first stage figured there is present in all of these forms a segmentation cavity (sg) situated not centrally but near the surface of the egg. The roof of the cavity is thin, being composed in the Amphibian embryo of epiblast alone, and in the Elasmobranch of epiblast and lower layer cells. The floor of the cavity is formed of so-called yolk, which forms the main mass of the embryo. In Amphibia the yolk is segmented. In Elasmobranchii there is at first a layer of primitive hypoblast cells separating the segmentation cavity from the yolk proper; this however soon disappears, and an unsegmented yolk with free nuclei fills the place of the segmented yolk of the Amphibia. The small cells at the sides of the segmentation cavity in Amphibia correspond exactly in function and position with the lower layer cells of the Elasmobranch blastoderm.

The relation of the yolk to the blastoderm in the Elasmobranch embryo at this stage of development very well suits the view of its homology with the yolk-cells of the Amphibian embryo. The only essential difference between the two embryos arises from the roof of the segmentation cavity being formed in the Elasmobranch embryo of lower layer cells, which are absent in the Amphibian embryo. This difference no doubt depends upon the greater quantity of yolk in the Elasmobranch ovum, and a similar distribution of the lower layer cells is found in Acipenser and in Petromyzon.

In the next stage for the Elasmobranch (fig. 173 and 174 B) and for the Amphibian (fig. 170 C) or better still Petromyzon (fig. 171) the agreement between the three types is again very close. For a small arc (x) of the edge of the blastoderm the epiblast and hypoblast become continuous, while at all other parts the epiblast, accompanied by lower layer cells, grows round the yolk or round the large cells which correspond to it. The yolk-cells of the Amphibian embryo form a comparatively small mass, and are therefore rapidly enveloped; while in the case of the Elasmobranch embryo, owing to the greater mass of the yolk, the same process occupies a long period. The portion of the blastoderm, where epiblast and hypoblast become continuous, forms the dorsal lip of an opening—the blastopore—which leads into the alimentary cavity. This cavity has the same relation in all the three cases. It is lined dorsally by lower layer cells, and ventrally by yolk-cells or what corresponds with yolk-cells; a large part of the ventral epithelium of the alimentary canal being in both cases eventually derived from the yolk. In Amphibia this epithelium is formed directly from the existing cells, while in Elasmobranchii it is derived from cells formed around the nuclei of the yolk.

As in the earlier stage, so in the present one, the anatomical relations of the yolk to the blastoderm in the one case (Elasmobranchii) are nearly identical with those of the yolk-cells to the blastoderm in the other (Amphibia).

The main features in which the two embryos differ, during the stage under consideration, arise from the same cause as the solitary point of difference during the preceding stage.

In Amphibia the alimentary cavity is formed coincidently with a true ingrowth of cells from the point where epiblast and hypoblast become continuous; and from this ingrowth the dorsal wall of the alimentary cavity is formed. The same ingrowth causes the obliteration of the segmentation cavity.

In Elasmobranchs, owing probably to the larger bulk of the lower layer cells, the primitive hypoblast cells arrange themselves in their final position during segmentation, and no room is left for a true invagination; but instead of this there is formed a simple space between the blastoderm and the yolk. The homology of this space with the primitive invagination cavity is nevertheless proved by the survival of a number of features belonging to the ancestral condition in which a true invagination was present. Amongst the more important of these are the following:—(1) The continuity of epiblast and hypoblast at the dorsal lip of the blastopore. (2) The continuous conversion of primitive hypoblast cells into permanent hypoblast, which gradually extends inwards towards the segmentation cavity, and exactly represents the course of the invagination whereby in Amphibia the dorsal wall of the alimentary cavity is formed. (3) The obliteration of the segmentation cavity during the period when the pseudo-invagination is occurring.

In the next stage there appear more important differences between the two types than in the preceding stages, though here again the points of resemblance predominate.

Figs. 170 D and 174 C represent longitudinal sections through embryos after the closure of the medullary canal. The neurenteric canal is established; and in front and behind the epithelium of the ventral wall of the mesenteron has begun to be formed.

The mesoblast is represented as having grown in between the medullary canal and the superjacent epiblast.

There are at this stage two points in which the embryo Elasmobranch differs from the corresponding Amphibian embryo. (1) In the formation of the neurenteric canal, there is no free passage leading into the mesenteron from the exterior as in Amphibia (fig. 170 D). (2) The whole yolk is not enclosed by the epiblast, and therefore part of the blastopore is still open.

The difference between Amphibia and Elasmobranchii in the first of these points is due to the fact that in Elasmobranchii, as in Amphioxus, the neural canal becomes first closed behind; and simultaneously with its closure the lateral parts of the lips of the blastopore, which are continuous with the medullary folds, meet together and shut in the hindmost part of the alimentary tract.

The second point is of some importance for understanding the relations of the formation of the layers in the amniotic and the non-amniotic Vertebrates. Owing to its large size the whole of the yolk in Elasmobranchii is not enclosed by the epiblast at the time when the neurenteric canal is established; in other words a small posterior and dorsal portion of the blastopore is shut off in the formation of the neurenteric canal. The remaining ventral portion becomes closed at a later period. Its closure takes place in a linear fashion, commencing at the hind end of the embryo, and proceeding apparently backwards; though, as this part eventually becomes folded in to form the ventral wall of the embryo, the closure of it really travels forwards. The process causes however the embryo to cease to lie at the edge of the blastoderm, and while situated at some distance from the edge, to be connected with it by a linear streak, representing the coalesced lips of the blastopore. The above process is diagrammatically represented in fig. 175 B; while as it actually occurs it is shewn in fig. 30, p. 63. The whole closure of the blastopore in Elasmobranchii is altogether unlike what takes place in Amphibia, where the blastopore remains as a circular opening which gradually narrows till it becomes completely enveloped in the medullary folds (fig. 175 A).

On the formation of the neurenteric canal the body of the embryo Elasmobranch becomes gradually folded off from the yolk, which, owing to its great size, forms a large sack appended to the ventral side of the body. The part of the somatopleure, which grows round it, is to be regarded as a modified portion of the ventral wall of the body. The splanchnopleure also envelops it, so that, morphologically speaking, the yolk lies within the mesenteron.

The Teleostei, so far as the first formation of the layers is concerned, resemble in all essential features the Elasmobranchii, but the neurenteric canal is apparently not developed (?), owing to the obliteration of the neural canal; and the roof of the segmentation cavity is formed of epiblast only.

* * * * *

In the preceding pages I have attempted to shew that the Amphibia, Acipenser, Petromyzon, the Elasmobranchii and the Teleostei agree very closely in the mode of formation of the gastrula. The unsymmetrical gastrula or pseudo-gastrula which is common to them all is, I believe, to be explained by the form of the vertebrate body. In Amphioxus, where the small amount of food-yolk present is distributed uniformly, there is no reason why the invagination and resulting gastrula should not be symmetrical. In true Vertebrates, where more food-yolk is present, the shape and structure of the body render it necessary for the food-yolk to be stored away on the ventral side of the alimentary canal. It is this fact which causes the asymmetry of the gastrula, since it is not possible for the part of the ovum, which will become the ventral wall of the alimentary tract, and which is loaded with food-yolk, to be invaginated in the same fashion as the dorsal wall.

Sauropsida. The comparison of the different types of the Ichthyopsida is fairly simple, but the comparison of the Sauropsida with the Ichthyopsida is a far more difficult matter. In all the Sauropsida there is a large food-yolk, and the segmentation agrees closely with that in the Elasmobranchii. It might have been anticipated that the resemblance would continue in the subsequent development. This however is far from being the case. The medullary plate, instead of lying at the edge of the blastoderm, lies in the centre, and its formation is preceded by that of a peculiar structure, the primitive streak, which, on the formation of the medullary plate, is found to lie at the hinder end of the latter and to connect it with the edge of the blastoderm.

The possibility of a comparison between the Sauropsida and the Elasmobranchii depends upon the explanation being possible of (1) the position of the embryo near the centre of the blastoderm, and (2) the nature of the primitive streak.

The answers to these two questions are, according to my view, intimately bound together.

I consider that the embryos of the Sauropsida have come to occupy a central position in the blastoderm owing to the abbreviation of a process similar to that by which, in Elasmobranchii, the embryo is removed from the edge of the blastoderm; and that the primitive streak represents the linear streak connecting the Elasmobranch embryo with the edge of the blastoderm after it has become removed from its previous peripheral position, as well as the true neurenteric part of the Elasmobranch blastopore.

This view of the nature of the primitive streak, which is diagrammatically illustrated in fig. 175, will be rendered more clear by a brief review of the early developmental processes in the Sauropsida.

After segmentation the blastoderm becomes divided, as in Elasmobranchii, into two layers. It is doubtful whether there is any true representative of the segmentation cavity. The first structure to appear in the blastoderm is a linear streak placed at the hind end of the blastoderm, known as the primitive streak (figs. 175 C, bl and 176, pr). At the front end of the primitive streak the epiblast and hypoblast become continuous, just as they do at the dorsal lip of the blastopore in Elasmobranchii. Continued back from this point is a streak of fused mesoblast and epiblast to the under side of which a linear thin layer of hypoblast is more or less definitely attached.

A further structure, best developed in the Lacertilia, appears in the form of a circular passage perforating the blastoderm at the front end of the primitive streak (fig. 176, ne). This passage is bounded anteriorly by the layer of cells forming the continuation of the hypoblast into the epiblast.

In the next stage the medullary plate becomes formed in front of the primitive streak (fig. 175 C), and the medullary folds are continued backwards so as to enclose the upper opening of the passage through the blastoderm. On the closure of the medullary canal (fig. 177) this passage leads from the medullary canal into the alimentary tract, and is therefore the neurenteric canal; and a postanal gut also becomes formed. The latter part of the above description applies especially to the Lizard: but in Chelonia and most Birds distinct remnants (vide pp. 162-164) of the neurenteric canal are developed.

On the hypothesis that the Sauropsidan embryos have come to occupy their central position, owing to an abbreviation of a process analogous to the linear closing of the blastopore behind the embryos of Elasmobranchii, all the appearances above described receive a satisfactory explanation. The passage at the front end of the primitive streak is the dorsal part of the blastopore, which in Elasmobranchii becomes converted into the neurenteric canal. The remainder of the primitive streak represents, in a rudimentary form, the linear streak in Elasmobranchii, formed by the coalesced edges of the blastoderm, which connects the hinder end of the embryo with the still open yolk blastopore. That it is in later stages not continued to the edge of the blastoderm, as in Elasmobranchii, is due to its being a rudimentary organ. The more or less complete fusion of the layers in the primitive streak is simply to be explained by this structure representing the coalesced edges of the blastopore; and the growth outwards from it of the mesoblast is probably a remnant of a primitive dorsal invagination of the mesoblast and hypoblast like that in the Frog.

The final enclosure of the yolk in the Sauropsida takes place at the pole of the yolk-sack opposite the embryo, so that the blastopore is formed of three parts, (1) the neurenteric canal, (2) the primitive streak behind this, (3) the blastopore at the pole of the yolk-sack opposite the embryo.

Mammalia. The features of the development of the placental Mammalia receive their most satisfactory explanation on the hypothesis that their ancestors were provided with a large-yolked ovum like that of the Sauropsida. The food-yolk must be supposed to have ceased to be developed on the establishment of a maternal nutrition through the uterus.

On this hypothesis all the developmental phenomena subsequently to the formation of the blastodermic vesicle receive a satisfactory explanation.

The whole of the blastodermic vesicle, except the embryonic area, represents the yolk-sack, and the growth of the hypoblast and then of the mesoblast round its inner wall represents the corresponding growths in the Sauropsida. As in the Sauropsida it becomes constricted off from the embryo, and the splanchnopleuric stalk of the sack opens into the ileum in the usual way.

In the formation of the embryo out of the embryonic area the phenomena which distinguish the Sauropsida from the Ichthyopsida are repeated. The embryo lies in the centre of the area; and before it is formed there appears a primitive streak, from which there grows out the greater part of the mesoblast. At the front end of the primitive streak the hypoblast and epiblast become continuous, though a perforated neurenteric blastopore has not yet been detected.

All these Sauropsidan features are so obvious that they need not be insisted on further. The embryonic evidence of the common origin of Mammalia and Sauropsida, both as concerns the formation of the layers and of the embryonic membranes, is as clear as it can be. The only difficulty about the early development of Mammalia is presented by the epibolic gastrula and the formation of the blastodermic vesicle (figs. 178 and 179). That the segmentation is a complete one is no doubt a direct consequence of the reduction of the food-yolk, but the growth of the epiblast cells round the hypoblast and the final enclosure of the latter, which I have spoken of as giving rise to the epibolic gastrula, are not so easily explained.

It might have been supposed that this process was equivalent to the growth of the blastoderm round the yolk in the Sauropsida, but then the blastopore ought to be situated at the pole of the egg opposite to the embryonic area, while, according to Van Beneden, the embryonic area corresponds approximately to the blastopore.

Van Beneden regards the Mammalian blastopore as equivalent to that in the Amphibia, but if the position previously adopted about the primitive streak is to be maintained, Van Beneden’s view must be abandoned. No satisfactory phylogenetic explanation of the Mammalian gastrula by epibole has in my opinion as yet been offered.

The formation of the blastodermic vesicle may perhaps be explained on the view that in the Proto-mammalia the yolk-sack was large, and that its blood-vessels took the place of the placenta of higher forms. On this view a reduction in the bulk of the ovarian ovum might easily have taken place at the same time that the presence of a large yolk-sack was still necessary for the purpose of affording surface of contact with the uterus.

The formation of the Mesoblast and of the Notochord.

Amphioxus. The mesoblast originates in Amphioxus, as in several primitive invertebrate types, from a pair of lateral diverticula, constricted off from the archenteron (fig. 180). Their formation commences at the front end of the body and is thence carried backwards, and each diverticulum contains a prolongation of the cavity of the archenteron. After their separation from the archenteron the dorsal parts of these diverticula become divided by transverse septa into successive somites, the cavities of which eventually disappear; while the walls become mainly converted into the muscle-plates, but also into the tissue around the notochord which corresponds with the vertebral tissue of the higher Chordata.

The ventral part of each diverticulum, which is prolonged so as to meet its fellow in the middle ventral line, does not become divided into somites, but contains a continuous cavity, which becomes the body cavity of the adult. The inner layer of this part forms the splanchnic mesoblast, and the outer layer the somatic mesoblast.

The notochord would almost appear to arise as a third median and dorsal diverticulum of the archenteron (fig. 180 ch). At any rate it arises as a central fold of the wall of this cavity, which is gradually constricted off from before backwards.

Urochorda. In simple Ascidians the above processes undergo a slight modification, which is mainly due (1) to a general simplification of the organization, and (2) to the non-continuation of the notochord into the trunk.

The whole dorsal wall of the posterior part of the archenteron is converted into the notochord (fig. 181 ch), and the lateral walls into the mesoblast (me); so that the original lumen of the posterior part of the archenteron ceases to be bounded by hypoblast cells, and disappears as such. Part of the ventral wall remains as a solid cord of cells (al´) The anterior part of the archenteron in front of the notochord passes wholly into the permanent alimentary tract.

The derivation of the mesoblast from the lateral walls of the posterior part of the archenteron is clearly comparable with the analogous process in Amphioxus.

Vertebrata. In turning from Amphioxus to the true Vertebrata we find no form in which diverticula of the primitive alimentary tract give rise to the mesoblast. There is reason to think that the type presented by the Elasmobranchii in the formation of the mesoblast is as primitive as that of any other group. In this group the mesoblast is formed, nearly coincidently with the hypoblast of the dorsal wall of the mesenteron, as two lateral sheets, one on each side of the middle line (fig. 182 m). These two sheets are at first solid masses; and their differentiation commences in front and is continued backwards. After their formation the notochord arises from the axial portion of the hypoblast (which had no share in giving rise to the two mesoblast plates) as a solid thickening (fig. 183 ch´), which is separated from it as a circular rod. Its differentiation, like that of the mesoblastic plates, commences in front. The mesoblast plates subsequently become divided for their whole length into two layers, between which a cavity is developed (fig. 184). The dorsal parts of the plates become divided by transverse partitions into somites, and these somites with their contained cavities are next separated from the more ventral parts of the plates (fig. 185 mp). In the somites the cavities become eventually obliterated, and from their inner sides plates of tissue for the vertebral bodies (fig. 186 Vr) are separated; while the outer parts, consisting of two sheets, containing the remains of the original cavity, form the muscle-plates (mp).

The undivided ventral portion gives rise to the general somatic and splanchnic mesoblast (fig. 185), and the cavity between its two layers constitutes the body cavity. The originally separate halves of the body cavity eventually meet and unite in the ventral median line throughout the greater part of the body, though in the tail they remain distinct and are finally obliterated. Dorsally they are separated by the mesentery. From the mesoblast at the junction of the dorsal and ventral parts of the primitive plates is formed the urinogenital system.

That the above mode of origin of the mesoblast and notochord is to be regarded as a modification of that observable in Amphioxus seems probable from the following considerations:—

In the first place, the mesoblast is split off from the hypoblast not as a single mass but as a pair of distinct masses, comparable with the paired diverticula in Amphioxus. Secondly, the body cavity, when it appears in the mesoblast plates, does not arise as a single cavity, but as a pair of cavities, one for each plate of mesoblast; and these cavities remain permanently distinct in some parts of the body, and nowhere unite till a comparatively late period. Thirdly, the primitive body cavity of the embryo is not confined to the region in which a body cavity exists in the adult, but extends to the summit of the muscle-plates, at first separating parts which become completely fused in the adult to form the great lateral muscles of the body.

It is difficult to understand how the body cavity could thus extend into the muscle-plates on the supposition that it represents a primitive split in the mesoblast between the wall of the gut and the body-wall; but its extension to this part is quite intelligible, on the hypothesis that it represents the cavities of two diverticula of the alimentary tract, from the muscular walls of which the voluntary muscular system has been derived; and it may be pointed out that the derivation of part of the muscular system from what is apparently splanchnic mesoblast is easily explained on the above hypothesis, but not, so far as I see, on any other.

Such are the main features, presented by the mesoblast in Elasmobranchii, which favour the view of its having originally formed the walls of the alimentary diverticula. Against this view of its nature are the facts (1) of the mesoblast plates being at first solid, and (2) of the body cavity as a consequence of this never communicating with the alimentary canal. These points, in view of our knowledge of embryological modifications, cannot be regarded as great difficulties in my hypothesis. We have many examples of organs, which, though in most cases arising as involutions, yet appear in other cases as solid ingrowths. Such examples are afforded by the optic vesicle, auditory vesicle, and probably also by the central nervous system of Osseous Fishes. In most Vertebrates these organs are formed as hollow involutions from the exterior; in Osseous Fishes, however, as solid involutions, in which a cavity is secondarily established.

There are strong grounds for thinking that in all Vertebrates the mesoblast plates on each side of the notochord originate independently, much as in Elasmobranchii, and that the notochord is derived from the axial hypoblast; but there are some difficulties in the application of this general statement to all cases. In Amphibia, Ganoids, and Petromyzon, where the dorsal hypoblast is formed by a process of invagination as in Amphioxus, the dorsal mesoblast also owes its origin to this invagination, in that the indifferent invaginated layer becomes divided into hypoblast and mesoblast. Amongst these forms the mesoblast sheet, when separated from the hypoblast, is certainly not continuous across the middle line in Petromyzon (Calberla) and the Newt (Scott and Osborn), and doubtfully so in the other forms. It arises, in fact, as in Elasmobranchii, as two independent plates. The fact of these plates originating from an invaginated layer can only be regarded in the light of an approximation to the primitive type found in Amphioxus.

In Petromyzon and the Newt the whole axial plate of dorsal hypoblast becomes separated off from the rest of the hypoblast as the notochord, and this mode of origin for the notochord resembles more closely that in Amphioxus than the mode of origin in Elasmobranchii.

In Teleostei, there is reason to think that the processes in the formation of the mesoblast accord closely with what has been described as typical for the Ichthyopsida, but there are still some points involved in obscurity.

Leaving the Ichthyopsida, we may pass to the consideration of the Sauropsida and Mammalia. In both of these types there is evidence to shew that a part of the mesoblast is formed in situ at the same time as the hypoblast, from the lower strata of segmentation spheres. This mesoblast is absent in the front part of the area pellucida, and on the formation of the primitive streak (blastopore), an outgrowth of mesoblast arises from it as in Amphibia, etc. From this region the mesoblast spreads as a continuous sheet to the sides and posterior part of the blastoderm. In the region of the embryo, its exact behaviour has not in some cases been quite satisfactorily made out. There are reasons for thinking that it appears as two sheets not united in the axial line in both Lacertilia (fig. 126) and Mammalia (fig. 187), and this to some extent holds true for Aves (vide p. 156). In Lacertilia (fig. 188) and Mammalia, the axial hypoblast becomes wholly converted into the notochord, which at the posterior end of the body is continued into the epiblast at the dorsal lip of the blastopore; while in Birds the notochord is formed by a very similar (fig. 189 ch) process.

The above processes in the formation of the mesoblast are for the most part easily explained by a comparison with the lower types. The outgrowth of the mesoblast from the sides of the primitive streak is a rudiment of the dorsal invagination of hypoblast and mesoblast found in Amphibia; and the apparent outgrowth of the mesoblast from the epiblast in the primitive streak is no more to be taken as a proof of the epiblastic origin of the mesoblast, than the continuity of the epiblast with the invaginated hypoblast and mesoblast at the lips of the blastopore in the Frog of the derivation of these layers from the epiblast in this type.

The division of the mesoblast into two plates along the dorsal line of the embryo, and the formation of the notochord from the axial hypoblast, are intelligible without further explanation. The appearance of part of the mesoblast before the formation of the primitive streak is a process of the same nature as the differentiation of hypoblast and mesoblast in Elasmobranchii without an invagination.

In the Sauropsida, some of the mesoblast of the vascular area would appear to be formed in situ out of the germinal wall, by a process of cell-formation similar to that which takes place in the yolk adjoining the blastoderm in Elasmobranchii and Teleostei. The mesoblast so formed is to be compared with that which arises on the ventral side of the embryo in the Frog, by a direct differentiation of the yolk-cells.

What was stated for the Elasmobranchii with reference to the general fate of the mesoblast holds approximately for all the other forms.

The Epiblast.

The epiblast in a large number of Chordata arises as a single row of more or less columnar cells. Since the epidermis, into which it becomes converted, is formed of two more or less distinct strata in all Chordata except Amphioxus and Ascidians, the primitive row of epiblast cells, when single, necessarily becomes divided in the course of development into two layers.

In some of the Vertebrata, viz. the Anurous Amphibia, Teleostei, Acipenser, and Lepidosteus, the epiblast is from the first formed of two distinct strata. The upper of these, formed of a single row of cells, is known as the epidermic stratum, and the lower, formed of several rows, as the nervous stratum. In these cases the two original strata of the epiblast are equivalent to those which appear at a later period in the other forms. Thus Vertebrates may be divided into groups according to the primitive condition of their epiblast, viz. a larger group with but a single stratum of cells at first; and a smaller group with two strata.

While there is no great difficulty in determining the equivalent parts of the epidermis in these two groups, it still remains an open question in which of them the epiblast retains its primitive condition.

Though it is not easy to bring conclusive proofs on the one side or the other, the balance of argument appears to me to be decidedly in favour of regarding the condition of the epiblast in the larger group as primitive, and its condition in the smaller group as secondary, and due to the throwing back of the differentiation of the epiblast to a very early period of development.

In favour of this view may be urged (1) the fact that the simple condition is retained in Amphioxus through life. (2) The correlation in Amphibia, and the other forms belonging to this group, between a closed auditory pit and the early division of the epiblast into two strata; there being no doubt that the auditory pit was at first permanently open, a condition of the epiblast which necessitates its never having an external opening must clearly be secondary. (3) It appears more likely that a particular genetic feature should be thrown back in development, than that such an important feature, as a distinction between two primary layers, should be absolutely lost during an early period of development, and then reappear in later stages.

The fact of the epiblast of the neural canal being divided, like the remainder of the layer, into nervous and epidermic parts, cannot, I think, be used as an argument in favour of the opposite view to that here maintained. It seems probable that the central canal of the nervous system arose phylogenetically as an involution from the exterior, and that the epidermis lining it is merely part of the original epidermis, which has retained its primitive structure as a simple stratum, but is naturally distinguishable from the nervous structures adjacent to it.

Where the epiblast is divided at an early period into two strata, the nervous stratum is always the active one, and takes the main share in forming all the organs derived from the layer.

Formation of the central nervous system. In all Chordata an axial strip of the dorsal epiblast, extending from the lip of the blastopore to the anterior extremity of the head, and known as the medullary plate, becomes isolated from the remainder of the layer to give rise to the central nervous axis.

According to the manner in which this takes place, three types may, however, be distinguished. In Amphioxus the axial strip becomes first detached from the adjoining epiblast, which then meets and forms a continuous layer above it (fig. 190 A and B np). The sides of the medullary plate, which is thus shut off from the surface, bend over and meet so as to convert the plate into a canal (fig. 190 C nc). In the second and ordinary type the sides of the medullary plate fold over and meet so as to form a canal before the plate becomes isolated from the external epiblast.

The third type is characteristic of Lepidosteus, Teleostei, and Petromyzon. Here the axial plate becomes narrowed in such a way that it forms a solid keel-like projection towards the ventral surface (fig. 191 Me). This keel subsequently becomes separated from the remainder of the epidermis, and a central canal is afterwards developed in it. Calberla and Scott hold that the epidermic layer of the skin is involuted into this keel in Petromyzon, and Calberla maintains the same view for Teleostei (fig. 32), but further observations on this subject are required. In the Teleostei a very shallow depression along the axis of the keel is the only indication of the medullary groove of other forms.

In Amphioxus (fig. 190), the Tunicata, Petromyzon (?), Elasmobranchii (fig. 182), the Urodela and Mammalia (fig. 187), the epiblast of the medullary plate is only formed of a single row of cells at the time when the formation of the central nervous system commences; but, except in Amphioxus and the Tunicata, it becomes several cells deep before the completion of the process. In other types the epiblast is several cells deep even before the differentiation of a medullary plate. In the Anura, the nervous layer of the epidermis alone is thickened in the formation of the central nervous system (fig. 72); and after the closure of the medullary canal, the epidermic layer fuses for a period with the nervous layer, though on the subsequent formation of the central epithelium of the nervous canal, there can be little doubt that it becomes again distinct.

It seems almost certain that the formation of the central nervous system from a solid keel-like thickening of the epidermis is a derived and secondary mode; and that the folding of the medullary plate into a canal is primitive. Apart from its greater frequency the latter mode of formation of the central nervous system is shewn to be the primitive type by the fact that it offers a simple explanation of the presence of the central canal of the nervous system; while the existence of such a canal cannot easily be explained on the assumption that the central nervous system was originally developed as a keel-like thickening of the epiblast.

It is remarkable that the primitive medullary plate rarely exhibits any indication of being formed of two symmetrical halves. Such indications are, however, found in the Amphibia (fig. 192 and fig. 72); and, since in the adult state the nervous cord exhibits nearly as distinct traces of being formed of two united strands as does the ventral nerve-cord of many Chætopods, it is quite possible that the structure of the medullary plate in Amphibia may be more primitive than that in other types[99].

Formation of the organs of special sense. The more important parts of the organs of smell, sight, and hearing are derived from the epiblast; and it has been asserted that the olfactory pit, optic vesicles and auditory pit take their origin from a special sense plate, continuous at first with this medullary plate. In my opinion this view cannot be maintained.

In the case of the group of forms in which the epiblast is early divided into nervous and epidermic layers, the former layer alone becomes involuted in the formation of the auditory pit and the lens, the external openings of which are never developed, while it is also mainly concerned in the formation of the olfactory pit.

Summary of the more important Organs derived from the three germinal layers.

The epiblast primarily gives origin to two very important parts of the body, viz. the central nervous system and the epidermis.

It is from the involuted epiblast of the neural tube that the whole of the grey and white matter of the brain and spinal cord appears to be developed, the simple columnar cells of the epiblast being directly transformed into the characteristic multipolar nerve cells. The whole of the sympathetic nervous system and the peripheral nervous elements of the body, including both the spinal and the cranial nerves and ganglia, are epiblastic in origin.

The epithelium (ciliated in the young animal) lining the canalis centralis of the spinal cord, together with that lining the ventricles of the brain, is the undifferentiated remnant of the primitive epiblast.

The epiblast also forms the epidermis; not however the dermis, which is of mesoblastic origin. The line of junction between the epiblast and the mesoblast coincides with that between the epidermis and the dermis. From the epiblast are formed all such tegumentary organs or parts of organs as are epidermic in nature.

In addition to the above, the epiblast plays an important part in the formation of the organs of special sense.

According to their mode of formation, these organs may be arranged into two divisions. In the first come the organs where the sensory expansion is derived from the involuted epiblast of the medullary canal. To this class belongs the retina, including the pigment epithelium of the choroid, which is formed from the original optic vesicle budded out from the fore-brain.

To the second class belong the epithelial expansions of the membranous labyrinth of the ear, and the cavity of the nose, which are formed by an involution of the epiblast covering the external surface of the embryo. These accordingly have no primary connection with the brain. ‘Taste bulbs’ and other terminal nervous organs, such as those of the lateral line in fishes, are also structures formed from the external epiblast.

In addition to these we have the crystalline lens formed of involuted epiblast as well as the cavity of the mouth and anus, and the glands derived from them. The pituitary body is also epiblastic in origin.

From the hypoblast are derived the epithelium of the digestive canal, the epithelium of the trachea, bronchial tubes and air cells, the cylindrical epithelium of the ducts of the liver, pancreas, thyroid body, and other glands of the alimentary canal, as well as the hepatic cells constituting the parenchyma of the liver, developed from the hypoblast cylinders given off around the primary hepatic diverticula.

Homologous probably with the hepatic cells, and equally of hypoblastic origin, are the spheroidal ‘secreting cells’ of the pancreas and other glands. The epithelium of the salivary glands, though these so closely resemble the pancreas, is probably of epiblastic origin, inasmuch as the cavity of the mouth is entirely lined by epiblast.

The hypoblast also lines the allantois. To these parts must be added the notochord and subnotochordal rod. From the mesoblast are formed all the remaining parts of the body. The muscles, the bones, the connective tissue and the vessels, both arteries, veins, capillaries and lymphatics with their appropriate epithelium, are entirely formed from the mesoblast.

The generative and urinary organs are entirely derived from the mesoblast. It is worthy of notice that the epithelium of the urinary glands, though resembling the hypoblastic epithelium of the alimentary canal, is undoubtedly mesoblastic.

From the mesoblast are lastly derived all the muscular, connective tissue, and vascular elements, as well of the alimentary canal and its appendages as of the skin and the tegumentary organs. Just as it is only the epidermic moiety of the latter which is derived from the epiblast, so it is only the epithelium of the former which comes from the hypoblast.

Growth in length of the Vertebrate Embryo.

With reference to the formation and growth in length of the body of the Vertebrate embryo two different views have been put forward, which can be best explained by taking the Elasmobranch embryo as our type. One of these views, generally held by embryologists and adopted in the previous pages, is that the Elasmobranch embryo arises from a differentiation of the edge of the blastoderm; which extends inwards from the edge for some little distance. This differentiation is supposed to contain within itself the rudiments of the whole of the embryo with the exception of the yolk-sack; and the hinder extremity of it, at the edge of the blastoderm, is regarded as corresponding with the hind end of the body of the adult. The growth in length takes place by a process of intussusception, and, till there are formed the full number of mesoblastic somites, it is effected, as in Chætopods, by the continual addition of fresh somites between the last-formed somite and the hind end of the body.

A second and somewhat paradoxical view has been recently brought into prominence by His and Rauber. This view has moreover since been taken up by many embryologists, and has led to strange comparisons between the formation of the mesoblastic plates of the Chætopods and the medullary folds of Vertebrata. According to this view the embryo grows in length by the coalescence of the two halves of the thickened edges of the blastoderm in the dorsal median line. The groove between the coalescing edges is the medullary groove, which increases in length by the continued coalescence of fresh portions of the edge of the blastoderm.

The following is His’ own statement of his view: “I have shewn that the embryo of Osseous Fishes grows together in length from two symmetrically-placed structures in the thickened edge of the blastoderm. Only the foremost end of the head and the hindermost end of the tail undergo no concrescence, since they are formed out of that part of the edge of the blastoderm which, together with the two lateral halves, completes the ring. The whole edge of the blastoderm is used in the formation of the embryo.”

The edges of the blastoderm which meet to form the body of the embryo are regarded as the blastopore, so that, on this view, the blastopore primitively extends for the whole length of the dorsal side of the embryo, and the groove between the coalesced lips becomes the medullary groove.

It is not possible for me to enter at any great length into the arguments used to support this position.

They may be summarised as (1) The general appearance; i.e. that the thickened edge of the blastoderm is continuous with the medullary fold.

(2) Certain measurements (His) which mainly appear to me to prove that the growth takes place by the addition of fresh somites between that last formed and the end of the body.

(3) Some of the phenomena of double monsters (Rauber).

None of these arguments appear to be very forcible, but as the view of His and Rauber, if true, would certainly be important, I shall attempt shortly to state the arguments against it, employing as my type the Elasmobranchii, by the development of which, according to His, the view which he adopts is more conclusively proved than by that of any other group.

(1) The general appearance of the thickened edge of the blastoderm becoming continuous with the medullary folds has been used as an argument for the medullary folds being merely the coalesced thickened edges of the blastoderm. Since, however, the medullary folds are merely parts of the medullary plate, and since the medullary plate is continuous with the adjoining epiblast of the embryonic rim, the latter structure must be continuous with the medullary folds however they are formed, and the mere fact of their being so continuous cannot be used as an argument either way. Moreover, were the concrescence theory true, the coalescing edges of the blastoderm might be expected to form an acute angle with each other, which they are far from doing.

(2) The medullary groove becomes closed behind earlier than in front, and the closure commences while the embryo is still quite short, and before the hind end has begun to project over the yolk. After the medullary canal becomes closed, and is continued behind into the alimentary canal by the neurenteric passage, it is clearly impossible for any further increase in length to take place by concrescence. If therefore His’ and Rauber’s view is accepted, it will have to be maintained that only a small part of the body is formed by concrescence, while the larger posterior part grows by intussusception. The difficulty involved in this supposition is much increased by the fact that long after the growth by concrescence must have ceased the yolk blastopore still remains open, and the embryo is still attached to the edge of the blastoderm; so that it cannot be maintained that the growth by concrescence has come to an end because the thickened edges of the blastoderm have completely coalesced.

The above are arguments derived simply from a consideration of the growth of the embryo; and they prove (1) that the points adduced by His and Rauber are not at all conclusive; (2) that the growth in length of the greater part of the body takes place by the addition of fresh somites behind, as in Chætopods, and it would therefore be extremely surprising that a small middle part of the body should grow in quite a different way.

Many minor arguments used by His might be replied to, but it is hardly necessary to do so, and some of them depend upon erroneous views as to the course of development, such as an argument about the notochord, which depends for its validity upon the assumption that the notochord ridge appears at the same time as the medullary plate, while, as a matter of fact, the ridge does not appear till considerably later. In addition to the arguments of the class hitherto used, there may be brought against the His-Rauber view a series of arguments from comparative embryology.

(1) Were the vertebrate blastopore to be co-extensive with the dorsal surface, as His and Rauber maintain, clear evidence of this ought to be apparent in Amphioxus. In Amphioxus, however, the blastopore is at first placed exactly at the hind end of the body, though later it passes up just on to the dorsal side (vide p. 4). It nearly closes before the appearance of the medullary groove or mesoblastic somites; and the medullary folds have nothing to do with its lips, except in so far as they are continuous with them behind, just as in Elasmobranchii.

(2) The food-yolk in the Vertebrata is placed on the ventral side of the body, and becomes enveloped by the blastoderm; so that in all large-yolked Vertebrates the ventral walls of the body are obviously completed by the closure of the lips of the blastopore, on the ventral side.

If His and Rauber are right the dorsal walls are also completed by the closure of the blastopore, so that the whole of the dorsal, as well as of the ventral wall of the embryo, must be formed by the concrescence of the lips of the blastopore; which is clearly a reductio ad absurdum of the whole theory. To my own arguments on the subject I may add those of Kupffer, who has very justly criticised His' statements, and has shewn that growth of the blastoderm in Clupea and Gasterosteus is absolutely inconsistent with the concrescence theory.

The more the theory of His and Rauber is examined by the light of comparative embryology, the more does it appear quite untenable; and it may be laid down as a safe conclusion from a comparative study of vertebrate embryology that the blastopore of Vertebrates is primitively situated at the hind end of the body, but that, owing to the development of a large food-yolk, it also extends, in most cases, over a larger or smaller part of the ventral side.

The origin of the Allantois and Amnion.

The development and structure of the allantois and amnion have already been dealt with at sufficient length in the chapters on Aves and Mammalia; but a few words as to the origin of these parts will not be out of place here.

The Allantois. The relations of the allantois to the adjoining organs, and the conversion of its stalk into the bladder, afford ample evidence that it has taken its origin from a urinary bladder such as is found in Amphibia. We have in tracing the origin of the allantois to deal with a case of what Dohrn would call ‘change of function.’ The allantois is in fact a urinary bladder which, precociously developed and enormously extended in the embryo, has acquired respiratory (Sauropsida) and nutritive (Mammalia) functions. No form is known to have been preserved with the allantois in a transitional state between an ordinary bladder and a large vascular sack.

The advantage of secondary respiratory organs during fœtal life, in addition to the yolk-sack, is evinced by the fact that such organs are very widely developed in the Ichthyopsida. Thus in Elasmobranchii we have the external gills (cf. p. 62). Amongst Amphibia we have the tail modified to be a respiratory organ in Pipa Americana; and in Notodelphis, Alytes and Cæcilia compressicanda the external gills are modified and enlarged for respiratory purposes within the egg (cf. pp. 140 and 143).

The Amnion. The origin of the amnion is more difficult to explain than that of the allantois; and it does not seem possible to derive it from any pre-existing organ.

It appears to me, however, very probable that it was evolved pari passu with the allantois, as a simple fold of the somatopleure round the embryo, into which the allantois extended itself as it increased in size and became a respiratory organ. It would be obviously advantageous for such a fold, having once started, to become larger and larger in order to give more and more room for the allantois to spread into.

The continued increase of this fold would lead to its edges meeting on the dorsal side of the embryo, and it is easy to conceive that they might then coalesce.

To afford room for the allantois close to the surface of the egg, where respiration could most advantageously be carried on, it would be convenient that the two laminæ of the amnion—the true and false amnion—should then separate and leave a free space above the embryo, and thus it may have come about that a separation finally takes place between the true and false amnion.

This explanation of the origin of the amnion, though of course hypothetical, has the advantage of suiting itself in most points to the actual ontogeny of the organ. The main difficulty is the early development of the head-fold of the amnion, since, from the position of the allantois, it might have been anticipated that the tail-fold would be the first formed and most important fold of the amnion.

Bibliography.

(239) F. M. Balfour. “A comparison of the early stages in the development of Vertebrates.” Quart. J. of Micr. Science, Vol. XV. 1875.
(240) F. M. Balfour. “A monograph on the development of Elasmobranch Fishes.” London, 1878.
(241) F. M. Balfour. “On the early development of the Lacertilia together with some observations, etc.” Quart. J. of Micr. Science, Vol. XIX. 1879.
(242) A. Götte. Die Entwicklungsgeschichte d. Unke. Leipzig, 1875.
(243) W. His. “Ueb. d. Bildung d. Haifischembryonen.” Zeit. f. Anat. u. Entwick., Vol. II. 1877. Cf. also His’ papers on Teleostei, Nos. 65 and 66.
(244) A. Kowalevsky. “Entwick. d. Amphioxus lanceolatus.” Mém. Acad. des Sciences St Pétersbourg, Ser. VII. Tom. XI. 1867.
(245) A. Kowalevsky. “Weitere Studien üb. d. Entwick. d. Amphioxus lanceolatus.” Archiv f. mikr. Anat., Vol. XIII. 1877.
(246) C. Kupffer. “Die Entstehung d. Allantois u. d. Gastrula d. Wirbelthiere.” Zool. Anzeiger, Vol. II. 1879, pp. 520, 593, 612.
(247) R. Remak. Untersuchungen üb. d. Entwicklung d. Wirbelthiere, 1850-1858.
(248) A. Rauber. Primitivstreifen u. Neurula d. Wirbelthiere. Leipzig, 1877.

The present section of this work would not be complete without some attempt to reconstruct, from the materials recorded in the previous chapters, and from those supplied by comparative anatomy, the characters of the ancestors of the Chordata; and to trace as far as possible from what invertebrate stock this ancestor was derived.

The second of these questions has been recently dealt with in a very suggestive manner by both Dohrn (No. 250) and Semper (Nos. 255 and 256), but it is still so obscure that I shall refrain from any detailed discussion of it.

While differing very widely in many points both Dohrn and Semper have arrived at the view, already tentatively put forward by earlier anatomists, that the nearest allies of the Chordata are to be sought for amongst the Chætopoda, and that the dorsal surface of the Chordata with the spinal cord corresponds morphologically with the ventral surface of the Chætopods with the ventral ganglion chain. In discussing this subject some time ago[100] I suggested that we must look for the ancestors of the Chordata, not in allies of the present Chætopoda, but in a stock of segmented forms descended from the same unsegmented types as the Chætopoda, but in which two lateral nerve-cords, like those of Nemertines, coalesced dorsally, instead of ventrally to form a median nervous cord. This group of forms, if my suggestion as to its existence is well founded, appears now to have perished. The recent researches of Hubrecht on the anatomy of the Nemertines[101] have, however, added somewhat to the probability of my views, in that they shew that in some existing Nemertines the nerve-cords approach each other very closely in the dorsal line.

With reference to the characters of the ancestor of the Chordata the following pages contain a few tentative suggestions rather than an attempt to deal with the whole subject; while the origin of certain of the organs is dealt with in a more special manner in the chapters on organogeny which form the second part of this work.

Before entering upon the more special subject of this chapter, it will be convenient to clear the ground by insisting on a few morphological conclusions to be drawn from the study of Amphioxus,—a form which, although probably in some respects degenerate, is nevertheless capable of furnishing on certain points very valuable evidence.

(1) In the first place it is clear from Amphioxus that the ancestors of the Chordata were segmented, and that their mesoblast was divided into myotomes which extended even into the region in front of the mouth. The mesoblast of the greater part of what is called the head in the Vertebrata proper was therefore segmented like that of the trunk.

(2) The only internal skeleton present was the unsegmented notochord—a fact which demonstrates that the skeleton is of comparatively little importance for the solution of a large number of fundamental questions, as for example the point which has been mooted recently as to whether gill-clefts existed at one time in front of the present mouth; and for this reason:—that from the evidence of Amphioxus and the lower Vertebrata[102] it is clear that such clefts, if they ever existed, had atrophied completely before the formation of cartilaginous branchial bars; so that any skeletal structures in front of the mouth, which have been interpreted by morphologists as branchial bars, can never have acted in supporting the walls of branchial clefts.

(3) The region which, in the Vertebrata, forms the œsophagus and stomach, was, in the ancestors of the Chordata, perforated by gill-clefts. This fact, which has been clearly pointed out by Gegenbaur, is demonstrated by the arrangement of the gill-clefts in Amphioxus, and by the distribution of the vagus nerve in the Vertebrata[103]. On the other hand the insertion of the liver, which was probably a very primitive organ, appears to indicate with approximate certainty the posterior limit of the branchial clefts.

With these few preliminary observations we may pass to the main subject of this section. A fundamental question which presents itself on the threshold of our enquiries is the differentiation of the head.

In the Chætopoda the head is formed of a præoral lobe and of the oral segment; while in Arthropods a somewhat variable number of segments are added behind to this primitive head, and form with it what may be called a secondary compound head. It is fairly clear that the section of the trunk, which, in Amphioxus, is perforated by the visceral clefts, has become the head in the Vertebrates proper, so that the latter forms are provided with a secondary head like that of Arthropods. There remain however difficult questions (1) as to the elements of which this head is composed, and (2) as to the extent of its differentiation in the ancestors of the Chordata.

In Arthropods and Chætopods there is a very distinct element in the head known as the procephalic lobe in the case of Arthropods, and the præoral lobe in that of Chætopods; and this lobe is especially characterized by the fact that the supraœsophageal ganglia and optic organs are formed as differentiations of part of the epiblast covering it. Is such an element to be recognized in the head of the Chordata? From a superficial examination of Amphioxus the answer would undoubtedly be no; but then it has to be borne in mind that Amphioxus, in correlation with its habit of burying itself in sand, is especially degenerate in the development of its sense-organs; so that it is not difficult to believe that its præoral lobe may have become so reduced as not to be recognizable. In the true Vertebrata there is a portion of the head which has undoubtedly many features of the præoral lobe in the types already alluded to, viz. the part containing the cerebral hemispheres and the thalamencephalon. If there is any part of the brain homologous with the supraœsophageal ganglia of the Invertebrates, and it is difficult to believe there is not such a part, it must be part of, or contain, the fore-brain. The fore-brain resembles the supraœsophageal ganglia in being intimately connected in its development with the optic organs, and in supplying with nerves only organs of sense. Its connection with the olfactory organs is an argument in the same direction. Even in Amphioxus there is a small bulb at the end of the nervous tube supplying what is very probably the homologue of the olfactory organ of the Vertebrata; and it is quite possible that this bulb is the reduced rudiment of what forms the fore-brain in the Vertebrata.

The evidence at our disposal appears to me to indicate that the third nerve belongs to the cranio-spinal series of segmental nerves, while the optic and olfactory nerves appear to me equally clearly not to belong to this series[104]. The mid-brain, as giving origin to the third nerve, would appear not to have been part of the ganglion of the præoral lobe.

These considerations indicate with fair probability that the part of the head containing the fore-brain is the equivalent of the præoral lobe of many Invertebrate forms; and the primitive position of the Vertebrate mouth on the ventral side of the head affords a distinct support for this view. It must however be admitted that this part of the head is not sharply separated in development from that behind; and, though the fore-brain is usually differentiated very early as a distinct lobe of the primitive nervous tube, yet that such differentiation is hardly more marked than in the other parts of the brain. The termination of the notochord immediately behind the fore-brain is, however, an argument in favour of the morphological distinctness of the latter structure.

The evidence at our disposal appears to indicate that the posterior part of the head was not differentiated from the trunk in lower Chordata; but that, as the Chordata rose in the scale of development, more and more centralizing work became thrown on the anterior part of the nervous cord, and pari passu this part became differentiated into the mid- and hind-brain. An analogy for such a differentiation is supplied in the compound subœsophageal ganglion of many Arthropods; and, as will be shewn in the chapter on the nervous system, there is strong embryological evidence that the mid- and hind-brains had primitively the same structure as the spinal cord. The head appears however to have suffered in the course of its differentiation a great concentration in its posterior part, which becomes progressively more marked, even within the limits of the surviving Vertebrata. This concentration is especially shewn in the structure of the vagus nerve, which, as first pointed out by Gegenbaur, bears evidence of having been originally composed of a great series of nerves, each supplying a visceral cleft. Rudiments of the posterior nerves still remain as the branches to the œsophagus and stomach[105].

The atrophy of the posterior visceral clefts seems to have taken place simultaneously with the concentration of the neural part of the head; but the former process did not proceed so rapidly as the latter, so that the visceral region of the head is longer in the lower Vertebrata than the neural region, and is dorsally overlapped by the anterior part of the spinal cord and the anterior muscle-plates (videfig. 47).

On the above view the posterior part of the head must have been originally composed of a series of somites like those of the trunk, but in existing Vertebrata all trace of these, except in so far as they are indicated by the visceral clefts, has vanished in the adult. The cranial nerves however, especially in the embryo, still indicate the number of anterior somites; and an embryonic segmentation of the mesoblast has also been found in many lower forms in the region of the head, giving rise to a series of cavities known as head-cavities, enclosed by mesoblastic walls which afterwards break up into muscles. These cavities correspond with the nerves, and it appears that there is a præmandibular cavity corresponding with the third nerve (fig. 193, 1pp) and a mandibular cavity (2pp) and a cavity in each of the succeeding visceral arches. The fifth nerve, the seventh nerve, the glossopharyngeal nerve, and the successive elements of the vagus nerve correspond with the posterior head-cavities.

The medullary canal. The general history of the medullary plate seems to point to the conclusion that the central canal of the nervous system has been formed by a groove having appeared in the ancestor of the Chordata along the median dorsal line, which caused the sides of the nervous plate, which was placed immediately below the skin, or may perhaps at that stage not have been distinctly differentiated from the skin, to be bent upwards; and that this groove subsequently became converted into a canal. This view is not only supported by the actual development of the central canal of the nervous system (the types of Teleostei, Lepidosteus and Petromyzon being undoubtedly secondary), but also (1) by the presence of cilia in the epithelium lining the canal, probably inherited from cilia coating the external skin, and (2) by the posterior roots arising from the extreme dorsal line (fig. 194), a position which can most easily be explained on the supposition that the two sides of the plate, from which the nerves originally proceeded have been folded up so as to meet each other in the median dorsal line[106].

The medullary plate, before becoming folded to form the medullary groove, is (except in Amphibia) without any indication of being composed of two halves. In both the embryo and adult the walls of the tube have however a structure which points to their having arisen from the coalescence of two lateral, and most probably at one time independent, cords; and as already indicated this is the view I am myself inclined to adopt; vide pp. 303 and 304.

The origin and nature of the mouth. The most obvious point connected with the development of the mouth is the fact that in all vertebrate embryos it is placed ventrally, at some little distance from the front end of the body. This feature is retained in the adult stage in Elasmobranchii, the Myxinoids, and some Ganoids, but is lost in other vertebrate forms. A mouth, situated as is the embryonic vertebrate mouth, is very ill adapted for biting; and though it acquires in this position a distinctly biting character in the Elasmobranchii, yet it is almost certain that it had not such a character in the ancestral Chordata, and that its terminal position in higher types indicates a step in advance of the Elasmobranchii.

On the structure of the primitive mouth there appears to me to be some interesting embryological evidence, to which attention has already been called in the preceding chapters. In a large number of the larvæ or embryos of the lower Vertebrates the mouth has a more or less distinctly suctorial character, and is connected with suctorial organs which may be placed either in front of or behind it. The more important instances of this kind are (1) the Tadpoles of the Anura, with their posteriorly placed suctorial disc, (2) Lepidosteus larva (fig. 195) with its anteriorly placed suctorial disc, (3) the adhesive papillæ of the larvæ of the Tunicata. To these may be added the suctorial mouth of the Myxinoid fishes[107].

All these considerations point to the conclusion that in the ancestral Chordata the mouth had a more or less definitely suctorial character[108], and was placed on the ventral surface immediately behind the præoral lobe; and that this mouth has become in the higher types gradually modified for biting purposes, and has been carried to the front end of the head.

The mouth in Elasmobranchii and other Vertebrates is originally a wide somewhat rhomboidal cavity (fig. 28 G); on the development of the mandibular and its maxillary (pterygo-quadrate) process the opening of the mouth becomes narrowed to a slit. The wide condition of the mouth may not improbably be interpreted as a remnant of the suctorial state. The fact that no more definite remnants of the suctorial mouth are found in so primitive a group as the Elasmobranchii is probably to be explained by the fact that the members of this group undergo an abbreviated development within the egg.

While the embryological data appear to me to point to the existence of a primitive suctorial mouth, very different conclusions have been put forward by other embryologists, more especially by Dohrn, which are sufficiently striking and suggestive to merit a further discussion.

As mentioned above, both Dohrn and Semper hold that the Vertebrata are descended from Chætopod-like forms, in which the ventral surface has become the dorsal. In consequence of this view Dohrn has arrived at the following conclusions: (1) that primitively the alimentary canal perforated the nervous system in the region of the original œsophageal nerve-ring; (2) that there was therefore an original dorsal mouth (the present ventral mouth of the Chætopoda); and (3) that the present mouth was secondary and derived from two visceral clefts which have ventrally coalesced.

A full discussion of these views[109] is not within the scope of this work; but, while recognizing that there is much to be said in favour of the interchange of the dorsal and ventral surfaces, I am still inclined to hold that the difficulties involved in this view are so great that it must, provisionally at least, be rejected; and that there are therefore no reasons against supposing the present vertebrate mouth to be the primitive mouth. There is no embryological evidence in favour of the view adopted by Dohrn that the present mouth was formed by the coalescence of two clefts.

If it is once admitted that the present mouth is the primitive mouth, and is more or less nearly in its original situation, very strong evidence will be required to shew that any structures originally situated in front of it are the remnants of visceral clefts; and if it should be proved that such remnants of visceral clefts were present, the views so far arrived at in this section would, I think, have to be to a large extent reconsidered.

The nasal pits have been supposed by Dohrn to be remnants of visceral clefts, and this view has been maintained in a very able manner by Marshall. The arguments of Marshall do not, however, appear to me to have any great weight unless it is previously granted that there is an antecedent probability in favour of the presence of a pair of gill-clefts in the position of the nasal pits; and even then the development of the nasal pits as epiblastic involutions, instead of hypoblastic outgrowths, is a serious difficulty which has not in my opinion been successfully met. A further argument of Marshall from the supposed segmental nature of the olfactory nerve has already been spoken of.

While most of the structures supposed to be remains of gill-clefts in front of the mouth do not appear to me to be of this nature, there is one organ which stands in a more doubtful category. This organ is the so-called choroid gland. The similarity of this organ to the pseudobranch of the mandibular or hyoid arch was pointed out to me by Dohrn, and the suggestion was made by him that it is the remnant of a præmandibular gill which has been retained owing to its functional connection with the eye[110]. Admitting this explanation to be true (which however is by no means certain) are we necessarily compelled to hold that the choroid gland is the remnant of a gill-cleft originally situated in front of the mouth? I believe not. It is easy to conceive that there may originally have been a præmandibular cleft behind the suctorial mouth, but that this cleft gradually atrophied (for the same reasons that the mandibular cleft shews a tendency to atrophy in existing fishes, &c.), the rudiment of the gill (choroid gland) alone remaining to mark its situation. After the disappearance of this cleft the suctorial mouth may have become relatively shifted backwards. In the meantime the branchial bars became developed, and as the mouth was changed into a biting one, the bar (the mandibular arch) supporting the then first cleft became gradually modified and converted into a supporting apparatus for the mouth, and finally formed the skeleton of the jaws. In the hyostylic Vertebrata the hyoid arch also became modified in connection with the formation of the jaws.

The conclusions arrived at may be summed up as follows:

The relations which exist in all jaw-bearing Vertebrates between the mandibular arch and the oral aperture are secondary, and arose pari passu with the evolution of the jaws[111].

The cranial flexure and the form of the head in vertebrate embryos. All embryologists who have studied the embryos of the various vertebrate groups have been struck with the remarkable similarity which exists between them, more especially as concerns the form of the head. This similarity is closest between the members of the Amniota, but there is also a very marked resemblance between the Amniota and the Elasmobranchii. The peculiarity in question, which is characteristically shewn in fig. 196, consists in the cerebral hemispheres and thalamencephalon being ventrally flexed to such an extent that the mid-brain forms the termination of the long axis of the body. At a later period in development the cerebral hemispheres come to be placed at the front end of the head; but the original nick or bend of the floor of the brain is never got rid of.

It is obvious that in dealing with the light thrown by embryology on the ancestral form of the Chordata the significance of this peculiar character of the head of many vertebrate embryos must be discussed. Is the constancy of this character to be explained by supposing that at one period vertebrate ancestors had a head with the same features as the embryonic head of existing Vertebrata?

This is the most obvious explanation, but it does not at the same time appear to me satisfactory. In the first place the mouth is so situated at the time of the maximum cranial flexure that it could hardly have been functional; so that it is almost impossible to believe that an animal with a head such as that of these embryos can have existed.

Then again, this type of embryonic head is especially characteristic of the Amniota, all of which are developed in the egg. It is not generally so marked in the Ichthyopsida. In Amphibia, Teleostei, Ganoidæ and Petromyzontidæ, the head never completely acquires the peculiar characteristic form of the head of the Amniota, and all these forms are hatched at a relatively much earlier phase of development, so that they are leading a free existence at a stage when the embryos of the Amniota are not yet hatched. The only Ichthyopsidan type with a head like that of the Amniota is the Elasmobranchii, and the Elasmobranchii are the only Ichthyopsida which undergo the major part of their development within the egg.

These considerations appear to shew that the peculiar characters of the embryonic head above alluded to are in some way connected with an embryonic as opposed to a larval development; and for reasons which are explained in the section on larval forms, it is probable that a larval development is a more faithful record of ancestral history than an embryonic development. The flexure at the base of the brain appears however to be a typical vertebrate character, but this flexure never led to a conformation of the head in the adult state similar to that of the embryos of the Amniota. The form of the head in these embryos is probably to be explained by supposing that some advantage is gained by a relatively early development of the brain, which appears to be its proximate cause; and since these embryos had not to lead a free existence (for which such a form of the head would have been unsuited) there was nothing to interfere with the action of natural selection in bringing about this form of head during fœtal life.

Postanal gut and neurenteric canal. One of the most remarkable structures in the trunk is the postanal gut (fig. 197). Its structure is fully dealt with in the chapter on the alimentary tract, but attention may here be called to the light which it appears to throw on the characters of the ancestor of the Chordata.

In face of the facts which are known with reference to the postanal section of the alimentary tract, it can hardly be doubted that this portion of the alimentary tract must have been at one time functional. This seems to me to be shewn (1) by the constancy and persistence of this obviously now functionless rudiment, (2) by its greater development in the lower than in the higher forms, (3) by its relation to the formation of the notochord and subnotochordal rod.

If the above position be admitted, it is not permissible to shirk the conclusions which seem necessarily to follow, however great the difficulties may be which are involved in their acceptance. These conclusions have in part already been dealt with by Dohrn in his suggestive tract (No. 250). In the first place the alimentary canal must primitively have been continued to the end of the tail; and if so, it is hardly credible that the existing anus can have been the original one. Although, therefore, it is far from easy, on the physiological principles involved in the Darwinian theory, to understand the formation of a new anus[112]; it is nevertheless necessary to believe that the present vertebrate anus is a formation acquired within the group of the Chordata, and not inherited from some older group. This involves a series of further consequences. The opening of the urinogenital ducts into the cloaca must also be secondary, and it is probable that the segmental tubes were primitively continued along the whole postanal region of the vertebrate tail, opening into the body cavity which embryology proves to have been originally present there. They are in fact continued in many existing forms for some distance behind the present anus. If the present anus is secondary, there must have been a primitive anus, which was probably situated behind the postanal vesicle; and therefore in the region of the neurenteric canal. The neurenteric canal is, however, the remnant of the blastopore (vide p. 277). It follows, therefore, that the vertebrate blastopore is probably almost, if not exactly identical in position with the primitive anus. This consideration may assist in explaining the remarkable phenomenon of the existence of the neurenteric canal. The attempt has already been made to shew that the central canal of the nervous system is really a groove converted into a tube and lined by the external epidermis. This tube (as may be concluded from embryological considerations) was probably at first open posteriorly, and no doubt terminated at the primitive anus. On the closure of the primitive anal opening, the terminal portions of the postanal gut and the neural tube, may conceivably have been so placed that both of them opened into a common cavity, which previously had communication with the exterior by the anus. Such an arrangement would necessarily result in the formation of a neurenteric canal. It seems not impossible that a dilated vesicle, often present at the end of the postanal gut (vide fig. 28*, p. 58), may have been the common cavity into which both neural and alimentary tubes opened[113]. Till further light is thrown by fresh discoveries upon the primitive condition of the posterior continuation of the vertebrate alimentary tract, it is perhaps fruitless to attempt to work out more in detail the above speculation.

Body cavity and mesoblastic somites. The Chordata, or at least the most primitive existing members of the group, are characterized by the fact that the body cavity arises as a pair of outgrowths of the archenteric cavity. This feature[114] in the development is a nearly certain indication that the Chordata are a very primitive stock. The most remarkable point with reference to the development of the two outgrowths is, however, the fact that the dorsal part of each outgrowth becomes separated from the ventral. Its walls become segmented and form the mesoblastic somites, which eventually, on the obliteration of their cavity, give rise to the muscle-plates and to the tissue surrounding the notochord. It is not easy to understand the full significance of the processes concerned in the formation of the mesoblastic somites (vide p. 296). The mesoblastic somites have no doubt a striking resemblance to the mesoblastic somites of the Chætopods, and most probably the segmentation of the mesoblast in the two groups is a phenomenon of the same nature; but the difference in origin between the two types of mesoblastic somites is so striking, and the development of the muscular system from them is so dissimilar in the two groups, as to render a direct descent of the Chordata from the Chætopoda very improbable. The ventral parts of the original outgrowth give rise to the permanent body cavity, which appears originally to have been divided into two parts by a dorsal and a ventral mesentery.

The notochord. The most characteristic organ of the Chordata is without doubt the notochord. The ontogenetic development of this organ probably indicates that it arose as a differentiation of the dorsal wall of the archenteron; at the same time it is not perhaps safe to lay too much stress upon its mode of development. Embryological and anatomical evidence demonstrate, however, in the clearest manner that the early Chordata were provided with this organ as their sole axial skeleton; and no invertebrate group can fairly be regarded as genetically related to the Chordata till it can be shewn to possess some organ either derived from a notochord, or capable of having become developed into a notochord. No such organ has as yet been recognized in any invertebrate group[115].

Gill-clefts. The gill-clefts, which are essentially pouches of the throat opening externally, constitute extremely characteristic organs of the Chordata, and have always been taken into consideration in any comparison between the Chordata and the Invertebrata.

Amongst the Invertebrata organs of undoubtedly the same nature are, so far as I know, only found in Balanoglossus, where they were discovered by Kowalevsky. The resemblance in this case is very striking; but although it is quite possible that the gill-clefts in Balanoglossus are genetically connected with those of the Chordata, yet the organization of Balanoglossus is as a whole so different from that of the Chordata that no comparison can be instituted between the two groups in the present state of our knowledge.

Other organs of the Invertebrata have some resemblance to the gill-clefts. The lateral pits of the Nemertines, which appear to grow out as a pair of œsophageal diverticula, which are eventually placed in communication with the exterior by a pair of ciliated canals (vide Vol. II. pp. 200 and 202), are such organs.

Semper (No. 256) has made the interesting discovery that in the budding of Nais and Chætogaster two lateral masses of cells, in each of which a lumen may be formed, unite with the oral invagination and primitive alimentary canal to form the permanent cephalic gut. The lateral masses of cells are regarded by him as branchial passages homologous in some way with those in the Chordata. The somewhat scanty observations on this subject which he has recorded do not appear to me to lend much support to this interpretation.

It is probable that the part of the alimentary tract in which gill-clefts are present was originally a simple unperforated tube provided with highly vascular walls; and that respiration was carried on in it by the alternate introduction and expulsion of sea water. A more or less similar mode of respiration has been recently shewn by Eisig[116] to take place in the fore part of the alimentary tract of many Chætopods. This part of the alimentary tract was probably provided with paired cæcal pouches with their blind ends in contiguity with the skin.

Perforations placing these pouches in communication with the exterior must be supposed to have been formed; and the existence of openings into the alimentary tract at the end of the tentacles of many Actiniæ and of the hepatic diverticula of some nudibranchiate Molluscs (Eolis, &c.[117]) shews that such perforations may easily be made. On the formation of such perforations the water taken in at the mouth would pass out by them; and the respiration would be localized in the walls of the pouches leading to them, and thus the typical mode of respiration of the Chordata would be established.

Phylogeny of the Chordata. It may be convenient to shew in a definite way the bearing of the above speculations on the phylogeny of the Chordata. For this purpose, I have drawn up the subjoined table, which exhibits what I believe to be the relationships of the existing groups of the Chordata. Such a table cannot of course be constructed from embryological data alone, and it does not fall within the scope of this work to defend its parts in detail.

The ancestral forms of the Chordata, which may be called the Protochordata, must be supposed to have had (1) a notochord as their sole axial skeleton, (2) a ventral mouth, surrounded by suctorial structures, and (3) very numerous gill-slits. Two degenerate offshoots of this stock still persist in Amphioxus (Cephalochorda), and the Ascidians (Urochorda).

The direct descendants of the ancestral Chordata, were probably a group which may be called the Protovertebrata, of which there is no persisting representative. In this group, imperfect neural arches were probably present; and a ventral suctorial mouth without a mandible and maxillæ was still persistent. The branchial clefts had, however, become reduced in number, and were provided with gill-folds; and a secondary head (vide p. 313), with brain and organs of sense like those of the higher Vertebrata, had become formed.

The Cyclostomata are probably a degenerate offshoot of this group.

With the development of the branchial bars, and the conversion of the mandibular bar into the skeleton of the jaws, we come to the Proto-gnathostomata. The nearest living representatives of this group are the Elasmobranchii, which still retain in the adult state the ventrally placed mouth. Owing to the development of food-yolk in the Elasmobranch ovum the early stages of development are to some extent abbreviated, and almost all trace of a stage with a suctorial mouth has become lost.

We next come to an hypothetical group which we may call the Proto-ganoidei. Bridge, in his memoir on Polyodon[118], which contains some very interesting speculations on the affinities of the Ganoids, has called this group the Pneumatocœla, from the fact that we find for the first time a full development of the air-bladder, though it is possible that a rudiment of this organ, in the form of a pouch opening on the dorsal side of the stomachic extremity of the œsophagus, was present in the earlier type.

Existing Ganoids are descendants of the Proto-ganoidei. Some of them at all events retain in larval life the suctorial mouth of the Protovertebrata; and the mode of formation of their germinal layers, resembling as it does that in the Lamprey and the Amphibia, probably indicates that they are not descended from forms with a large food-yolk like that of Elasmobranchii, and that the latter group is therefore a lateral offshoot from the main line of descent.

Of the two groups into which the Ganoidei may be divided it is clear that certain members of the one (Telcostoidei), viz. Lepidosteus and Amia, shew approximations to the Teleostei, which no doubt originated from the Ganoids; while the other (Selachoidei or Sturiones) is more nearly related to the Dipnoi. Polypterus has also marked affinities in this direction, e.g. the external gills of the larva (vide p. 118).

The Teleostei, which have in common a meroblastic segmentation, had probably a Ganoid ancestor, the ova of which were provided with a large amount of food-yolk. In most existing Teleostei, the ovum has become again reduced in size, but the meroblastic segmentation has been preserved. It is quite possible that Amia may also be a descendant of the Ganoid ancestor of the Teleostei; but Lepidosteus, as shewn by its complete segmentation, is clearly not so.

The Dipnoi as well as all the higher Vertebrata are descendants of the Proto-ganoidei.

The character of the limbs of higher Vertebrata indicates that there was an ancestral group, which may be called the Proto-pentadactyloidei, in which the pentadactyle limb became established; and that to this group the common ancestor of the Amphibia and Amniota belonged.

It is possible that the Plesiosauri and Ichthyosauri of Mesozoic times may have been more nearly related to this group than either to the Amniota or the Amphibia. The Proto-pentadactyloidei were probably much more closely related to the Amphibia than to the Amniota. They certainly must have been capable of living in water as well as on land, and had of course persistent branchial clefts. It is also fairly certain that they were not provided with large-yolked ova, otherwise the mode of formation of the layers in Amphibia could not be easily explained.

The Mammalia and Sauropsida are probably independent offshoots from a common stem which may be called the Protoamniota.

Bibliography.

(249) F. M. Balfour. A Monograph on the development of Elasmobranch Fishes. London, 1878.
(250) A. Dohrn. Der Ursprung d. Wirbelthiere und d. Princip. d. Functionswechsel. Leipzig, 1875.
(251) E. Haeckel. Schöpfungsgeschichte. Leipzig. Vide also Translation. The History of Creation. King and Co., London, 1876.
(252) E. Haeckel. Anthropogenie. Leipzig. Vide also Translation. Anthropogeny. Kegan Paul and Co., London, 1878.
(253) A. Kowalevsky. “Entwicklungsgeschichte d. Amphioxus lanceolatus.” Mém. Acad. d. Scien. St Pétersbourg, Ser. VII. Tom. XI. 1867, and Archiv f. mikr. Anat., Vol. XIII. 1877.
(254) A. Kowalevsky. “Weitere Stud. üb. d. Entwick. d. einfachen Ascidien.” Archiv f. mikr. Anat., Vol. VII. 1871.
(255) C. Semper. “Die Stammesverwandschaft d. Wirbelthiere u. Wirbellosen.” Arbeit. a. d. zool.-zoot. Instit. Würzburg, Vol. II. 1875.
(256) C. Semper. “Die Verwandschaftbeziehungen d. gegliederten Thiere.” Arbeit. a. d. zool.-zoot. Instit. Würzburg, Vol. III. 1876-1877.

I. THE MODE OF ORIGIN AND HOMOLOGIES OF THE GERMINAL LAYERS.

It has already been shewn in the earlier chapters of the work that during the first phases of development the history of all the Metazoa is the same. They all originate from the coalescence of two cells, the ovum and spermatozoon. The coalesced product of these cells—the fertilized ovum—then undergoes a process known as the segmentation, in the course of which it becomes divided in typical cases into a number of uniform cells. An attempt was made from the point of view of evolution to explain these processes. The ovum and spermatozoon were regarded as representing phylogenetically two physiologically differentiated forms of a Protozoon; their coalescence was equivalent to conjugation: the subsequent segmentation of the fertilized ovum was the multiplication by division of the organism resulting from the conjugation; the resulting organisms, remaining, however, united to form a fresh organism in a higher state of aggregation.

In the systematic section of this work the embryological history of the Metazoa has been treated. The present chapter contains a review of the cardinal features of the various histories, together with an attempt to determine how far there are any points common to the whole of these histories; and the phylogenetic interpretation to be given to such points.

Some years ago it appeared probable that a definite answer would be given to the questions which must necessarily be raised in the present chapter; but the results of the extended investigations made during the last few years have shewn that these expectations were premature, and in spite of the numerous recent valuable contributions to this branch of Embryology, amongst which special attention may be called to those of Kowalevsky (No. 277), Lankester (Nos. 278 and 279), and Haeckel (No. 266), there are few embryologists who would venture to assert that any answers which can be given are more than tentative gropings towards the truth.

In the following pages I aim more at summarising the facts, and critically examining the different theories which can be held, than at dogmatically supporting any definite views of my own.

In all the Metazoa, the development of which has been investigated, the first process of differentiation, which follows upon the segmentation, consists in the cells of the organism becoming divided into two groups or layers, known respectively as epiblast and hypoblast.

These two layers were first discovered in the young embryos of vertebrated animals by Pander and Von Baer, and have been since known as the germinal layers, though their cellular nature was not at first recognised. They were shewn, together with a third layer, or mesoblast, which subsequently appears between them, to bear throughout the Vertebrata constant relations to the organs which became developed from them. A very great step was subsequently made by Remak (No. 287), who successfully worked out the problem of vertebrate embryology on the cellular theory.

Rathke in his memoir on the development of Astacus (No. 286) attempted at a very early period to extend the doctrine of the derivation of the organs from the germinal layers to the Invertebrata. In 1859 Huxley made an important step towards the explanation of the nature of these layers by comparing them with the ectoderm and endoderm of the Hydrozoa; while the brilliant researches of Kowalevsky on the development of a great variety of invertebrate forms formed the starting point of the current views on this subject.

The differentiation of the epiblast and hypoblast may commence during the later phases of the segmentation, but is generally not completed till after its termination. Not only do the cells of the blastoderm become differentiated into two layers, but these two layers, in the case of a very large number of ova with but little food-yolk, constitute a double-walled sack—the gastrula (fig. 198)—the characters of which are too well known to require further description. Following the lines of phylogenetic speculation above indicated, it may be concluded that the two-layered condition of the organism represents in a general way the passage from the protozoon to the metazoon condition. It is probable that we may safely go further, and assert that the gastrula reproduces, with more or less fidelity, a stage in the evolution of the Metazoa, permanent in the simpler Hydrozoa, during which the organism was provided with (1) a fully developed digestive cavity (fig. 198 b) lined by the hypoblast with digestive and assimilative functions, (2) an oral opening (a), and (3) a superficial epiblast (d). These generalisations, which are now widely accepted, are no doubt very valuable, but they leave unanswered the following important questions:
(1) By what steps did the compound Protozoon become differentiated into a Metazoon?
(2) Are there any grounds for thinking that there is more than one line along which the Metazoa have become independently evolved from the Protozoa?
(3) To what extent is there a complete homology between the two primary germinal layers throughout the Metazoa?

Ontogenetically there is a great variety of processes by which the passage from the segmented ovum to the two-layered or diploblastic condition is arrived at.

These processes may be grouped under the following heads:

1. Invagination. Under this term a considerable number of closely connected processes are included. When the segmentation results in the formation of a blastosphere, one half of the blastosphere may be pushed in towards the opposite half, and a gastrula be thus produced (fig. 199, A and B). This process is known as embolic invagination. Another process, known as epibolic invagination, consists in epiblast cells growing round and enclosing the hypoblast (fig. 200). This process replaces the former process when the hypoblast cells are so bulky from being distended by food-yolk that their invagination is mechanically impossible.

There are various peculiar modifications of invagination which cannot be dealt with in detail.

Invagination in one form or other occurs in some or all the members of the following groups:

The Dicyemidæ, Calcispongiæ (after the amphiblastula stage) and Silicispongiæ, Cœlenterata, Turbellaria, Nemertea, Rotifera, Mollusca, Polyzoa, Brachiopoda, Chætopoda, Discophora, Gephyrea, Chætognatha, Nematelminthes, Crustacea, Echinodermata, and Chordata.

The gastrula of the Crustacea is peculiar, as is also that of many of the Chordata (Reptilia, Aves, Mammalia), but there is every reason to suppose that the gastrulæ of these groups are simply modifications of the normal type.

2. Delamination. Three types of delamination may be distinguished:
a. Delamination where the cells of a solid morula become divided into a superficial epiblast, and a central solid mass in which the digestive cavity is subsequently hollowed out (fig. 201).

b. Delamination where the segmented ovum has the form of a blastosphere, the cells of which give rise by budding to scattered cells in the interior of the vesicle, which, though they may at first form a solid mass, finally arrange themselves in the form of a definite layer around a central digestive cavity (fig. 202).
c. Delamination where the segmented ovum has the form of a blastosphere in the cells of which the protoplasm is differentiated into an inner and an outer part. By a subsequent process the inner parts of the cells become separated from the outer, and the walls of the blastosphere are so divided into two distinct layers (fig. 205).

Although the third of these processes is usually regarded as the type of delamination, it does not, so far as I know, occur in nature, but is most nearly approached in Geryonia (fig. 203).

The first type of delamination is found in the Ceratospongiæ, some Silicispongiæ (?), and in many Hydrozoa and Actinozoa, and in Nemertea and Nematelminthes (Gordioidea?). The second type occurs in many Porifera [Calcispongiæ (Ascetta), Myxospongiæ], and in some Cœlenterata, and Brachiopoda (Thecidium).

Delamination and invagination are undoubtedly the two most frequent modes in which the layers are differentiated, but there are in addition several others. In the first place the whole of the Tracheata (with the apparent exception of the Scorpion) develop, so far as is known, on a plan peculiar to them, which approaches delamination. This consists in the appearance of a superficial layer of cells enclosing a central yolk mass, which corresponds to the hypoblast (figs. 204 and 214). This mode of development might be classed under delamination, were it not for the fact that the early development of many Crustacea is almost the same, but is subsequently followed by an invagination (fig. 208), which apparently corresponds to the normal invagination of other types. There are strong grounds for thinking that the tracheate type of formation of the epiblast and hypoblast is a secondary modification of an invaginate type (vide Vol. II. p. 457).

The type of some Turbellaria (Stylochopsis ponticus) and that of Nephelis amongst the Discophora is not capable of being reduced to the invaginate type.

The development of almost all the parasitic groups, i.e. the Trematoda, the Cestoda, the Acanthocephala, and the Linguatulida, and also of the Tardigrada, Pycnogonida, and other minor groups, is too imperfectly known to be classed with either the delaminate or invaginate types.

It will, I think, be conceded on all sides that, if any of the ontogenetic processes by which a gastrula form is reached are repetitions of the process by which a simple two-layered gastrula was actually evolved from a compound Protozoon, these processes are most probably of the nature either of invagination or of delamination.

The much disputed questions which have been raised about the gastrula and planula theories, originally put forward by Haeckel and Lankester, resolve themselves then into the simple question, whether any, and if so which, of the ontogenetic processes by which the gastrula is formed are repetitions of the phylogenetic origin of the gastrula.

It is very difficult to bring forward arguments of a conclusive kind in favour of either of these processes. The fact that delaminate and invaginate gastrulæ are in several instances found coexisting in the same group renders it certain that there are not two independent phyla of the Metazoa, derived respectively from an invaginate and a delaminate gastrula[119].

The four most important cases in which the two processes coexist are the Porifera, the Cœlenterata, the Nemertea, and the Brachiopoda. In the cases of the Porifera and Cœlenterata, there do not appear to me to be any means of deciding which of these processes is derived from the other; but in the Nemertea and the Brachiopoda the case is different. In all the types of Nemertea in which the development is relatively not abbreviated there is an invaginate gastrula, while in the types with a greatly abbreviated development there is a delaminate gastrula. It would seem to follow from this that a delaminate gastrula has here been a secondary result of an abbreviation in the development. In the Brachiopoda, again, the majority of types develop by a process of invagination, while Thecidium appears to develop by delamination; here also the delaminate type would appear to be secondarily derived from the invaginate.

If these considerations are justified, delamination must be in some instances secondarily derived from invagination; and this fact is so far an argument in favour of the more primitive nature of invagination; though it by no means follows that in the invaginate process the steps by which the Metazoa were derived from the Protozoa are preserved.

It does not, therefore, seem possible to decide conclusively in favour of either of these processes by a comparison of the cases where they occur in the same groups.

The relative frequency of the two processes supplies us with another possible means for deciding between them; and there is no doubt that here again the scale inclines towards invagination. It must, however, be borne in mind that the frequency of the process of invagination admits of another possible explanation. There is a continual tendency for the processes of development to be abbreviated and simplified, and it is quite possible that the frequent occurrence of invagination is due to the fact of its being, in most cases, the simplest means by which the two-layered condition can be reached. But this argument can have but little weight until it can be shewn in each case that invagination is a simpler process than delamination; and it is rendered improbable by the cases already mentioned in which delamination has been secondarily derived from invagination.

If it were the case that the blastopore had in all types the same relation to the adult mouth, there would be strong grounds for regarding the invaginate gastrula as an ancestral form; but the fact that this is by no means so is an argument of great weight in favour of some other explanation of the frequency of invagination.

The force of this consideration can best be displayed by a short summary of the fate of the blastopore in different forms.

The fate of the blastopore is so variable that it is difficult even to classify the cases which have been described.

(1) It becomes the permanent mouth in the following forms[120]:
 Cœlenterata.—Pelagia, Cereanthus.
 Turbellaria.—Leptoplana (?), Thysanozoon.
 Nemertea.—Pilidium, larvæ of the type of Desor.
 Mollusca.—In numerous examples of most Molluscan groups, except the Cephalopoda.
 Chætopoda.—Most Oligochæta, and probably many Polychæta.
 Gephyrea.—Phascolosoma, Phoronis.
 Nematelminthes.—Cucullanus.

(2) It closes in the position where the mouth is subsequently formed.
 Cœlenterata.—Ctenophora (?).
 Mollusca.—In numerous examples of most Molluscan groups, except the Cephalopoda.
 Crustacea.—Cirripedia (?), some Cladocera (Moina) (?).

(3) It becomes the permanent anus.
 Mollusca.—Paludina.
 Chætopoda.—Serpula and some other types.
 Echinodermata.—Almost universally, except amongst the Crinoidea.

(4) It closes in the position where the anus is subsequently formed.
 Echinodermata.—Crinoidea.

(5) It closes in a position which does not correspond or is not known to correspond[121] either with the future mouth or anus.—Porifera—Sycandra. Cœlenterata—Chrysaora*, Aurelia*. Nemertea*—Some larvæ which develop without a metamorphosis. Rotifera*. Mollusca—Cephalopoda. Polyzoa*. Brachiopoda—Argiope, Terebratula, Terebratulina. Chætopoda—Euaxes. Discophora—Clepsine. Gephyrea—Bonellia*. Chætognatha. Crustacea—Decapoda. Chordata.

The forms which have been classed together under the last heading vary considerably in the character of the blastopore. In some cases the fact of its not coinciding either with the mouth or anus appears to be due simply to the presence of a large amount of food-yolk. The cases of the Cephalopoda, of Euaxes, and perhaps of Clepsine and Bonellia, are to be explained in this way: in the case of all these forms, except Bonellia, the blastopore has the form of an elongated slit along the ventral surface. This type of blastopore is characteristic of the Mollusca generally, of the Polyzoa, of the Nematelminthes, and very possibly of the Chætopoda and Discophora. In the Chætognatha (fig. 209 B) the blastopore is situated, so far as can be determined, behind the future anus. In many Decapoda the blastopore is placed behind, but not far from, the anus. In the Chordata it is also placed posteriorly to the anus, and, remarkably enough, remains, in a large number of forms, for some time in connection with the neural tube by a neurenteric canal.

The great variations in the character of the gastrula, indicated in the above summary, go far to shew that if the gastrulæ, as we find them in most types, have any ancestral characters, these characters can only be of the most general kind. This may best be shewn by the consideration of a few striking instances. The blastopore in Mollusca has an elongated slit-like form, extending along the ventral surface from the mouth to the anus. In Echinodermata it is a narrow pore, remaining as the anus. In most Chætopoda it is a pore remaining as the mouth, but in some as the anus. In Chordata it is a posteriorly-placed pore, opening into both the archenteron and the neural canal.

It is clearly out of the question to explain all these differences as having connection with the characters of ancestral forms. Many of them can only be accounted for as secondary adaptations for the convenience of development.

The epibolic gastrula of Mammalia (vide pp. 215 and 291) is a still more striking case of a secondary embryonic process, and is not directly derived from the gastrula of the lower Chordata. It probably originated in connection with the loss of food-yolk which took place on the establishment of a placental nutrition for the fœtus. The epibolic gastrula of the Scorpion, of Isopods, and of other Arthropoda, seems also to be a derived gastrula. These instances of secondary gastrulæ are very probably by no means isolated, and should serve as a warning against laying too much stress upon the frequency of the occurrence of invagination. The great influence of the food-yolk upon the early development might be illustrated by numerous examples, especially amongst the Chordata (vide Chapter XI.).

If the descendants of a form with a large amount of food-yolk in its ova were to produce ova with but little food-yolk, the type of formation of the germinal layers which would thereby result would be by no means the same as that of the ancestors of the forms with much food-yolk, but would probably be something very different, as in the case of Mammalia. Yet amongst the countless generations of ancestors of most existing forms, such oscillations in the amount of the food-yolk must have occurred in a large number of instances.

The whole of the above considerations point towards the view that the formation of the hypoblast by invagination, as it occurs in most forms at the present day, can have in many instances no special phylogenetic significance, and that the argument from frequency, in favour of invagination as opposed to delamination, is not of prime importance.

A third possible method of deciding between delamination and invagination is to be found in the consideration as to which of these processes occurs in the most primitive forms. If there were any agreement amongst primitive forms as to the type of their development this argument might have some weight. On the whole, delamination is, no doubt, characteristic of many primitive types, but the not infrequent occurrence of invagination in both the Cœlenterata and the Porifera—the two groups which would on all hands be admitted to be amongst the most primitive—deprives this argument of much of the value it might otherwise have.

To sum up—considering the almost indisputable fact that both the processes above dealt with have in many instances had a purely secondary origin, no valid arguments can be produced to shew that either of them reproduces the mode of passage between the Protozoa and the ancestral two-layered Metazoa. These conclusions do not, however, throw any doubt upon the fact that the gastrula, however evolved, was a primitive form of the Metazoa; since this conclusion is founded upon the actual existence of adult gastrula forms independently of their occurrence in development.

Though embryology does not at present furnish us with a definite answer to the question how the Metazoa became developed from the Protozoa, it is nevertheless worth while reviewing some of the processes by which this can be conceived to have occurred.

On purely à priori grounds there is in my opinion more to be said for invagination than for any other view.

On this view we may suppose that the colony of Protozoa in the course of conversion into Metazoa had the form of a blastosphere; and that at one pole of this a depression appeared. The cells lining this depression we may suppose to have been amœboid, and to have carried on the work of digestion; while the remaining cells were probably ciliated. The digestion may be supposed to have been at first carried on in the interior of the cells, as in the Protozoa; but, as the depression became deeper (in order to increase the area of nutritive cells and to retain the food) a digestive secretion probably became poured out from the cells lining it, and the mode of digestion generally characteristic of the Metazoa was thereby inaugurated. It may be noted that an intracellular protozoon type of digestion persists in the Porifera, and appears also to occur in many Cœlenterata, Turbellaria, &c., though in most of these cases both kinds of digestion probably go on simultaneously[122].

Another hypothetical mode of passage, which fits in with delamination, has been put forward by Lankester, and is illustrated by fig. 205. He supposes that at the blastosphere stage the fluid in the centre of the colony acquired special digestive properties; the inner ends of the cells having at this stage somewhat different properties from the outer, and the food being still incepted by the surface of the cells (fig. 205, 3). In a later stage of the process the inner portions of the cells became separated off as the hypoblast; while the food, though still ingested in the form of solid particles by the superficial cells, was carried through the protoplasm into the central digestive cavity. Later (fig. 205, 4), the point where the food entered became localised, and eventually a mouth became formed at this point.

The main objection which can be raised against Lankester’s view is that it presupposes a type of delamination which does not occur in nature except in Geryonia.

Metschnikoff has propounded a third view with reference to delamination. He starts as before with a ciliated blastosphere. He next supposes the cells from the walls of this to become budded off into the central cavity, as in Eucope (fig. 202), and to lose their cilia. These cells give rise to an internal parenchyma, which carries on an intracellular digestion. At a later stage a central digestive cavity is supposed to be formed. This view of the passage from the protozoon to the metazoon state, though to my mind improbable in itself, fits in very well with the ontogeny of the lower Hydrozoa.

Another view has been put forward by myself in the chapter on the Porifera[123], to the effect that the amphiblastula larva of Calcispongiæ may be a transitional form between the Protozoa and the Metazoa, composed of a hemisphere of nutritive amœboid cells, and a hemisphere of ciliated cells. The absence of such a larval form in the Cœlenterata and higher Metazoa is opposed, however, to this larva being regarded as a transitional form, except for the Porifera.

It is obvious that so long as there is complete uncertainty as to the value to be attached to the early developmental processes, it is not possible to decide from these processes whether there is only a single metazoon phylum or whether there may not be two or more such phyla. At the same time there appear to be strong arguments for regarding the Porifera as a phylum of the Metazoa derived independently from the Protozoa. This seems to me to be shewn (1) by the striking larval peculiarities of the Porifera; (2) by the early development of the mesoblast in the Porifera, which stands in strong contrast to the absence of this layer in the embryos of most Cœlenterata; and above all, (3) by the remarkable characters of the system of digestive channels. A further argument in the same direction is supplied by the fact that the germinal layers of the Sponges very probably do not correspond physiologically to the germinal layers of other types. The embryological evidence is insufficient to decide whether the amphiblastula larva is, as suggested above, to be regarded as the larval ancestor of the Porifera.

Homologies of the germinal layers. The question as to how far there is a complete homology between the two primary germinal layers throughout the Metazoa was the third of the questions proposed to be discussed here.

Since there are some Metazoa with only two germinal layers, and other Metazoa with three, and since, as is shewn in the following section, the third layer or mesoblast can only be regarded as a derivative of one or both the primary layers, it is clear that a complete homology between the two primary germinal layers does not exist.

That there is a general homology appears on the other hand hardly open to doubt.

The primary layers are usually continuous with each other, near one or both (when both are present) the openings of the alimentary tract.

As a rule an oral and anal section of the alimentary tract—the stomodæum and proctodæum—are derived from the epiblast; but the limits of both these sections are so variable, sometimes even in closely allied forms, that it is difficult to avoid the conclusion that there is a border-land between the epiblast and hypoblast, which appears by its development to belong in some forms to the epiblast and in other forms to the hypoblast. If this is not the case it is necessary to admit that there are instances in which a very large portion of the alimentary canal is phylogenetically an epiblastic structure. In some of the Isopods, for example, the stomodæum and proctodæum give rise to almost the whole of the alimentary canal with its appendages, except the liver.

The origin of the Mesoblast. A diploblastic condition of the organism preceded, as we have seen, the triploblastic. The epiblast during the diploblastic condition was, as appears from such forms as Hydra, especially the sensory and protective layer, while the hypoblast was the secretory and assimilating layer; both layers giving rise to muscular elements. It must not, however, be supposed that in the early diploblastic ancestors there was a complete differentiation of function, but there is reason to think that both the primary layers retained an indefinite capacity for developing into any form of tissue[124]. The fact of the triploblastic condition being later than the diploblastic proves in a conclusive way that the mesoblast is a derivative of one or both the primary layers. In the Cœlenterata we can study the actual origin from the two primary layers of various forms of tissue which in the higher types are derived from the mesoblast[125]. This fact, as well as general à priori considerations, conclusively prove that the mesoblast did not at first originate as a mass of independent cells between the two primary layers, but that in the first instance it gradually arose as differentiations of the two layers, and that its condition in the embryo as an independent layer of undifferentiated cells is a secondary condition, brought about by the general tendency towards a simplification of development, and a retardation of histological differentiation[126].

The Hertwigs have recently attempted (No. 271) to distinguish two types of differentiation of the mesoblast, viz. (1) a direct differentiation from the primitive epithelial cells; (2) a differentiation from primitively indifferent cells budded off into the gelatinous matter between the two primary layers.

It is quite possible that this distinction may be well founded, but no conclusive evidence of the occurrence of the second process has yet been adduced. The Ctenophora are the type upon which special stress is laid, but the early passage of amœboid cells into the gelatinous tissue, which subsequently become muscular, is very probably an embryonic abbreviation; and it is quite possible that these cells may phylogenetically have originated from epithelial cells provided with contractile processes passing through the gelatinous tissue.

The conversion of non-embryonic connective-tissue cells into muscle cells in the higher types has been described, but very much more evidence is required before it can be accepted as a common occurrence.

In addition to the probably degraded Dicyemidæ and Orthonectidæ, the Cœlenterata are the only group in which a true mesoblast is not always present. In other words, the Cœlenterata are the only group in which there is not found in the embryo an undifferentiated group of cells from which the majority of the organs situated between the epidermis and the alimentary epithelium are developed.

The organs invariably derived, in the triploblastic forms, from the mesoblast, are the vascular and lymphatic systems, the muscular system, and the greater part of the connective tissue and the excretory and generative (?) systems. On the other hand, the nervous systems (with a few possible exceptions) and organs of sense, the epithelium of most glands, and a few exceptional connective-tissue organs, as for example the notochord, are developed from the two primary layers.

The fact of the first-named set of organs being invariably derived from the mesoblast points to the establishment of the two following propositions:—(1) That with the differentiation of the mesoblast as a distinct layer by the process already explained, the two primary layers lost for the most part the capacity they primitively possessed of giving rise to muscular and connective-tissue differentiations[127], to the epithelium of the excretory organs, and to generative cells. (2) That the mesoblast throughout the triploblastic Metazoa, in so far as these forms have sprung from a common triploblastic ancestor, is an homologous structure.

The second proposition follows from the first. The mesoblast can only have ceased to be homologous throughout the triploblastica by additions from the two primary layers, and the existence of such additions is negatived by the first proposition.

These two propositions, which hang together, are possibly only approximately true, since it is quite possible that future investigations may shew that differentiations of the two primary layers are not so rare as has been hitherto imagined.

Ranvier[128] finds that the muscles of the sweat-glands are developed from the inner part of the layer of epiblast cells, invaginated to form these glands.

Götte[129] describes the epiblast cells of the larva of Comatula as being at a certain stage contractile and compares them with the epithelio-muscular cells of Hydra. These cells would appear subsequently to be converted into a simple cuticular structure.

It is moreover quite possible that fresh differentiations from the two primary layers may have arisen after the triploblastic condition had been established, and by the process of simplification of development and precocious segregation, as Lankester calls it, have become indistinguishable from the normal mesoblast. In spite of these exceptions it is probable that the major part of the muscular system of all existing triploblastic forms has been differentiated from the muscular system of the ancestor or ancestors (if there is more than one phylum) of the triploblastica. In the case of other tissues there are a few instances which might be regarded as examples of an organ primitively developed in one of the two primary layers having become secondarily carried into the mesoblast. The notochord has sometimes been cited as such an organ, but, as indicated in a previous chapter, it is probable that its hypoblastic origin can always be demonstrated.

The nervous system, although imbedded in mesoblastic derivates in the adults of all the higher triploblastica, retains with marvellous constancy its epiblastic origin (though it is usually separated from the epiblast prior to its histogenic differentiation); yet in the Cephalopoda, and some other Mollusca, the evidence is in favour of its developing in the mesoblast. Should future investigations confirm these conclusions, a good example will be afforded of an organ changing the layer from which it usually develops[130]. The explanation of such a change would be precisely the same as that already given for the mesoblast as a whole.

The actual mode of origin of various tissues, which in the true triploblastic forms arise in mesoblast, can be traced in the Cœlenterata[131]. In this group the epiblast and hypoblast both give rise to muscular and connective-tissue elements; and although the main part of the nervous system is formed in the epiblast, it seems certain that in some types nerves may be derived from the hypoblast[132]. These facts are extremely interesting, but it is by no means certain that any conclusions can be directly drawn from them as to the actual origin of the mesoblast in the triploblastic forms, till we know from what diploblastic forms the triploblastica originated. All that they shew is that any of the constituents of the mesoblast may have originated from either of the primitive layers.

For further light as to the origin of the mesoblast, it is necessary to turn to its actual development.

The following summary illustrates the more important modes in which the mesoblast originates.

1. It grows inwards from the lips of the blastopore as a pair of bands. In these cases it may originate (a) from cells which are clearly hypoblastic, (b) from cells which are clearly epiblastic, (c) from cells which cannot be regarded as belonging to either layer.

Mollusca.—Gasteropoda, Cephalopoda, and Lamellibranchiata. In Gasteropoda and Lamellibranchiata the mesoblast sometimes originates from a pair of cells at the lips of the blastopore, though very probably some of the elements subsequently come from the epiblast; and in Cephalopoda it begins as a ring of cells round the edge of the blastoderm.

Polyzoa Entoprocta.—It originates from a pair of cells at the lips of the blastopore.

Chætopoda.—Euaxes. It arises as a ridge of cells at the lips of the blastopore (fig. 200).

Gephyrea.—Bonellia. It arises (fig. 206) as an infolding of the epiblastic lips of the blastopore.

Nematelminthes.—Cucullanus. It grows backwards from the hypoblast cells at the persistent oral opening of the blastopore.

Tracheata.—Insecta. It grows inwards from the lips of the germinal groove (fig. 207), which probably represent the remains of a blastopore. Part of the mesoblast is probably also derived from the yolk-cells. A similar though more modified development of the mesoblast occurs in the Araneina (fig. 214).

Crustacea.—Decapoda. It partly grows in from the hypoblastic lips of the blastopore, and is partly derived from the yolk-cells (fig. 208).

2. The mesoblast is developed from the walls of hollow outgrowths of the archenteron, the cavities of which become the body cavity.

Brachiopoda.—The walls of a pair of outgrowths form the whole of the mesoblast.

Chætognatha.—The mesoblast arises in the same manner as in the Brachiopoda (fig. 209).

Echinodermata.—The lining of the peritoneal cavity is developed from the walls of outgrowths of the archenteron, but the greater part of the mesoblast is derived from the amœboid cells budded off from the walls of the archenteron (fig. 210).

Enteropneusta (Balanoglossus).—The body cavity is derived from two pairs of alimentary diverticula, the walls of which give rise to the greater part of the mesoblast.

Chordata.—Paired archenteric outgrowths give rise to the whole mesoblast in Amphioxus (fig. 211), and the mode of formation of the mesoblast in other Chordata is probably secondarily derived from this.

3. The cells which will form the mesoblast become marked out very early, and cannot be regarded as definitely springing from either of the primary layers.

Turbellaria.—Leptoplana (fig. 212), Planaria polychroa (?).

Chætopoda.—Lumbricus, &c.

Discophora.

It is very possible that the cases quoted under this head ought more properly to belong to group 1.

4. The mesoblast cells are split off from the epiblast.

Nemertea.—Larva of Desor. The mesoblast is stated to be split off from the four invaginated discs.

5. The mesoblast is split off from the hypoblast.

Nemertea.—Some of the types without a metamorphosis.

Mollusca.—Scaphopoda. It is derived from the lateral and ventral cells of the hypoblast.

Gephyrea.—Phascolosoma.

Vertebrata.—In most of the Ichthyopsida the mesoblast is derived from the hypoblast (fig. 213). In some types (i.e. most of the Amniota) the mesoblast might be described as originating at the lips of the blastopore (primitive streak).

6. The mesoblast is derived from both germinal layers.

Tracheata.—Araneina (fig. 214). It is derived partly from cells split off from the epiblast and partly from the yolk-cells; but it is probable that the statement that the mesoblast is derived from both the germinal layers is only formally accurate; and that the derivation of part of the mesoblast from the yolk-cells is not to be interpreted as a derivation from the hypoblast.

Amniota.—The derivation of the mesoblast of the Amniota from both the primary germinal layers is without doubt a secondary process.

The conclusions to be drawn from the above summary are by no means such as might have been anticipated. The analogy of the Cœlenterata would lead us to expect that the mesoblast would be derived partly from the epiblast and partly from the hypoblast. Such, however, is not for the most part the case, though more complete investigations may shew that there are a greater number of instances in which the mesoblast has a mixed origin than might be supposed from the above summary.

I have attempted to reduce the types of development of the mesoblast to six; but owing to the nature of the case it is not always easy to distinguish the first of these from the last four. Of the six types the second will on most hands be admitted to be the most remarkable. The formation of hollow outgrowths of the archenteron, the cavities of which give rise to the body cavity, can only be explained on the supposition that the body cavity of the types in which such outgrowths occur is derived from diverticula cut off from the alimentary tract. The lining epithelium of the diverticula—the peritoneal epithelium—is clearly part of the primitive hypoblast, and this part of the mesoblast is clearly hypoblastic in origin.

In the case of the Chætognatha (Sagitta), Brachiopoda, and Amphioxus, the whole of the mesoblast originates from the walls of the diverticula; while in the Echinodermata the walls of the diverticula only give rise to the vaso-peritoneal epithelium, the remainder of the mesoblast being derived from amœboid cells which spring from the walls of the archenteron before the origin of the vaso-peritoneal outgrowths (figs. 199 and 210).

Reserving for the moment the question as to what conclusions can be deduced from the above facts as to the origin of the mesoblast, it is important to determine how far the facts of embryology warrant us in supposing that in the whole of the triploblastic forms the body cavity originated from the alimentary diverticula. There can be but little doubt that the mode of origin of the mesoblast in many Vertebrata, as two solid plates split off from the hypoblast, in which a cavity is secondarily developed, is an abbreviation of the process observable in Amphioxus; but this process approaches in some forms of Vertebrata to the ingrowth of the mesoblast from the lips of the blastopore.

It is, therefore, highly probable that the paired ingrowths of the mesoblast from the lips of the blastopore may have been in the first instance derived from a pair of archenteric diverticula. This process of formation of the mesoblast is, as may be seen by reference to the summary, the most frequent, including as it does the Chætopoda, the Mollusca, the Arthropoda, &c.[133]

While there is no difficulty in the view that the body cavity may have originated from a pair of enteric diverticula in the case of the forms where a body cavity is present, there is a considerable difficulty in holding this view, for forms in which there is no body cavity distinct from the alimentary diverticula.

Of these types the Platyelminthes are the most striking. It is, no doubt, possible that a body cavity may have existed in the Platyelminthes, and become lost; and the case of the Discophora, which in their muscular and connective tissue systems as well as in the absence of a body cavity resemble the Platyelminthes, may be cited in favour of this view, in that, being closely related to the Chætopoda, they are almost certainly descended from ancestors with a true body cavity. The usual view of the primitive character of the Platyelminthes, which has much to support it, is, however, opposed to the idea that the body cavity has disappeared.

If Kowalevsky[134] is right in stating that he has found a form intermediate between the Cœlenterata and the Platyelminthes, there will be strong grounds for holding that the Platyelminthes are, like the Cœlenterata, forms the ancestors of which were not provided with a body cavity.

Perhaps the triploblastica are composed of two groups, viz. (1) a more ancestral group (the Platyelminthes), in which there is no body cavity as distinct from the alimentary, and (2) a group descended from these, in which two of the alimentary diverticula have become separated from the alimentary tract to form a body cavity (remaining triploblastica). However this may be, the above considerations are sufficient to shew how much there is that is still obscure with reference even to the body cavity.

If embryology gives no certain sound as to the questions just raised with reference to the body cavity, still less is it to be hoped that the remaining questions with reference to the origin of the mesoblast can be satisfactorily answered. It is clear, in the first place, from an inspection of the summary given above, that the process of development of the mesoblast is, in all the higher forms, very much abbreviated and modified. Not only is its differentiation relatively deferred, but it does not in most cases originate, as it must have done to start with, as a more or less continuous sheet, split off from parts of one or both the primary layers. It originates in most cases from the hypoblast, and although the considerations already urged preclude us from laying very great stress on this mode of origin, yet the derivation of the mesoblast from the walls of archenteric outgrowths suggests the view that the whole, or at any rate the greater part, of the mesoblast primitively arose by a process of histogenic differentiation from the walls of the archenteron or rather from diverticula of these walls. This view, which was originally put forward by myself (No. 260), appears at first sight very improbable, but if the statement of the Hertwigs (No. 270), that there is a large development of a hypoblastic muscular system in the Actinozoa, is well founded, it cannot be rejected as impossible. Lankester (No. 279), on the other hand, has urged that the mode of origin of the mesoblast in the Echinodermata is more primitive; and that the amœboid cells which here give rise to the muscular and connective tissues represent cells which originally arose from the whole inner surface of the epiblast. It is, however, to be noted that even in the Echinodermata the amœboid cells actually arise from the hypoblast, and their mode of origin may, therefore, be used to support the view that the main part of the muscular system of higher types is derived from the primitive hypoblast.

The great changes which have taken place in the development of the mesoblast would be more intelligible on this view than on the view that the major part of the mesoblast primitively originated from the epiblast. The presence of food-yolk is much more frequent in the hypoblast than in the epiblast; and it is well known that a large number of the changes in early development are caused by food-yolk. If, therefore, the mesoblast has been derived from the hypoblast, many more changes might be expected to have been introduced into its early development than if it had been derived from the epiblast. At the same time the hypoblastic origin of the mesoblast would assist in explaining how it has come about that the development of the nervous system is almost always much less modified than that of the mesoblast, and that the nervous system is not, as might, on the grounds of analogy, have been anticipated, as a rule secondarily developed in the mesoblast.

The Hertwigs have recently suggested in their very interesting memoir (No. 271) that the Triploblastica are to be divided into two phyla, (1) the Enterocœla, and (2) the Pseudocœla; the former group containing the Chætopoda, Gephyrea, Brachiopoda, Nematoda, Arthropoda, Echinodermata, Enteropneusta and Chordata; and the latter the Mollusca, Polyzoa, the Rotifera, and Platyelminthes.

The Enterocœla are forms in which the primitive alimentary diverticula have given origin to the body cavity, while the major part of the muscular system has originated from the epithelial walls of these diverticula, part however being in many cases also derived from the amœboid cells, called by them mesenchyme, by the second process of mesoblastic differentiation mentioned on p. 347.

In the Pseudocœla the muscular system has become differentiated from mesenchyme cells; while the body cavity, where it exists, is merely a split in the mesenchyme.

It is impossible for me to attempt in this place to state fully, or do justice to, the original and suggestive views contained in this paper. The general conclusion I cannot however accept. The views of the Hertwigs depend to a large extent upon the supposition that it is possible to distinguish histologically muscle cells derived from epithelial cells, from those derived from mesenchyme cells. That in many cases, and strikingly so in the Chordata, the muscle cells retain clear indications of their primitive origin from epithelial cells, I freely admit; but I do not believe either that its histological character can ever be conclusive as to the non-epithelial origin of a muscle cell, or that its derivation in the embryo from an indifferent amœboid cell is any proof that it did not, to start with, originate from an epithelial cell.

I hold, as is clear from the preceding statements, that such immense secondary modifications have taken place in the development of the mesoblast, that no such definite conclusions can be deduced from its mode of development as the Hertwigs suppose.

In support of the view that the early character of embryonic cells is no safe index as to their phylogenetic origin, I would point to the few following facts.

(1) In the Porifera and many of the Cœlenterata (Eucope polystyla, Geryonia, &c.) the hypoblast (endoderm) originates from cells, which according to the Hertwigs’ views ought to be classed as mesenchyme.

(2) In numerous instances muscles which have, phylogenetically, an undoubted epithelial origin, are ontogenetically derived from cells which ought to be classed as mesenchyme. The muscles of the head in all the higher Vertebrata, in which the head cavities have disappeared, are examples of this kind; the muscles of many of the Tracheata, notably the Araneina, must also be placed in the same category.

(3) The Mollusca are considered by the Hertwigs to be typical Pseudocœla. A critical examination of the early development of the mesoblast in these forms demonstrates however that with reference to the mesoblast they must be classed in the same group as the Chætopoda. The mesoblast (Vol. II. p. 227) clearly originates as two bands of cells which grow inwards from the blastopore, and in some forms (Paludina, Vol. II. fig. 107) become divided into a splanchnic and somatic layer, with a body cavity between them. All these processes are such as are, in other instances, admitted to indicate Enterocœlous affinities.

The subsequent conversion of the mesoblast elements into amœboid cells, out of which branched muscles are formed, is in my opinion simply due to the envelopment of the soft Molluscan body within a hard shell.

In addition to these instances I may point out that the distinction between the Pseudocœla and Enterocœla utterly breaks down in the case of the Discophora, and the Hertwigs have made no serious attempt to discuss the characters of this group in the light of their theory, and that the derivation of the Echinoderm muscles from mesenchyme cells is a difficulty which is very slightly treated.

II. Larval forms: their nature, origin and affinities.

Preliminary considerations. In a general way two types of development may be distinguished, viz. a fœtal type and a larval type. In the fœtal type animals undergo the whole or nearly the whole of their development within the egg or within the body of the parent, and are hatched in a condition closely resembling the adult; and in the larval type they are born at an earlier stage of development, in a condition differing to a greater or less extent from the adult, and reach the adult state either by a series of small steps, or by a more or less considerable metamorphosis.

The satisfactory application of embryological data to morphology depends upon a knowledge of the extent to which the record of ancestral history has been preserved in development. Unless secondary changes intervened this record would be complete; it becomes therefore of the first importance to the embryologist to study the nature and extent of the secondary changes likely to occur in the fœtal or the larval state.

The principles which govern the perpetuation of variations which occur in either the larval or the fœtal state are the same as those for the adult condition. Variations favourable to the survival of the species are equally likely to be perpetuated, at whatever period of life they occur, prior to the loss of the reproductive powers. The possible nature and extent of the secondary changes which may have occurred in the developmental history of forms, which have either a long larval existence, or which are born in a nearly complete condition, is primarily determined by the nature of the favourable variations which can occur in each case.

Where the development is a fœtal one, the favourable variations which can most easily occur are—(1) abbreviations, (2) an increase in the amount of food-yolk stored up for the use of the developing embryo. Abbreviations take place because direct development is always simpler, and therefore more advantageous; and, owing to the fact of the fœtus not being required to lead an independent existence till birth, and of its being in the meantime nourished by food-yolk, or directly by the parent, there are no physiological causes to prevent the characters of any stage of the development, which are of functional importance during a free but not during a fœtal existence, from disappearing from the developmental history. All organs of locomotion and nutrition not required by the adult will, for this reason, obviously have a tendency to disappear or to be reduced in fœtal developments; and a little consideration will shew that the ancestral stages in the development of the nervous and muscular systems, organs of sense, and digestive system will be liable to drop out or be modified, when a simplification can thereby be effected. The circulatory and excretory systems will not be modified to the same extent, because both of them are usually functional during fœtal life.

The mechanical effects of food-yolk are very considerable, and numerous instances of its influence will be found in the earlier chapters of this work[135]. It mainly affects the early stages of development, i.e. the form of the gastrula, &c.

The favourable variations which may occur in the free larva are much less limited than those which can occur in the fœtus. Secondary characters are therefore very numerous in larvæ, and there may even be larvæ with secondary characters only, as, for instance, the larvæ of Insects.

In spite of the liability of larvæ to acquire secondary characters, there is a powerful counterbalancing influence tending towards the preservation of ancestral characters, in that larvæ are necessarily compelled at all stages of their growth to retain in a functional state such systems of organs, at any rate, as are essential for a free and independent existence. It thus comes about that, in spite of the many causes tending to produce secondary changes in larvæ, there is always a better chance of larvæ repeating, in an unabbreviated form, their ancestral history, than is the case with embryos, which undergo their development within the egg.

It may be further noted as a fact which favours the relative retention by larvæ of ancestral characters, that a secondary larval stage is less likely to be repeated in development than an ancestral stage, because there is always a strong tendency for the former, which is a secondarily intercalated link in the chain of development, to drop out by the occurrence of a reversion to the original type of development.

The relative chances of the ancestral history being preserved in the fœtus or the larva may be summed up in the following way:—There is a greater chance of the ancestral history being lost in forms which develop in the egg; and of its being masked in those which are hatched as larvæ.

The evidence from existing forms undoubtedly confirms the a priori considerations just urged[136]. This is well shewn by a study of the development of Echinodermata, Nemertea, Mollusca, Crustacea, and Tunicata. The free larvæ of the four first groups are more similar amongst themselves than the embryos which develop directly, and since this similarity cannot be supposed to be due to the larvæ having been modified by living under precisely similar conditions, it must be due to their retaining common ancestral characters. In the case of the Tunicata the free larvæ retain much more completely than the embryos certain characters such as the notochord, the cerebrospinal canal, etc., which are known to be ancestral.

Types of Larvæ.—Although there is no reason to suppose that all larval forms are ancestral, yet it seems reasonable to anticipate that a certain number of the known types of larvæ would retain the characters of the ancestors of the more important phyla of the animal kingdom.

Before examining in detail the claims of various larvæ to such a character, it is necessary to consider somewhat more at length the kind of variations which are most likely to occur in larval forms.

It is probable a priori that there are two kinds of larvæ, which may be distinguished as primary and secondary larvæ. Primary larvæ are more or less modified ancestral forms, which have continued uninterruptedly to develop as free larvæ from the time when they constituted the adult form of the species. Secondary larvæ are those which have become introduced into the ontogeny of species, the young of which were originally hatched with all the characters of the adult; such secondary larvæ may have originated from a diminution of food-yolk in the egg and a consequently earlier commencement of a free existence, or from a simple adaptive modification in the just hatched young. Secondary larval forms may resemble the primary larval forms in cases where the ancestral characters were retained by the embryo in its development within the egg; but in other instances their characters are probably entirely adaptive.

Causes tending to produce secondary changes in larvæ.—The modes of action of natural selection on larvæ may probably be divided more or less artificially into two classes.
1. The changes in development directly produced by the existence of a larval stage.
2. The adaptive changes in a larva acquired in the ordinary course of the struggle for existence.

The changes which come under the first head consist essentially in a displacement in the order of development of certain organs. There is always a tendency in development to throw back the differentiation of the embryonic cells into definite tissues to as late a date as possible. This takes place in order to enable the changes of form, which every organ undergoes, in repeating even in an abbreviated way its phylogenetic history, to be effected with the least expenditure of energy. Owing to this tendency it comes about that when an organism is hatched as a larva many of the organs are still in an undifferentiated state, although the ancestral form which this larva represents had all its organs fully differentiated. In order, however, that the larva may be enabled to exist as an independent organism, certain sets of organs, e.g. the muscular, nervous, and digestive systems, have to be histologically differentiated. If the period of fœtal life is shortened, an earlier differentiation of certain organs is a necessary consequence; and in almost all cases the existence of a larval stage causes a displacement in order of development of organs, the complete differentiation of many organs being retarded relatively to the muscular, nervous, and digestive systems.

The possible changes under the second head appear to be unlimited. There is, so far as I see, no possible reason why an indefinite number of organs should not be developed in larvæ to protect them from their enemies, and to enable them to compete with larvæ of other species, and so on. The only limit to such development appears to be the shortness of larval life, which is not likely to be prolonged, since, ceteris paribus, the more quickly maturity is reached the better it is for the species.

A very superficial examination of marine larvæ shews that there are certain peculiarities common to most of them, and it is important to determine how far such peculiarities are to be regarded as adaptive. Almost all marine larvæ are provided with well-developed organs of locomotion, and transparent bodies. These two features are precisely those which it is most essential for such larvæ to have. Organs of locomotion are important, in order that larvæ may be scattered as widely as possible, and so disseminate the species; and transparency is very important in rendering larvæ invisible, and so less liable to be preyed upon by their numerous enemies[137].

These considerations, coupled with the fact that almost all free-swimming animals, which have not other special means of protection, are transparent, seem to shew that the transparency of larvæ at all events is adaptive; and it is probable that organs of locomotion are in many cases specially developed, and not ancestral.

Various spinous processes on the larvæ of Crustacea and Teleostei are also examples of secondarily acquired protective organs.

These general considerations are sufficient to form a basis for the discussion of the characters of the known types of larvæ.

The following table contains a list of the more important of such larval forms:
 Dicyemidæ.—The Infusoriform larva (vol. II. fig. 62).
 Porifera.—(a) The Amphiblastula larva (fig. 215), with one-half of the body ciliated, and the other half without cilia; (b) an oval uniformly ciliated larva, which may be either solid or have the form of a vesicle.
 Cœlenerata.—The planula (fig. 216).
 Turbellaria.—(a) The eight-lobed larva of Müller (fig. 222); (b) the larva of Götte and Metschnikoff, with some Pilidium characters.
 Nemertea.—The Pilidium (fig. 221).
 Trematoda.—The Cercaria.
 Rotifera.—The Trochosphere-like larvæ of Brachionus (fig. 217) and Lacinularia.
 Mollusca.—Mollusca.—The Trochosphere larva (fig. 218), and the subsequent Veliger larva (fig. 219).
 Brachiopoda.—The three-lobed larva, with a postoral ring of cilia (fig. 220).
 Polyzoa.—A larval form with a single ciliated ring surrounding the mouth, and an aboral ciliated ring or disc (fig. 228).
 Chætopoda.—Various larval forms with many characters like those of the molluscan Trochosphere, frequently with distinct transverse bands of cilia. They are classified as Atrochæ, Mesotrochæ, Telotrochæ (fig. 225 A and fig. 226), Polytrochæ, and Monotrochæ (fig. 225 B).
 Gephyrea Nuda.—Larval forms like those of preceding groups. A specially characteristic larva is that of Echiurus (fig. 227).
 Gephyrea Tubicola.—Actinotrocha (fig. 230), with a postoral ciliated ring of arms.
 Myriapoda.—A functionally hexapodous larval form is common to all the Chilognatha (vol. II. fig. 174).
 Insecta.—Various secondary larval forms.
 Crustacea.—The Nauplius (vol. II. fig. 208) and the Zoæa (vol. II. fig. 210).
 Echinodermata.—The Auricularia (fig. 223 A), the Bipinnaria (fig. 223 B), and the Pluteus (fig. 224), and the transversely-ringed larvæ of Crinoidea (vol. II. fig. 268). The three first of which can be reduced to a common type (fig. 231 C).
 Enteropneusta.—Tornaria (fig. 229).
 Urochorda (Tunicata).—The tadpole-like larva (vol. III. fig. 8).
 Ganoidei.—A larva with a disc with adhesive papillæ in front of the mouth (vol. III. fig. 67).
 Anurus Amphibia.—The tadpole (vol. III. fig. 80).

Of the larval forms included in the above list a certain number are probably without affinities outside the group to which they belong. This is the case with the larvæ of the Myriapoda, the Crustacean larvæ, and with the larval forms of the Chordata. I shall leave these forms out of consideration.

There are, again, some larval forms which may possibly turn out hereafter to be of importance, but from which, in the present state of our knowledge, we cannot draw any conclusions. The infusoriform larva of the Dicyemidæ, and the Cercaria of the Trematodes, are such forms.

Excluding these and certain other forms, we have finally left for consideration the larvæ of the Cœlenterata, the Turbellaria, the Rotifera, the Nemertea, the Mollusca, the Polyzoa, the Brachiopoda, the Chætopoda, the Gephyrea, the Echinodermata, and the Enteropneusta.

The larvæ of these forms can be divided into two groups. The one group contains the larva of the Cœlenterata or Planula, the other group the larvæ of all the other forms.

The Planula (fig. 216) is characterised by its extreme simplicity. It is a two-layered organism, with a form varying from cylindrical to oval, and usually a radial symmetry. So long as it remains free it is not usually provided with a mouth, and it is as yet uncertain whether or no the absence of a mouth is to be regarded as an ancestral character. The Planula is very probably the ancestral form of the Cœlenterata.

The larvæ of almost all the other groups, although they may be subdivided into a series of very distinct types, yet agree in the possession of certain common characters[138]. There is a more or less dome-shaped dorsal surface, and a flattened or concave ventral surface, containing the opening of the mouth, and usually extending posteriorly to the opening of the anus, when such is present.

The dorsal dome is continued in front of the mouth to form a large præoral lobe.

There is usually present at first an uniform covering of cilia; but in the later larval stages there are almost always formed definite bands or rings of long cilia, by which locomotion is effected. These bands are often produced into arm-like processes.

The alimentary canal has, typically, the form of a bent tube with a ventral concavity, constituted (when an anus is present) of three sections, viz. an œsophagus, a stomach, and a rectum. The œsophagus and sometimes the rectum are epiblastic in origin, while the stomach always and the rectum usually are derived from the hypoblast[139].

To the above characters may be added a glass-like transparency; and the presence of a widish space possibly filled with gelatinous tissue, and often traversed by contractile cells, between the alimentary tract and the body wall.

Considering the very profound differences which exist between many of these larvæ, it may seem that the characters just enumerated are hardly sufficient to justify my grouping them together. It is, however, to be borne in mind that my grounds for doing so depend quite as much upon the fact that they constitute a series without any great breaks in it, as upon the existence of characters common to the whole of them. It is also worth noting that most of the characters which have been enumerated as common to the whole of these larvæ are not such secondary characters as (in accordance with the considerations used above) might be expected to arise from the fact of their being subjected to nearly similar conditions of life. Their transparency is, no doubt, such a secondary character, and it is not impossible that the existence of ciliated bands may be so also; but it is quite possible that if, as I suppose, these larvæ reproduce the characters of some ancestral form, this form may have existed at a time when all marine animals were free-swimming, and that it may, therefore, have been provided with at least one ciliated band.

The detailed consideration of the characters of these larvæ, given below, supports this view.

This great class of larvæ may, as already stated, be divided into a series of minor subdivisions. These subdivisions are the following:

1. The Pilidium Group.—This group is characterised by the mouth being situated nearly in the centre of the ventral surface, and by the absence of an anus. It includes the Pilidium of the Nemertines (fig. 221), and the various larvæ of marine Dendrocœla (fig. 222). At the apex of the præoral lobe a thickening of epiblast may be present, from which (fig. 232) a contractile cord sometimes passes to the œsophagus.

2. The Echinoderm Group.—This group (figs. 223, 224 and 231 C) is characterised by the presence of a longitudinal postoral band of cilia, by the absence of special sense organs in the præoral region, and by the development of the body cavity as an outgrowth of the alimentary tract. The three typical divisions of the alimentary tract are present, and there is a more or less developed præoral lobe. This group only includes the larvæ of the Echinodermata.

3. The Trochosphere Group.—This group (figs. 225, 226) is characterised by the presence of a præoral ring of long cilia, the region in front of which forms a great part of the præoral lobe. The mouth opens immediately behind the præoral ring of cilia, and there is very often a second ring of short cilia parallel to the main ring, immediately behind the mouth. The function of the ring of short cilia is nutritive, in that its cilia are employed in bringing food to the mouth; while the function of the main ring is locomotive. A perianal patch or ring of cilia is often present (fig. 225 A), and in many forms intermediate rings are developed between the præoral and perianal rings.

The præoral lobe is usually the seat of a special thickening of epiblast, which gives rise to the supraœsophageal ganglion of the adult. On this lobe optic organs are very often developed in connection with the supraœsophageal ganglion, and a contractile band frequently passes from this region to the œsophagus.

The alimentary tract is formed of the three typical divisions.

The body cavity is not developed directly as an outgrowth of the alimentary tract, though the process by which it originates is very probably secondarily modified from a pair of alimentary outgrowths.

Paired excretory organs, opening to the exterior and into the body cavity, are often present (fig. 226 nph).

This type of larva is found in the Rotifera (fig. 217) (in which it is preserved in the adult state), the Chætopoda (figs. 225 and 226), the Mollusca (fig. 218), the Gephyrea nuda (fig. 227), and the Polyzoa (fig. 228)[140].

4. Tornaria.—This larva (fig. 229) is intermediate in most of its characters between the larvæ of the Echinodermata (more especially the Bipinnaria) and the Trochosphere. It resembles Echinoderm larvæ in the possession of a longitudinal ciliated band (divided into a præoral and a postoral ring), and in the derivation of the body cavity and water-vascular vesicle from alimentary diverticula; and it resembles the Trochosphere in the presence of sense organs on the præoral lobe, in the existence of a perianal ring of cilia, and in the possession of a contractile band passing from the præoral lobe to the œsophagus.

5. Actinotrocha.—The remarkable larva of Phoronis (fig. 230), known as Actinotrocha, is characterised by the presence of (1) a postoral and somewhat longitudinal ciliated ring produced into tentacles, and (2) a perianal ring. It is provided with a præoral lobe, and a terminal or somewhat dorsal anus.

6. The larva of the Brachiopoda articulata (fig. 220).

The relationships of the six types of larval forms thus briefly characterised have been the subject of a considerable amount of controversy, and the following suggestions on their affinities must be viewed as somewhat speculative. The Pilidium type of larva is in some important respects less highly differentiated than the larvæ of the five other groups. It is, in the first place, without an anus; and there are no grounds for supposing that the anus has become lost by retrogressive changes. If for the moment it is granted that the Pilidium larva represents more nearly than the larvæ of the other groups the ancestral type of larva, what characters are we led to assign to the ancestral form which this larva repeats?

In the first place, this ancestral form, of which fig. 231 A is an ideal representation, would appear to have had a dome-shaped body, with a flattened oral surface and a rounded aboral surface. Its symmetry was radial, and in the centre of the flattened oral surface was placed the mouth, and round its edge was a ring of cilia. The passage of a Pilidium-like larva into the vermiform bilateral Platyelminth form, and therefore it may be presumed of the ancestral form which this larva repeats, is effected by the larva becoming more elongated, and by the region between the mouth and one end of the body becoming the præoral region, and by an outgrowth between the mouth and the opposite end developing into the trunk, an anus becoming placed at its extremity in the higher forms.

If what has been so far postulated is correct, it is clear that this primitive larval form bears a very close resemblance to a simplified free-swimming Cœlenterate (Medusa), and that the conversion of such a radiate form into the bilateral took place, not by the elongation of the aboral surface, and the formation of an anus there, but by the unequal elongation of the oral face, an anterior part, together with the dome above it, forming a præoral lobe, and a posterior outgrowth the trunk (figs. 226 and 233); while the aboral surface became the dorsal surface.

This view fits in very well with the anatomical resemblances between the Cœlenterata and the Turbellaria[141], and shews, if true, that the ventral and median position of the mouth in many Turbellaria is the primitive one.

The above suggestion as to the mode of passage from the radial into the bilateral form differs largely from that usually held. Lankester[142], for instance, gives the following account of this passage:

≴It has been recognised by various writers, but notably by Gegenbaur and Haeckel, that a condition of radiate symmetry must have preceded the condition of bilateral symmetry in animal evolution. The Diblastula may be conceived to have been at first absolutely spherical with spherical symmetry. The establishment of a mouth led necessarily to the establishment of a structural axis passing through the mouth, around which axis the body was arranged with radial symmetry. This condition is more or less perfectly maintained by many Cœlenterates, and is reassumed by degradation of higher forms (Echinoderms, some Cirrhipedes, some Tunicates). The next step is the differentiation of an upper and a lower surface in relation to the horizontal position, with mouth placed anteriorly, assumed by the organism in locomotion. With the differentiation of a superior and inferior surface, a right and a left side, complementary one to the other, are necessarily also differentiated. Thus the organism becomes bilaterally symmetrical. The Cœlentera are not wanting in indications of this bilateral symmetry, but for all other higher groups of animals it is a fundamental character. Probably the development of a region in front of, and dorsal to the mouth, forming the Prostomium, was accomplished pari passu with the development of bilateral symmetry. In the radially symmetrical Cœlentera we find very commonly a series of lobes of the body-wall or tentacles produced equally—with radial symmetry, that is to say—all round the mouth, the mouth terminating the main axis of the body—that is to say, the organism being ‘telostomiate.’ The later fundamental form, common to all animals above the Cœlentera, is attained by shifting what was the main axis of the body—so that it may be described now as the ‘enteric’ axis; whilst the new main axis, that parallel with the plane of progression, passes through the dorsal region of the body running obliquely in relation to the enteric axis. Only one lobe or outgrowth of those radially disposed in the telostomiate organisms now persists. This lobe lies dorsally to the mouth, and through it runs the new main axis. This lobe is the Prostomium, and all the organisms which thus develop a new main axis, oblique to the old main axis, may be called prostomiate.”

It will be seen from this quotation that the aboral part of the body is supposed to elongate to form the trunk, while the præoral region is derived from one of the tentacles.

Before proceeding to further considerations as to the origin of the Bilateralia, suggested by the Pilidium type of larva, it is necessary to enter into a more detailed comparison between our larval forms.

A very superficial consideration of the characters of these forms brings to light two important features in which they differ, viz.:

(1) In the presence or absence of sense organs on the præoral lobe.

(2) In the presence or absence of outgrowths from the alimentary tract to form the body cavity.

The larvæ of the Echinodermata and Actinotrocha (?) are without sense organs on the præoral lobe, while the other types of larvæ are provided with them. Alimentary diverticula are characteristic of the larvæ of the Echinodermata and of Tornaria.

If the conclusion already arrived at to the effect that the prototype of the six larval groups was descended from a radiate ancestor is correct, it appears to follow that the nervous system, in so far as it was differentiated, had primitively a radiate form; and it is also probably true that there were alimentary diverticula in the form of radial pouches, two of which may have given origin to the paired diverticula which become the body cavity in such types as the Echinodermata, Sagitta, etc. If these two points are granted, the further conclusions seem to follow—(1) that the ganglion and sense organs of the præoral lobe were secondary structures, which arose (perhaps as differentiations of an original circular nerve ring) after the assumption of a bilateral form; and (2) that the absence of these organs in the larvæ of the Echinodermata and Actinotrocha (?) implies that these larvæ retain, so far, more primitive characters than the Pilidium. The same may be said of the alimentary diverticula. There are thus indications that in two important points the Echinoderm larvæ are more primitive than the Pilidium.

The above conclusions with reference to the Pilidium and Echinoderm larvæ involve some not inconsiderable difficulties, and suggest certain points for further discussion.

In the first place it is to be noted that the above speculations render it probable that the type of nervous system from which that found in the adults of the Echinodermata, Platyelminthes, Chætopoda, Mollusca, etc., is derived, was a circumoral ring, like that of Medusæ, with which radially arranged sense organs may have been connected; and that in the Echinodermata this form of nervous system has been retained, while in the other types it has been modified. Its anterior part may have given rise to supraœsophageal ganglia and organs of vision; these being developed on the assumption of a bilaterally symmetrical form, and the consequent necessity arising for the sense organs to be situated at the anterior end of the body. If this view is correct, the question presents itself as to how far the posterior part of the nervous system of the Bilateralia can be regarded as derived from the primitive radiate ring.

A circumoral nerve-ring, if longitudinally extended, might give rise to a pair of nerve-cords united in front and behind—exactly such a nervous system, in fact, as is present in many Nemertines[143] (the Enopla and Pelagonemertes), in Peripatus[144], and in primitive molluscan types (Chiton, Fissurella, etc.). From the lateral parts of this ring it would be easy to derive the ventral cord of the Chætopoda and Arthropoda. It is especially deserving of notice in connection with the nervous system of the above-mentioned Nemertines and Peripatus, that the commissure connecting the two nerve-cords behind is placed on the dorsal side of the intestine. As is at once obvious, by referring to the diagram (fig. 231 B), this is the position this commissure ought, undoubtedly, to occupy if derived from part of a nerve-ring which originally followed more or less closely the ciliated edge of the body of the supposed radiate ancestor.

The fact of this arrangement of the nervous system being found in so primitive a type as the Nemertines tends to establish the views for which I am arguing; the absence or imperfect development of the two longitudinal cords in Turbellarians may very probably be due to the posterior part of the nerve-ring having atrophied in this group.

It is by no means certain that this arrangement of the nervous system in some Mollusca and in Peripatus is primitive, though it may be so.

In the larvæ of the Turbellaria the development of sense organs in the præoral region is very clear (fig. 222 B); but this is by no means so obvious in the case of the true Pilidium. There is in Pilidium (fig. 232 A) a thickening of epiblast at the summit of the dorsal dome, which might seem, from the analogy of Mitraria, etc. (fig. 233), to correspond to the thickening of the præoral lobe, which gives rise to the supraœsophageal ganglion; but, as a matter of fact, this part of the larva does not apparently enter into the formation of the young Nemertine (fig. 232). The peculiar metamorphosis, which takes place in the development of the Nemertine out of the Pilidium[145], may, perhaps, eventually supply an explanation of this fact; but at present it remains as a still unsolved difficulty.

The position of the flagellum in Pilidium, and of the supraœsophageal ganglion in Mitraria, suggests a different view of the origin of the supraœsophageal ganglion from that adopted above. The position of the ganglion in Mitraria corresponds closely with that of the auditory organ in Ctenophora; and it is not impossible that the two structures may have had a common origin. If this view is correct, we must suppose that the apex of the aboral lobe has become the centre of the præoral field of the Pilidium and Trochosphere larval forms[146]—a view which fits in very well with their structure (figs. 226 and 233). The whole of the questions concerning the nervous system are still very obscure, and until further facts are brought to light no definite conclusions can be arrived at.

The absence of sense organs on the præoral lobe of larval Echinodermata, coupled with the structure of the nervous system of the adult, points to the conclusion that the adult Echinodermata have retained, and not, as is now usually held, secondarily acquired, their radial symmetry; and if this is admitted it follows that the obvious bilateral symmetry of Echinoderm larvæ is a secondary character.

The bilateral symmetry of many Cœlenterate larvæ (the larva of Æginopsis, of many Acraspeda, of Actinia, &c.), coupled with the fact that a bilateral symmetry is obviously advantageous to a free-swimming form, is sufficient to shew that this supposition is by no means extravagant; while the presence of only two alimentary diverticula in Echinoderm larvæ is quite in accord with the presence of a single pair of perigastric chambers in the early larva of Actinia, though it must be admitted that the derivation of the water-vascular system from the left diverticulum is not easy to understand on this view.

A difficulty in the above speculation is presented by the fact of the anus of the Echinodermata being the permanent blastopore, and arising prior to the mouth. If this fact has any special significance, it becomes difficult to regard the larva of Echinoderms and that of the other types as in any way related; but if the views already urged, in a previous section on the germinal layers, as to the unimportance of the blastopore, are admitted, the fact of the anus coinciding with the blastopore ceases to be a difficulty. As may be seen, by referring to fig. 231 C, the anus is placed on the dorsal side of the ciliated band. This position for the anus adapts itself to the view that the Echinoderm larva had originally a radial symmetry, with the anus placed at the aboral apex, and that, with the elongation of the larva on the attainment of a bilateral symmetry, the aboral apex became shifted to the present position of the anus.

It may be noticed that the obscure points connected with the absence of a body cavity in most adult Platyelminthes, which have already been dealt with in the section of this chapter devoted to the germinal layers, present themselves again here; and that it is necessary to assume either that alimentary diverticula, like those in the Echinodermata, were primitively present in the Platyelminthes, but have now disappeared from the ontogeny of this group, or that the alimentary diverticula have not become separated from the alimentary tract.

So far the conclusion has been reached that the archetype of the six types of larvæ had a radiate form, and that amongst existing larvæ it is most nearly approached in general shape and in the form of the alimentary canal by the Pilidium group, and in certain other particulars by the Echinoderm larvæ.

The edge of the oral disc of the larval archetype was probably armed with a ciliated ring, from which the ciliated ring of the Pilidium type and of the Echinodermata was most likely derived. The ciliated ring of the Pilidium varies greatly in its characters, and has not always the form of a complete ring. In Pilidium proper (fig. 232 A) it is a simple ring surrounding the edge of the oral disc. In Müller’s larva of Thysanozoon (fig. 222 B) it is inclined at an axis to the oral disc, and might be called præoral, but such a term cannot be properly used in the absence of an anus.

The Echinoderm ring is oblique to the axis of the body, and, owing to the fact of its passing ventrally in front of the anus, must be called postoral.

The next point to be considered is that of the affinities of the other larval types to these two types.

The most important of all the larval types is the Trochosphere, and this type is undoubtedly more closely related to the Pilidium than to the Echinoderm larva. Mitraria amongst the Chætopods (fig. 233) has, indeed, nearly the form of a Pilidium, and mainly differs from a Pilidium in the possession of an anus and of provisional bristles; the same may be said of Cyphonautes (fig. 234) amongst the Polyzoa.

The existence of these two forms appears to shew that the præoral ciliated ring of the Trochosphere may very probably be derived directly from the circumoral ciliated ring of the Pilidium; the other ciliated rings or patches of the Trochosphere having a secondary origin.

The larva of the Brachiopoda (fig. 220), in spite of its peculiar characters, is, in all probability, more closely related to the Chætopod Trochosphere than to any other larval type. The most conspicuous point of agreement between them is, however, the possession in common of provisional setæ.

Echinoderm larvæ differ from the Trochosphere, not only in the points already alluded to, but in the character of the ciliated band. The Echinoderm band is longitudinal and postoral. As just stated, there is reason to think that the præoral band of the Trochosphere and the postoral band of the Echinoderm larva are both derived from a ciliated ring surrounding the oral disc of the prototype of these larvæ (vide fig. 231). In the case of the Echinodermata the anus must have been formed on the dorsal side of this ring, and in the case of the Trochosphere on the ventral side; and so the difference in position between the two rings was brought about. Another view with reference to these rings has been put forward by Gegenbaur and Lankester, to the effect that the præoral ring of the Trochosphere is derived from the breaking up of the single band of most Echinoderm larvæ into the two bands found in Bipinnaria (vide fig. 223) and the atrophy of the posterior band. There is no doubt a good deal to be said for this origin of the præoral ring, and it is strengthened by the case of Tornaria; but the view adopted above appears to me more probable.

Actinotrocha (fig. 230) undoubtedly resembles more closely Echinoderm larvæ than the Trochosphere. Its ciliated ring has Echinoderm characters, and the growth along the line of the ciliated ring of a series of arms is very similar to what takes place in many Echinoderms. It also agrees with the Echinoderm larvæ in the absence of sense organs on the præoral lobe.

Tornaria (fig. 229) cannot be definitely united either with the Trochosphere or with the Echinoderm larval type. It has important characters in common with both of these groups, and the mixture of these characters renders it a very striking and well-defined larval form.

Phylogenetic conclusions. The phylogenetic conclusions which follow from the above views remain to be dealt with. The fact that all the larvæ of the groups above the Cœlenterata can be reduced to a common type seems to indicate that all the higher groups are descended from a single stem.

Considering that the larvæ of comparatively few groups have persisted, no conclusions as to affinities can be drawn from the absence of a larva in any group; and the presence in two groups of a common larval form may be taken as proving a common descent, but does not necessarily shew any close affinity.

There is every reason to believe that the types with a Trochosphere larva, viz. the Rotifera, the Mollusca, the Chætopoda, the Gephyrea, and the Polyzoa, are descended from a common ancestral form; and it is also fairly certain there was a remote ancestor common to these forms and to the Platyelminthes. A general affinity of the Brachiopoda with the Chætopoda is more than probable. All these types, together with various other types which are nearly related to them, but have not preserved an early larval form, are descended from a bilateral ancestor. The Echinodermata, on the other hand, are probably directly descended from a radial ancestor, and have more or less completely retained their radial symmetry. How far Actinotrocha[147] is related to the Echinoderm larvæ cannot be settled. Its characters may possibly be secondary, like those of the mesotrochal larvæ of Chætopods, or they may be due to its having branched off very early from the stock common to the whole of the forms above the Cœlenterata. The position of Tornaria is still more obscure. It is difficult, in the face of the peculiar water-vascular vesicle with a dorsal pore, to avoid the conclusion that it has some affinities with the Echinoderm larvæ. Such affinities would seem, on the lines of speculation adopted in this section, to prove that its affinities to the Trochosphere, striking as they appear to be, are secondary and adaptive. From this conclusion, if justified, it would follow that the Echinodermata and Enteropneusta have a remote ancestor in common, but not that the two groups are in any other way related.

General conclusions and summary. Starting from the demonstrated fact that the larval forms of a number of widely separated types above the Cœlenterata have certain characters in common, it has been provisionally assumed that the characters have been inherited from a common ancestor; and an attempt has been made to determine (1) the characters of the prototype of all these larvæ, and (2) the mutual relations of the larval forms in question. This attempt started with certain more or less plausible suggestions, the truth of which can only be tested by the coherence of the results which follow from them, and their capacity to explain all the facts.

The results arrived at may be summarised as follows:

1. The larval forms above the Cœlenterata may be divided into six groups enumerated on pages 370 to 373.

2. The prototype of all these groups was an organism something like a Medusa, with a radial symmetry. The mouth was placed in the centre of a flattened ventral surface. The aboral surface was dome-shaped. Round the edge of the oral surface was a ciliated ring, and probably a nervous ring provided with sense organs. The alimentary canal was prolonged into two or more diverticula, and there was no anus.

3. The bilaterally symmetrical types were derived from this larval form by the larva becoming oval, and the region in front of the mouth forming a præoral lobe, and that behind the mouth growing out to form the trunk. The aboral dome became the dorsal surface.

On the establishment of a bilateral symmetry the anterior part of the nervous ring gave rise (?) to the supraœsophageal ganglia, and the optic organs connected with them; while the posterior part of the nerve-ring formed (?) the ventral nerve-cords. The body cavity was developed from two of the primitive alimentary diverticula.

The usual view that radiate forms have become bilateral by the elongation of the aboral dome into the trunk is probably erroneous.

4. Pilidium is the larval form which most nearly reproduces the characters of the larval prototype in the course of its conversion into a bilateral form.

5. The Trochosphere is a completely differentiated bilateral form, in which an anus has become developed. The præoral ciliated ring of the Trochosphere is probably directly derived from the ciliated ring of Pilidium, which is itself the original ring of the prototype of all these larval forms.

6. Echinoderm larvæ, in the absence of a nerve-ganglion or special organs of sense on the præoral lobe, and in the presence of alimentary diverticula, which give rise to the body cavity, retain some characters of the prototype larva which have been lost in Pilidium. The ciliated ring of Echinoderm larvæ is probably derived directly from that of the prototype by the formation of an anus on the dorsal side of the ring. The anus was very probably originally situated at the aboral apex.

Adult Echinoderms have probably retained the radial symmetry of the forms from which they are descended, their nervous ring being directly derived from the circular nervous ring of their ancestors. They have not, as is usually supposed, secondarily acquired their radial symmetry. The bilateral symmetry of the larva is, on this view, secondary, like that of so many Cœlenterate larvæ.

7. The points of similarity between Tornaria and (1) the Trochosphere and (2) the Echinoderm larvæ are probably adaptive in the one case or the other; and, while there is no difficulty in believing that those to the Trochosphere are adaptive, the presence of a water-vascular vesicle with a dorsal pore renders probable a real affinity with Echinoderm larvæ.

8. It is not possible in the present state of our knowledge to decide how far the resemblances between Actinotrocha and Echinoderm larvæ are adaptive or primary.

Bibliography.

(257) Allen Thomson. British Association Address, 1877.
(258) A. Agassiz. “Embryology of the Ctenophoræ.” Mem. Amer. Acad. of Arts and Sciences, Vol. X. 1874.
(259) K. E. von Baer. Ueb. Entwicklungsgeschichte d. Thiere. Königsberg, 1828-1837.
(260) F. M. Balfour. “A Comparison of the Early Stages in the Development of Vertebrates.” Quart. Journ. of Micr. Sci., Vol. XV. 1875.
(261) C. Claus. Die Typenlehre u. E. Haeckel’s sg. Gastræa-theorie. Wien, 1874.
(262) C. Claus. Grundzüge d. Zoologie. Marburg und Leipzig, 1879.
(263) A. Dohrn. Der Ursprung d. Wirbelthiere u. d. Princip des Functionswechsels. Leipzig, 1875.
(264) C. Gegenbaur. Grundriss d. vergleichenden Anatomie. Leipzig, 1878. Vide also Translation. Elements of Comparative Anatomy. Macmillan & Co., 1878.
(265) A. Götte. Entwicklungsgeschichte d. Unke. Leipzig, 1874.
(266) E. Haeckel. Studien z. Gastræa-theorie, Jena, 1897; and also Jenaische Zeitschrift, Vols. VIII. and IX. 1874-5.
(267) E. Haeckel. Schöpfungsgeschichte. Leipzig. Vide also Translation, The History of Creation. King & Co., London, 1878.
(268) E. Haeckel. Anthropogenie. Leipzig. Vide also Translation, Anthropogeny. Kegan Paul & Co., London, 1878.
(269) B. Hatschek. “Studien üb. Entwicklungsgeschichte d. Anneliden.” Arbeit. a. d. zool. Instit. d. Univer. Wien. 1878.
(270) O. and R. Hertwig. “Die Actinien.” Jenaische Zeitschrift, Vols. XIII. and XIV. 1879.
(271) O. and R. Hertwig. Die Cœlomtheorie. Jena, 1881[148].
(272) O. Hertwig. Die Chætognathen. Jena, 1880.
(273) R. Hertwig. Ueb. d. Bau d. Ctenophoren. Jena, 1880.
(274) T. H. Huxley. The Anatomy of Invertebrated Animals. Churchill, 1877.
(274*) T. H. Huxley. “On the Classification of the Animal Kingdom.” Quart. J. of Micr. Science, Vol. XV. 1875.
(275) N. Kleinenberg. Hydra, eine anatomisch-entwicklungsgeschichtliche Untersuchung. Leipzig, 1872.
(276) A. Kölliker. Entwicklungsgeschichte d. Menschen u. d. höh. Thiere. Leipzig, 1879.
(277) A. Kowalevsky. “Embryologische Studien an Würmern u. Arthropoden.” Mém. Acad. Pétersbourg, Series VII. Vol. XVI. 1871.
(278) E. R. Lankester. “On the Germinal Layers of the Embryo as the Basis of the Genealogical Classification of Animals.” Ann. and Mag. of Nat. Hist. 1873.
(279) E. R. Lankester. “Notes on Embryology and Classification.” Quart. Journ. of Micr. Sci., Vol. XVII. 1877.
(280) E. Metschnikoff. “Zur Entwicklungsgeschichte d. Kalkschwämme.” Zeit. f. wiss. Zool., Vol. XXIV. 1874.
(281) E. Metschnikoff. “Spongiologische Studien.” Zeit. f. wiss. Zool., Vol. XXXII. 1879.
(282) A. S. P. Packard. Life Histories of Animals, including Man, or Outlines of Comparative Embryology. Holt and Co., New York, 1876.
(283) C. Rabl. “Ueb. d. Entwick. d. Malermuschel.” Jenaische Zeitsch., Vol. X. 1876.
(284) C. Rabl. “Ueb. d. Entwicklung. d. Tellerschnecke (Planorbis).” Morph. Jahrbuch, Vol. V. 1879.
(285) H. Rathke. Abhandlungen z. Bildung und Entwicklungsgesch. d. Menschen u. d. Thiere. Leipzig, 1833.
(286) H. Rathke. Ueber die Bildung u. Entwicklungs. d. Flusskrebses. Leipzig, 1829.
(287) R. Remak. Untersuch. üb. d. Entwick. d. Wirbelthiere. Berlin, 1855.
(288) Salensky. “Bemerkungen üb. Haeckels Gastræa-theorie.” Archiv. f. Naturgeschichte, 1874.
(289) E. Schäfer. “Some Teachings of Development.” Quart. Journ. of Micr. Science, Vol. XX. 1880.
(290) C. Semper. “Die Verwandtschaftbeziehungen d. gegliederten Thiere.” Arbeiten a. d. zool.-zoot. Instit. Würzburg, Vol. III. 1876-7.