The material for the investigations which are described in this article was collected at Princeton, N.J., except a few specimens of Triton alpinus kindly given me by Professor H. F. Osborn. Through a failure to obtain adult specimens at the time when I obtained the embryos, there remains room for doubt as to the exact species of the embryos. The Ambly-stoma embryos correspond exactly to Clarke’s1 description of the embryos of Amblystoma punctatum, but there is one difference in the appearance of the egg-membranes which leads me to think that this may be a different species from that described by Clarke. It is, perhaps, A. bicolor, for which Jordan2 gives only the habitat New Jersey. The Frog embryos are either Rana halecina or R. palustris. I judge them to be the former. In the stages of development with which my work has dealt there are probably no specific differences in the embryos.

The developing eggs of this species of Amblystoma seem to present a remarkable case of symbiosis. The eggs, surrounded by their gelatinous matrix, appear as a white mass floating on the surface of the water. (I found them in a small swampy pool on elevated ground.) In the first lot that I collected the medullary plates were just forming, and the two membranes surrounding each egg appeared perfectly homogeneous and transparent. In the second lot, collected some days later, the embryos were somewhat elongated, and the medullary canal had apparently just closed. In this lot the internal membrane of each egg was coloured a uniform light green by the presence in the membrane of a large number of minute globular green Algæ. Neither in the spaces adjoining the internal membrane, nor in the external membrane, nor in the matrix, was there any colouration or trace of this Alga. The external membrane was transparent and the matrix white and translucent as before. In a third lot, collected when the balancers and gills had appeared, these conditions were the same, except that the Algæ had increased in number and the colour was a much darker green. I have not discovered how the Algæ enter the membrane, nor what physiological effect they have on the respiration of the embryo, but it seems probable that in this latter respect they may have an important influence.

Clarke has given a detailed description of the external appearance and changes of the embryo of Amblystoma punctatum, so that for this part I may refer the reader to his work. As might have been expected, Amblystoma and Triton present much similarity in their development, while both differ in about the same degree from the Frog. The chief points of which I shall treat are the central nervous system, the hypophysis, and some other parts and appendages of the head. The comparison of the embryos of the different groups affords some light in the attempt to explain the development of some of the more complicated parts. In order to avoid repetitions I have not separated the descriptions of parts which are intimately related to each other in the process of their development. Much of what I have written will have been readily inferred by embryologists, though the embryology of the genus on which most of my work is based has not hitherto been worked out. But as some of my conclusions are different from those previously expressed, I have thought best to give in detail a description of the facts that the reader may thereby test my conclusions.

The first differentiation of the central nervous system of Amblystoma appears as figured in the sagittal and transverse sections (figs. 1, 2 A, 6 A). The transverse section is cut through the middle dorsal region. By the thickening of the dorsal epiblast there are formed two broad epiblastic plates (M.P.), connected with each other on the median line by a thinner portion of epiblast. A slight longitudinal groove (G.) is formed by the sinking inward of this thinner median portion of epiblast. Directly underneath this groove there is a longitudinal fold in the hypoblast, which causes a conspicuous median groove in the roof of the alimentary tract. The dorsal part of this hypoblastic fold touches the epiblast along the median line; and the part adjacent to the epiblast is the part which later forms the notochord. On each side of the hypoblastic fold, and apparently fused with it, lie the two layers of the mesoblast (So., Sp.). An examination of all my sections shows that the condition of the mesoblast at this point is the same in Amblystoma as Hertwig discovered it to be in Triton.1Some of my sections show a space between the two layers of mesoblast continuous with the archenteron. I have not found any trace, however, of mesoblast originating from any other part of the hypoblast or from the yolk.

The sagittal section (fig. 1) passes through the epiblastic groove (G.) and the dorsal groove in the hypoblast, thus cutting only the thin median part of the epiblast and the hypoblast. The sections on either side of this median section pass through the thicker part of the body wall, which contains also the mesoblast. In section, fig. 1, it may be seen that the thinner epiblast in the median line does not extend to the extreme anterior end of the rudiment of the nervous system, but that it ceases in the head region, while anteriorly the lateral medullary plates unite with each other undiminished in thickness, thus forming an anterior medullary plate (a. M. P.). The distinction between anterior and lateral plates is arbitrary and adopted only for convenience. They might be described as one thickened epiblastic plate, bent in such a manner that the curved part lay in the head region, while the two straight parallel ends lay one on each side of the dorsal median line. The distal periphery of the anterior medullary plate is a curve. Fig. 6 A represents a section through the anterior plate of the same embryo from which fig. 2 A was taken. It may be seen here that there is no sign of a bilateral division of the neural rudiment at its anterior end.

The further development of the medullary plates in the dorsal region is shown in figs. 3 B, 4 c, 5 D, Pl. XXVII. The lateral edges of the plates roll slightly upwards, forming the folds (M.F.). At the same time the median edges become pressed together, thus causing the floor of the median groove (G.) to sink farther inwards. The space between the medullary folds (M. F.) gradually decreases in size as the folds approach the median line. There is no very marked increase in the height of the folds. The originally dorsal surfaces of the medullary plates, bending inward, become pressed together in a vertical median plane under the groove, G. Across this line of median vertical contact there is no fusion of the cells. A heavy pigment marks this line (g., fig. 5 D) as continuous with the pigmented surface. Beneath the lower extremity of the line g. a small ridge of cells, continuous with the lateral halves of the neural rudiment, marks the original median connection of the medullary plates. By a comparison of the figures illustrating this period of development it may be seen that the cells of the neural rudiment gradually become smaller (owing to division and loss of yolk), that the whole organ becomes more compact and occupies much less space in the end than in the beginning. The primarily broad space enclosed laterally between the medullary folds (M. F.) diminishes in size until it becomes a small rounded groove in the dorsal part of the neural rudiment, as represented at fig. 5D. In a section through the cervical region of the same embryo (fig. 10 D) the epiblast has met above the groove, thus forming a relatively very small neural canal. As the epiblast of the two sides fuses above the canal the lumen of the latter becomes in some places suppressed, leaving as its only remnant a heavy accumulation of pigment.

In the species of Frog examined by me the lumen of the neural canal at this period of development becomes suppressed throughout the posterior part, thus differing in this respect from Goette’s account of Bombinator. Towards its posterior extremity the neural rudiment of the Frog closely resembles that of an osseous fish at the same period. This manner of development of the neural rudiment in Amphibians presents a stage intermediate to the condition of Elasmobranchs and Reptiles on the one hand and the condition of Petromyzon and the Teleosts on the other.

After the fusion of the epiblast dorsal to the neural rudiment the neural canal opens from before backwards along the pigmented line (g., fig. 5 D) which has previously been described. The canal, at first somewhat irregular, becomes in cross-section dorso-ventrally elongated. The walls of the neural tube become bilaterally symmetrical, and are thickest laterally. A transverse section through an older embryo of Amblystoma (fig. 11) shows the result of these changes. In the Frog embryo the appearance is fundamentally the same.

In the cephalic region the development of the neural rudiment differs from that in the dorsal region. In order to clearly understand this difference, it is necessary to bear in mind that the anterior medullary plate (a. M. P., figs. 6 A, and 7 B) is not a paired continuation of the dorsal medullary plates, but is a transverse curved plate connecting the two dorsal plates. It is also necessary to distinguish the modifications caused by the cranial flexure, in order to recognise the homology between the parts in the dorsal and cephalic regions. It will be seen in the sagittal section (fig. 1) and in the transverse section (fig. 6 A) that at first the anterior medullary plate (a. M. P.) is externally flattened. In fig. 7 B, where the dorsal medullary folds have appeared (compare fig. 3 B), the lateral edges of the anterior medullary plate turn slightly upward (where the same letters are affixed to the numbers the sections are from the same embryo). In fig. 8 c this upward bending of the lateral edges has increased, and in fig. 9 D the edges meet dorsally. During this process the median part of the anterior medullary plate (A. F.) departs from its original slanting position (fig. 1), and comes to lie nearly parallel to the dorsal surface of the embryo, though at a lower level. The floor of the dorsal medullary groove extends forwards nearly horizontal as far as the region of the midbrain; there it bends downwards almost at a right angle, and joins the posterior edge of the medially horizontal anterior plate. Thus is formed the primary cranial flexure before the medullary folds have fused above the neural canal. In the head this fusion takes place later from behind forwards, curving down to the anterior edge of the anterior medullary plate (A. F, fig. 9 D). The cranial flexure therefore is not simply a bend in the floor of the primitive neural tube, but is also a bend in the dorsal surface. It involves the anterior part of the neural tube in a bend about equal to a right angle. The line of fusion of the medullary folds in the head is homologous with the same fusion in the dorsal region. The morphologically dorsal surface of the neural tube extends therefore throughout the region of the fusion to the anterior edge of the anterior medullary plate. Taking into account the anterior bending of the axis of the neural tube, its morphologically anterior surface would be represented by the anterior medullary plate, which extends from the above-mentioned vertical portion of the floor to the anterior end of the dorsal fusion.

The anterior medullary plate of Amblystoma is homologous with the anterior medullary fold of the Lizard, and for the purpose of indicating this homology I have marked it in the drawings as the anterior medullary fold. In both cases it forms the primitive morphologically anterior surface of the brain. There is a marked difference between this anterior brain-surface in the Lizard and the same part in Amblystoma. In the Lizard the anterior brain-surface comes to lie at a right angle to the axis of the dorsal part of the neural tube, and faces posteriorly; in Amblystoma it lies parallel to the axis of the dorsal part of the neural tube, and faces ventrally. This difference seems to be due to the different methods according to which in the two forms the medullary folds unite to form the medullary tube. In Amblystoma the condition is caused in the following manner. In the primitive neural rudiment there is a thinner median portion of epiblast lying between the dorsal medullary plates and behind the anterior medullary plate. As the distal lateral edges of the neural rudiment approach each other different effects are produced in the region of the thin median epiblast and in the anterior plate. In the first-named region, as the lateral edges of the medullary plates approach each other, their median edges are compressed, and as the width of the neural rudiment decreases its median thickness increases. In the anterior plate there is no thin median portion and no thickening resulting from compression, therefore as the lateral edges approach each other the median portion must bend downward. In this manner the median portion of the anterior plate comes to lie at a much lower level than the floor of the neural tube in the dorsal region. The cranial flexure is the result of the presence of an anterior medullary plate, and, as I have elsewhere pointed out, this seems to be the case also in the Lizard.

In the Frog the anterior medullary plate forms a fold directly comparable to the medullary folds in the dorsal region. The anterior fold is, however, much more prominent than the folds in the dorsal region. Fig. 19 represents a median sagittal section of a Frog embryo at this stage. The lateral sections of this embryo show that the anterior fold (A. F.) is laterally and posteriorly continuous with the paired medullary folds, thus enclosing anteriorly and laterally the anterior enlargement of the neural groove (F. B.). This anterior enlargement is the first rudiment of the vesicle of the fore-brain. The cranial flexure in this embryo is in process of formation; when the medullary folds in the head later meet dorsally, the cranial flexure is complete. The presence of an elevated anterior fold in the Frog, and its absence in Amblystoma, is not so much due to absolute difference in the form of the neural rudiment as to the relative growth of the surrounding parts. In Amblystoma the presence of the hypoblast and anterior end of the alimentary tract beneath the anterior medullary plate (fig. 9 D) prevents the latter from appearing as a fold raised above the head surface. But at a later period the hypoblast disappears from beneath the anterior plate, and the external surface of the anterior plate is then covered only with epiblast (fig. 12 E).

The disappearance of the hypoblast and alimentary cavity from beneath the anterior medullary plate, or rather the (morphologically) anterior surface of the brain, is due to the more rapid growth of the brain, especially an increase of length, by which the fore-brain advances to a position in front of the anterior end of the alimentary cavity. At a very early stage the anterior end of the alimentary cavity is enclosed only by hypoblast and epiblast (Ep., Hyp., fig. 1). A fusion of these two layers soon takes place at this point, and indicates the eventual position of the mouth-opening. As the fore-brain is projected anterior to this mouth-fusion, the epiblast dorsal to the fusion is brought into close contact with the anterior surface of the brain (fig. 12 E). Figs. 12 E and 13 E represent two nearly sagittal sections of the same embryo, one section passing through the oral fusion and hypophysis-rudiment, the other passing through the notochord and pineal rudiment. The age and general condition of development of this embryo will be best understood by comparing these sagittal sections with sections 14 r, 15 F, 16 F, which are horizontal and taken from an embryo of the same age. The anterior part of the alimentary canal is distended into a large pharyngeal branchial cavity (fig. 12 E). The hypoblast of the anterior wall of this cavity touches the nearly vertical floor of the fore-brain which forms the wall of the infundibulum. The lower anterior wall of the pharyngeal cavity is fused with the epiblast at M., forming the oral fusion. A wedge-shaped mass of epiblast (Hph.) extends inward between the oral fusion and the wall of the infundibulum; this is the rudiment of the hypophysis. It is not necessary to interpret this condition as an ingrowth of the epiblast. I am inclined to think that the wedge-like shape of the epiblastic mass is due to the pressure of the more rapidly growing brain. It is evident from this section that at this stage of the development of Amblystoma there is no appearance of a stomodæum or epiblastic mouth-cavity. From this time on the rudiment of the lower jaw begins to extend forward, and grows beyond the oral fusion and hypophysis toward the nasal tip of the head. The epiblast retains its connection with the hypoblast, and also for a time with the hypophysis; thus the epiblast posterior to its point of fusion with the hypoblast is pressed close against the epiblast anterior to the point of fusion. These two united layers of epiblast form an apparently solid mass extending from the hypoblast to the surface of the head (M.). This stage is illustrated in the nearly median sagittal section, fig. 17 G. The hypophysis (Hph.) has broken loose from the in-folded mass of epiblast, and still remains adjacent to the posterior wall of the infundibulum (In.). The point for the ultimate external opening of the mouth (M.) has been moved by the growth of the lower jaw, forward to a position anterior to region of the optic chiasma (Ch.). The position of the perfected mouth-opening is shown in fig. 18. The condition of the rudiment of the mouth, as represented in fig. 17 G, is that which has been described by other writers as a solid ingrowth of epiblast or a stomodæum; but it is evident from the above-described manner of development that the term ingrowth leads to a false conception as to the origin of the part referred to. The primary development of the hypophysis, and the growth forward of the lower jaw, are fundamentally the same in Amblystoma as I found them in the Lizard.

During the process above described, the parts of the brain and the hypophysis and notochord change their positions with relation to each other. Fig. 13 E shows the anterior end of the notochord, which in this embryo is at some distance from the hypophysis, while the floor of the hind-brain (H. B.) is widely separated from the infundibulum. There is a median thickening of the hypoblast extending from the anterior end of the notochord down to the hypophysis. This thickening seems to disappear very quickly after formation. It seems possible that this median thickening may be homologous with that foremost part of the notochord which in the Lizard and in the Mole extends as far as the epiblast at the hypophysis. In the Anura at an early stage there is a layer of mesoblast extending across the median line between the anterior end of the notochord and the hypophysis-rudiment. Why the meso-blastic product of the hypoblast along the median line at this region does not become differentiated into notochord in the Amphibia, as it does in the Lizard and the Mole, may be explained perhaps by the changes which immediately succeed this stage—changes which would be hindered by a developed notochord in this region. The changes thus referred to are exhibited in fig. 17 G. Here the secondary cranial flexure has appeared in the hind-brain, and the floor of the hind-brain is pushed against the infundibulum, causing the latter to be slightly compressed. At the same time the bending floor of the hind-brain has pushed the notochord downward, so that the anterior end of the developed notochord touches the hypophysis. (These changes of position are of course to be understood only in terms of relative topography as the absolute changes of location cannot be ascertained. Thus, the changes might be accounted for by supposing the secondary cranial flexure to lift the anterior part of the brain and head upward; but the former view lends itself more readily to the explanation of the facts, and admits of more extended homologies).

As nearly as can be judged from the more limited number of my specimens of Triton, the method of development during the above-described stages is exactly the same in Triton as in Amblystoma; though I should add that my youngest stage of Triton corresponds with the stage of Amblystoma represented in figs. 12 E to 16 F. From this stage onward my series of the embryos of the two genera run about parallel, and a great similarity continues to exist throughout all the stages which I have examined.

The development of the hypophysis and mouth iu the Frog differs in a marked manner from the development of the same organs in Amblystoma and Triton. The same fundamental principles seem to obtain in both methods of development, but the difference is apparently due to a different proportional rate of growth of the parts adjacent to each other. The development of these parts in the Frog is illustrated in figs. 19—23. These sections are sagittal, or nearly sagittal, and all meet the median vertical plane in the centre of the mouth-fusion. In fig. 19, between, the lip of the anterior medullary fold (A. F.) and the mouth-fusion, lies the epiblast which is to form the hypophysis. In this embryo the cranial flexure is not yet complete, and the alimentary cavity extends forward beyond the anterior fold. The rudiment of the hypophysis lies therefore immediately exterior to the anterior fold. In a somewhat older embryo (fig. 20) the brain is enclosed, and has increased so much in size that it projects forward anterior to the mouth-fusion (M). The increase of the cranial flexure has caused a change in the position of the anterior fold. In fig. 19 the anterior fold occupies a vertical position, and in fig. 20 it occupies a horizontal position (A. F.), forming in both cases the morphologically anterior wall of the brain. In embryos slightly younger than the one represented by fig. 20 serial sections show that the dorsal linear opening of the central nervous system extends as far as the horizontal anterior fold to about the point indicated by o. g. in fig. 20. When this opening becomes closed by the dorsal median fusion of the lateral walls, the line of fusion remains marked by the accumulated mass of epidermoidal pigment. This pigmented line is cut at o.g. in fig. 20, very near the end which indicates the boundary of the anterior fold. It may be seen from the figures that the change of position of the anterior fold is accompanied by a corresponding change in the position of the hypophysis rudiment (Hph.), so that the latter continues in the same topographical relation to the anterior fold. The rudiment of the hypophysis extends a short distance posterior to the limit of the anterior fold (o. g.).

The next three stages (figs. 21, 22, 23) illustrate the further development of these parts. The most striking changes are the increase of the cranial flexure and the growth of the dorsal part of the fore-brain. (This latter is not so well shown in fig. 21 owing to the obliquity of that section.) It is evident that these changes would cause a relative change of position of the point marked o.g. in fig. 20. In two of the sections (figs. 22, 23) may be seen a slight groove (o.g.) in the morphologically anterior surface of the brain. This groove lies at first between the optic stalks, and ultimately just anterior (or morphologically dorsal) to the chiasma. I have not been able to absolutely demonstrate that the groove (o. g.) is developed from the point o.g. in fig. 20, but the evidence in favour of the view that such is the case seems to me so strong that I have been forced for the present to accept that conclusion. In the Lizard the primitive opening of the brain extends down the anterior surface of the brain to a point between the optic stalks, and in the Lizard there is also a similar groove at that point. In the present case we have only to imagine that owing to the increase of the cranial flexure and the growth of the fore-brain the point o.g., fig. 20, has receded relatively in a posterior direction, until it reached the point o.g., fig. 23. In figs. 22 and 23 such a relative posterior recession of the groove o. g. is perfectly evident. This relative recession is due chiefly to the greater growth in the region in front of the point o. g. It will be seen that in all these five embryos (19—23) the posterior end of the hypophysis-rudiment lies at about the same distance behind the region of the point o.g., but the lower jaw advances continually until it extends anteriorly beyond the posterior end of the hypophysis and beyond the point o. g. This process of growth is essentially the same in the Frog as in Amblystoma and Triton; but in the Frog the growth of the dorsal part of the fore-brain and the growth forward of the lower jaw take place at the same time, and in nearly the same extent, thus making the hypophysis appear as an ingrowth, whereas it is simply that part of the epiblast which has retained its original position with relation to the brain, and which has become surrounded and embedded by the expansion of the adjacent parts. There is another point of difference between the hypophysis of the Frog and the hypophysis of the Urodele embryos. In Amblystoma and Triton the hypophysis at the very beginning of its differentiation lies immediately adjacent to the posterior wall of the infundibulum, and later the anterior end of the notochord touches its posterior side. In the Frog the hypophysis at first does not reach the posterior wall of the infundibulum. As it begins to loose its connection with the epiblast it gradually comes to lie nearer the posterior wall of the infundibulum, and finally lies slightly ventral to the anterior end of the notochord, the latter being pressed against the infundibulum. Thus a nearly similar condition results from two apparently different methods of development. In Amblystoma the position of the hypophysis is the result (mechanically) chiefly of a forward movement of the anterior part of the brain. It seems most probable that the case is the same in the Frog, but that the forward movement of the anterior part of the brain takes place at a later date.

To Goette’s1 description of the other parts of the brain of Anura during these stages I have nothing to add. There are a few points, however, which may be mentioned for the sake of orientation as to the stages of development of the embryos here Preferred to. In a transverse section through the head of an embryo at the stage of fig. 20 the lumen of the fore-brain appears triangular, with one angle representing the dorsal crest of the brain, and the side opposite that angle representing the morphologically anterior wall of the brain. The lateral angles of the lumen are the beginnings of the optic outgrowths. In an embryo at the stage of fig. 21 the optic outgrowths are somewhat prolonged, and the lumen is drawn out laterally in them. In the embryo of fig. 22 the optic outgrowths are bent backwards and upwards, and in the embryo of fig. 23 the eye has progressed so far that the lens has appeared.

In the Amblystoma embryo of series D, in which the forebrain is not yet enclosed, there is no trace of the optic vesicles. The next older stage of Amblystoma among my specimens is illustrated in figs. 12 E—16 F. The condition of the cranial flexure is shown at 12 E. In the anterior wall of the brain may be seen the optic groove (o. g.), and behind the latter is the anterior fold (A. F.). Immediately posterior to the anterior fold is the rudiment of the infundibulum. At this stage the primary triple division of the brain is not yet very pronounced, and there is no trace of nerve-fibres in the brain. The position of the rudiment of the epiphysis (Eph., fig. 13 E) indicates the posterior extent of the primary fore-brain. In fig. 16 F the fore-brain is represented in section parallel to its morphologically anterior surface very near the latter, and in the region of the optic stalks (Ey.). In this section the lateral thickenings of the brain wall in front of the optic stalks are the rudiments of the corpora striata, which appear much earlier in Amblystoma than in the Lizard. Fig. 15 F represents a horizontal section of the embryo passing through the dorsal part of the pharyngeal cavity and through the mid-brain above the region of the infundibulum. This section shows the rudiments of the eyes (Ey.), which as yet possess no lens. Fig. 14 F represents a horizontal section through the hind-brain and dorsal medulla. This section shows the rudiments of the fifth, seventh and eighth, ninth and tenth cranial nerves. In three places the hind-brain shows a marked dilation of its lumen, and the lateral walls of the brain pass around these dilated parts undiminished in thickness. Opposite these dilated parts of the lumen arise the three chief nerve-roots of the hind-brain. The most anterior dilation corresponds to the fifth nerve-root (n. V). The next dilation corresponds to the common root of the seventh and eighth nerves (n. VIII and VII), and the posterior dilation corresponds to the root of the tenth nerve (n. X). The rudiment of the ear (E.) lies between the regions of the posterior and middle dilations, and immediately behind the ear arises the root of the ninth nerve (n. IX). These dilated parts of the hind-brain in Amblystoma resemble in some degree what I have described as the neuromeres in the hind-brain of the Lizard, except that in Amblystoma they are fewer in number, and certain intermediate neuromeres appear to have been suppressed. I am inclined to think that the large quantity of yolk present in these parts in Amblystoma has considerably changed their appearance and development. These dilations of the hind-brain have disappeared in Amblystoma, as in the Lizard, by the time the nerve-fibres of the brain have appeared. It will be seen in fig. 14 r that the cranial nerves meet and fuse with the epiblast. This fusion I think corresponds with what has been described by Miss Johnson and Miss Sheldon5 as the first or dorsal fusion of the cranial nerves with the epiblast. These authors have described this fusion for the fifth, seventh, and ninth nerves, and supposed it for the vagus. My section shows the correctness of their supposition. The vagus retains for some time this fusion with the epiblast, and from the point of fusion there soon grows posteriorly a large linear thickening of the epiblast, which forms the lateral nerve. This in its earlier stage is very conspicuous, but soon becomes much smaller. I have not been able to trace the different steps between what the above-named authors have called the “first (dorsal) fusion” and the “second (ventral) fusion.” One of my series of sections of Triton alpinus shows the condition described by them as the “second (ventral) fusion.” In this series the distal ends of the two primary branches of the fifth nerve touch the epiblast and appear to be fused with the same.

The further development of the brain is shown in figs. 17 G and 18. The irregular appearance of these sections is due to the fact that they are neither exactly median nor exactly vertical; they cross the median vertical plane in a line drawn through the epiphysis (Eph.) and the region of the optic chiasma (Ch.) and hypophysis (Hph.). The morphologically anterior surface of the brain has remained in about the same position that it occupies in fig. 12 E, but the floor of the hind-brain is bent downward and is pressed against the infundibulum. Just anterior to the epiphysis (Eph.) is a deep fold, extending transversely across the dorsal wall of the brain, and thus dividing off the secondary fore-brain. There is another longitudinal and median fold, extending from this transverse fold forward to the anterior surface of the brain; thus dividing the secondary fore-brain into the two hemispheres. This longitudinal fold is not so deep as the transverse fold. Fig. 35 represents a section transverse to the long axis of an embryo of the same stage as fig. 17 G. This section is behind the deepest extent of the median longitudinal fold, but still shows the transverse fold. The rudiments of the corpora striata, which are already evident at the stage of fig. 13 E, st., are shown again in transverse section in fig. 35. The corpora striata extend parallel to each other on each side of the median line, along the morphologically anterior surface of the brain, and are limited ventrally by the optic groove (o. g., fig. 17 G). Immediately ventral to the optic groove is seen the remnant to the anterior fold, containing a bundle of transverse nerve-fibres, of which a part form the optic chiasma (Ch.). In an exactly median vertical section of the brain of an embryo at the stage of fig. 17 G, this remnant of the anterior fold would be the thickest portion of the brain wall, being about as thick as the lateral walls of the medulla. The thickness of the floor of the hind-brain in the median line is shown in fig. 24 G, W. H. B.

Before the embryo of Amblystoma has reached the stage of development represented by fig. 17 G, the first development of nerve-fibres has taken place in the central nervous system. The arrangement of these nerve-fibres corresponds very closely to the first arrangement of the nerve-fibres in the Lizard, and the arrangement seems to be identically the same in Triton and Rana. The nerve-fibres in the neural tube of the dorsal region first appear as two flat bands of longitudinal fibres, lying next the lateral surfaces of the tube. Fig. 34 shows a section of the neural tube of Amblystoma in the anterior dorsal region. The band of longitudinal fibres (L. F.) extends nearer to the ventral median than to the dorsal median surface of the tube. Goette has described these fibres as originating in the external halves of the peripheral cells throughout this portion of the tube; while the internal half of each cell, with the nucleus, becomes one of the cells of the grey matter. These points I have not been able to follow out with the material at my command. Shortly after the longitudinal fibres have appeared another system of fibres arises—the transverse fibres or ventral commissure (T. F). These fibres appear as polar outgrowths of the cells which lie internal to the longitudinal band. They pass ventrally along the inner surface of the longitudinal band, and cross transversely the ventral surface of the neural tube immediately inside the cuticula. Both of these systems of nerve-fibres develope later in the posterior than in the anterior part of the central nervous system. The transverse fibres extend as a continuous ventral commissure as far forward as the point where the floor of the mid-brain bends ventralwards into the posterior wall of the infundibulum. This is shown in median vertical section in fig. 24 G. The lateral bands of longitudinal fibres extend forward through the hind- and mid-brain, showing the same relations as in the dorsal region (fig. 34). On passing from the mid-brain to the fore-brain the lateral bands follow the curve of the cranial flexure; and on reaching the morphologically anterior surface of the brain, they cross it, blending with each other immediately ventral to the optic stalks. The lateral bands thus blend into an anterior band, which is cut transversely into the median vertical sections, 17 G and 18, at Ch. This anterior band comprises a bundle of fibres, which I would roughly estimate to be about twenty times as large as the bundle of fibres which appears shortly afterwards on each optic stalk. The course of the lateral band (L. F.) in the mid- and fore-brain is shown in the lateral vertical section fig. 32 G; the dotted line indicates the lower median contour of the brain. Fig. 33 shows the anterior band (A. F.) of the Frog just behind the optic stalks. This section is cut transverse to the long axis of the embryo. No fibres appear in the region of the infundibulum which lies between the anterior band and the anterior edge of the above-described continuous ventral commissure. Of the brain commissures (not including the anterior band) the posterior commissure is the first to appear. It developes about the time that the ventral commissural system appears. The posterior commissure is shown at P. C., figs. 18, 32 G, and 35. It crosses the dorsal surface of the brain immediately posterior to the epiphysis. Its fibres seem to be not continuous with the fibres of the lateral bands, but, as far as they can be traced, they cross the course of the lateral bands; losing themselves, however, in the region of the latter. The anterior commissure developes relatively much earlier in Amblystoma than in the Lizard. It first arises as two lateral symmetrical bundles of fibres, passing along the exterior surfaces of the corpora striata and intersecting the lateral bands just posterior to the optic stalks (A, C., fig. 32 G). This section shows that these fibres are not continuous with the fibres of the lateral bands. A part of these bundles of fibres crosses the anterior surface of the brain a short distance dorsal to the optic groove at the point A. C. in figs. 18 and 30 H. The rest of these fibres continue on toward the roots of the olfactory nerves, n. I, fig. 29 H. A short time after the anterior band has appeared, there appears on the morphologically anterior surface of each optic stalk a small growth of nerve-fibres, developing as far as can be seen, in exactly the same manner as the development of the fibres of the lateral longitudinal bands. These optic fibres appear at the point n. II, in fig. 32 G (Amblystoma), and are shown in fig. 33 (Frog), where they are cut nearly longitudinally. The latter section shows that no fibres appear in the posterior wall of the optic stalk (op.). Medianly, the optic fibres meet and blend with the anterior band; distally, they pass unbroken into the inner surface of the eye-cup (fig. 33). I have not followed the later growth of the optic nerve in the Amphibia, but I judge from the close similarity between this stage and a stage in the Lizard, that the development of the optic nerve in the Amphibia is throughout about the same as I have described it for the Lizard.1

Figs. 27 H—30 H show four horizontal sections through the head of an embryo of Amblystoma at an age corresponding to that of fig. 18. These sections show the nerve-fibres of the brain at a more advanced period than that above described. Of these sections, 27 H is cut nearest the dorsal surface of the head, and on the left side passes above the lateral band of longitudinal fibres (L. F.) in the region of the secondary cranial flexure just in front of the ear. On the same side of the section the lateral band in the hind-brain is seen to be continuous with the lateral band in the mid-brain (L′. F.). In front of the mid-brain is seen the posterior part of the cerebral hemispheres. The next more ventral section (28 H) passes through the thalamencephalon and through the fold which separates the infundibulum (In.) from the hind-brain. In the hind-brain may be seen the transverse fibres of the ventral commissure (T. F.). These are also visible (T. F.) in section 29 H, the hind-brain in this section being cut tangentially to its ventral convexity. In this same section may be seen on the right hand side the connection between the lateral band and those fibres which run dorsally along the corpora striata. One part of these fibres forms the anterior commissure as above mentioned (fig. 30 H, A. C.); while the other part continues onward to the region of the olfactory nerve (n. I), and here blends with a superficial layer of nerve-fibres, which covers the lateral dorsal part of each hemisphere, and extends so far upwards and backwards as to appear in section 27 H. Fig. 30 H shows the brain in section very near its anterior surface. At A. F. may be seen the fibres of the anterior band, with the fibres of the optic nerve (n. II) blending with its dorsal edge. At A. C. may be seen the fibres of the anterior commissure. Between the thickening of the anterior band and the anterior commissure appears the optic groove (o.g.). Orientation as to the direction of this section through the brain may be easily acquired by comparing it with fig. 18. The section 30 n would be perfectly horizontal in the fig. 18. Thus it enters the brain at the hinder edge of the anterior band and passes forward at an acute angle to the morphologically anterior surface of the brain. In this way the fibres passing from the region of the lateral bands to the anterior commissure are cut obliquely. The relations of these fibres to the lateral bands are shown in fig. 32 G. Here it appears that they do not bend and run with the lateral bands, but may be traced for some distance, crossing the latter at right angles. The anterior commissure is at first undivided and lies next to the surface of the brain, but in the latest stage which I have examined an internal part has become divided off from the superficial part (fig. 18). This internal part I judge to be the corpus callosum.

The growth of the hind-brain, together with its change of form, has in this oldest stage brought the cranial nerves of this region much nearer together. These conditions are illustrated in figs. 27 H and 28 H. The nerve-roots which are present form very large ganglia. The common ganglion of the seventh and eighth nerves (n. VIH, VII) lies relatively much nearer the root of the fifth nerve (n. V) than it did at the time of its first appearance. The roots of the ninth and tenth nerves appear to have fused in a common ganglion (n. X, IX). This may be due to the great growth of the auditory vesicle pushing the root of the ninth nerve backward. I have been unable to find in these stages any traces of the third, fourth, and sixth nerves. In the Lizard the third nerve developes as soon as the other ventral roots of the nervous system; the sixth nerve developes somewhat later than the other cranial nerves, except the fourth, which first appears at a stage much later than the present stage of Amblystoma. The olfactory nerve (n. I) is shown in fig. 29 H, entering the olfactory sac (N. a.). The course of this nerve from its origin in the fore-brain is backwards and downwards. The fibres of the optic nerve are also shown in fig. 30 H, entering the brain at n. II, where they join the forward or dorsal edge of the anterior band of fibres (A. F.).

The gill-clefts develope in Amblystoma after the usual manner from before backward. The first or hyoid cleft (I) does not break through, but forms like the others a laterally extended hypoblastic pouch (figs. 15 F, and 26). In the case of the hyoid this pouch extends in a ventral and median direction, forming a groove which meets a similar groove from the opposite side. The median portion of this groove is shown in the longitudinal vertical section of fig. 18, Th. From comparisons with the work of other writers I suppose this part marked Th. to be the rudiment of the thyroid gland, though in this case I have traced the development no farther. Whether this relation of the thyroid rudiment to the hyoid clefts can be considered as an argument for the phylogenetic origin of the thyroid gland from the ventral coalition of the hyoid clefts, is, I think, doubtful. The ventral groove may be the result of the early development of the tongue-rudiment. In the Lizard the hyoid clefts are widely open to the outside, and the thyroid rudiment appears between the transverse areas of the hyoid and first branchial clefts. The thyroid rudiment in the Lizard has no apparent connection with the hyoid clefts.

In the stage represented in fig. 15 r, the hyoid (I) and the first two branchial cleft-rudiments (II, III) have appeared; in the stage of fig. 26 five in all have appeared (I—V), but none of them have as yet broken through. These stages show the development of the head-cavities or mesoblastic somites of the head. The anterior somite is the first to develope, and appears just behind the eye. The other somites are separated off from this first one by the successive development of the hyoid and branchial clefts. These somites of the head do not attain a characteristic development as cavities as is the case with Elasmobranchs and the Lizard. Nevertheless there is here a tendency in that direction, and sometimes a slight cavity appears as in fig. 26, H.C. Where this happens it is generally in the most anterior somites.

Previous to the breaking open of the gill-clefts there appears on each side of the mandibular arch a small thickening and protrusion of the epiblast. These protuberances appear long before any of the external gills of the other arches. They become later rod-like structures, and are then easily recognised as the organs which Clarke has called “balancers.” An examination of their structure and relations shows them to be homologous with the external gills. They are supplied with blood by the most anterior or mandibular fork of the ventral aorta, and a branch of the fifth nerve may be traced down to the neighbourhood of the base of each balancer. Fig. 25 shows a vertical section passing longitudinal to the axis of the embryo and nearly longitudinal to the balancer (bl.). The balancer consists of a cylinder of compact epiblastic tissue, growing slightly thicker toward the distal end, where it forms a thick epiblastic cap similar to the epiblastic cap generally observed on the limb-rudiments of vertebrate embryos. Internally the balancer is nearly hollow, but is generally divided longitudinally through the greater part of its proximal length by a thin membranous network of which the function is probably to separate the courses of the arterial and venous blood. The large amount of blood which passes through the balancer indicates that it subserves in part a respiratory function. Balfour, following the account of Groette, has stated that the mandibular artery is never developed in Amphibians. In Amblystoma I find the mandibular artery developed, though in a less degree than the posterior branchial arteries. Owing to the early disappearance of the balancers this artery probably atrophies at an early date. Though my sections of Triton were not so favorable to the observation of this point, yet I find traces there of the existence of a mandibular artery.

At a stage when the branchial clefts have broken through and the cartilaginous skeleton has appeared (fig. 30 H), the relation of the balancer to the mandible becomes even more pronounced. The quadrate cartilage sends out laterally a crescent-shaped process immediately above the articulation with the Meckelian cartilage. This process extends to the base of the balancer separating the two blood-vessels which pass to and from the balancer. This process appears crescent-shaped only in vertical longitudinal section, and the posterior blood-vessel lies partly enclosed in the crescent. The process is shown at p. in the horizontal section, fig. 30H. Here also may be seen a bundle of muscle-cells extending from the pterygoid muscle (m.) into the base of the balancer (bl.). Another band of apparently undifferentiated muscular elements passes from the end of the above-mentioned process down into the balancer. Section 30H is cut through the base of the balancer, the free end of which extends below the plane of the section.

The balancers of Triton are of the same character as those of Amblystoma, but in Triton they appear to be not quite so highly developed.

Clarke observed the use of the balancers in the living embryos, and came to the conclusion that the chief function of the organs was as a means of support for the embryos to prevent them from sinking into the slime on the bottom of the pools in which they live. My own observations on the living embryos have led me to the same conclusion. It seems therefore that we have in this case not only the peculiarity of a homologue of the external gills arising from the mandibular arch, but also a homologue of the external gills becoming metamorphosed into an organ for the support of the body. It is also noteworthy that the balancers drop off after the limbs have appeared.

If we seek among the Anura for organs homologous with these balancers of the Urodela, the only organs which we can fix upon with any degree of probability are the suckers of the tadpole. Balfour has stated that these suckers arise on the hyoid arch, but in the embryos of Anura which I have examined they appear immediately posterior to the mouth-fusion (fig. 20) long before any trace of a division into visceral arches has appeared. I think for this reason that the suckers cannot properly be assigned to the hyoid arch. Balfour, in describing the embryology of the Urodela (‘Comp.Embry.’), says, “Stalked suckers of the same nature as the suckers of Anura are formed on the ventral surface behind the mouth.” The balancers in the two forms of Urodela which I have examined possess none of the characteristics of suctorial organs. Yet the balancers of the Urodela and the suckers of the Anura serve ultimately the same purpose, namely, to prevent the embryos from sinking into the soft organic mud usually found in the bottom of the pools which they inhabit, immersion in which would undoubtedly prove fatal to large numbers. An examination of a larger number of species may bring to light intermediate forms of these organs which would prove a more direct homology between the balancers and suckers.

The condition of the branchial apparatus and the skeleton of the head at a time when the branchial clefts have opened and shortly after the cartilage has appeared, is shown in figs. 27H—31, and also fig. 18. Fig. 31 is an approximate reconstruction from drawings of sections in three planes at right angles to each other; it represents the skeleton of the left side of the head. The four branchial clefts (II—V) are situated between the hyoid arch (I′) and the posterior branchial arch (V′ fig. 30 H). Each of the posterior four cartilaginous arches (II′—V′, fig. 38 H) supports an external gill. The cartilaginous hyoid arch has no external gill, but supports an opercular fold (o.f., fig. 30 H) which extends transversely across the ventral side of the head (fig. 18, o.f.) and a short distance up the lateral sides, partly overlapping the external gills. The cartilaginous bars of the visceral skeleton are of unequal length. Only the hyoid and the first two branchial bars extend to the median line, where they unite in a basi-hyo-branchial plate of cartilage (B. Hy., figs. 31 and 18). From this basi-hyobranchial plate there extends in a ventral and posterior direction a long curved process of which the flattened end touches the pericardium. The posterior two branchial bars (IV′, V′) each unite with the next preceding bar as shown in fig. 31. The hyoid bar does not extend dorsalwards more than half as far as the first branchial bar, so that in the sections of series H it first appears in the section 30 H. Each of these five bars is supported dorsally by a small muscle; the muscles are shown at m′. m′. in fig. 27 H. Dorsally the four branchial bars are united by a continuous piece of cartilage, c. b., fig. 27 H. None of these bars are articulated into different pieces, but the hyoid and first two branchial bars show ventrally a rudimentary beginning of an articulation.

The cranial skeleton of Amblystoma at this stage shows certain peculiarities, the homologies of which I am unable to determine in other forms. A general idea of the shape of the skeleton may be derived from fig, 31 and series H. Each lateral half of the cranial skeleton, together with the corresponding quadrate, appears as one continuous piece of cartilage. Of this piece the parts corresponding to the trabecular (tbr.) and parachordal (prc.) cartilages are easily distinguished; the former lying along the anterior surface of the brain, and the latter lying along the floor of the hind-brain adjacent to the notochord. The trabeculæ do not meet anteriorly. From the anterior end of the parachordal region there extends in a dorso-lateral direction a small bar of cartilage (x, figs. 31 and 28 H). This is met by another bar of cartilage (y), which extends upward and backward from the trabecula at the region of the optic nerve. These two bars (x and y) form thus a triangle, of which the base is the posterior part of the trabecula. At the junction of the anterior bar (y) with the trabecula there is a foramen through which passes the optic nerve (n. II, fig. 31). The greater part of the bar marked y. appears to pass dorsal to the optic nerve (fig. 31). This bar is a relatively thin piece, and separates the eyeball from the thalamencephalon. The posterior bar (x) of the triangle lies in the lateral groove between the floor of the hind-brain and the infundibulum, that is, lateral to the fold caused by the primary cranial flexure. At the dorsal apex of this cartilaginous triangle the cartilage is continuous with the dorsal proximal part of the quadrate cartilages (Q.). This is seen at Q. x, y, in figs. 27 H and 28 H. The dorsal part of the quad-rate cartilage is rather thin, and lies transversely with its lateral edge curved posteriorly toward the otic cartilage; farther ventralwards its section is shown in figs. 29 H and 30 H. The Meckelian cartilage shows no unusual peculiarities. Anteriorly it is connected with the cartilage of the opposite side by a short band of undifferentiated connective tissue.

What the significance of this manner of development of the chondrocranium may be, or how much importance should be attached to it, I am unable to say, as I have observed it only at this one stage.

At this stage well-characterised rudiments of teeth have appeared (d., figs. 30 H and 18). They are present in a semicircle above the Meckelian cartilages. They do not appear in a single row, but in several irregular rows. In the same manner they appear just ventral to the trabeculæ cranii along those parts of the trabeculæ which lie anterior to the optic nerve. In a cross section of the several irregular rows of teeth the teeth seem to radiate from the bar of cartilage on which they rest. In this respect the trabeculæ cranii and the Meckelian cartilages present the same appearance.

The central nervous system of Amphibians first appears as a transverse epiblastic thickening dorsal to the mouth-fusion, and continuous with paired elongated epiblastic thickenings lying dorsally on each side of the median line.

The primary cranial flexure is due to the presence of the transverse epiblastic thickening (anterior medullary plate).

The transverse epiblastic thickening forms, when the brain is enclosed, that part of the brain wall which lies between the infundibulum and the optic groove (i. e. the depression just dorsal to the chiasma of the optic nerves).

The first nerve-fibres which develope in the brain appear on what was originally the internal surface of the primitive epiblastic thickenings running longitudinally in the dorsal region and uniting continuously in the region of the primitive transverse thickening.

A subsequent development of nerve-fibres gives rise to a continuous ventral commissure extending through the floor of the mid-brain and hind-brain and spinal cord; and to the anterior and posterior commissures of the brain.

The fibres of the optic nerves are intimately connected with and are developed in the same manner as the main bundle of fibres in the region of the primitive transverse epiblastic thickening.

The hypophysis of Amblystoma presents a form of development intermediate to that of the Lizard and that of the Frog.

The balancers of Amblystoma may be considered as external gills of the mandibular arch which have become metamorphosed into embryonic organs of support.

P.S.—In his work entitled ‘Untersuchungen über die ver-gleichende Anatomie des Gehirns,’ Dr. Ludwig Edinger has described a Commissur der basalen Vorderhirnbündel, which he says appears in all classes of Vertebrates. The position of this Commissur in the adult brain immediately behind the optic chiasma is identical with that of the anterior band of nerve-fibres (A. F.), which I have described in the embryonic condition. The relatively large size and pronounced character of the anterior band in both Reptilian and Amphibian embryos lead me to think that it was once of primary importance, and that the Commissur in the adult brain is probably a rudiment of the same with changed relations and functions.

Illustrating Mr. Henry Orr’s paper “Note on the Development of Amphibians, chiefly concerning the Central Nervous System; with Additional Observations on the Hypophysis, Mouth, and the Appendages and Skeleton of the Head.”

Where a number of figures represent sections of the same individual embryo, all those figures have the same letter affixed to their numbers.

All figures of sections have been drawn with the Abbey camera lucida and a Zeiss’s microscope, so that in figures magnified to the same degree the size of the parts may be directly compared. (Z. 2, A, means Zeiss’s ocular 2, and objective A, &c.)

Index Letters.

A. C. Anterior commissure of fore-brain. A. F. Anterior medullary fold. A. F’. Anterior band of nerve-fibres, continuous with the lateral bands, L. F. a. M. P. Anterior medullary plate. B. By. Basi-hyobranchial plate of cartilage. Bl. Region of the blastopore. II. Balancers, c. b. Cartilaginous bar connecting dorsally the cartilaginous gill-arches. Ch. Optic chiasma united with the anterior band of nerve-fibres (A. F′.). d. Dental rudiments. E. Ear. Ep. Epiblast. Eph. Epiphysis cerebri. Ey. Eye. F. B. Forebrain. G—g. See explanation of Figs. 2 A—5 D. H. B. Hind-brain. H. C. Head cavity. Bph. Hypophysis. Bt. Heart. Byp. Hypoblast. In. Infundibulum. L. Lens of eye. L. F. Primary longitudinal fibres of central nervous system; L. F′. the same in the region of the thalamcncephalon. M. Mouth-fusion or mouth. m. and nJ. Pterygoid and branchial muscles. M. B. Mid-brain. Md. Medulla spinalis. hies. Mesoblast, hl. F. Lateral medullary fold. mk. Meckelian cartilage, M. P. Lateral medullary plate. N. Notochord or rudiment of the same. Na. Nasal sac. n. I, n. II—n. X. Olfactory, optic, and succeeding cranial nerves, o.f. Opercular fold of the hyoid arch. o.g. Optic groove, o.p″. Posterior wall of the optic stalk. p. Lateral cartilaginous process of the quadrate at the base of the balancer. P. C. Posterior commissure of the brain, p. g. Rudiment of the pectoral girdle. PB. Pharyngeal cavity. So. Somatopleure of mesoblast. Sp. Splanchnopleure of mesoblast. St. Corpora striata. tbr. Trabeculæ cranii. T. F. Transverse nerve-fibres forming a continuous ventral commissure. Th. Rudiment of the thyroid gland, w. H. B. Floor of hind-brain, x—y. See explanation in text, p. 319. X. Yolk. I—V. Hyoid and branchial clefts. I′—F′. Cartilaginous gill-arches.

PLATE XXVII.

FIG. 1.—Median-longitudinal and nearly vertical section of the egg of Amblystoma, at a time when the medullary plates have first appeared. It shows the anterior epiblastic thickening (a. M. P.), which unites anteriorly the two dorsal medullary plates. Also the thinner median portion (g. g.) between the dorsal medullary plates. Bl. Region of the blastopore. N. Undifferentiated hypoblastic tissue of the notochord. (Z. 2, A A.)

FIGS. 2 A, 3 B, 4 C, and 5 D.—Transverse sections through the anterior dorsal region of embryos of Amblystoma, showing successive stages of development. The first of the series (2 A) is at a stage corresponding with that of Fig. 1. G. The thinner median portion of epiblast between the dorsal medullary plates which becomes pushed downwards, so that the surfaces immediately lateral to it become pressed together along the line g. (Z. 2, A A.)

FIGS. 6 A, 7 B, 8 c, 9 D.—Transverse sections through the anterior region of the head of the same embryos respectively as Figs. 2 A—5 D. These sections show the anterior medullary plate (a. M. P. or A. F) which connects the lateral dorsal medullary plates. A. L. Anterior end of the alimentary cavity. (Z. 2, A A.)

FIG. 10 D.—Transverse section through the posterior region of the head to show the reduction of the lumen of the neural canal. (Z. 2, A A.)

FIG. 11.—Transverse section through the cervical region of an embryo of Amblystoma, somewhat more advanced than that of series D, showing the change of shape in the neural tube and canal. (Z. 2, A A.)

PLATE XXVIII.

FIGS. 12 E and 13 E.—Longitudinal and nearly vertical sections of an embryo of Amblystoma (at a stage represented by Clarke’s Fig. 14). Fig. 12E passes through the vertical plane at the hypophysis (Hph.) and the dorsal notochord; Fig. 13 E at the anterior end of the notochord and the epiphysis (eph.). (Z. 2, A A.)

FIGS. 14 F, 15 F, and 16 F.—Horizontal sections of an embryo of Amblystoma at the same stage as the preceding two figures. Fig. 14 F passes through the hind-brain and part of the dorsal medulla. Fig. 15 F passes through the mid-brain and the dorsal part of the alimentary cavity. Fig. 16 E passes through the hypophysis and the anterior part of the fore-brain. (Z. 2, A A.)

FIGS. 17 G and 18.—Longitudinal median vertical sections of two embryos of Amblystoma. 17 G is older than the stage of series E and F, and 18 is older than 17 G. These sections, together with 12 E, show the development of the lower jaw, the formation of the mouth, and the hypophysis. (Z. 2, A A.)

FIGS. 19—23 inc.—Longitudinal median vertical sections of successive stages of Frog-embryos, showing the formation of the fore-brain, the hypo-physis, and the mouth-fusion, with their relative changes of position. (Z. 2, A A.)

FIG. 24 G.—Part of a longitudinal median vertical section of an embryo of Amblystoma, showing the floor of the hind-brain and the fold between the hind-brain and infundibulum, which is caused by the primary cranial flexure. (Z. 2, A.)

FIG. 25.—Taken from a series of longitudinal vertical sections of an embryo of Amblystoma; it shows the balancer in nearly longitudinal section, and also a superficial portion of the mandibular arch. (Z. 2, A.)

PLATE XXIX.

FIG. 26.—Horizontal section of an embryo of Amblystoma at a stage between the stages E—F and the stage G. This figure shows the first formation of the optic lens, also the formation of the gill-clefts and the somites of the bead. (Z. 2, A A.)

FIGS. 27 n—30 H incl.—Horizontal sections of an embryo of Amblystoma at the same stage as the embryo of Fig. 18. Of these sections, 27 H is the most dorsal and 30 H the most ventral, the others being intermediate in the order of their numbers. The sections show the development of the nerve-fibres in the brain, and the early development of the cartilaginous skeleton and the branchial apparatus. (Z. 2, A A.)

FIG. 31.—Approximate reconstruction of an early stage of the cranial and visceral skeleton of Amblystoma, made from drawings of series of sections cut in three planes at right angles to each other. It shows the skeleton of the head viewed from the left side, also the shape and relative position of the left rudiment of the pectoral girdle (P. g.).

FIG. 32 G.—Lateral longitudinal vertical section of the brain of an embryo of Amblystoma, cut through the left side, showing the course of the nervefibres at the time of their first development in the brain. (Z. 2, A.)

FIG. 33.—Part of a transverse section through the fore-brain of a Frogembryo, showing the fibres of the lateral and anterior band (A F. and A. F.), the latter crossing the anterior surface of the brain; also the fibres developing on the anterior wall of the optic stalk (n. II). The posterior wall of the optic stalk (o.p″.) is free from fibres. (Z. 4, A.)

FIG. 34.—Transverse section of the spinal cord in the dorsal region of an embryo of Amblystoma at the stage of series H. It shows in cross-section the longitudinal nerve-fibres of the lateral band (L. F.), also the transverse fibres of the ventral commissure (T F.). (Z. 4, A.)

FIG. 35.—Transverse section passing through the mid-brain and secondary fore-brain of an embryo of Amblystoma at the same stage as the embryo of series G. It shows the corpora striata (st.) and the transverse fold which separates the secondary fore-brain from the thalamencephalon. P. C. shows the position of the posterior commissure. (Z. 2, A A.)

1

S. F. Clarke, “Development of Amblystoma punctatum, Part I, External,” ‘Studies from the Biological Laboratory of the John Hopkins University,” No. ii, 1880.

2

D. S. Jordan, ‘Manual of the Vertebrates of the Northern United States, &c.,’ 1876.

1

For a statement of Hertwig’s results, see A. 0. Haddon, ‘Au Introduction to the Study of Embryology,’ 1887.

1

Goette, ‘Die Entwickelungsgeschichte der Unke,’ Leipzig, 1875.

1

Johnson and Sheldon, “Notes on the Development of the Newt (Triton cristatus),” ‘Quart. Journ. Micr. Sci.,’ vol. xxvi, N. S., 1886.

1

Orr, “Contribution to the Embryology of the Lizard,” ‘Journal of Morphology,’ vol. i, No. 2, 1887.