Talk:Book - Contributions to Embryology Carnegie Institution No.36

From Embryology



By Florence R. Sabin, Professor of Histology in the Johns Hopkins University.

With six plates and one text-figure.



Introduot ion :;;-.. 215

Methods 217

Methods of nutrition of early embryos 220

Appearances of the exocoelom in the living blastoderm 222

Differentiation of angioblasts from mesoderm 227

Method of formation of blood-vessels from sohd angioblasts 231

Origin of blood-islands 236

Cycles in the development of the vascular system 241

Cycles in cell division 252

Origin of the heart and aorta 255

Conclusions 256

Bibliography 257

Explanation of plates 259





By Florence R. Sabin.


In the .study of tho origin of ondotholium and hlood-colls it i.s obviou.s that any extension of the method of observing them in hving form i.s an advance. Until recently the tadpole's tail constituted the main subject for the study of gro^^ih of the vascular system in a living form. From the time of the discovery by Kolhker that it contains two kinds of capillaries, those carrjong blood and those carrying hnmph, the tadpole's tail has been studied over and over again in connection ^snth the question of the growth of vessels. The final studies on the living tadpole, made bj' E. R. Clark (1909, 1918), have demonstrated conclusively that at certain stages both blood-capillaries and l}TTiph-cai)illaries in the area of the tail grow by the method of sprouting; that is, by the division of and increase in the cj^oplasm of the endothehum lining the vessels, and not by the addition of new mesenchjTQal cells to their walls.

An extension of the study of the vascular .system to a living form was then made by Stockard (1915), who watched blood-vessels arise in developing fish embryos. This author showed that the first sign of blood-vessels is the elongation of certain mesenchymal cells into spindle-shaped cells which could be identified as angioblasts. This I beUeve to be a fundamental point — that the vascular system begins by the differentiation of certain cells which we may term angioblasts, or according to Ranvier, vasoformative cells. Stockard then subjected the fish to experimental conditions, by which he made an analy.sis of the place of origin of red blood-cells. In fish embryos that had been treated by certain chemicals he produced an abnormality consisting of a lack of union between the venous end of the heart and the vitelUne veins, so that there was no circulation and consequently no moving of the blood from place to place. From these experiments he concluded that in the fish all of the cells that form red blood-cells are located in the intermediate cell-mass in the posterior part of the embryo, and that none of them arise in the head. In this conclusion his results have been questioned by Reagan (1917) in a study on the same form. Certainly in the chick it can be definitely proved that the endothehum of the vessels gives rise to red blood-cells, and that this process can be seen to take place not only in every part of the area vasculosa, but T\ithin the embryo itseK, the process having been observed in the aorta in the living chick.

Observations on blood-vessels in living chicks are by no means new; in fact it is obvious that many of the conclusions in the monograph of His, published in 1S6S, arc based upon studies of liviM<i; clucks immediately after removing them from the shell; wliile Duval in his atlas, published in 1889, called attention to the advantage of studying the beating of the heart in living chicks in a watch crystal. Indeed this advantage has been quite apparent to the whole group of embryologists using the same material. As a result of the develoi)ment of different artificial solu- tions ujwn which the method of tissue culture depends, there has naturally followed a marked increase in the studies of Uving embryos of the higher forms; for examj^le, the series of observations made by E. R. and E. L. Clark (1912, 1915) on the super- ficial lymphatics in chicks held in the shell and kept alive by Rusch solution, and the demonstration by Brachet (1913) that a mammaUan egg could be kept aUve for at least 48 hours in l)lood-i)lasma. These studies, though depending upon the development of the media wliich are used in tissue-culture, are not specifically tissue-cultures, for by that term is meant the growth of tissues on a coversHj) in artificial media, so arranged that the development of their cells can be watched under the microscope.

The first direct appUcation of the method of tissue-culture to the entire blas- toderm of the chick — that is, the study of the blastoderm on a coverslip so that its cells could be clearlj- followed — was made by McWhorter and Whip])le in 1912. These investigators followed the technique of Burrows and Carrel, developed from the work of Harrison, using clotted chicken-plasma in which to grow the specimens. In this medium they kept chicks under observation for a considerable period of time — namely, from the stage of about 2 or 3 somites up to the stage of 17 or 18 somites. In my experiments I have followed the method of W. H. and M. R. Lewis, using Locke-Lewis solution instead of the clotted plasma. This tcchnicjue is much more simple and, I beheve, offers many advantages, in spite of the fact that the embryos do not Uve nearly as long. I have made no attempt to determine the possible duration of life of the specimens in the media, but have watched them on an average of 4 to 5 hours, during which time the maximum number of new somites is 3, with an average of 2. The greater ease of the method renders possible much more extensive observations. My series numbers 266.

With this method of studying the blastoderm of the chick, it is po.ssible to trace the formation of blood-vessels back to their very beginning and to ob.serve them develojiing by differentiation from mesoderm to a new tyi)e of cell, the angioblast. These angioblasts form soUd clumps, then bands or cords of cells, wliich soon unite in a plexus by the well-known method of sprouting. The central part of the sub- stance, either of the isolated clumps or of the plexus of angioblasts, then liquefies, leaving the rim to form the endothelial wall of the vessel. In this hciuefaction both cyto])lasm and nuclei are involved, and the resulting fluid constitutes the first blood- phi>ma. Moreover, angiol)lasts not only give rise to endothelium and blood-plasma, but also produce the red blood-coriniscles. Thus in the chick angioblasts are cells which differentiate from mesoderm and form endothehum, blood-j)lasma, and red blood-cells.

That angioblasts differentiate in the area vasculosa of the chick has long been known; indeed, the idea is involved in our knowledge of the origin of the so-called blood-islands dating l)ack to the early embryologists, notably Wolff and Pander. In this study I propose to sharpen the conception of the differentiation of angio- blasts and to Hmit the term blood-island to those masses of cells which actually produce hemoglobin and become red blood-corpuscles. It will be shown that though angioblasts, immediately after their differentiation, begin to sjjread by sprouting, the primitive vessels do not invade the embryo in the manner beheved by His, but that there is a progressive differentiation of angioblasts in silu, starting in the area opaca of the cliick and gradually extending toward the embryo. The heart, most of the dorsal aorta, and even a part of the ventral aorta of the head, can be seen in the living chick to differentiate in situ. Thus studies in the living blastoderm confirm the experiments of Hahn (1909), in which he destroyed the area vasculosa on one side of a very young blastoderm and subsequently found an aorta on that side. These experiments were confirmed later by McWhorter and INIillcr (1914) and the .same point was brought out by Reagan (1915), who i.solated the head-fold of a young embryo and found vessels in the isolated segment. It may thus be considered as settled that blood-vessels arise from cells which differ- entiate from mesoderm, and that these cells (angioblasts) differentiate not only throughout the area, but also throughout the body of the embryo itself. The ultimate i)eriod at which angioblasts cease to differentiate from mesenchyme must be regarded as still unknown.


My early studies were made in the following Locke-Lewis solution: NaCl 0.9, KCl 0.042, CaCU 0.024, NaHCOs 0.02, Glucose 0.25— with chicken bouUlon.

It is convenient to make stock solutions of the salts four times the desired strength and mix 25 c.c. of each. To this mixture is added the glucose and 20 c.c. of chicken bouillon. Tliis is made according to the method for media and carefully neutralized with sodium bicarbonate. The bouillon has also 0.50 grams of sodium chloride to each 100 c.c. in order not to dilute the Locke solution too much. The mixture of Locke solution and bouillon (Locke-Lewis solution) is then divided into test tubes and heated in an Arnold sterilizer. It may be put through a Bergfelt filter instead, but this has no advantage except to prevent the solution from becom- ing spoiled during sterilization bj' the formation of a precipitate, an accident which occasionally happens.

This Locke-Lewis fluid is, of course, a solution in which the exact amount of salts is unknown, owing to the addition of the bouillon; but since a given solution of bouillon can be kept sterile and made to serve for a very large number of exT^eri- ments, there is a possibility of making certain experiments with a solution which varies onh- in the known amount of salts added. This solution has been shown by W. H. and M. R. Lewis to be excellent for the growth of tissues of chicks that have been incubated for from 6 to 12 days, but in studying the entire blastoderm in this fluid I noted that in specimens of the latter half of the second day, by wiiich time circulation had been established, nearly all of the hemoglobin w-as laked out of the corpuscles. Following the studies of Bialaszewicz (1912), w-ho showed that an unincubated hen-egg contains 1.07 per cent of sodium cliloride, and that the amount decreases as the blastoderm develops, I used 1.07, 1.06 and 1.05 per cent.

With these higher percentages of sodium chloride it was evident at once, not only that the hemoglobin was not laked out of the red l)lood-cells, l)ut that the solu- tion was a more favorable one for all of the cells of the young blastoderm. To deter- mine the best solution for each of the early stages of the cliick would necessitate a much longer series of experiments than I have carried through. So far my best results have been with solutions containing 1.06 or 1.05 per cent of sodium chloride in specimens of less than 20 somites. In specimens having between 20 and 30 somites a solution containing 1.07 or 1.06 per cent of sodium chloride will show a considerable amount of crenation of the free red blood-corpuscles, and hence is too strong. It is interesting to note that the young red cells still attached to the islands, or just freed from them, do not tend to crenate as much as the older cells, indicating that there is a shght difference between them. The concentration of the sodium chloride is a very important point for the successful study of the young chick- embryos because all of the cells are very sensitive to it.

The technique of preparing the specimens is as follows: The blastoderm is cut out in a warm box under sterile precautions, leaving a narrow rim around the area opaca; it is then placed in a dish of warm, sterile solution and freed from the yolk and the vitelline membrane, after which it is transferred to a dish of fresh solution and floated on a sterile covershp. This must be free from grease, as for blood-smears. The coverslip is then placed on a piece of blotting paper and the rim willed perfectly dry. I mount the blastoderms, for the most part, with the endod(u-m against the covershp. This has a twofold advantage: in the first place, the great majority of the vessels develop ventral to the mesoderm and hence are nearer to the endoderm than to the ectoderm; in the second place, the cut edges of the blas- toderm alwaj^s roll dorsalward and hence it is easier to get flat mounts on the ventral surface. Moreover, the endoderm sticks to the glass much more closely than the ectoderm, a matter of great importance in using a thin medium Uke the Locke- Lewis solution. If the specimen sags away from th{> glass it interferes very much with the development of the blastoderm. It must therefore b(> sjjread out carefully and the edges pulled out to the dry rim of the coverslip, leaving room enough for the rim of vaseline which seals the mount but wliich must not touch the specimen at any point. If in the mounting of the specimen a little of the blastoderm comes in contact with the vaseline it should be remounted, or else a symmetrical point of the specimen be brought in contact with the vaseline. Otherwise surface tension will greatly distort or even destroy the eml)ryo. The covers are mounted on a hanging-drop slide and the amount of fluid which remains on the blastoderm is sufficient to keep the embryo alive.

Inasmuch as the endoderm of the area opaca is many times as thick as that of the area pellucida, it is clear that there must be a rim of tissue just at the border between the two which will not be in exact contact with the glass. To make this rim as narrow as possible I stretch the specimen on the glass to a considerable extent. If one has a large number of specimens it will Ix'come clear that there is a very great variation in the width of the area pellucida, and since it is only tliis area that can be seen distinctly in the living specimen, there is considerable variation in the value of different specimens. The stretcliing of the specimen flattens the embryo some- what, so that the circulation, if begun, is often not reestabhshed immediately on the coversUp. In this connection it is interesting to note that the circulation is more often impeded in the left side than in the right, since the heart curves to the right, and hence the left vitelUne vein is the longer and becomes compressed across the body of the embryo. In sjnte of this mechanical disadvantage a good circulation is often (indeed usually) reestablished and maintained for 3 or 4 hours. The sUde must, of course, be kept at a temperature of 39° C, either in a stage incubator or in a warm box.

The average life of the specimen is about 5 hours. The heart can often be made to beat, after it has stopped, by a bath in fresh solution, but I have never seen it beat for more than an hour after this procedure. Cell division does not when the heart stops, but I have made no tests to determine the actual length of hfe of cells in these prejiarations. The most striking sign of the death of the cell in these specimens is that the resting nuclei, which have been practicalh' invisible in the total mounts, become almost as plain as if they had been treated with an acid.

The specimens are fixed by floating the cover-slip on Bouin's fluid of 75 parts of saturated aqueous picric acid, 20 parts of formol and 5 parts of glacial acetic acid. The picric acid is removed by repeated changes of 70 per cent alcohol without the use of any of the lower grades of alcohol or of water. If placed in water the specimens swell and sag away from the glass. I change the alcohol several times the first day, then keep the specimens in 70 per cent alcohol until white, when they are placed in 80 per cent. I have used absolute alcohol, methyl alcohol, and Helly's fluid as fixatives, but with less success.

For stains I have had the best results with hematoxyhn alone, or with hema- toxylin and a counterstain of eosin (6 parts), orange G (4 parts), and aurantia (1 part). In staining with hematoxyhn the specimens must be hurried through water and not left too long in the dilute stain or they will swell, as in water. The 3'oung blastoderms react intensely to hematoxyhn and nothing in the tissues reacts to an acid stain except the globules of yolk. A blastoderm, therefore, \\ath- out any counterstain can be analyzed in a total preparation to a considerable extent over the area opaca, while it is seldom possible to focus sharply enough through its thick endoderm in the living specimen. In specimens which have hemoglobin in the cells a counterstain helps to bring out the contrast between angioblasts without any hemoglobin and the true blood-islands. I have used iron hematoxylin, but with less success, for the total mounts. Eosin-azur brings out the basophihc granules of the angioblasts and of the j'oung blood-islands better than hematoxyhn, but it is difficult to get good, permanent preparations of total mounts with this stain, since it is impossible to differentiate accurateh' for a specimen wliich varies so greatly in thickness. Alum-carmine is also an excellent stain for these specimens.

If it is desired to mount a specimen with the ectoderm against the cover-shp, a method which offers advantages for stud\ang the area opaca, or if the specimen is to be cut into sections, the blastoderm should be removed from the glass by a single stroke with a Gillette blade under 95 per cent alcohol. If removed in any of the lower grades of alcohol the specimen will shrink and wrinkle, but after the blastoderm is well hardened in 95 per cent alcohol it will remain perfectly flat through the processes of embedding. If a specimen is thus fixed and dehydrated on the glass, cells which have been stuched in the living form can be readily identi- fied in the total mounts or even in sections. I clear the mounts in loenzene and oil of wintergreen. The only obvious shrinkage is in passing into the oil, then in an occasional specimen the ectoderm and mesoderm may crack, while the endoderm usually stays intact against the glass.

In describing the specimens I shall use the number of somites as an indication of the stage of development. It has become evident, however, that there is a very wide variation in the stage of development of the vascular system with a given number of somites, and moreover, that there is a great diiTerence in the develop- ment of the vessels during the interval between the formation of one somite and that of the next. To show the variation of the vessels with reference to the somites, angioblasts may occasionally be seen in a specimen of two somites in the area pellucida, but they are not usually present there until the chick has 4 somites, and never in great numbers until the stage of 5 somites. Again, a blastoderm of 14 somites usually has angioblasts but no vessels in the posterior imrt of the area pellucida, but I have a specimen of 12 somites in which angioblasts in this area have developed into vessels, the endothehum of which is giving rise to blood-islands.


In the study of the blastoderm there is one phenomenon associated with the nutrition of the early embryo which must be clearly recognized, not only on account of its importance physiologically, but because in the Uving specimen it simulates so closely the formation of blood-vessels. This phenomenon I have called endo- dermal blisters. The early blastoderms receive nourishment in two ways: (1) by wandering cells heavily laden with yolk, which become detached from the tliick endoderm of the area opaca and wander throughout the embryo. These cells were described by O. Van der Stricht in 1892 (p. 217) and have been called wandering endodcrmal cells by Maximow and Danchakoff; (2) by the absorption of fluid substances from the yolk, which occurs in definitely localized areas. If examina- tion be made of any collection of young chick blastoderms that have been stained and mounted with the endoderm against the cover-slip, it will be noted that large numbers of them show hazy Unes, especially in the posterior part of the area pel- lucida (fig. 2, i)late 1).

The appearance of the bhsters is far more striking in the living form than in fixed specimens; in the former they simulate vessels to such an extent that almost anyone studying hving si)ecimens for the first time would regard them as the beginning of the vascular system. As has been said, angioblasts are not found con- stantly in the area pellucida until the chick has 5 somites; an occasional blastoderm may show the very first angioblasts there at the stage of 2 somites, more often at the stage of 3 or 4 somites, but in all of these the forerunners of blood-vessels are solid clumps of cells without any lumen and hence not simulating vessels at all, as can be seen on plate 6, figure 27. In such a specimen as the one shown on plate 1, figure 2, where the somites are just beginning, I do not think there is any possibility of the presence of angioblasts in the area pellucida, and these spaces can not be the forerunners of the vascular system. The small dark spots seen in the area pellucida (plate 1, fig. 2) are either globules of free yolk or, in one or two cases, a precii)itate of the stain. In the living blastoderms in the early stages these blisters, with their highly refractive contours, make the most striking picture in the entire blastoderm. They appear as isolated spaces of varied size and shape; sometimes there are many small vesicles, or again there are a few large confluent ones resembUng multilocular cysts. Their walls are either a fine, sharp, refractive line, in which case the spaces are much distended, or the border appears to be wider, limited by a definite row of cells with an occasional nucleus looking exacth^ hke the nuclei of endothelium, bulging into the lumen. Such blisters are to be seen in section in plate 4, figure 14, from a chick with no somites and of nearly the same stage as the one shown on plate 1, figure 2.

The fluid in these spaces is held in place by irregular threads of endodermal cells which stretch toward the mesoderm; as can be imagined, these threads break readih' and hence the spaces may change with great rapidity, even disappearing entirely within a few minutes. In order to find them in sections one must fix a specimen while they are still visible, as was done with that shown on plate 4, figure 14. Moreover, the identification in sections of bhsters that had been recorded in the specimen at the time of fixation is possible only in a technique such as I have described, in which the specimen is kept on the cover-slip throughout dehydration. If the blastoderm be removed with care from the glass the blisters are not usually broken.

In the Living specimen one can prove definitely that these structures are blebs of endoderm by means of the focusing screw, as their contours can be made to lead over to the thin film of endoderm that covers them. Moreover, if they are not too thick one can focus through them to the mesoderm beneath. An excellent descrip- tion of this phenomenon is given in the article of McWhorter and Whipple (1912, page 125), except that these authors interpret it as the beginning of blood-vessels; that is, they include these structures with undoubted vessels which they followed in later stages. They describe them as under the ectoderm rather than under the endoderm, and indeed similar bhsters do occur under the ectoderm. By my method of studying the specimen with the endoderm against the cover-shp the bhsters of the endoderm are much jjlainer, and I think that they are much more frequent. McWhorter and Whipple also state that they are not to be found after the stage of 10 somites. I find them frequently in the older specimens, for example at the stage of 17 or 18 somites; and in specimens under 5 somites, I would say that they are almost constant.

It will be seen readily that these bhsters tend to disappear under the experi- mental conditions of growth on a cover-shp and that they can not form again.

Thus, to show them best in total blastoderms the s]iecimens should be fixed as soon as they will stick closely on a cover-shp, wliich is approximately from one-half to one hour. By the time a specimen has been stretched on a cover-slip for four or five hours the blisters have usually all disappeared. If one wishes to convince oneself that these structures are not artefacts, but represent a true physiological process, it is necessary only to cut out a few early blastoderms and examine them inmiedi- ately in Locke solution under a l)inocular microscope, wliicli will show tlie endo- dermal bhsters with great clearness.

The contents of blisters is nearly alwaj^s in complete solution, suggesting that they contain a fluid wlaich has been produced by the digestive activity of the endodermal cells. O. Van der Stricht (1899, page 345) described evidences that the cells of the endoderm show a secretory activity, producing the first tissue fluid for the embryo. On the left side of plate 4, figure 14, the blister contains a single wandering cell from the endoderm, while on the right side the space appears empty. The bhsters on plate 4, figure 14, are opposite the intermediate zone of the endoderm which marks the border of the area pellucida. The section shows well the three zones of the endoderm, the thin wall of the mid-hne, the interme- diate zone and the area opaca. In this section the cells of the endoderm over the bhsters are full of globules of yolk. In many specimens this endoderm has been stretched until it is a very thin, fine fine.

The endodermal bhsters, which may be found in any blastoderm of the first two days of incubation and perhaps even later, represent the jjrimitive method by which the embryo is supplied with fluid and nourishment in the early stages, and hence are especially frequent and important in the early stages before the blood- vessels develop. They are areas of absorption of the fluids wliich bathe the tissues. They represent the absorption by endodermal cells of the fluid of the subgerminal cavity and are of great importance as a demonstration of the method of nutrition of the early stages. They show how the tissue-fluid arises, j^recedes the blood- plasma, and bathes aU of the cells of the growing blastoderm.

All series of sections of chicks throughout the second day of incubation show also the origin of the wandering endodermal ccUs. The structure of the endoderm of the area opaca is that of a network of cytoplasm with nuclei in the meshes and enormous droplets of yolk in the vacuoles. From this network many individual cells become packed with droplets of yolk and free themselves. The two types of endodermal cells, those fixed in the network and the wandering type, are shown on plate 5, figure 21, and on plate 6, figure 29. In the latter figure the label en. c. indicates one that has wandered to a position dorsal to the mesoderm. These endodermal cells may attain enormous size and may wander to any point in the substance of the embryo. Later they may be found witliin th(> blood-ve.ssols and may circulate with the blood-cells.


The second jjlienomeiion with which one must become thoroughly familiar in studying the living blastoderm is the varied appearance of the developing exocce- lom. The subject is taken up here from the standpoint of the appearance of the different areas of the coelom, as they can be made out in the developing chick, with reference to tlie origin of the blood-vessels. At certain stages the exoccjelom in the living chick simulates the blood-vessels, although formed by an entirely different process.

The specimen shown in figure 2, plate 1, is in what may be termed the .second stage in the development of the mesoderm of the cliick. As is well known, in early chick blastoderms of the first day of incubation there is a stage in which the primi- tive streak has formed, but before the appearance of a head-fold, when the mesoderm is spread out uniformly through the area pellueida and the area o])aca except in the pro-amnion. At the stage shown in figme 2, plate 1, it is evident that the mesoderm is more dense opposite the posterior half of the area opaca, and that the outer rim of the area opaca is much mottled. These patches of cells would, of course, be uni- versally termed the blood-islands of Wolff, but I wish to question the use of the term blood-islands for them. In sections the mottUng is seen to be due to isolated patches of mesoderm eloseh^ attached to the ectoderm (plate 4, fig. 14). Moreover, it can be noted that though the dorsal margin of the ectoderm has a smooth contour, the ventral border has many filamentous processes pointing toward the endoderm, and that here and there one sees a nucleus in this ragged border of the ectodermal cells which might be interpreted as the beginning of a new group of mesodermal cells from ectoderm.

To the evidence that mesoderm may arise from ectoderm in situ, along the edge of the area opaca, may be added *hat in later stages, after the sinus terminahs has formed, one can still find quite isolated masses of new mesodermal cells lateral to the sinus. Vialleton (1892) called attention to these small masses of cells, attached to the ectoderm, which are to be found entirely outside the area vasculosa in a chick of 11 somites, and interpreted them as the blood-islands of Wolff. During the next year O. Van der Stricht pubUshed a note in which he questioned the con- clusions of Vialleton, on the ground that these masses of cells had none of the char- acteristics by which they could be interpreted as blood-islands. This view coin- cides with my own interpretation of them — i. e., that, like the primitive mesoderm within the area vasculosa, thej' constitute undifferentiated masses of cells and thus are not to be identified as blood-islands; and that they are destined to form two different tissues, the two layers of cells which form the cxocoelom, and the clumps of angioblasts. They are of especial interest as bearing on the (question of the origin of the mesoderm.

Thus the evidence seems to me to point toward the view that while the primi- tive mesoderm of the chick and the mass of mesoderm of the area vasculosa arise at the border of the primitive streak or in the axis of the embrj'o for the head- mesoderm, there may be a continual, considerable increase in the mesoderm of the area opaca by new cells which are differentiated in situ from the ectoderm along the border of the area vasculosa. This is in accord with the view of His and O. \'an der Stricht, as opposed to that of v. Kolhker, Hertwig, and Rabl. As was noted by O. Van der Stricht (1895, page 184), no matter how intimate may seem the relation between the mesodermal masses that develop into angioblasts and the endoderm there is no sign that the endoderm gives rise to angioblasts.

A point which I wish to emphasize is that at the stage shown in figure 2, plate 1. the masses of mesoderm which are so conspicuous in the posterior zone of the blastoderm are as yet without differentiation. They have been called blood-islands, and it will be difficult to change a name with such a long history. I beUeve, however, that it would be possible to show that these uniform masses of cells are destined, as stated above, to differentiate into two different structures, neither one of wliich are blood-islands; ?. e., two layers of cells placed along the dorsal border of the masses, which are the forerunners of the mesoderm which will border the coelom, and ventral clumps of angioblasts. Moreover, the angioblasts must first make blood-vessels in the chick before any clumps of cells form hemoglobin and become the. ancestors of red blood-corpuscles. There is no differentiation in most of the primitive masses of mesoderm before the somites develop, as inchoated in figure 14, l)late 4, where the dark spots along the ventral border of the mesoderm, seen espe- cially well on the left side, are merely cells in division. In this section the ectoderm is readily distinguishable, for the reason that its cells form a definite row with many nuclei equally spaced. The mesodermal masses have larger, irregularly placed nuclei, and more vacuoles and more droplets of yolk in the cytoplasm.

The next stage in the development of the coelom is shown in plate 1, figure 3, a blastoderm of 2 somites. This is from a chick which was grown for 3f hours on a cover-sUp. When first taken from the egg it had many endodermal blis- ters, but they had all disappeared before the specimen was fixed. It showed a primitive streak and a head-fold but no somites. While it was growing on the cover-slip the first cleft, which determines the first and second somites, appeared, and when fixed, the first and second somites were clear. They are still so delicate as to be seen but faintly in the photograph. The latter discloses an interesting point in connection with the primitive streak, namely, a transverse WTinkhng of the streak which is a constant phenomenon in the blastoderms of early stages when grown on a cover-slip. It indicates, I think, that the streak is a zone of active growth in length, so that the fixing of the margins of the blastoderm to the cover-slip consequently produces a wrinkling of the tissues at that point. The same wrinkling of the primitive streak is shown in section in figure 14, plate 4. The specimen shown in figure 3, plate 1, was growing vigorously when it was fixed, for the entire endoderm is caught in the i)hase of nuclear division. There is a defect in a portion of the mesoderm of the posterior part of the area opaca which is not an uncommf)n abnormality in these chicks that have been grown on a cover-slip. The defect apparent in the mesoderm on the right side of the photograi)h is not a real one, but a zone where the cells have reacted less to hematoxylin. The specimen was chosen in spite of these defects because it shows so well the formation of the exocce- lom in the area oj)aca.

Throughout the area opaca the mesoderm has formed a network of hollow vesicles so closely i)acked together that the interspaces are mere lines of cells. In this particular specimen almost all of the primitive mesoderm, as seen in figure 2, plate 1, has been involved in making these vesicles, and very few of the cells have become angioblasts. About the middle of the area opaca, on either side, are certain denser masses of cells which can be identified as angioblasts, but such a specimen emphasizes the value of not calhng the first isolated masses of mesoderm blood- islands or even angioblasts, in spite of the fact that so much of the primitive meso- derm is not always involved in forming the coelom. It will be noticed that these angioblasts he opposite the septa of the vesicles, as was pointed out by Ruckert.

In figure 4, plate 1, is a specimen with 4 somites showing still more differentia- tion. Here the primitive streak is almost without wrinkles, for the specimen was grown on a cover-shp for only 35 minutes. There are two small endodermal blisters in the margin between the area opaca and the area pellucida at the posterior end of the blastoderm. In the zone of the amnio-cardiac vesicles the coelom is in the form of the large, closely packed vesicles, like those of the area opaca in figure 3, plate 1. This type of vesicle for the coelom is constant over the area of the amnio- cardiac vesicles and over the area opaca.

The middle zone of the area pellucida presents an entirely different appearance, here is to be seen another and an important phase in the development of the exocoe- lom, a phase with which it is essential to be thoroughlj' famihar in relation to the study of the vascular system in the living form. It is a stage in which there are no large, definite vesicles, closely packed together Uke those just described, but rather where there is a delicate network of mesoderm with wide gaps where the mesoderm fails altogether. Such an area is shown in figure 16, plate 4, a photograph of a section passing through the first somite of a chick of 5 somites. About the middle of this section there are two very plain gaps in the mesoderm. The same point is well shown in figure 27, plate 6, from the same area at a later stage, in which there is a network of angioblasts as well as a network of the exocoelom.

It will be seen in the section on plate 4, figure 16, that the dehcate network of mesoderm shown in figure 4, plate 1, over the middle part of the area pellucida, is in the form of two layers with an excessively narrow sht between them, quite different from the wide vesicles of the amnio-cardiac region shown in figure 15, plate 4. In fact, the two sections shown in figures 15 and 16, plate 4, are to be compared with the appearance of the mesoderm in the anterior and middle part of the area pellucida in figure 4, plate 1. In the anterior half of the area opaca, on the other hand, two different structures are visible in figure 4, plate 1 : the large vesicles of the coelom, and soUd bands of cells which tend to Ue opposite the edges of the vesicles and which show particularly well on the right side. These bands or cords of cells are angioblasts and, in the specimen itself, are very clear. The posterior part of the area opaca, on the other hand, has dense masses of mesodermal cells which, in the total preparation, seem entirely undifferentiated; that is to say, one can not make out the difference between the coelomic mesoderm and the angio- blasts. In a section of the same stage, however, as shown in figure 29, plate 6, it can be seen that the dorsal cells of the mesodermal masses are just forming two definite layers {mes.) which are the forerunners of the hning of the exocoelom. This is the very beginning of thedifferentiation of the mesoderm of this area into a dorsal or ccelomic part and a ventral or vascular layer (a).

While it is true that in an occasional section of a specimen of the stage shown in figure 2, plate 1, the very beginning of this differentiation of the cojlomic meso- derm can be found in the posterior area, it is not constant or extensive there until the stage of 4 somites. The fact that the formation of the exocoelom is initiated by the arrangement of the mesoderm into two layers of cells was described antl illus- trated by O. \'an der Strieht for the rabbit in 1895. This author stated that the mesodermal cells arrange themselves into two layers, a dorsal layer, the forerunner of the somatopleure, and a ventral layer, the forerunner of the splanchnopleure; and that at the start there are no spaces between these two layers, but gradually there appear isolated clefts which flow together to make the cavity of the coelom. It can thus be made very clear that the ca?lom forms from the arrangement of the meso- derm into two layers, which then split apart without the destruction of any of the cells. It is this lack of destruction of tissue that I wish to emphasize, since it brings out in strong contrast the fact that the coelom forms by a method entirely different from the methods by which blood-vessels on the one hand, and the cerebro- sinnal spaces on the other, are formed.

These three structures, blood-vessels, ccelom and the cerebro-sj^inal spaces, each have a different embryological laistory and can not be too strongly contrasted. I shall show that the lumen of a blood-vessel forms by the solution or Uquefaction of the central part of a solid mass of protoplasm and thus can not be considered as having any relation to tissue-spaces. According to the work of Weed, the arach- noidal spaces form in a mass of mesenchjTne. It is interesting to note that in the stages which we are considering in these early chick blastoderms there is no true mesenchyme, but simply mesoderm in layers and angioblasts in solid clumjis or masses. Dr. Weed has proved that the arachnoidal spaces come from tissue spaces in a mesenchyme by gradual increase in the size of the mesh of the mesenchjTne brought about by the dilatation of existing spaces and by the breaking of some of the strands of the mesenchyme cells. In this process the residual mesenchyme cells ultimately flatten out to make a lining for the cerebro-spinal channels. The same process is to be seen in the formation of the perioticular spaces in the inner ear (Streeter, 1918). On the other hand, the cells which go to make up the ctrlom hue up in the form of two layers, wliich then split apart without any destruction of tissue. Thus, embryologically, these three structures are entirely different, and are not analogous in any sense. Blood-vessels develop from dense, solid masses of cells; the ctt'lom develops from cells tliat invaginate to forma mesodermal cavity or, as in thechick, from mesoderm which forms two layers subsequently splitting aiwrt; while the arachnoidal channels come from the spaces of a typical mesenchyme. Blood-vessels are not derived from tissue-spaces; while, on the t)ther hand, the ca4om and the arachnoidal spaces both come from tissue-spaces, but by different processes.

In a study of the differentiation of angioblasts in the hving chick, one must be able to distinguish them readily from the ajjpearances of the ccelom, as seen in figure 4, plate 1, and tlu; point becomes still sharper in the stage of 5 somites

where there are more angioblasts to be seen against the background of the ccelom. This is shown on plate 1 , figure 5, and plate 2, figure 6, and also in a drawing from a small part of the area pellucida of one of them, figure 27, plate 6. Thus there is a stage when the coelom appears as a plexus of undifferentiated mesoderm dorsal to the plexus of angioblasts.

The plexus of the exoca>lom, as shown in figure 27, plate 6, is a dehcate network of protoplasm, with nuclei in the nodes and vacuoles and delicate fibrils in the cj-to- plasm. This network looks not at all uiiUke a typical mesenchj-me except that it is always in definite layers. Here and there are large interspaces which are entirely devoid of mesoderm. Against this dehcate network can be seen the mas.sive bands of angioblasts, still connected in places with the parent mesoderm beneath by bands of cells, and with each other by the sprouts so characteristic of angioblasts. The difference between these sharp, definite sprouts (representing the method by which the angioblasts join each other) and the delicate fibrils in the mesh of the mesoderm is well shown in the drawing. Throughout the stages of from 3 to 8 somites these networks must be studied together. The existence of the two net- works, that of the coelom and the more ventral plexus of angioblasts, was recognized by His (1868).

It is not my purpose to follow^ any further in this paper the development of the ccelom, because, as the chick grows older, it becomes less and less a source of con- fusion in connection with the studj^ of the vascular system. Up to the stage of 5 to 7 somites one must be thoroughly familiar wdth the different appearances of the coelom in the different parts of the area vasculosa, but as time goes on the ventral space occupied by the vascular layer grows wdder and the two structures thus cease to be so nearh^ at the same focussing level; thus the difficulty in keeping them quite distinct practically disappears. In general, however, in studying the hving chick there are three different zones of mesoderm in the area pellucida against which the angioblasts must be studied; first, the area of large vesicles which, bj' fusion into a single cavity, give rise to the amnio-cardiac cavities; second, the fine network of mesoderm of the middle and posterior zones of the area pellucida; and third, the axial zone of the somites and the dense, undifferentiated mesoderm posterior to them. This posterior, axial zone of massive mesoderm is so dense that only an occasional specimen will show the angioblasts with any great clearness against this background. It is, however, an important area for the study of the aorta. Fortu- nately, the angioblasts of the axial zone opposite the interspaces between the somites are especially clear in the h\ang chick, while the angiobhists of the posterior zone over the undifferentiated mesoderm, opposite which the posterior end of the aorta develops, can also be seen much more clearly in the hving specimen in which the cells are dividing.

DIFFERENTIATION OF ANGIOBLASTS FROM MESODERM. At a given stage in the development of the mesoderm of the chick, certain cells differentiate from the mesoderm to become the forerunners of the vascular system. The sequence of this differentiation will be taken up later.

but for the i^resent I will describe the jirocess wlierc it can be seen best in the hving chick — that is, in the posterior part of the area i^elhicida. This will become clear by comparing figures 4 and 5 on plate 1 and all of the figures on plate 2. In this posterior part of the area pellucida the angioblasts do not begin quite as early as in the more anterior part, but they are always more massive. Thus in figure 4, plate 1. there are a few small, isolated clumps of angioblasts in the anterior part of the area pellucida, while in figure 5 of the same plate there is a definite plexus of them in the posterior part. The characteristics of this jjlexus arc best shown on plate G, figure 27, a drawing of the area included in the square on plate 1, figure 5.

In the living chick the bands of angioblasts are very readily distinguished from the network of the ccelom because they are much more refractive than the undiffer- entiated mesoderm. This refractivity is much increased during the period of cell division, so occasionally when one first takes out a blastoderm the entire network of angioblasts will appear very brilliant; or again, bands of angioblasts, which are at first rather dull, will gradually develop a high refractivity. This refractivity is due to a change in the cytoplasm of cells which precedes the division of nuclei by jjerhaps an hour or more. It is characteristic also of mesenchyme cells, as has been described by W. H. and M. R. Lewis, but the phenomenon becomes even more striking when one is dealing with such massive structures as the angioblasts. Since all of the angioblasts of an area such as the posterior zone of the area pellucida pass into tliis refractive phase at the same time, a living specimen in which the angio- blasts are about to divide becomes a very brilUant object. Besides a greater refrac- tivity, angioblasts have a denser cytoplasm than the mesoderm. This is due to large numbers of granules in the cytoplasm. These granules are the azur-granules which have been brought out by Maximow as characteristic of young red blood- cells. Due to their presence, the cytoplasm of the angioblasts is very dense in the hving form and stains intensely in fixed specimens with all of the basic dj^es like hematoxyUn. This reaction to dyes is also intensified at the time of division.

Besides this, there is another very important characteristic of the angioblast, namely: the behavior of the cells after division. In watching the living specimen one selects the bands about to divide by the high refractivity of the cytoplasm and then in about an hour the nuclear figures begin to appear, one after another, in quick succession. The only nuclear figure that can be made out in these masses in the living specimen is the metaphase. After the cells have divided the difference between the mesenchyme and the angioblasts is very striking. The mesenchyme cells become excessively irregular, then separate, and the delicate strands of their cytoi)lasm are reprotluced; while the two angioblasts stay together, forming soUd masses in which no cell outhnes can be discerned in th(> living specimen. A single, well-differentiated, resting angioblast can be distinguished from mesenchyme. One can not always be sure of a single cell in process of division because both mesen- chyme cells and the angioblasts are highly refractive at that time; but two angio- blasts can always be recognized by this characteristic formation of apparently syncytial masses. Such a clump of two cells is shown in figure 2(5, plate 6.

In figure 27, plate 6, as well as in figure 24, plate 5, can be seen another charac- teristic of angioblasts. Immediately after their differentiation they show a remark- able tendency to send out exceedingly deUcate sprouts of cytoplasm toward similar cells. These processes always emanate from a nuclear area, so that when one finds a clump of cells connected with a vessel by slender, solid processes, it is an indica- tion of new cells which have joined the wall. Such sprouts are also shown on the Uttle group of two angioblasts in figure 26, plate 6. By means of this sprouting masses of isolated angioblasts soon form a plexus; thus, even at the stage of 5 somites, at which angioblasts are just differentiating in the area pellucida, there is a very extensive plexus of these bands. In figure 27, plate 6, all of these characteristics are evident. In this specimen the mass of angioblasts showed the protoplasmic changes which precede nuclear division, and the specimen was fixed just as the stage of the metaphase became visible in a few of the nuclei. The bands of angio- blasts thus stand out very clearly against the more deUcate network of the ccelom beneath, ?s is plain in the photograph of this section (plate 1, figure 5). Here and there the bands of angioblasts themselves still show a httle of the character of the original mesoderm; that is to saj'-, thej^ are in the process of differentiating. Again in many places the bands of angioblasts are still connected by bands of cells with the mesoderm beneath, so that the angioblasts fade into the mesoderm, while in other places they are fully formed and fully connected with each other by characteristic, deUcate sprouts, figure 27, plate 6. In such a preparation there can be no doubt as to the relation of the angioblasts to the mesoderm, and this relation is as clear in the Uving specimen as in the fixed preparation. The origin of the angioblasts from mesoderm was clearly brought out by O. Van der Stricht (1895, page 182) and agrees with the views of Riickert, MolUer, Maximow, Danchakoff, and others.

The characteristics of the soHd bands of angioblasts, as seen in the Uving form, are brought out in figure 20, plate 4, a drawing of a plexus of angioblasts taken from the posterior part of the area peUucida in a chick of 12 somites, while it was growing on the cover-sUp. The position is just lateral to the square on plate 2, figure 8, from a sUghtly younger specimen. In the Uving blastoderm the bands of angioblasts appear Uke a complete sjTicj'tium. The interspaces are, of course, indistinct when the angioblasts are in focus, because thej^ rejjresent the laj'er of endoderm above the angioblasts and the layer of mesoderm beneath. There is never any difficult}' in distinguishing the definite layer of endoderm by changing the focus, either in the Uving specimen or in the total preparations. Under certain conditions, however, the endoderm may prove to be a very confusing factor in these studies. For instance, occasionally the endodermal ceUs may contain so many droplets of j^olk that one can not focus through them clearly. Such specimens never clear up and are useless for a study of the blood-vessels beneath. Or, for some unknown reason, the cells of the endoderm may become excessively vacuolated, in which case they are difficult to focus through, both in the Uving form and in the fixed specimens. Again, the entire endoderm may divide and its cytoplasm in con- sequence become too opaque to see through. In my early studies I concluded that these specimens had died, but finally saw some of the nuclear figures, and found

that it was only Jiocossary to wait until the phase of division was over, when the endoderm would agjiin clear uj). In the specimen from which figure 3, plate 1, is taken, practically every cell of the endoderm has a nuclear figure.

As can be seen in figure 20, plate 4, the nuclei in the living, resting cells can not be made out in these dense bands. The structure has the ap))earance of a syn- cytium; there is some network, occasional globules of yolk, and granules in the cytoplasm. The granules I take to be the basopliilic granules and the mitochondria. The only time when these bands of angioblasts look as if they were made u]) of individual cells is when they are passing through the stage of cell division. During this time, both in the living blastoderm and in fixed specimens, each nucleus is surrounded by its own zone of cytoplasm, as is clearly shown in figure 24, plate 5, from a fixed specimen. This chick had 10 somites when taken from the shell, and the heart was just beginning to beat; when fixed it had 11 somites. The heart stopi)ed beating after H hours, but the embryo lived and was fixed 2 hours later during a cycle of cell division. The blastoderm is shown in figure 8, plate 2, and the area just referred to (fig. 24, plate 5) is within the rectangle drawn on the photograph. The drawdng is reversed in position from the photograph. This specimen is one in which the c>i;oplasm shows the changes of cell di^'ision and brings out the fact that after the cytoplasmic period of prejiaration all the nuclei do notfUvideat exactly the same moment; hence some of the nuclei are in the prophase, others in the metaphase, and a few in the anaphase. In the living si)ecimen it can be seen that one cell after another divides until all have divided at the end of the cycle.

In figure 24, plate 5, it is clear that the band is still soUd; the process of Uque- f action (w^hich will be described in the next section) has not yet begun, and after all nuclei have divided the band will again look like syncytium as complete as that shown in figure 20, ])late 4. This is the more certain in that the specimen from which figure 24, plate o, is taken has 11 somites and is thus at a stage when the angioblasts of the posterior rim of the area pellucida are in process of growth rather than in the stage of transformation into vessels. The earliest specimen in w'hich I have seen the formation of a lumen in this posterior zone was one of 12 somites; usually, the condition is just beginning at the stage of 13 somites.

It will be noticed in figure 24, plate 5, that along the lower left margin there is a nucleus wliich has elongated slightly and looks as if it had undergone a slight cUfTerentiation into the type of nucleus characteristic of endothelial cells. \A'hen the vacuolation takes place there is a real differentiation of the cells along the border into endothelium, both as regards the nuclei and the cytoplasm ; that is, the nuclei elongate and the cytoj)lasm becomes less granular. If one did not know the actual history of a specimen such as is shown in figure 24, j^late 5, one might inter- pret its appearance as evidence that vessels form from angioblasts l)y a breaking apart of the individual cells of the solid mass, according to the generally accepted idea; but I am convinced from a study of the hving form that the essential points in the process of vessel formation are a differentiation of the cells on the margin into endothelium and a liciuefaction of part of the original jirotnplasm of the mass to make blood-pla.sma, and that this rounding up of tlic cells of the mass is a part of

tlu" phcnoim^iion of cell division. It can readily bo seen that there be a stage in the differentiation of a single angioblast from mesoderm when it is difficult to determine the fate of a given cell. This point can be seen in almost any chick of about 1 1 somites in the posterior part of the area pellucida, where there Ls a very active zone of the differentiation of new angioblasts (figs. 8 and 9, plate 2). Thus, for example, on the right side of figure 24, plate 5, there is a single cell which, while nf)t quite so granular as the main mass of angioblasts, still looks more hke an angio- blast than like a mesodermal cell. I interpret this cell to be an angioblast about to join the main nr.ass. The same point can be seen in figure 27, plate 6, where there may be some difficulty in determining the actual fate of some of the cells that con- nect the angioblasts with the mesoderm beneath. As soon as a cell is completely differentiated into an angioblast, and certainly after its first division, there is no difficulty in identification. For examj^le, the small clump of cells ventral to the mesoderm shown in Van der Stricht's figure 2 (1895, page 209), from the blastoderm of a rabbit, can with assurance be identified as a grouj) of angioblasts, just as there is no question of the fate of (he clump of two cells shown in my figure 26, plate 6.

The next stage in the development of blood-vessels is the formation of a lumen in these solid bands of angioblasts. In his original description of the development of blood-vessels His (18(38) asserts that they are at first solid and then, by some method which he could not make out, acquire a lumen; but that after the lumen is formed the resulting endothelial cells are less granular than the original solid masses. Shortly afterward, in 1871, this process was correctly analyzed by Klein. This work I am quoting from the Jahresberichte of 1872, since the original was not accessible. Klein states that certain cells of the deeper layers of the mesoderm become hollow through vacuolization. Through the enlargement of these vacuoles vesicles are formed, the walls of which become endothelium and then, by subse- tiuent division of the endothelium, there develop within the vesicles masses of cells, partly colored yellow with hemoglobin and partly uncolored. Thus in this early work is given perhaps the best description of the actual processes by which vessels can be seen to form in the Uving chick embryo. I agree with every one of these points except as regards the early stages ; i.e., that during the second half of the second day of incubation all of the cells attached to the inner wall of the vessels develop hemoglobin.

All subsequent investigators of blood-vessel development in the chick have described the vacuoles as being seen in the primitive masses of cells which produce blood-vessels; for example, O. Van der Stricht, Ruckert, ]MolUer, Maximow, Dan- chakoff, and Lillie all have called attention to this phenomenon, but the importance of the process first described by Klein is now brought out convincingly in these observations on the Uving form. If one watches such a band of angioblasts as that seen in figure 20, plate 4, vacuoles will begin to appear in the sohd mass, often just under the edge, but many times threctly in the center. It is a very striking process in the Uving chick and I endeavored to obtain drawings of it, but the changes in form arc so rapid that it is very difficult to record them well. By the time a tracing is made with a camera lucida the form is not quite the same and yet the changes can be easily followed with the eye, being not nearly as rapid as the changes occur- ring in the endodormal l)listcrs. Angioblasts may be completely transformed into vessels during the time that the blastoderms are growing on the cover-sUp — that is to say, in 3 or 4 hours — and the i)rocess is extensive in one hour.

In watching a hving specimen it can be readily seen that the vacuoles increase in size; many of them flow together and ultimately one may watch the entire center disappear from such a band as is shown in figure 20, plate 4. In figure 25, plate 5, from a chick of 13 somites, can be seen an earlj^ stage of the process of the solution of the center of a band of angioblasts. This figure is taken from the angioblasts of the posterior part of the area pellucida. Here and there are large vacuoles just under the edges, while the center of the mass is intact. It will be noticed that some of the vacuoles are against the nuclei, a very common occurrence. I consider it important as showing that the vacuolation or Uquefaction is a real intracellular process. The usual conception of the formation of a blood-vessel out of angio- blasts is that the mass breaks up by the separation of the indi^•idual cells in the center of the mass, wliilc the cells on the edge flatten out to form an endotheUal border. Instead, there appears to be no flattening out of the border cells, but rather vacuoles, which occur just under the borders, leave a rim of cytoplasm while the center of the mass disappears. There is a real differentiation of the cells thus left along the edges of the vessel, because they no longer look just like the original angioblasts. Their nuclei elongate sUghtly and their cj'toplasm becomes less gran- ular, so that in the living specimen endothelium comes to resemble ground glass (figs. 18 and 19, plate 4). These endothelial cells, however, retain the power of reproducing the more granular type of cytoplasm, which they do in giving rise to blood-islands. During this process of hquefaction at the center of these angio- blastic masses, the vessel formed is almost never any larger than the original mass, indicating that the absorption of the fluid present in the surrounding tissue is but slight.

A later stage in the process is shown in figure 23, plate 5, from a ])la8toderm of the same stage. In this specimen the center of the mass is disapjiearing and the vacuolation is much more extensive. Here the edges of the vacuoles are more ragged and little shreds of tissue can be seen in them. On the left border of the mass is to be seen a chain of angioblasts in single file. In this chain the same process is going on between two nuclei, giving the clearest jMcture of it in a single cell. The same point is shown in the upjx'r right sprout of figure 25, j^late 5. There is one place where the process of vacuolization can be always found witliin single cells — ■ namely, at certain stages over the amnio-cardiac vesicles. Here, as will be de- scribed later, angioblasts always form in long, slender bands (fig. 7, plate 2) and in these bands one can always find the liquefaction of chains of single angioblasts. I am quite sure, therefore, that the lumen of a vessel is made within the cytoplasm of a cell and nf)t entirel}' or even mainly l\v the separation of cells. It can be readily realized that one often sees the separation of individual cells in such a process.

During the phase of division the cells of the bands always look as if they were about to separate into individual cells, as suggested in figure 24, plate 5, but if such a specimen be watched until it passes into a resting stage the mass will again take on the appearance of a syncytium. When a .single nucleus is caught in the stage of division, Uke the one in figure 25, plate 5, it looks also as if the cell were to become separated from the mass.

There is another phenomenon very characteristic of these developing bands of angioblasts which suggests the idea of the breaking apart of the cells to form a vessel, i. e., cell death. Often while the vacuolation is going on, or just before it begins in a mass, occasional cells stand out by virtue of certain special characteris- tics. This is true also of the blood-islands which develop subse(iuently. These characteristics are, first, that the cytoplasm becomes perfectly clear except for a few irregular ma.sses, a phenomenon verj' familiar in certain cells in blood-smears; second, the nuclei show a sharp contour which is unusual in a living cell. In the hving specimen the dead cells stain with a dilute solution of neutral red, while the living cells do not react at all to the dye. In fixed specimens they show picnotic or fragmented nuclei characteristic of dead cells.

A still later stage in the process of Uquefaction of angioblasts, by which the lumen of vessels is formed, is shown in figure 28, plate 6, a section from a chick of 11 somites. Here the process is in the edge of the area opaca since all of the angio- blasts of the posterior i^art of the area pellucida are still solid at this stage. In tliLs section the cytoplasm just above the label L has almost Uquefied, but in the center of the vessel there is still a Uttle clump pf cytoplasm labeled A, showing a picnotic nucleus, an indication that there are some degenerative processes going on in the protoplasm. Between the labels A and L in the same vessel the endothelial border itself is very ragged, showing also some degeneration of cytoplasm.

In certain areas, especially in chicks a Uttle older (for examjjle, during the fourth day of incubation), there are numerous small, completely isolated vesicles, filled with free blood-cells, wliich look just as if thej' had become separated from an originally solid mass. Such tiny vesicles are always to be found dorsal to the amnio-cardiac vesicles, opposite the delicate ventral bands of this area which are shown in figure 7, i)late 2. Tliis specimen has these small vesicles wliich are filled with dead cells, as .'^hown by their picnotic nuclei. In other specimens the vesicles are filled with normal red cells, and I beheve are formed, Uke the rest of the blood- vessels, by a partial solution of the center of the mass and fill up subsequently by division of the cells of their walls — though there might well be some separating of individual cells in the transformation of the solid center of the angioblasts into the lumen of a vessel. However, in all of the masses of angioblasts which I have seen transformed into a vessel in the living specimen, there has been a considerable amount of the liquefaction of cytoplasm.

Just !is cell division progresses in cycles, so this process of Uquefaction progresses by stages, and if one finds it in a single band of angioblasts all of the bands in that area wiU show the same. In general one can see the process, in any blastoderm of 6 or 7 somites, taking place in the area pellucida opposite the venous end of the heart, in the zone in which the viteUine veins are developing. In a cliick of 8 or 9 somites it can almost always be found going on in the lower end of the aorta, opposite the undifferentiated mesoderm and just posterior to the somites; while chicks with from 12 to 14 somites will nearly alwaj's show the process in the pos- terior zone of the area pellucida. This zone is the best in wliich to observe it in the hving form.

From these observations it appears to be clear that the lumen of a vessel is ])ro- duced by the hquefaction of the central mass of angioblasts to make blood-i^lasma, and that wliile whole cells are destroyed and some of the original cells may separate from the original mass, the essential process is intracellular and the formation of the blood plasma by hquefaction is one of its important effects. This fact is brought out most strikingly when the lumen of a vessel is formed in chains of single angiol)lasts and where there can be no qu(!stion of the lumen developing in jiotential clefts between cells, since it is within the body of a single cell. The same fact is actually shown bj^ noting that many of the vesicles begin against the nuclei in the sj'ncytial masses. Thus the first blood-plasma is the result of the destruction of large masses of angioblasts bj^ a true cytolysis, and one of the important functions of these primitive masses of angioblasts is the formation of plasma. Since the tissue fluid is formed before the vessels begin, the endothelium, from the beginning, is a mem- brane which separates two different fluids, plasma and tissue fluid.

As has been said, the process of hquefaction of protoplasm is shown in section in figure 28, plate 6, which is from a vessel just under the edge of the area opaca of a chick of 1 1 somites. In this section the center of the larger mass has almost dis- appeared, while the endotheUal border is seen full of tiny vacuoles. In the upper part of the section is a blood-island. The lower vessel, on the other hand, shows again the liciuefaction of the angioblasts beginning along the edge of the main mass. The endoderm in this section is also interesting, showing both the syncytial net- work of the cytoplasm and some of the wandering endodermal cells which are becoming free from the network.

The idea that the lumen of a vessel can be considered as intracellular is not new. It was first expressed by Strieker (1860), who conceived the idea from studying the tiny sprouts to be made out along the capillaries in the tail of the living tadpole. Concerning it, he says :

"Wenn ich sage, die Wiinde der Capillargefiisse sind Pr()to])lasina. daiiii muss ich wohl selbst zuKcljon, dass sie aus Zollon hostohen; nichts dostoworiigcr liegt pin liefer und diirch- grelfender Unterschied zwischen dom, was uns an den barvon klar und luiwidcrlcglich vor Augcn tritt, und zwischen dem, was in neuster Zeit aus der Silberniothode dcducirt wiirde. Xach dicsor Deduction sollen die Capillaren aus Zollen zusajninengefvigt spin, und das Bhit in jpnen mithin intprcpllular flipsspn; nach dem was sicli an dpr L;u-vp prgipbt, ist pin Capillarrohr Protoplasma in R(")hrpnfnrm, Protoplasma, welches im Inncrn ausliohit ist, und wo das Blut gicichsam intraopihilar flpisst" (page 7).

It was then brought out in a series of interesting papers pubhshed in the early seventies. As has been said, it was most clearly and correctly presented by Klein in 1871, and later was expressed by lialfour (1873), Schaefer (1874), Ranvier (1875), Leboucq (187G), and Wissozky (1877). Ranvier studied the formation of blood- vessels in the omentum of the guinea-pig and in the blastoderm of the chick, and described the vessels as coming from vasoformative cells. He states:

"C'est g^neralenient dans les points nodaux que se produisent les premiere.s cavites vasculaires et les ilols sanguins. Les premieres cavites vasculaires sont d'abord des creux remplis de liquide qui s'agrandissent et s'allongent pour canaliser les branches du reseau. Les noyaux et le protoplasma refoules a la peripherie constituent les premiers elements de la paroi du vaisseau. Ces elements, agissant a la maniere des cellules glandulaires, .secre- tent un liquide, premier plasma du sang, qui distend peu a peu les branches du reseau pour leur donner le diajnetre considerable dent nous avons deja parle. Les ilots sanguins se forment aux depens de certaines cellules des cordons vasculaires prunitifs qui sont mises en liberte dans leur interieur au moment de leur canalisation. Ces cellules, relativement peu nombreuses, sont spheriques et contiennent d'abord un seul noyau (page 640)."

With this description I agree, except in considering that the i)rocess bj' which plasma is formed is one of hquefaction of protoplasm rather than a process analogous to secretion. Schaefer studied the formation of capillaries in the skin of a newborn rat and described certain vasoformative cells with vacuoles within the cytoplasm in which developed new, disc-Uke, adult red corpuscles. The latter point is, of course, entirely out of harmony with our present ideas concerning the origin of red cells; it is more in harmony with our ideas of the destruction of red corpuscles than of their origin.

The work of Wissozky is very interesting. This author began with the study of the reaction of hemoglobin-bearing cells to eosin seen in drops of fresh blood. From this he went on to making flat preparations of the embryonic membranes at the edge of the placenta of the rabbit, fixed the membrane in toto, brushed ofif the epitheUum and stained the whole with hematoxyhn and eosin. He made similar preparations from the allantois of the chick and the rabbit, and described the forma- tion of single angioblasts, wliich he called hematoblasts, described how they formed a network with numerous processes which he interpreted as indicating amoeboid acti\dty, and, finally how vacuoles formed in these soUd bands. Thus he saj's:

"An irgend einer Stelle des soliden haematoblastischen Stranges erscheint zuerst ein durchsichtger farbloser Streifen, welcher gewohnlich bogenformig ist; dieser Streifen erweitert sich ferner, uinmit die Gestalt eines Halbmonder an, seine Enden nahern sich mehr und mehr, um endlich zusanmaen zu fiiessen. Auf dicse Weise entstehen in dem Protoplasma der haematoblastischen Strange die bcschriebenen Lucken, in welchen die embryonalen Blutkorkerchen liegen, umgehen von durchsichtigen, farblosen Ringen."

With his description he gives a beautiful figure of this vacuolation going on in bands of single angioblasts, very hke my figure 25, plate 5.

All of the characteristics of angioblasts are beautifully shown in two figures of Maximow (1909, ])late xviii, figs. 1 and 8). These masses of cells (the first taken from the area opaca of a rabbit embryo and the second from the forerunner of the endocardium of the heart of a rabbit) show the azurophile c^-toplasm and the tendency to form syncytial masses. This is especially true in his figure of the forerunners of the endocardium. In the first figure are shown the irregular processes of the cells which I have found dejiended so much upon the concentration of the sur- rounding fluid. The angioblast in the second figure shows the beginning of the vacuolation by which the solid masses are converted into vessels.


The angioblasts not only give rise to endothelium and to the blood-plasma, but produce red blood-cells or, strictly si:)eaking, erythr()l)lasts. In the two figures from specimen 174 (figs. 23, jilate 5, and fig. 26, plate 6) is shown the striking con- trast between a vessel in which the central mass is going to liquefy entirely and an adjacent vessel in which some of the original mass is going to form a blood-island. These vessels are just along the border of the dense mesoderm of the posterior zone in a chick of 14 somites. Figure 11, plate 3, is a photograph of this blastoderm, and both drawings are taken from within the square on the ])hotograi)h. The angio- blasts in question are thus going to form a part of the lower end of the aorta. The aorta follows the lateral line of the somites and below the last somite curves out- ward, following the lateral edge of the dense, axial mesoderm. Outside this zone in this particular specimen, there is a zone of blood-islands in the vessels, and mesial to this area masses of angioblasts are becoming vessels, while the masses on the border between the two areas show both processes. In figure 23, plate 5, the central mass is deUcate and the vacuolation very extensive, and from observations on the living one would exj^ect that the lumen would be complete in a short time. In the other drawing, on the other hand, a part of the cytoplasm has become much more dense than the original angioblasts. This more deeply staining mass would appear slightly tinged with yellow in the living specimen, due to the presence of hemoglobin in the cells. In other words, it has become a blood-island. I have watched the formation of such a blood-island from the stage of solid angioblasts, as shown in figure 20, plate 4, so that I know it is not to be interpreted as a new outgrowth from endotheUum. The blood-island in figure 26, plate 6, is still attached to the wall by guy ropes of the original angioblasts, as well as by a solid base. In this manner exceedingly large and irregular blood-islands form. Such islands are not different in their behavior from those which arise along the walls of empty vessels from a division of the endothehal cells.

In figure 26, jilate 6, there is also another process to be seen, namely, that of differentiation of new angioblasts. To the right of the main vessel is a clump of two angioblasts wliich have not as yet joined the neighboring vessels nor begun to form a lumen of their own. These two cells were a little farther along the edge of the vessel than is shown in the drawing; they were shifted slightly in the drawing in order to make; them come into the same field. Since these cells are going to join the end of the aorta, they are evidence of the fact that the aorta difTerentiates in situ.

It is now necessary to describe how blood-islands form in vessels in which there has been a complete solution of the central cytoplasm, without any of the original angioblasts remaining except those that make an endothehal border for the vessel. In watching the living sj^ecimen the change in the apjiearance of the vessels after the lumina have formed is very striking. Wliile they are in the form of solitl angioblasts, it is the angioblasts that make the striking feature of a specimen, as shown in figure 20, plate 4. After the lumen has formed it is the exact reverse, for then the interspaces are conspicuous and simulate vessels. This, I think, will be clear from a careful examination of the photograph shown in figure 11, plate 3. In this figure there are pale rings in the area pellucida, showing especially well at about the middle of the area. These rings are the intersj^aces and the dull network between them are the vessels. The six sharply outUned spots near the undifferentiated mesoderm, i)osterior to the somites, are defects in the mesoderm similar to seen in figure 9. In figure 11 the blood-islands can be clearly seen in the gray bands, that is, in the lumina of the vessels.

In the Uving specimen the endotheUum of a fully formed vessel has the appear- ance of ground glass, as represented in figures 18 and 19, ])late 4. The contrast between the two stages, first, the interspaces between angioblasts, and second, the interspaces between blood-vessels, is brought out by com])aring figures 20 and 18, plate 4, Loth of which are taken from living specimens and represent the actual appearance as nearly as possible. The confusion in regard to determining the inter- spaces from the lumen of the vessels (which is inevitable to one looking at such a specimen for the first time) at a stage before there is any circulation is entirely eliminated after the blood-cells move in response to the beat of the heart. In fixed preparations the chance of confusion is not great.

In an area in which many new blood-islands are forming, one can often find unicellular blood-islands, two of which are shown in figure 18, plate 4. Again, such an island is shown in a section in figure 21, plate 5. One specimen which was grow- ing on the cover-shp had so many of these unicellular islands that there appeared to be almost a dupUcation of the endothelium. In such specimens, fixed just during the phase of cell division, I find the nuclear spindles placed perpendicular to the wall of the vessel, so that the inner cell which is going to make the blood-island projects directly into the lumen of the vessel at the very start. On the other hand, the nuclear spindles usually he in the plane of the endotheUal wall, as can be seen in any specimen in wliich the endotheUum is dividing to increase the lining of the vessels.

These small unicellular blood-islands develop a granular cytoplasm, as can be made out in figure 21, plate 5. Thus, from the evidence of the living specimen, the small masses of cells, even the unicellular ones, are properly called blood-islands, except for the fact that, strictly speaking, they are not islands at all, being alwaj's attached to the walls of vessels. Not only are the islands yellow, but the cells are uniformly yellow with hemoglobin, showing that all of them become erji:hroblasts. This was i^ointed out by O. Van der Stricht (1892) in a study of the development of the blood in the chick. He says (page 216) :

"Des le premier stade, toutes les cellules sanguines present ent un aspect particulier. Leur protoplasma est d'un jaune fonce plus compact que celui des elements voisuis. Chargees done d'une quantito d'hemoglobine plus ou moins considerable, elles ont, des leur origine, les caracteres du corpuscule rouge."

Tlic young islands, as seen in figure 18, plate 4, have a smooth contour, even a verj' definite border. Neither in the living nor in the fixed specimen can one see anj' cell outUnes within the mass. Nevertheless each nucleus has its own zone of cji;oplasm which closes in about it when the cells divide. Just as all of the angioblasts divide at the same time (fig. 24, plate 5), so all of the cells in all of the islands of a given area become highly refractive at the same time. This period of high refractivity lasts about an hour or more, and then the observer sees one nucleus after another pass into the metaphase with the chromosomes on the spindle. This is shown in the small drawing (fig. 18); the specimen was fixed just as the record was made on the drawing of the division of the nuclei, and the nuclei were found in the stiiined specimen in the metaphase. A photograph of this specimen is shown in figure 12, plate 3. Throughout their early stages the contours of the blood-islands are all round and smooth, except for a very interesting phenomenon which is often to be observed, namely, that deUcate sprouts of cytoplasm, exactly hke those by which the original angioblasts fuse to make a plexus, are put out from the islands. These sprouts creep along the inner wall of the vessels and attach themselves like guy ropes to the wall. They show a very marked tendency to join other neighboring blood-islands, and I have often seen sprouts from two islands (like the small island on the left of figure 18, plate 4, and the nearby unicellular island) join by sjjrouts which meet half way between the masses. In this way com- pound islands are formed, cells filUng in along the guy ropes until the whole mass becomes a single island. At other times these new sprouts simply seem to serve as extra guy ropes by which the islands are more firmly anchored to the wall. In the living chick the presence of a circulation in a given area does not interfere with the development of the blood-islands, though it is clear that the venous zone of the area vasculosa is less active in the jjroduction of new islands than the more jios- terior arterial zone when the circulation is later estabhshed. I have not noticed that a specimen showing a very marked tendency to form islands is one in which there is vigorous circulation in the area in question. The large compound islands often completely shut off the circulation in a given channel, and in the living speci- men I have observed that blood-cells which have been circulating get caught by the bridges crossing the lumen and become incorjiorated into a growing island.

This property of the blood-islands to send out the guy roj)es by which they become anchored in many places to the neighboring wall is duplicated by the Kupffer cells in the Uver. In sections of the liver of a rabbit that has received repeated doses of trypan-blue one can get an exact duphcation of the picture of numerous unicellular blood-islands seen in the young blastoderm of the chick. In examining sections of the liver which have been stained with carmine or any nuclear dye after repeated doses of tryi)an-blue, one will often find the nucleus of the parent endothelial cell behind a Kupffer cell exactly as that shown in figure 21, plate 5, in the unicellular island labeled B. i., from a chick of 17 .somites. Thus it seems to me clear that Kupffer cells do not make an endothelial lining of the capil- laries of the liver but constitute another generation of cells from the endothelial wall of capillaries, exactly hke the l)lood-islands in the embryo cliick; and that

they are anchored out into the lumen of the capillaries by numerous guy rojjes of cytoplasm. Instead of developing hemoglobin like the analogous cells of the embryo, they develop phagocytic powers to a high degree.

These sprouts of cytoplasm, proceeding from both the angioblasts and the cells of the blood-islands within the lumen of vessels, are referred to many times in the literature as evidence of amffboid activity. In reality, both types of cells have but a slight i^ower of movement from place to place, and the sprouts represent a rather- remarkable degree of attraction of similar cells. By means of these proce.sses isolated masses of angioblasts are brought into a plexus, compound blood-islands are formed, and blood-islands are, as it were, more securel}- anchored to the inner wall of a vessel.

In figure 18, plate 4, a certain number of er\-throblasts are seen in the lumina of the vessels. These cells, which have beceme free from the islands, continue to divide, even while they are circulating. These circulating cells make it quite easy to determine the lumen of a vessel in the hving cliick, and, after studying the blasto- derms in which the circulation has been established, one will never have any difficulty in identifying the lumen. In the specimen shown in figure 12, plate 3, blood was being pumped into the area of the omphalomesenteric arteries, which are just beginning to be indicated in the figure, but there was no movement of the blood in the area represented in figure 18, plate 4. In figure 19, plate 4, is shown a later stage in the formation of blood-islands. This was also drawn while the cells were in the phase of division, and hence each cell in the mass stands out indi\'idually. In the fixed specimen the nuclei of more than half of the cells are in the prophase of division. In the resting stage of these older islands the cells are no more definite in the center of the mass than is shown in figure 18, plate 4. The border of the islands, however, now displays the contours of the individual cells instead of the sharp, smooth contour of figure 18. If islands, such as the one shown in figure 19, plate 4, are watched in the living specimen one will see the cells (one after another) free themselves from the edges. This process is surprisingly slow. I have seen it take U hours for a single cell to become separated from an island.

These preparations give a good chance to test out the idea as to whether the jirimitive mesamoeboid cells are really amoeboid at all. The freeing of an individual cell from an island of course involves power on the part of the individual cells to move, but as seen in the living form tliis movement is very slow and not associated with much, if any, change in the shape of the cell. Moreover, a cell that has just become free from an island, provided there is no current fluid by which it can be carried away, will stay close to the island where it was formed, and one has to watch closely to detect the sUght changes in its contour. This is very different from the rapid changes characteristic of the wliite corpuscles. I am of the opinion that the marked, blunt processes which are found in specimens of erythroblasts of young chick embryos are associated with a reaction to the concentration of the fluid in which they are placed, because these processes can be much increased by simply increasing the concentration of the sodium chloride in the solution. I therefore conclude that the young erythroblasts are sensitive to osmotic pressure, and that

the younger and older cells of a gi^•en sjiecimen vary in their reaction to a given concentration of salt; but that the erythroblasts as a whole are characterized by their very slight power of amoeboid movement. Their contours change slightly, as seen in a Uving specimen, but the cells themselves move from place to place exceedingly slowly when not swept along in a current of fluid. They are elastic but not very motile.

In regard to the breaking uj) of the blood-i.sland.s, the only proces.s wliich I have actuiilly seen in the hving cliick is the slow freeing of individual cells from the edge of the mass, but so many fixed specimens look as if islands had been caught just as all the cells were going to break apart at once, that I can readily beUeve this does actually take place.

We have thus described the processes b}^ wliich two different structures develop out of the primitive mesoderm, the ccelom on the one hand and the blood-vessels on the other. We have shown that blood-vessels arise from cells called angioblasts, wlxich differentiate from mesoderm and produce endothelium, blood-plasma, and red blood-corpuscles. We have emphasized the importance of the destruction of cells in the production of the first blood-plasma and have shown that tliis plasma is different in origin from the tissue-fluid. Moreover, it has become clear that it is inadvisable to identify as blood-islands the primitive masses of mesoderm which are to give rise to both of these structures, for the reason that they must first si)lit into cells which will form two layers for the ccelom and those which will develop a different type of cytoplasm and form clumps of cells, the forerunners of vessels. Besides this, these original cells are not all the forerunners of blood-cells, but rather are masses which are to be further differentiated into those which form endothelium, with the potentiaUty of producing cells which can themselves make hemoglobin and those that become blood-islands. Since one can now distinguish all these types of cells, a more restricted terminology would seem to be of value. That is to saj% the original masses of cells in the blastoderm, long known as blood-islands, we might call by the general term primitive viesoderm, and distinguish three tj^pes of cells, i. e., (1) angioblasts; (2) the cells forming blood-islands (cells anchored to the endotheUal hning of vessels, which develop hemoglobin in their mass and which are derived either directly from angioblasts or from endothelium); and (3) primitive erythrol)lasts or cells wliich have become free from the islands but which go on dividing actively within the lumen of the vessels.

In his studies on Uving fish embryos Stockard (1915) did not find evidence that endothelium can produce blood-cells in that form. He says (p. 229) :

"There are numerous descriptions and illustrations of the origin of blood-cells from the vessel linings in the literature of the last twenty-five years, since Schmidt in 1S1>2 described the transformation of individual endothelial cells into white antl red l)lood- corpuscles. Yet again, I believe that the really skeptical reader will not be at all convinced that such a thing really ever takes i)lace, from the evidence jiresented in the literature, certainly not from any of the illustrations that have been made of this j)rocess. No real vascular endothelial cell has ever been actually observed to metamorphose into a blood-cell, or to divide off another cell which forms a blood-cell, and until such a direct observation is forthcoming one can only question the accuracy of the interpretation of the various obser- vations up to now recorded."

It is this exact observation that has now been made in the blastoderm of a chick by the application of the method of tissue-culture. One can actually see the division of an endothcUal cell into two daughter-cells, one of which remains in the wall of the vessel as a part of its endothelial hning, while the other protrudes into the lumen and becomes a unicellular island, the forerunner of a of erythro- bkists. That endothelium gives rise to erythroblasts may therefore be accepted as proved in the case of th(; chick. The stuches herein recorded do not include any observations of the origin of the wliite blood-cells, since there are no cells in the chick of the second day incubation that can be identified as the forerunners of the white cells, all of the cells in the islands and free in the vessels having hemoglobin.

In 1892 Schmidt described the origin of both red and white blood-cells from the endotheUum of the vessels of the Uver and spleen in human embryos. He described localized areas of mitosis in the endotheUum of the capillaries of the human Uver during development and interpreted them correctly as giving rise to clumps of blood-cells. These he interpreted as both red and white cells. In his own words he concludes (p. 220) :

"In der embryonalen Leber findet eine mit der Gefiissentwicklung im Zusammenhang stehende Neubildung weisser und rother Blutkorperchen statt. Die ersteren werden von den Endothelien der Capillaren durch karyokinetische Theilung producirt und pflanzen sich selbst durch Mitose welter fort. Die rothen entstehen aus den farblosen durch Auftreten von Haemoglobin im Protoplasma und besitzen ebenfalls die Fiihigkeit iiquivalenter Theilung durch Mitose."

Since that time many authois have given evidences of the origin of blood- cells from endothelium, both in birds and in manomals, as wiU be described later in connection with the origin of blood-cells from the endothelium of the aorta.

CYCLES IN THE DEVELOPMENT OF THE VASCULAR SYSTEM. It will be interesting to take up what I shall call the cycles in the development of the vascular system, and which I shaU subsequently show are due to successive cycles in cell di\asion. This subject is iUustrated in a series of photographs, plates 1 to 3. At the stage shown in figure 2, plate 1, when there is no head-fold, the mesoderm is almost undifferentiated. In the area peUucida there are two zones of mesoderm, an axial, dense mass, and a lateral, less dense zone. In the lateral part there is little indication of the zones which will ultimately divide into three parts; an anterior zone of the amino-cardiac vesicles, a middle zone which will ultimately Ue opposite the venous end of the heart, and in which the vitelUne veins will develop, and a posterior zone in which the omphalo-mesenteric arteries will appear. Over the area opaca the mesoderm is dense, and over the posterior area, especially, it is much mottled. It has already been brought out that the mass of mesoderm at this stage is but slightly differentiated into the cells that go to make up the exoccelom and those that become angioblasts. In the figure of a shghtly older stage given by Riickert (1906, fig. 880, p. 1210) can be seen the differentiation between angioblasts and the exoccelom, the faint rings around the dark spots repre- senting vesicles of the exoccelom beneath, that is dorsal to the angioblasts, if I inter- pret the drawing correctly.

Ill spite of the general view that angiobhists differentiate first in the most posterior zone of the area opaca, I am of the opinion that they tend to show first in a more anterior zone, not only in the area pcllucida but also in the area opaca. Ruckert describes the order as being the reverse in the two areas; that is, first in the posterior jiart of the area oi)aca and last in the ijosterior ])art of the area pellu- cida. Throughout the early development angioblasts are more massive in the posterior part of both the area opaca and the area pellucida, but in each area they are always a little farther advanced in the more anterior zone.

At the stage of 2 somites (as shown in fig. 3, plate 1) there is but Uttle change in the mesoderm of the area pellucida, but in the area opaca the great mass of primitive mesoderm has formed large vesicles, the forerunners of the primitive exocoelom. Opposite the middle region of the area opaca, esjiecially, are small masses of dense cells, the first angioblasts. As described by Ruckert, these tend to he opposite the walls of two adjacent vesicles of the exocoelom, so that they alter- nate with its cavities. I have other specimens of 2 somites which show a greater number of angioblasts in this middle zone, but still the same lack of differentiation in the posterior zone. In one specimen of 2 somites I can identifj- as angioblasts one or two clumps of cells in the area pellucida; again, specimens of 3 somites may how a few angioblasts in the area pelludica, but usually they are not well marked there until the chick has 4 somites, and I beheve are never abundant until the stage of 5 somites is reached.

The stage of 4 somites is especially interesting on account of a differentiation in the zones of the exocoelom. In figure 4, plate 1, from a chick with 4 somites, it will be seen that over the amnio-cardiac vesicles of the area pellucida, and also over the entire anterior half of the area opaca, the layers of cells which make the exocoelom have differentiated into large vesicles. In the posterior half of the area opaca this differentiation is only just beginning, as can be seen in figure 29, plate 6, in which it is clear that along the dorsal border of masses of mesoderm there are two layers of cells, the forerunners of the walls of the exocoelom; while the ventral, solid mass is now made up of angioblasts. Toward the end of the stage of 4 somites this differentiation becomes much more marked. The vesicles of the coelom become larger and the angioblasts more massive. In the area pellucida the zone of the amino-cardiac vesicles is well marked, as is also the middle zone of the exocoelom, where will be seen the dehcate plexus of mesoderm characteristic of tliis area. In this middle zone are a few angioblasts near the edge of the area opaca. This is the first zone of the area pellucida in which angioblasts aj^pear. The more posterior part of the area shows no differentiation of the mesoderm, except into the more dense axial zone. In the axial line the only hulication of the vascular system is a slight tliickening of the splanchnic mesoderm wliich marks the very l)eginning of the myocardium.

In the history of the vascular system the stage of 5 somites is very important and is shown in two photographs, figures 5 and 6, the former being the less devel- oped. On the left side all of the points are obscured by yolk, but the angioblasts are very clear on the right side. The entire area opaca is occupied by two structures,

the large vesicles of the coelom (which do not show in the photograph) and masses of solid ansioblasts. There has been no liquefaction of the angioblasts. The area pollucida has three or four masses of angioblasts in the zone of the amnio-cardiac vesicles, all of them lying opposite the borders of the vesicles of the coelom. Oppo- site the venous end of the heart in the mid-zone the angioblasts are too delicate to show at this magnification, but they are present, and some of them have even formed tiny vessels. In the posterior zone of the area pollucida the angiobla.sts are more massive and have already been shown in a drawing to illustrate their contrast to the co'lom (fig. 27, i)late 6). In the a.\ial Hne between the mycardium and the edge of the head-fold are a few tiny angioblasts, too small to be seen, which are the forerunners of the endocardium. These are shown in figure 1, in the text, at the stage of 6 somites. Between the myotomes a higher magnification shows the first angioblasts of the aorta.

In contrast to tliis specimen of 5 somites is one in figure 0, plate 2. In this blastoderm it is clear that the area opaca has now become di\ided into two zones, an outer and an inner. The two leaders on the left hand indicate the width of the zones. Over the whole outer zone the angioblasts have become transformed into vessels by the liquefaction of most of the angioblasts. This is illustrated in figure 22, plate 5, which is a drawing of the area outUned in figure 6, plate 2. In this drawing it will be seen that there are vessels on the left side forming a blood-island while on the right similar masses of cells lead directly over into a plexus of solid angioblasts wliich have not yet become vessels. The large blood-islands can be readily compared with the same mass in the photograph. In the specimen it can be noticed that this blood-island is perceptibly darker than the angiobla.sts just internal to it ; this is because the island has enough hemoglobin in its cells to be made out in the counterstain. By a study of the photograph it can be seen that a considerable amount of the original mass of angioblasts has been used up in the formation of the plasma in the outer zone; perhaps less than half of the original mass has remained in the form of blood-islands.

There is undoubtedly a considerable variation in regard to the time at which the angioblasts of the outer rim develop into vessels; in some of my specimens of 6 somites the change has not been made and Riickert (1906) states that vessels with lumina are to be found only in chicks with about 7 somites, and not an^-^N-here in the area vasculosa at the stage of 6 somites (p. 1224, fig. 890). In my specimen of 5 somites, on the other hand, the transformation of the entire mass of angio- blasts into vessels in the outer rim of the area opaca has taken place, and I judge that this is about the earUest stage in wliich one is Ukely to find the condition. As has been said, in this outer zone it is obvious that the amount of blood-islands left is certainlj' not more than about half of the original mass of angioblasts, indicating that a considerable amount of the original mass of angioblasts has been used up in the formation of endothelium and plasma. I beUeve that this is the usual condi- tion in the formation of the early vessels. The vessels will soon fill up almost com- pletely with new masses of blood-cells, but wherever I have observed the actual liquefaction of the angioblasts in living chick at least half of the cytoplasm has liquefiod (fig. 20, i)late 6).

In connection with a specimen such as that shown in figure G, jjlate 2, it is interesting to consider at what point in the development of the chick the hemoglobin appears. It is, I think, generally admitted that the first masses of cells, which were called blood-islands by the early embryologists, in the stages of 1, 2 and 3 somites have no hemoglobin at all. In the area pellucida in a living sjjccimen one can dis- tinguish hemoglobin by a slight yellow color under the microscoi)e, although the amount is too small to fi.x and stain. On the other hand, in the cells of the area opaca, which, in these hanging-drop preparations, must always be seen against a background of endodermal cells packed with yolk, one can not be so sure of detecting the first traces of hemoglobin by this color reaction. In specimens of 6 somites, however, in which there has been no liquefaction of the angioblasts to form blood- vessels in the outer rim of the area opaca, it is certain that one can see no traces of the 3'ellow color in the blastoderm in a hanging-drop preparation. In this con- nection it is interesting to note the work of two Russians on the time of appearance of the hemoglobin. These works I have not seen in the original but know only through a quotation by Ma.ximow (1909, p. 465) :

"So fand Smiechowski (1892) dass das Haemoglobin sich optisch und cliemisch erst in Huhnerembryonen nachweisen liisst, die schon 12 differenzierte Sogmente besitzen. Wulf (1897) der das Haemoglobin speziell mittelst des Spektrokops suchte, fand die ersten Spuren erst beim Hiihnerembryo mit 6 Segmenten, wahrend das voile Haemoglobinspectnun erst mit 9 Paar Segmenten erschien."

The time at which hemoglobin appears, in terms of the number of somites, probably varies within wide limits, but the evidence all tends to indicate that it occurs after the time when the vessels differentiate out of angioblasts, and that hemoglobin is not formed until blood-plasma is produced from the hquefaction of some of the protoplasm of angioblasts. This point seems to me to emphasize again the disadvantage of calUng the masses of primitive mesoderm blood-islands. All of this evidence is in harmony with the point of view of Madame Danchakoff, now accepted, that in the chick red blood-cells form only within the lumen of a vessel, and suggests at least that the production of some plasma by the hquefaction of certain cells has something to do with the capacity of the cells in the islands to develop hemoglobin. Recently, Madame Danchakoff (1918) has brought forth evidence of the power of endotheUum in the chick to produce red cells, even in zones where the endotheUum has been so injured that it does not completely inclose a lumen. These observations are especially interesting in connection with the origin of red cells in mammals. It is generally beUeved that hemoglobin-bearing cells in mammals develoj) outside of vessels, which would indicate that the theory concerning the importance of endothehum and of plasma in the production of red cells can not be generally ai)i)lied.

According to Maximow (1909), there are two different types of red blood-cells in mammals: "Sehr merkwiirdig ist die Tatsache, dass es beim Siiugetier zwei so scharf gcschiedene Typen von roten Blutzellen gibt; die primitiven und die

definitiven oder sekundaren" (p. 489). He described the first or primitive blood- cells 11.S arising witliin the vessels in connection with endothelium, like those of l)irds. These primitive (Tvthroblasts he believes die out completely and are replaced by secondary red cells, which arise from indifferent mesenchyme cells outside of vessels. These mesenchyme cells become Ijinphocytes, which in turn give rise to erj^hro- blasts (p. 547).

In the article of ]Mollier (1909), in which he brings out the idea of the extra- vascular origin of erythroblasts in the developing Uver, certain figures (e. g., figs. 7 and 8) can be interpreted easily as a solid mass of angioblasts in wliich the process of Uquefaction is going on with the production of hemoglobin-bearing cells. Indeed, this interpretation also fits in with his conclusions (p. 519) :

"In der embryonalcn Leber der Siiugetiere werden Blutzellen gebildet aus einem indiflferenten Material', dem Reticulum, das vom \'isceralen Blatt des Mesoderms gebildet sich zu P]ndothelien, Blutzellen und Stutzgewehe differenziert. Die Blutzellenbildung erfolgt ausscrhalb dor Gefiisslichtung im Reticulum. Die blutbildenden Gefa.ssanlagen haben alio rcticulfire Wand. Die Blutzellen wandern nicht selbstiindig durch geschlossenes Endothel in die Blutbahn ein, es reisst das Kndothel zu diesem Zwecke auch nicht ein, sondern es bleibt die retikuliire Gefiisswand solange bestehen, als die Blutbildung anhalt."

In studying sections of the developing Uver in very young pig embrj'os, it seems to me that one can identify solid masses of cells between the columns of Uver cells as analogous to the soUd masses of angioblasts to be seen in blastoderms of the chick. These arc the solid, blood-forming capillaries in the sense of O. Van der 8tricht. Thus, in view of the great discrepancy in regard to the formation of erythroblasts, namely, that in birds they form within the lumen of vessels and in mammals outside of the lumen, it seems important to test the identification of angioblasts in the case of mammals, especially in connection with Mollier's observa- tion that the waUs of vessels in the liver remain reticular as long as blood is being produced there. The question is, therefore, can it be shown for mammals that ceUs which may be identified as angioblasts produce erythroblasts, and that associated with the process there is a certain destruction, Uquefaction, or vacuolation (see MolUer's text-figures 7 and 8) of cytoplasm, such as can be determined for birds?

To return to the specimen shown in figure 6, plate 2, the area peUucida shows three zones: anterior, middle, and posterior. These zones are more striking stiU in the other photographs on plate 2, especially in figure 7. Over the anterior zone the vesicles of the amnio-cardiac cavities have become very large, with fine, sharp boundaries. Besides these boundaries there are numerous smaU, isolated clumps of angioblasts. The photograph does not enable one to distinguish these two structures, but they would not be confused in the specimen. Over this area angio- blasts are much more scanty than farther posterior, and hence it always happens that they remain very much longer as isolated masses. This is simply due to the fact that the distances between the masses are greater and hence it takes longer for the sprouts to bridge the gaps. WTien the vacuolation occurs in these isolated clumps there results the formation of isolated vesicles which may be found any- where in the area peUucida, but more commonly in this region. They were noted in


this area by McWhorter and Whipijlc in their study on the liviiifi; cliick blastoderm. These authors studied them from the standpoint of tlie orifjin of the vessels from tissue-spaees. They are, on the other hand, formed from anj^ioblasts by the licjue- faction of the center of the mass and will join the main mass of the vessels by the process of sprouting. At the stage of 5 somites these masses of angioblasts are all soUd. Over the middle zone of the area pellucida, which will become the area of the vitelhne veins, there is an extensive mass of solid angioblasts, becoming less extensive farther posteriorly. This illustrates that the middle zone opposite the venous area of the heart is always the precocious area in the development of the vascular system in the chick. The myocardium is just indicated in this specimen, and a verj^ few angioblasts, which are the first forerunners of the heart, maj^ be seen between the myocardium and the endoderm, both in the total specimen in figure 6, and in sections from the same stage. Their position is incUcated in the diagram of figure 1 in the text, from a chick of 6 somites. No angioblasts along the margin of the somites in the position of the future aorta can be made out in this particular specimen, but they could be in the other specimens of this stage.

Thus, to sum up the stage of 5 somites, it is important as showing the first blood-vessels from the primitive angioblasts, making an outer rim of vessels over the entire area opaca, the forerunner of the marginal sinus. It marks thus the stage of the first blood-plasma and of the first erythroblasts which can be identified in the form of blood-islands in these primitive vessels. It is the stage in which angioblasts are first found in any great numbers in the area pellucida, as well as in the stage in which the first angioblasts can be seen in the axial line, constituting the forerunners of the endocardium and the aorta.

During the stages of 6, 7, and 8 somites very interesting changes take place. These are illustrated in a blastoderm with 7 somites in figure 7. In the area opaca the blood-vessels of the outer rim have developed great masses of blood-cells. These are no longer in the form of blood-islands attached to the wall, but have become free cells which in many i:)laces, and especially at the outer margin, com- pletely fill the lumina of the vessels. In the inner margin of the area opaca, on the other hand, the angioblasts are just beginning to pass into the stage of liquefaction to form vessels.

The area pellucida likewise shows very interesting features. Over the amino- cardiac vesicles, especially of the left side, can be seen fine parallel lines of angio- blasts, which are also characteristic of this area. have been retouched in the photogra[)h. They are on the ventral surface of the amino-cardiac vesick^s and are the forerunners of the anterior veins of Poj)off. They are especially interesting because one can always find in them examples of the Ucjuefaction of the cytoplasm in single chains of angioblasts, as shown in figure 25, plate 5, which seems to me to be the best jiroof that the lumen of the vessels may be considered as intracellular. This specimen contains a number of isolated clumps of angiolilasts over the dorsal surface of the amino-cardiac ve.sicles and even extending into the middle zone of the area pellucida. These clumps of angioblasts dorsal to the somat{)j)leure, which are particularly abundant in this specimen, are constant. They are shown in Lillie's figure 08 B (1908) from a cliick of 10 somites, and in Van der Stricht's figure 3 (1895, p. 210). In this specimen (fig. 25, plate 5) many of the clumps have become tiny vesicles bj^ the Hquefaction of their centers, and several show a few red blood- corpuscles. These vesicles over the somatopleure remain isolated for a long time, simpl)' because they are few and far apart. I have a specimen of the fourth day of incubation, grown on a cover-slip, which has as many as 10 or 12 of them. Some of them, I think, are degenerate, but it soon becomes difficult to follow them in the living specimens on account of the increasing thickness of the blastoderms. Their especial interest Ues in noting the early appearance of vessels in the somatopleure in the chick.

In the middle or venous zone of the area pellucida the angioblasts of this specimen have become vessels. They form a delicate plexus and the vessels are consequently small and inconsjMcuous in the photograph. In the living chick, how- ever, the process of hquefaction can be easily followed. Most of the vessels of this area are emjity, that is as far as cells are concerned, but here and there are a few small clumps of red-blood cells, showing that the angioblasts have the potentiaUty to produce cells bearing hemoglobin. The posterior zone of the area pellucida is especially interesting in this specimen, as it happens to be one in which all of the angioblasts exist in the form of isolated masses of cells. There are numerous deh- cate sprouts from these masses, but for the most part these have not yet joined similar masses of cells.

Text-figi'RE 1. — Diagram showing the position of the angioblasts which are the forerunners of the endocardium in a chick (No. 206) with 6 somites, incubated for 24 hours and .SO minutes and then grown in Locke-Lewis solution in which there was 1.0.5 per cent XaCl. The diagram shows the actual masses of angioblasts making up the endocardium, and is to be compared with the photograph on plate 1, figure 5. X135. A, angioblasts of the endocardiiun; En., line of the endoderm; M., myocardium.

The stage of 6 somites is the best for studj-ing the differentiation of the heart and aorta from angioblasts. This stage has been extremely well described by Wilhams (1910-11). Figure 1 in the text shows the edge of the head-fold, and the margin of the myocardium just before it fuses with the sj-mmetrical fold of the opposite side, while between the two are the early chains of angioblasts with tiny vesicles wliich are destined to make the endocardium of the heart. This figure corresponds to the description of the heart of the Selachian given by 0. Van der Stricht (1896). The endocardium continues to receive new angioblasts from the zone of the myocardium, certainly throughout the second day of incubation. These can readily be seen in total mounts of chicks with 17 to 20 somites spanning the wide gap between the myocardium and the endocardium, which is so characteristic of these early stages. The wide gap is not due to skrinkage, because it can be easily

determined with the focusing screw in the living specimen. The endocardium, then, is constantly increased in two ways: first by the division of its cells, and second by the addition of new angioblasts to the outside.

The aorta in tlie stage of 6 somites looks much like the endocardium shown in text-figure 1, except that the clumps of angioblasts are more isolated. During the stages from 6 to 9 somites the angioblasts along the axial line unite to become a complete vessel. The liquefaction of angioblasts of the dorsal aorta opposite the lower somites can be followed in clucks of 8 to 9 somites, and it is very interesting to note that if a chick of about 9 somites is watched one will usualh' find a few small blood-islands giving rise to cry throblasts in the lower end of the aorta. At this stage only small islands are formed there, never larger than two or three cells and these from the original angioblasts rather than from a new i)r()liferati()n of the endotliolium.

That the endothelium does proliferate later in the lower aorta to form blood- islands has now been abundantly proved. This was first discovered by Madame Danchakofi" (1907) in a study of the development of blood in the chick. She described that in chicks of from 4 to 5 days of incubation there was an intensive growth of the endotheUum of the vessels, especially great in the lower aorta, giving clumps of .young indifferent elements Uke the primitive blood-islands. These masses of cells she described as becoming free both within and without the lumina of the vessels, and as giving rise to both red and white blood-cells. In 1909 Maxi- mow described these masses of cells in the aorta of the cat and the rabbit (p. 517), while in 1911 and 1912 they were described by Minot in the hinnan embryo. The question brought uj) by the latter author as to whether these masses of cells or blood-islands could be proved to be derived from endotheUum has, I tliink, been entirely settled by these observations in the Uving form. The blood-islands in the aorta have been described more recently by Emmel (1915-1916) in other mammals, especially in pig embryos, and by Jordan (1916-1917) in pig and mongoose embryos. All of these observers describe the islands as giving rise to a primitive mesamoeboid cell capable of producing both red and white corpuscles. This point seems to be now the most significant (question in regard to these structures. It is generally admitted that in the chick the red cells develop witliiu the lumina of vessels, and I think these observations on the living form make it possible to sharpen this con- ception by stating that red cells in the chick during the first two days of incubation come either from the original angioblasts or from the subsequent division of endothe- lium, and that the develojjment of hemoglobin in some cells within the vessels is preceded by the hciuefaction of some protojilasm to make i)lasma. Moreover, it may be said that all of the cells that become free within the lumina of the vessels in the first two days of incubation become red cells. This, of, does not include certain wandering cells from the endodcrm, or germ cells which may get into the ves- sels. That all of the primitive blood-cells are erythroblasts was pointed out by O. Van d(;r Stricht in 1892 for the chick and in 1895 for the ral)bit. Thus the red cells develop intravascularly in the chick because it is angioblasts that give rise to them. The proliferation of endotheUum to make blood-islands, now abundantly proved for the chick and shown to occur in mammals, becomes especially interesting

in connection with the question of the ultimate fate of these cells. All of the cells in these islands in the chick of 2 days of incubation become erythroblasts. All of the observers of this process in mammals agree in considering these masses to be made of primitive hemoblasts or mesamcrboid cells, in the sense of Minot, or primitive lymphocytes in the of Maximow; that is, capable of jiroducing both red cells and lymphocytes. That they produce erythrocytes is generally conceded, and the evidence that they also produce white corpuscles, that is IjTnphocytes, is that in many of them none of the signs of hemoglobin can be made out.

The next phases in the development of the vascular system are brought out in two photographs from chicks with 11 somites, figures 8 and 9, plate 2. In both of these figures there is a massive plexus of solid angioblasts in the posterior part of the area jjcllucida, which is characteristic of this stage. The blastoderm in figure 8, plate 2, had 10 somites when the specimen was taken from the shell, and the heart was just twitcliing. It shows a great advance over the stage of 7 .somites, for the myocardium has fused across the mid-Une, the heart has become a vessel, and con- trary to the usual form has curved to the left side of the embryo (right side of the photograph). The specimen was fixed during the phase of cell-division for the angioblasts, as is show-n in figure 24, plate 5. In this specimen the two .sides are not s>Tnmetrical in regard to the area opaca. On the right side it will be noticed that there is not a very sharp contrast between the outer and inner rims of the posterior part of the area opaca. This is because the outer rim is full of blood-cells and the inner rim is still made of solid angioblasts. On the left side of the figure there is a part of the inner rim where the angioblasts have become vessels and almost all of the angioblasts have liquefied, the small dark spots representing tiny blood-islands. Figure 9, plate 2, is a photograph of a cliick of the same stage, in which all of the angioblasts of the inner rim of the area opaca have hquefied. Tliis specimen also had 10 somites when taken from the shell, and the heart was just twitching; it stopped beating very soon, but the cells continued to hve, as they were in process of division when the blastoderm was fixed. The specimen has some abnormalities of the brain and several defects in the mesoderm, but it is very striking for the mass of angioblasts in the posterior part of the area i)ellucida and for the almost complete hquefaction of the angioblasts of the inner rim of the area opaca. From the study of these h\ing forms, I believe tliis verj' complete hquefaction of the angioblastsis the rule rather than the exception — that it to say, it is the rule in the formation of the vessels in these early stages.

To return to figure 8, plate 2, vessels have formed throughout the area pellucida, except in the posterior part, but they are so delicate as not to show in the photograph. They are empty except for a few free blood-cells wliich have formed in the area since there has been no circulation at this stage. These corpuscles may oscillate back and forth with the beating of the heart, but are not moved through the heart until about the stage of 16 or 17 somites, when the circulation begins. A study of this figure will show that if a transverse section be taken through the posterior part of such an embryo one can see all the processes in the formation of blood-vessels and of


blood-cells in one section. Thus, in the photograph of such ii section as that shown in figure 17, it is plain that the vessels of the outer rim, constituting the sinus terniinalis, are well formed and contain free blood-cells. Just internal are vessels with blood-islands attached, while in tliis section the angiol)lasts just at the border of the area opaca are in process of liciuefaction to form vessels. (Tliis zone, outlined in the photograph, is shown in i)late 6, fig. 28.) Witliin the area pellucida arc tiny clumps of sohd angioblasts that have not yet begun to liquefy.

Thus we have seen that at the stage of 4 somites the area opaca contains soUd masses of angioblasts. At the stages of 5 to 7 somites these angioblasts may be forming vessels in the outer rim of tliis area, while the stage of from 7 to 1 1 somites will show this outer rim of vessels filUng up with blood-cells and the angioblasts of the iimer rim of the area opaca htjuefying to form vessels. In the posterior part of the area pellucida, on the other hand, the stages of 5 to 11 somites are marked by the active production of new angioblasts, while at the stages of 6 to 9 somites vessels in the anterior zone can be seen to form from angioblasts.

The next specimen in the series (fig. 10, plate 3) is from a cliick of 14 somites and is evidently quite characteristic for this stage, as it is so much Ukc Lillie's figure 45 (1908, p. 88) and Rlickert's figure 886 (1906, p. 1214). This specimen repre- sents a stage before the circulation has begun, although the heart was beating vigorously. The entire area opaca is covered with well-formed vessels. In the outer half the terminal sinus, as can be plainly seen, is quite filled up with free erythro- blasts, more so in my figure than in the other two. The blood in this peripheral zone is now readj^ to be moved forward into the venous end of the area vasculosa, which is indeed the first movement of the blood after the circulation begins. The inner margin of the area opaca, on the other hand, shows vessels well formed and practically devoid of free erythroblasts, but with numerous young, attached blood- islands. The actual vessels show better in figure 11, plate 3, from another chick of the same stage. They also show in the same manner in the drawings given by Ruckert (1906, figure 886,) and Lillie (1908, figure 45), where the interspaces are pale rings and the vessels are a gray network. This is just the way the vessels themselves appear in the specimens, and at first sight almost everyone takes the interspaces to be the lumina of the vessels — not only in a living blastoderm, but also in the fixed specimen. Having once seen the blood circulating in the vessels, however, one will not be confused on this point. Clinging to the walls of the vessels and projecting into the lumina are numerous small islands. In all four figures these extend well into the region opposite; the venous end of the heart.

In figure 10, jjlate 3, throughout the area pellucida the vessels are well formed except in the posterior region, where they are still solid angioblasts; this is also true of Lillie's and Riickert's figures just mentioned. In none of the three figures are these angioblasts conspicuous, a fact which I interpret to mean that all three speci- mens happened to be fixed during the resting phase. In my own, I know this to be the case. In this sj^ecimen li(iuofaction is just beginning in the margin of the pos- terior rim between the area pellucida and the area opaca. In figure 11, |)late 3, on the other hand, tliis posterior zone has well-formed vessels except over the undifferentiated zone of axial mesoderm, where vessels are forming. The transi- tion i.s shown in two drawings, figures 23, plate 5, and figure 26, plate 6.

I have given two j)hotographs of blastoderms after the circulation has begun. The first is from a cliick with 17 somites (figure 12, plate 3), the second from one with 18 somites (figure 13, plate 3). Both show that the ma-ss of blood in the outer rim of the area opaca is drawn forward into the veins by the beat of the heart. In figure 12, plate 3, there is a new generation of tiny blood-islands just beginning in this depleted area. Such new islands are also shown in figure 21, plate 5, frf)m a section through the marginal vein in a specimen of the same stage, i. e., 17 somites. Thej' show very interestingly that a new cycle of blood-islands Ls beginning in the outer area; this means that a wave of blood-islands will again sweep across the area vasculosa. In both of the specimens there are masses of free blood in the inner rim of the area opaca. These are due not alone to an old generation of blood- islands just breaking up, but also to the fact that the omphalo-mesenteric arteries, which are clearly seen in the older specimen (figure 13, i)late 3), there is some abnormal heaping up of red cells in these cliicks that are grown on cover-shps. Both specimens show tliat now the jiosterior zone of the area pellucida has vessels throughout the outer rim, the new angioblasts being in the axial zone the undifferentiated mesoderm. These new angioblasts belong to the posterior end of the aorta. In the vessels that lie in the arch of the area pellucida posterior to the omphalo-mesenteric arteries are to be seen the blood-i.slands. Figure 13, plate 3, shows about the maximum number of islands wliich I have found in this area. This blastoderm has certain especially interesting points: (1) That there are numer- ous islands in the arteries in direct Hne with the active circulation, and (2) that in this particular specimen the zone of new islands extends far f(jrward into the venous portion of the area pellucida.

From the figures representing stages with 11, 14, 17, and 18 somites, it is clear, I think, that any section through the lower end of the spinal cord in any of these stages would show all of the processes by which blood-vessels and blood-cells are formed. Thus, if a section were taken through the lower end of the central nervous system in figure 12, plate 3, one would see three generations of blood-islands. In the vessels of the marginal sinus one would find little new islands with two or three cells, hke those in figure 21, plate 5; in the inner rim of the area opaca old islands just about to break up, and in the area pellucida islands intermediate between these two extremes. Thus it is clear that after the vessels have formed in the outer part of the posterior zone of the area pellucida, as at the stage of 17 somites, one wave after another of blood-islands sweeps across the area vasculosa, beginning with the outer margin, so that one can find zones of young islands alternating with zones of old ones. In one of my specimens, in which one of these cycles of blood formation was just beginning, so many new unicellular islands were forming in the area pellucida that it seemed ahnost as if there was an entire duphcation of the endothelium. Again, in some of the specimens of 17 somites a cycle of the islands has just been completed and the walls of the vessels are almost bare of islands. If one considers the posterior half of the area vasculosa during the second day of incu-


bation four zones arc to be made out : an outer and an inner zone of the area opaca and an outer and axial zone of the area pelhieida. In the outer and the inner zones of the area opaca one will always find two different generations of blood-islands; if one zone has young islands the other will have an older generation . In the area l^ellucida, during these stages, the observer can follow one generation of islands after another in the vessels of the posterior arch, while the posterior axial zone in the stages which we are considering continues to be an area for the differentiation of new angiolilasts with but small contributions to the number of red cells.

At the stage when the omphalo-mesenteric arteries are forming, figure 13, plate 3, another factor must be recognized in the Uving specimen, namely, that along the vessels the mesenchyme cells begin to form in chains along the outer wall which must be distinguished from the endothelium. These cells rej^resent the addi- tion of mesenchyme to the wall of the vessel in the process of development from a capillary into an arteiy. The}' are the forerunners of the media and adventia. In the living chicks of 2 daj's of incubation I have never observed any indication of an endothehal cell leaving the wall of a vessel, a condition described for later stages by Madame Danchakoff (1909) in connection with the formation of blood-cells.


These cjdes in the development of the vascular system are dependent on the fact of cycles in cell division. In these living specimens it has become clear that, in the case of three tissues at least, there are definite cycles of cell division, i. e., in the nervous system, in the endoderm, and in the vascular sj'stem. I have not yet studied the ectoderm in this regard, except in the nervous system, and have not found the process so easy to follow in the mesoderm hning the coelom. In the case of the endoderm attention has already been called to the fact that when the entire endoderm passes into the refractive state of the cytoplasm which precedes cell divi.sion, the specimen can not be studied for vessels. If such a specimen is fixed at once nothing in the stained preparation will indicate that the cells were about to divide; but if fixation is delayed until an occasional nucleus can be seen in the metaphase, it will be observed that nearly all of the nuclei in the endoderm show division. I have many specimens showing this cycle of division in the endo- derm and in the nervous system. The phenomenon is illustrated in connection with the blood-islands in figures 18 and 19, plate 4. In such specimens as the one shown in figure 18, plate 4, if one watches the living island until all of the cells have divided, the entire island will appear to be of the same size as at the beginning, but every cell will be half as large as it was originally. This appearance proves that all of the cells divide in one cycle, notwithstanding the fact that every nucleus does not show a sjjindle at exactly the same moment. It shows strildngly also that growth (specifically increase in size) occurs in the phases between cell division. Moreover, if one takes the zones of develojjment which have just been described, namely, the outer ancHnner zone of the area opaca and the outer and axial zone of the ])osterior part of the area jjellucida, all of the cells — either of angioblasts or of blood-islands, whichever happen to be present in any one of these four zones — will be at the same


phase; that is to say, all dividing, resting, or undergoing Uquefaction at the same time. I can not make this out to be true of the endothehum after the vessels have formed. In other words, there is not sufficient change in the appearance of the cytoplasm of the finished endothelium in the hving specimen to indicate whether the cells are going to divide or not, and in fixed specimens one finds only scattered nuclei with mitotic figures. However, a specimen in which there are any endotheUal nuclei with figures will show many of them. Therefore, recognition of this process of cycles in cell division depends on finding the changes in the cytoplasm which precede the nuclear changes in the living form.

That erythroblasts keep on chviding in cycles after they are free from the islands is suggested by identifying several stages of nuclear figures and young cells half the normal size in the circulating blood; that is to say, if one finds one nucleus in the metaphase there will be many in the same phase; or one may find groups in the metaphase and other groups in the prophase. They serve to emphasize the fact that in an embryo after the stage of 5 somites there are two sources of red blood- cells, the endothehal cells of the vessels and free erythroblasts.

The question as to whether or not the same marked cycles of cell division can be made out for the mesoderm, is an interesting one. In many of the early prepara- tions nearly every cell in the myotomes or very extensive masses of the dense axial mesoderm may be found in division; but I have no specimens proving that the entire mesoderm of the embryo or of the membranes di\'ides at one time, as is the case for the endoderm and the angioblasts. In later stages, when angioblasts are differentiating in large numbers, there may be some difficulty in distinguishing a single angioblast in division from a mesodermal cell in division. This is due to the fact that the cytoplasm of the mesodermal cell rounds up around its nucleus during the phase of di\'ision and shows also some increase m density, so that it maj^ simu- late the angioblasts. There is no difficulty in detecting a fuUy differentiated, resting angioblast, nor any clump of two or more angioblasts. A specimen such as that shown in figure 9, plate 2, for example, has a very large number of single dividing ceUs near dense clumps of angioblasts, which I interpret as cells which are just becoming angioblasts, though it may be admitted that in watching such a hving specimen in wliich there is a question of the differentiation of large numbers of new unicellular angioblasts, the final proof of the nature of the cells would be their behavior after they had finished their first phase of cell-division. Had the cji;o- plasm remained rounded up, and the new ceUs remained together, they would soon join a neighboring band of angioblasts; on the other hand, had the two cells separ- ated and put out the deUcate exoplasm characteristic of mesoderm, the cells would still be undifferentiated.

A question of great importance is whether or not anj' red blood-cells can be seen to differentiate outside the lumen of a vessel. All of mj' eAidence tends to show that in these stages the red cells develop only within the vessels. This is in entire agreement with the view of ^Madame Danchakoff (1909), who has shown that in the chick the development of the red cells is intravascular, while that of the granular, white corpuscles is extravascular. In the hving blastoderm the only


specimens in which the matter could be called into question is such a one as that just described, in which there are many cells that I term unicellular angioblasts. These can always be found at the stage of 11 somites in the posterior jxirt of the area pellucida. When such a blastoderm is fixed one might bring up the cjuestion as to whether any of the cells outside the vessels might not contain hemoglobin, but I think that the answer is negative, on the ground that at this stage no hemoglobin-bearing cells are forming within the vessels of tliis area. The question could be definitely settled by more careful records than I have yet made of the evidences of hemoglobin in the li\ing cells.

In later stages, on the other hand, after a considerable mass of blood has been formed, the presence of true erythroblasts in the interspaces is a very common occurrence in all of the specimens, just as it is a famiUar phenomenon in all sections of embryos. I interpret these cells as having escaped from the lumina of the vessels, just as they have been interpreted in the study of sections, and for the same reasons. In the first place, there must be some rupturing of the delicate walls of endotheUum in the mounting; often the specimens must be shaken a little in the solution to free them from the vitelUne membrane, and it is hardly possible that any artificial medium would not make some change in these thin membranes. Then, just as Madame Danchakoff (1909, p. 125) finds in sections, I find in the living specimens that these cells tend to degenerate. Moreover, I have never observed them to move toward the vessels, indicating that, like the erythroblasts within the vessels, they have but Uttle motiUty. The question as to whether any red cells of the chick ever differentiate outside the lumen of a vessel is one of very considerable importance on account of the theory that they do so differentiate in mammals. So far the evidence in the study of these living forms is that in the chick erythroblasts differentiate only within the lumen of a vessel. I have made but few pecimens of 3 or 4 days of incubation. The method can be applied up to the fourth day, and in one such specimen there is an isolated vesicle near one of the branches of the artery which is packed with erythroblasts, showing that the formation of new angioblasts giving rise to vesicles is going on, and again emphasizing the intravascular forma- tion of the red cells.

Another point of significance, which these observations on the living specimens sec^ms to me to settle definitely, is that all of the cells of the blood-islands of the chick during the first two days of incubation become red cells; that is, they all develoj) hemoglobin. This is very striking in the living chick in which every single cell of an island can be seen to be j'ellow, and all of tlie cells of a given island are uniformly yellow. In fact, all of the cells of these islands appear alike except for an occasional dead cell among them. These, as has been said, react to neutral red, and in fixed preparations show picnotic and fragmented nuclei. Thus the only cells which by any chance could be confused with white blood-cells can be jjroved to be dead cells. The granular corj)uscles can be id(Mitified in the l1lood-^•essels in the hving chick at the stage of the fourth day of incubation and their origin must therefore be taken up in later stages.



It may be interesting to sum up what can be seen of the origin and develop- ment of the heart and the aorta in these hving blastoderms. As has been shown, it is possible to find the first angioblasts that differentiate from mesoderm in the wall of the mj'ocardium before the two mj'ocardia have fused the mid-Une. These cells appear first at the stage of 5 somites. At the same time isolated clumps of cells appear along the ventro-lateral margin of the somites, the forerunners of the aorta. These clumps of isolated angioblasts Uquefy to form vesicles, which grad- ually unite to form the endocardium and the aorta. By the time the chick has 9 somites a complete aorta can be seen along the ventro-lateral margin of the somites, but this vessel is constantly increased by the addition of new angioblasts. The stage of 9 somites is especiallj^ favorable for watching this process. Indeed, I have found it rare that a chick of that stage does not show a few clumps of angioblasts joining the aorta along its mesial border opposite the last two somites. Moreover, practically all chicks with from 14 to 17 somites wdll show new angioblasts joining the aorta opposite the undifferentiated mesoderm at the lower end of the embrj'o. Therefore, any illustrations of an injection of the lower end of the aorta (such as are shown in Evans's [1909] fig. 1) should have added to them a few masses of solid angioblasts about to join the wall of the plexus in order to completely' represent the aorta in that area. It is interesting to note that in this area, when a mass of angioblasts joins the wall of a vessel, sprouts from the older vessel and from the 3'oung masses of cells meet half way between the two structures, and then the new cells seem to be gradually drawn into the vessel wall. On the other hand, in the growth of a plexus the new cells may remain at some distance from the older vessel, while the protoplasmic bridge develops into a vessel connecting the two lumina. This is a different process from the drawing of new ceUs into the wall of a vessel, such as can be seen so readily in the case of the endocardium and the wall of the aorta.

The study of the heart in these specimens is interesting. The early stages have already been mentioned. There are at first isolated clumps of angioblasts which form tiny vesicles, as shown in text-figure 1. These vesicles then unite to make a plexus. The development of the later stages must be watched through the myocar- dium and hence it can not be seen as clearl}' as the vessels themselves. After the endocardium has formed a complete vesicle the heart curves usually to the right, by an occasional anomaly to the left, as seen in figure 8. The very first beats of the heart can be made out in these hanging-drop specimens. They occur at the stage of 10 somites and always in the same position. The first twitching is along the right margin (left side of the photograph), beginning just at the lower border above the junction and between the viteUine vein and the ventricle. It is interesting to note that there is no movement whatever in the vein, the entire twitching being confined to the ventricle proper. That is to say, the myocardium over the ventricle is formed long before there is any muscle at all in the wall of the veins. The beat is at first slow but rhA,i;hmical, and gradually involves the entire wall of the ventricle, spreading from the posterior to the anterior end. The heart beats throughout the stages of from 10 to about 16 somites without actually pumping any blood into the aorta. At the stage of 16 to 17 somites circulation begins.



These studies have shown that throuffhout the first two days of incubation there is a continual differentiation of mesenchimal cells into angioblasts. In the chick these cells arise first in the embryonic membranes, but bj^ the stage of 5 somites they begin to differentiate actively in the embryo itself and both the endocardium and aorta can be seen to differentiate in situ. Moreover, angioblasts continue to differentiate in the wall of the yolk-sac during the third and fourth days of incubation.

The characteristics of these new angioblasts have been made sufficiently sharp so that they can be identified in sections. In this way we can be sure that not only the aorta but the primitive vessels of the embryo, such as the vessels along the nervous system and the cardinal veins, differentiate in situ. These studies must leave open the question as to the time, if ever, when angioblasts cease entirely to differentiate and all of the growth comes to be from the walls of previously formed vessels, as was observed by Clark (1909) for the tadpole's tail. The place at which to attack this problem again seems to me to be in further studies on the regeneration of vessels in healing wounds.

In these observations on the living blastoderm of the chick it has been shown that it is possible to make a sharper distinction between the cells wliich cUfferentiate to form the ccelom and those which form the vascular system. It therefore seems better to use the term primitive mesoderm for the masses of cells which will give rise to both structures. Indeed, the relationships can be made quite clear by limiting the term blood-island to those masses of cells that actually develop hemoglobin and become erythroblasts, while the indifferent masses, which will give rise to mesoderm on the one hand and angioblasts on the other, are given a less specific name.

The exocoelom forms by a splitting apart of the two layers of cells which come from the primitive mesoderm. Blood-vessels, on the other hand, arise by the differentiation of a new type of cell from this same primitive mesoderm. This has a different cytoplasm from the original mesodermal cell, is more granular, more basophihc, and has new qualities, namely, the tendency to form solid masses which appear Uke a syncytium, the centers of

which hquefy to form blood-plasma, and a Undifferentiated mesenchyme cell

marked tendency to put out delicate ^^AngioWasi

sprouts by wliich it joins similar masses. Endotheliu;r^d<^lasma ^od-isla.,d(erylhroblart)

These cells, or angioblasts, give rise to endo- \"02:45, 27 March 2012 (EST)~--— -^

,,,. ,1 I'l 1 111 11 Endothelium Blood-i8land(erythrobla9l)

thehum, blood-islands, and blood-plasma.

All of the cells of the blood-islands of the first two days of incubation become erythroblasts. The lumina of the vessels is not formed of tissue-spaces, but rather by a process of cytolysis in the center of masses of cells or even within the cyto- plasm of a single cell. An endothelial cell differentiates in its turn from the original angioblast. It can give rise to other endothelial cells or to erythryoblasts.

The observations herein recorded do not bear on the question of the origin of white blood-cells, because there are no cells in the chick of the second day of incuba- tion that can be identified as the ancestors of the white cell. They do show, however, that the ancestry of the red cells can be outUned as shown by the diagram above.



In this scheme I have used the term blood-island in the sense of hemoglobin- bearing cells attached to the inner wall of a vessel, while erythrohlast is used in the usual sense of free hemoglobin-l^earing cells that are continuing to divide. For the cliick the (juestion conc(>rning the origin of the white cells must be concerned with two possibilities; first, whether all of the white cells develop subsequent!}' from the mesenchyme outside the vessels that does not go through the stage of differentiating into angioblasts; second, whether in later stages the endothelium produces true Ij-mphocytes that do not develop hemoglobin.


AssHETON, Richard, 1916. Growth in length. Cam- bridge University Pr&ss.

BiALASzEwicz, L., 1912. Untersuchungen Uber die osmo- tischen Verhiiltnisse bei der Entwicklung der Frosch- und Hiihnerembryonen. Vorl. MitteQ. Bull, intern. .\cad. Sc. Cracovie, S(5r. B, 1-11.

Brachet, a., 1913. Recherches sur le ddterminisme h6r6- ditaii-e de I'ocuf des mammif ^es. Developpement "in vitro" de jeunes v^sicules blastodermiques de lapin. Arch, de Biol., t. 28, p. 447-.503.

Clark, E. L., 1912. General observations on early super- ficial lymphatics in living chick embryos. Anat. Rec, vol. 6, p. 247-251.

— — ■ — , 191.5. Observations of the lymph flow and the associated changes in the cjirly superficial lym- phatics of chick embryos. Anat. Rec, vol. 9, p. 65-67.

Clark, E. R., 1909. Observations on Uving, growing lymphatics in the tail of the frog larva. Anat. Rec, vol. 3, p. 183-198.

, 1912. Further observations on living, growing

lymphatics; their relation to the mesenchyme cells. Amer. Jour. Anat., vol. 13, p. 351-379.

, 1915. Studies on the growth of blood-vessels, by

observation of hving tadpoles and by experiments on chick embryos. Anat. Rec, vol. 9, p. 67-68.

, 1918. Studies on the growth of blood-vessels in

the tail of the frog larva — by obser\-ation and experiment on the hving animal. Amer. Jour. Anat., vol. 23, p. 37-88.

Clark, E. R. and E. L., 1912. Observations on the development of the earliest lymphatics in the region of the posterior lymph heart in the hving chick embryos. Anat. Rec, vol. 6, p. 253-259.

Danxhakoff, v., 1907. Ueber das erste .\ufreten der Blut-elemente in Huhnerembryo. Folia Ha-ma- tologia, vol. 4, Supplement; p. 159- 166.

, 1908. Untersuchungen uber die Entwickelung

des Blutes und Bindegewebes bei den Vogeln. 1. Die erste Enstehung der Blutzellen beimHiihner- embryo und der Dottersack als blutbildones Or- gan. .\nat. Hefte, Erste .\bth., Bd. 37, p. 471-589.

— , 1909. Untersuchungen fiber die Entwicklung

von Blut imd Bindegewebe bei Vogeln. Das lock- ere Bindegewebe des Hiihnchens im fetalen Leben. Arch. f. mikr. Anat. u. Entwicklngsgesch., Bd. 73, p. 117-181.

, 1915-16. Origin of the blood cells. Develop- ment of the hiematopoietic organs and regenera- tion of the blood cells from the standpoint of the monophyletic school. ,\nat. Rec, vol. 10, p. 397^13.

Dajv'chakoff, v., 191.5-16. Concerning the conception of potentialities in the embrj'onic cells. Ibid., p. 41.5-416.

, 1916. Equivalence of different hematopoietic an-

lages (by method of stimulation of their stem cells). 1. Spleen. Amer. Jour. Anat., vol. 20, p. 2.5.5-308.

, 1918. Cell potentialities and differential factors

considered in relation to erythropoiesis. Amer. Jour. Anat., vol. 24, p. 1-30.

Duval, M., 1889. Atlas d'embryologie. Paris.

Emmel, V. E., 1915. The cell clusters in the dorsal aorta of the pig embryo. .\nat. Rec, vol. 9, p. 77-78.

, 1916. The cell clusters in the dorsal aorta of

mamniaUan embryos. Amer. Jour. Anat., vol. 19, p. 401^16.

Evans, H. E., 1909. On the development of the aortte, cardinal veins, and other blood vessels of verte- brate embryos from capillaries. Anat. Rec, vol. 3, p. 498-518.

Hahn, H., 1909. Experimentelle Studien uber die Ents- tehung des Blutes und der ersten Gefasse beim Huhnchen. Arch. f. Entwicklungsmechanik der Organismen, Bd. 27, p. 337-433.

His, W., 1868. Untersuchungen uber die erste .\nlage des Wirbelthiereleibes. Leipzig.

, 1900. Lecithoblast und Angioblast der Wirbel-

thiere. .\bhandl. d. math.-phys. CI. d. k sachs. Gesellsch. d. Wissensch., Bd. 26, p. 171-328.

JoRD.\N, H. E., 1916. The microscopic structure of the yolk sac of the pig embryo, with special reference to the origin of the erjthrocytes. .\mer. Jour. .\nat., vol. 19, p. 277-299.

, 1916. Evidence of hemogenic capacity of endo-

thehum. .^at. Rec, vol. 10, p. 417-420.

, 1917. Hemopoiesis in the mongoose embrj-o,

with special reference to the activity of the endo- thelium, including that of the yolk-sac. Pubhca- tion No. 251 of the Carnegie Institution of Wash- ington, p. 291-312.

Jordan, B.. E., and J. C. Flippin, 1913. Hematopoi- esis in Chelonia. Foha Haematologica, vol. 15, p. 1-22.

Klein, E., 1871. Das mittlere Keimblatt in seinen Beziehimgun zur Entwicklung der Blutgefasse und Blutkorperchen im Hiihnerembryonen. Wiener akad. Sitzungsberichte, Bd. 63, p. 374-380.

Lewis, M. R., 1916. Sea water as a medium for tissue cultures. .\nat. Rec, vol. 10, p. 287-299.

Lewis, M. R. and W. H., 1911. The growth of embry- onic chick tissues in artificial media, agar and bouillon. Johns Hopkins Hosp. Bull., vol. 22, p. 126-127.



Lewis, M. R. and W. H., 19U. The cultivation of tissues from chiek embryos in solutions of NaCl, CaCls, KCl, ami XaHC(h. Anat. Roc., vol. 5, p. 277-293.

, lOl.'i. Mitochondria (and other cytoplasmic

structures) in tissue cultures. Amer. Jour. Anat., vol. 17, p. 339-401. LiLLiE, F. R., 1908. The development of the chick.

New York. McWhoktek, J. K., and A. O. Whipple, 1912. The de- velopment of the blastoderm of the chick in vitro. Anat. Rec, vol. 6, p. 121-139. McWhorter, J. E., 1914. (See Miller and McWliorter). M.\xiMow, A., 1909. Untersuchuiipen iibcr Blut und Hiiidogewebe. 1. Die friihcstcn Kntwicklungs- stailian der Blut- und Hindgcwebszollen boim SiiuKetioronibryo, bis zum .\ufang(l('r Hlutbildung in der Leber. Arch. f. mikr. Anat. u. Eiitwick- lungsgescli., Bd. 73, p. 444-561.

Miller, A. M., and J. E. McWhouter, 1914. Experi- ments on the development of blood-vessels in the area pellucida and embryonic body of the chick. Anat. Rec, vol. S, p. 203-227.

MiNOT, C. S., 1911-12. Development of the blood. Haiidbuch d. Entwicklngsgesch. d. Menschen (Keil)i4 & Mall), Bd. 2, 1911; Leipzig. Also Manual of Human Embryology (Keibel & Mall), Philadelphia, vol. 2, 1912.

Moluer, S., 1906 (See Riickert and Mollier).

, 1909. Die Blutbildung in der cmbryonalen

Jycber des Menschen und der Saugetiere. Arch. f. mikr. Anat., Bd. 74, p. 474-524.

R.\nvier, L., 1875. Trait«^ technique d'histologie. Paris.

Reagan, F. P., 1915. Vascularization phenomena in fragments of embryonic bodies completely iso- lated from yolk-sac bla.stodcrm. Anat. Rec, vol. 9, p. 329-341,

, 1915-16. A further study of the origin of blood

va-scular ti.ssues in chemically treated teleost embryos, with especial reference to haematopoesis in the anterior mesenchyme and in the heart. Anat. Rec, vol. 10, p. 99-llS.

, 1917. Experimental studies on the origin of vas- cular endothelium and of erythrocytes. Amer. Jour. Anat., vol. 21, p. 39-130.

Reagan, F. P., and J. M. Thorington, 1915-16. The vascularization of the cmbryoic body of hybrid teleosts without circulation. Anat. Rec, vol. 10, p. 79-98.

RUcKERT, J., and S. Mollier, 1906. Die er-ste Entste- hung der Gefasse und des Blutes bei Wirbeltiercn. Handbuch der vergleichenden und experimen- tellcn Entwickelungslehre der Wirbeltiere (Hert- wig). Jena.

Sabin, F. R., 1915. The method of growth of the lym- phatic system. The Harvey Lectures, Series 11, p. 124-145.

, 1917. Origin and <levelopment of the primitive

vessels of the chick and of the pig. Contribu- tions to Embryologj', vol. 6; Publication No. 226, Carnegie Inst. Wash., p. 61-124.

, 1917. Preliminary note on the differentiation of

angioblasts and the method by which they pro- duce blood-vessels, blood-plasma and red blood- celU as seen in the living chirk. .\nat, Rec, vol. 13, p. 199-204.

ScHAEFER, E. A., 1912. Quain's Elements of Anatomy.

Microscopic Anatomy, 11th ed., vol. 2, Part 1. Schmidt, M. H., 1892. IVber Blutzellenbildung in Leber und Milz unter normalen und pathologischcn \'er- hiiltnissen. Beitriige zur pathologischcn Anat- mie und Pathologic, Bd. 11, p. 199-233. Smiechowski, H., 1892. I'eber das erste Auftreten des Hiimoglobins bei Hiihnerembryonen. Inaug. Diss., Dorpat. Stockard, G. R., 1915. The origin of blood and vascular endothehum in embryos without a circulation of the blood and in the normal embryo. Amer. Jour. Anat., vol. IS, p. 227-327.

, 1915. A study of the wandering mesenchymal

cells on the living yolk-sac and their developmental products: chromatophorcs, vascular endothelium and blood cells. Amer. Jour. Anat., vol. 18, p. 525-594. Streeter, G. L., 1917. The development of the scala tympani, scala vestibuli and perioticular cistern in the human embryo. Amer. Jour. Anat., vol. 21, p. 299-320. Stricker, S., 1866-70. Studien iiber den Bau und das Leben der capillaren Blutgefasse. Untersuch. z. Naturl. d. Mensch. u. d. Thiere, Bd. 10. Van der Stricht, O., 1891. Le dcveloppement du sang dans le foie embryonnaire. Arch, de Biol., t. 11, p. 19-113.

, 1892. Nouvelles recherches sur la genese des

globules rouges et des globules blancs du sang. Arch, de Biol., t. 12, p. 199-344.

, 1893. Sur I'existence d'tlots cellulaires h. la p^ri-

ph^rie du blastoderme de poulet. Anat. Anz., Bd. 8, p. 266-271.

, 1895. De la premiere origine du sang et des

capillaires sanguins dans I'aire vasculaire du lapin. Comptes rend. d. 1. Soc. d. Biol., t. 12, S^rie 10, p. 181-185.

, 1895. La premifere apparition de la cavity coelo-

mique dans I'aire ciiibronnaire du lapin. Compt. rend, de la Soc. ,1. HIdI., t. 12, p. 207-211.

, 1896. Originc des globules sanguins, de I'aorte et

de I'endocarde chez les erabryons de S61aciens, Compt. rend, de la Soc. d. Biol., t. 13, S6rie 10. p. 287-290.

, 1899. L'origine des premieres cellules sanguines

et premiers vaisseaux sanguins dans I'aire vascu- laire de chauves-souris. Bull. d. I'Acad. Roy. de . MM. d. Bclgique, t. 13, Scf-rie 4, p. 336-349. Vialleton, L., 1892. Sur l'origine des germes vascu- laires dans I'cmbryon du poulet. Anat. Anz., Bd. 7, p. 624-627. Weed, L. H., 1917. The development of the cerebro- spinal spaces in pig and in man. Contributions to Embrj'ologj', vol. 5, Pub. No. 225, Carnegie Inst, of Wash. Williams, L. W., 1910-11. The somites of the chick.

Amer. Jour. .\nat., vol. 11, p. 5.5-l(X). WissozKy, N., 1877. Heberdsis Eosin als Reagensauf Hiim- oglobin und die Bildung von Blutgefjussen und Blutkcirperchen bei Saugethier->m<l Hiihnerem- bryonen. Arch. f. mikr. Anat., Bd. 13, p. 479-496. Wi'LF, 1897. Mikroskopische Bcobachtungen iiber die Entwicklung des Hiimoglobins beim Hiihnerem- bryo. Inaug. Diss., St. Petersburg.


Plate 1.

Fio. 2. Blastoderm of a chick (No. 141) with 1 somito, incubated for 28 hours and then grown for 22 nninutes in Locke-Lewis solution in which there was 1.06 per cent of NaCl. It had no somites when taken from the shell. The specimen is viewed from the endodermal surface, as Ls the case for all of the photographs on plates 1, 2, and 3. The lines of the endodermal blisters in the posterior part of the areJ^ pellucida on the right side have been retouchetl on the photograph. The line across the figure is approximately the level of the section shown on plate 4, figure 14, from a chick of about the same stage and pas-ses through the endo- dermal blisters. X24. .S'. 1, first somite.

Fig. 3. Bhustoderm of a chick (Xo. 212) with 2 somites, incubated for 25 hours and then grown for 3 hours and 4.5 minutes in Locke-LewLs solution containing 1.05 per cent NuCl and only 0.014 per cent of CaClj. If had no somites when taken from the shell; 2 are shown faintly in the photograph. X30. A., angioblasts; S. 1, first somite.

Fig. 4. Blastoderm of a chick (Xo. 201) with 4 somites, incubated for 26 hours and 40 minutes and then grown for 35 minutes in Locke-Lewis solution having 1.06 per cent XaCl and only 0.025 per cent of KCI. It had 3 somites when taken from the shell. The more anterior line across the figure is approximately the level of the section shown on plate 4, figure 15, and passes through the large vescicles of the mesoderm characteristic of the amnio-eardiac vescicles. The more posterior line shows the level of the section on pUiteO, figure 29. X24.

Fig. 5. BLustoderm of a chick (Xo, 198) with 5 somites, incubated for 26 hours and then grown for an hour in Locke- Lewis solution containing 1.06 per cent of XaCl and only 0.025 per cent of KCI. It had 4 somites when taken from the shell. The line across the figure shows the level of the section shown on plate 4, figure 16, from another specimen of about the same stage. It passes through the delicate network of the coelom character- istic of the middle and posterior part of the area pellucida. The square on the right side shows the zone which has been drawn for plate 6, figure 27, to show the appearance of angioblasts against the delicate network of mesoderm characteristic of the coelom. On the left side of the photograph there is a large amount of free yolk. Stained with hematoxyUn alone. X18.

Plate 2.

Fig. 6. Blastoderm of a chick (Xo. 213) with 5 somites, incubated for 25 hours and 10 minutes and then grown for 3 hours and 45 minutes in Locke-Lewis solution containing 1.05 per cent X^aCl and only 0.014 per cent of CaCh. It had 3 somites when taken from the shell. It shows especially well four zones in connection with the vascular system: (1) an outer zone of the area opaca, indicated by the leader Xo. 1, in which all of the angioblasts have become vessels; (2) an inner zone of the area opaca, indiciited by the leader X'o. 2, in which the angioblasts are still solid; (3) an ou*er zone of the area pellucida with solid angiobla.sts; (4) an axial zone of dense mesoderm. The rectangle on the right side shows the area which was drawn for plate 5, figure 22, inclosing the transition between the outer and inner zones of the area opaca. X18.

Fig. 7. Blastoderm of a chick (Xo. 238) with 7 somites, incubated for 29 hours and 30 minutes, and then grown for 35 minutes in Locke-Lewis solution in which there was 1.06 per cent of XaCl. It had 6 somites when taken from the shell. It shows a number of clumps of quite isolated angioblasts in the posterior part of the area peUucida, and over the amnio-cardiac vescicles there are the long, slender chains of angioblasts characteristic of that region. They have been retouched in the photograph. X15.

Fig. 8. Blastoderm of a chick (Xo. 177) with 11 somites incubated for 52 hours and 10 minutes and then grown for 3 hours and 35 minutes in Locke-Lewis solution containing 1.06 per cent of XaCl and 0.052 per cent of KCI. It had 10 somites when taken from the shell and the heart was just twitching along the outer border. It shows masses of soUd angioblasts in the posterior part of the area peUucida which are in the phase of division. Contrary to the usual position, the heart projects to the left side of the embrj'o, right side of the photograph. The rectangle indicates the position of plate 5, figure 24. X15.

Fig. 9. Bla.stoderm of a chick (Xo. 151 ) with 1 1 somites, incubated for 42 hours and 40 minutes and then grown for 3 hours and 40 minutes in Locke- Lewis solution in which there was 1.06 per cent of XaCl. It had 10 somites when taken from the shell and the heart was just twitching. The specimen shows an abnormal form of the brain and there are two large defects in the mesoderm, one on the right side and the other at the posterior end. It is e-spccially fine for the plexus of solid angioblasts in the posterior part of the area peUucida; in fact shows the maximum amount of angioblasts in any of my specimens. It also shows the outer and inner zones of the area opaca especiaUy weU, for the vessels of the outer zone are packed with blood-ceUs and the angiobla.sts of the inner zone have just Uquefied. The line across the figure shows the level of plate 4, figure 17. X17.

Plate 3.

Fig. 10. Blastoderm of a chick (X'o. 1.50) with 14 somites, incubated for 42 hours and then grown for 2 hours in Locke-Lewis solution in which there was 1.06 per cent of X'aCl. It had 11 or possibly 12 somites when taken from the sheU and now the fourteenth is just indicated in the undifferentiated mesoderm. From this stage on the number of somites anterior to the cardiac fold can not be made out for certain in the living chick and records should state the number posterior to the cardiac fold. The heart was beating but there was no circu- lation. The photograph shows four zones in connection with the vascular system, in a transverse Une just posterior to the dilated end of the spinal cord; an outer zone in the area opaca in which the vessels are packed with free red blood-corpuscles; an inner zone in the area opaca in which there are blood-vessels containing blood islands; an outer zone of the area peUucida with soUd angioblasts; and an axial zone in which there are more deUcate angioblasts which do not show at this magnification. 13 X.

Fig. 11. Bla.stoJorm of a chick (No. 174) with 14 somites, inoubateil for 52 hours and 4.") minutes and then grown for

2 hours and 1.5 minutes in Locke-Lewis solution having 1.06 per cent of NaCl and only 0.014 per cent of CaCU. It had 12 somites when taken from the shell. It shows the plexus of ves.sels in the entire area pellucida and demonstrates that the blood-islands in the posterior part of the area pellucida are within the ve-s-scls. The interspaces are pale rings and the vessels are a gray plexus, while the dark spots in the plexus are the blood-islands. The heart was beating but there was no circulation, and hence the veins are empty except for a few clumps of red cells, which have developed in them. In the vitelline veins close to the heart the streaks are due to the fact that the interspaces are so close that the endothelium of two adjacent vessels touch and therefore appear as Unes. The rectangle covers the area from which two drawings were made, namely figures 23 and 26. Xll.

Fig. 12. Bhistodcrm of a chick (No. 145) with 17 somites, incubated for 43 hours and 15 minutes and then grown for 2 hours in Locke-Lewis solution having 1.06 per cent of NaCl. It has blood-vessels throughout the area pellucida. ThLs is the specimen in which the islands were drawn in the living form shown in figure 18 from the zone indicated by the line No. 1. It had 17 somites when taken from the shell and the number did not The heart was beating and the circulation had begun, but was not reestablishe<l on the coverslip. It is also the specimen in which the blood of the outer zone of the area opaca has been drawn forward into the veins, and in which a generation of new blood-islands is beginning in this area. X8.6.

Fig. 13. Blastoderm of a chick (No. 183) with 18 .somites, incubated for 52 hours and 45 minutes, and then grown 4 hours in Locke-Lewis solution containing 1.06 per cent of NaCl and only 0.014 per cent of CaCb. It had 18 somites when taken from the shell and the number did not increase. The heart was beating vigorously and the circulation was reestabUshed on the coverslip. It shows three areas in which blood-cells are some- what abnormally massed; first, in the anterior veins; second, in the vitelline veins close to the heart; thir<i, opposite the emphalo-mesenteric arteries. It has blood-vessels throughout the area pellucida and shows a large number of blood-islands in the posterior part of the area pellucida. X7.2.

Plate 4.

Fig. 14. Photograph of a section through a blastoderm (No. 144) with no somites to show endodermal blisters. The section passes through the lower part of the primitive streak at approximately the level of the transverse Une on plate 1, figure 2. The chick was incubated for 26 hours and 45 minutes, was then grown for 2 hours and 10 minutes in Locke-Lewis solution having 1.06 per cent of NaCl, and fixed while a few endodermal bhsters were present. There is a small blister on the left side which contains a wandering endodermal cell and a larger empty one on the right side. The mesoderm is in soUd undifferentiated masses close to or touch- ing the ectoderm. X60.

Fig. 15. Photograph of a section through the aranio-cardiac vescicles of a chick (No. 115) with 3 somites which had been incubated for 23 hours and 25 minutes and then grown for 2 hours in Locke-Lewis solution. It had

3 somites when taken from the shell and the number did not increase. The section is taken at approxi- mately the level of the line across plate 1, figure 4. The section shows an open neural tube. The large, closely packed vescicles characteristic of the development of the amnio-cardiac vescicles are clearly shown, with small clumps of angioblasts opposite the walls of these vescicles. X60. A. angioblasts; C, ccelom.

Fio. 16. Photograph of a section through the first somite from a blastoderm (No. 155) with 5 somites, incubated for 27 hours and 20 minutes and then grown for 2 hours in Locke-Lewis solution having 1.06 per cent of NaCl. It was taken at approximately the level of the transverse line on plate 1, figure 5, which is a specimen of the same stage. The section shows the type of the eoelom, which is characteristic of the middle and pos- terior zones of the area puUucida. In the area pellucida the wide gaps between the vescicles of the coelom, where there are no cells except the ectoderm and endoderm, are very jilain. X60.

Fig. 17. Photograph of a section of a blastoderm (^Xo. 95) with 11 somites, incubated for 42 hoius and 30 minutes and then grown for 2 hours in Locke-Lewis solution. It had 9 somites when taken from the shell. The photograph is given especially to show t he jjosition of the drawing on plate 6, figure 28, which is taken within the scjuare. It shows all the of formation of the blood-vessels and blood-islands in a single section. At the outer edge of the areu opaca are blood-vessels, the sinus marginalis containing free blood-corpuscles. Within the square at the inner border of the area opaca are both blood-islands in a vessel and angioblasts liquefying to form the lumen of a vessel, while in the area pellucida are new clumps of angioblasts. X48. A, angioblasts; I, lumen of the sinus marginalis.

Fig. 18. Blood-islands in the vessels of the area pellucida of a chick (No. 145) of 17 somites, drawn from the living specimen. It is designed to .show as nearly as pd.ssibli' the appearance of the living tis.sues. The region from which the drawing was taken is shown by the leader No. 1, plate 3, figure 12, from a photograph of the same specimen, but which does not show the exact form of the drawing since the specimen was fixe<l 2 hours and 15 minutes after the outlines for the drawing were made. The lumina of the vessels is made to appear like ground, which represents their actual appearance. The small drawing at the bottom of the figure shows the phase of division of the nuclei, which took place while the drawing was being made. At the top of the figure arc two unicellular blood-islands and some free red blood-corpuscles in the lumen of the vessel. On the right side is a large blood-island attached to the endothelium of the ves.sel. X4.50. li. i., young blood- island shown in the rest ing phase, the small drawing below being the same island during division; c, endothe- lium; i., interspace between th(! vessels lus it appears in the living specimen; it represents the mesoderm beneath the vessels, not analyzed with reference to its cc^lls, and is bordered with a rim of endothelium; I., luraen of a ves.sel.

Fig. 10. Blooil-islamls from the aroa pcUucifla of a chick CSo. 13.5) with 10 somites, drawn to show the appearance of the living spcciincii. The cliick W!i.s incubated for 42 hours and 20 minutes in Locke- Lewis solution having 1.07 per cent of NaCl. The heart was be.atinp; and the circulation was establi.shefi. The islands were taken from the posterior part of the area pellucida at about the same region as the drawing of figure 18 of the same plate. The specimen sliows a late stage in the development of the islands; the cell outlines are an indication that the islands were undergoing division, though the chromosomes were not seen in the Uving specimen. In the fixed specimen more than half of the nuclei are in the prophase, which can not be made out in these thick specimens while living. X4.50. B. i., blood islands; e., endothelium; i, interspace; I, lumen of the plexus of vessels.

Fig. 20. Plexus of angioblasts from a chick (N'o. 161) with 13 somites, drawn to show the appearance of the living specimen. The chick had been incubated for .TO hours and then grown for 3 hours and 4.5 minutes in Locke- Lewis solut'on in which there was 1.00 per cent of N'aCl. It had 12 somites when the egg was taken from the shell and the drawing was made at that stage. The hejirt was beating, but there wa-s no circulation. The drawing is to be compared with the blastoderms shown on plate 2, figures 8 and 0, and on plate 3, figure 10; and is from the zone just lateral to the rectangle shown on plate 2, figure 8. The drawing shows the character of the bands of angiobl.osts before there is any liquefaction whatever as seen seen in the living chick. X47.5. A, angioblasts; i, interspace.

Plate 5.

Fig. 21. Section tlirough the marginal sinus of a chick (No. 96) with 18 somites, which had been incubated for 43 hours and then grown for 2 hovu-s in Locke-Lewis solution. It had 17 somites when taken from the shell, and when fixed the eighteenth was just appearing. It is to show a new generation of bloo<l-islands beginning in the marginal sinus, like those of the total preparation shown on plate 3, figure 12. The heart was beating and the circulation well established. The ectodc-m is not included in the drawing. X700. B. i., unicellular blood-island in the edge of the marginal sinus; to the left is a larger blood-island, also attached to the endo- thelium, and the sinus has many free erythroblasts; c, exoccelom; en., endoderm; en. c, endodcrmal cell full of yolk and ready to become a wandering ceD; mcs., mesoderm along the border of the exoccelom.

Fig. 22. Drawing of the transition zone between the outer and inner zones of the are.a opaca in a chick (So. 213) with 5 somites. The total blastoderm from which the fLrawing was taken is shown on plate 2, figure 6. The place of the drawing is shown by the rectangle in the photograph, but the position is reversed. On the left side, outer zone of the embryo, the ves.soLs have formed and show blood-i.slands attached to their walls, while on the right side the hollow ves.sels k^ad over into the plexus of solid angioblasts. X330. A, angif)- blasts, the suggestion of cell outlines indicating cell division; b. i., blood-island within the lumen of the vessel; i, interspace; I, lumen of a vessel.

Fig. 23. Drawing of a mass of angioblasts in which the liquefaction of the cytoplasm is very extensive, from a chick (No. 174) with 14 somites. It is from the blastoderm shown on plate 3, figure 11. Both this figure and the one on plate 6 (fig. 26) were taken within the rectangle on the photograph. They are both vessels which make up the lower end of the developing aorta. This figure shows a mass of angioblasts which is going to liquefy entirely, while the other shows a partial transformation of angioblasts into erythroblasts. The vacuoles vary greatly in size, and many of them are immediately against the nuclei. X920.

Fig. 24. Drawing of soUd angioblasts from the posterior part of the area pellucida from a chick (No. 177) with 11 somites. It is from the blastoderm shown on plate 2, figure 8, the position of the drawing being shown by the rectangle on the photograph. The drawing is reversed from the photograph. The angioblasts are entirely solid and show no sign of liquefaction. The mass was imdergoing division and different types of nuclear figures are plain. At the lower left-hand border a single nucleus h>as elongated to make a typical endothelial nucleus, while on the right-hand border a single cell become an angiobliist and is about to join the main mass. X920.

Fig. 2.5. Drawing of angioblasts taken during the process of liquefaction from a chick (No. 172) with 13 somites. The specimen had been incubated for 48 hours and 30 minutes in Locke-Lewis solution having 1.06 per cent of NaCl and only 0.014 per cent of CaCL. The drawing is taken from the arch of the area pellucida poste- rior to the end of the spinal cord in a specimen nearly like that on plate 3, figure 10. The chick had 11 somites when taken from the shell and the heart was just beating. The entire center of the mass is soUd, but there are vacuoles forming along the edges, leaving an endothelial border. In the upper right-hand process the lumen of the vessel is forming within a single angioblast between two nuclei. X920.

Plate 6.

Fig. 26. Drawing of a vessel in which a part of the original angiohlast-s has developed hemoglobin and is thus forming blood-islands instead of liquefying; from a chick (No. 174) with 14 somites. It is from the blastoderm .shown on plate 3, figiue 11. Both this figure and that shown on plate 5 (fig. 23) were taken within the rectangle on the photograph. They are both vessels that form a part of the lower end of the developing aorta. The clump of two cells to the right are new angioblasts; in the specimen they are along the vessel wall just out- side of the area drawn, but are about the same distance from the main vessel; they were shifted in the drawing, so as to come witliin the field. X920.

Fig. 27. Drawing of a plexus of angioblasts as seen against the developing coelom in the area pellucida of a chick (No. 198) with 5 somites. It is from the blastoderm shown on plate 1, figure ,5, and covers the area of the square on that photograph. The angioblasts are shown as dense bands against the more deUcate background of the ccclom. In several places there are transition zones between the two structures. There are numerous gaps in the mesodermal layer. X4(X). A, angioblasts; mes., mesoderm.

Fio. 28. Section throiigh the inner e<J(!e of the area opara of a chick (No. 95) with 1 1 somites, which was incubated for 42 hours ami 30 minutes and then grown for 2 hours in Locke-Lewis solution. The chick had 9 somites when taken from the shell, and the eleventh was just appearing when it was fixed. The heart was not beat ing when the egg was opened, but twitched slightly on the coverslip. The section is appro.xiniately at the level of the line across plate 2, figure 9, which is a specimen of about, the same stage. A photograi)h of the section from which the drawing was made is given on plate 4, figure 17, w'ith a square to indicate the zone. It shows the process of liquefaction of angioblasts to make vessels, as it can be seen in sections, as well as a blood- island attached to the endothelium of the same vessel. It shows the same processes illustrated in total prej)- arations on plate 5 (fig. 23) and plate 6 (fig. 26). X700. A, angioblasts in the process of liquefying, show- ing pycnotic nuclei; 6. i., blood-island; c, exoccclom; ec, ectoderm; en., endoderm; I., lumen of ves.sel; mes., mesoderm.

Fig. 29. Section through the posterior part of the area opaca of a chick (No, 148) with 4 somites, which had been incubated for 30 hours and 15 minutes, and then grown for an hour in I>ocke-Lewis solution having 1.06 per cent of XaCl. It is taken at approximately the level of the lower line on plate 1, figure 4, from a si)ecimen of about the same stage. The drawing shows the differentiation of the primitive mesoderm into the two layers of celLs which form the coclom and the more ventral solid clumps of angiobhists. X620. A, angio- blasts; c, ca'lom; ec, ectoderm; en. c, endodermal cell which has wandered to a position dorsal to the meso- derm; mes., mesoderm.