T H E TRANSFORMATION OF ADIPOSE TISSUE INTO HEMOCYTOPOIETIC TISSUE H. E. JORDAN Laboratory of Histology and Embryology, University of Virginia The usual sites of extramedullary blood formation in man and mammals are the spleen and the lymph nodes. These organs in man frequently become hemopoietically active in certain cases of anemia and leukemia. The myeloid nietaplasia apparently represents a reaction compensatory to a morbid bone marrow. Additional ectopic hemocytopoietic areas in the adult may occur in the periportal regions of the liver and in the kidneys, where the reticular connective tissue constitutes the source of the differentiating hemoblasts. I n many cases of extramedullary blood development in man, the condition may extend also in variable degree to certain lobules of adipose tissue, more commonly those of the prevertebral abdominal region and those adjacent t o the kidneys. This investigation is concerned especially with a study of the histogenetic process by which adipose tissue in relation to the kidneys and certain lymph nodes becomes converted into erythrocytopoietic tissue. The matter is of especial importance because of its bearing on the problem of the manner in which hypoplastic (fatty) marrow changes to the hyperplastic condition. The fundamental questions iiivolved in this process concern the source of the erythrocyte precursor, whether endothelial cell, lymphocyte or reticular cell ; and the condition of the lining of the erythrocytopoietic capillaries and sinusoids, whether complete or fenestrated. MATERIALS AND METHODS The normal material includes kidneys of the cat in which the vascular system was injected with carmine gelatin. The 461 T H E A N A T O X I C A L REOORD, VOL. 59, NO. 4 462 H. E. JORDAN injection passed to the capillary system of the tunica adiposa of the capsule. This adipose tissue supplies the material for the study of the fat-cell-capillary relationship in normal fat. The pathological material showing extramedullary hemocytopoiesis in adipose tissue includes the kidneys and lymph nodes from a case of adenocarcinoma of the prostate. The tumor had metastasized widely and extensively, having invaded all abdominal organs except the spleen. The bone marrow was largely replaced by the tumor tissue and almost all of the lymph nodes were greatly enlarged and entirely replaced by the tumor. However, in the case of a few of the larger lymph nodes a small polar area remains free of the carcinoma. Such areas have in part the appearance of hemal node tissue, in part they resemble more closely red bone marrow. In the latter case there is considerable fat. Furthermore, the fatty tissue adjacent to many of these lymph nodes shows a variable degree of myeloid (erythroid) metaplasia. These fatty areas supply the material for the study of the process of origin of the hemocytopoietic tissue. The material was fixed with Helly’s fluid, and stained either with the Giemsa stain o r hematoxylin and eosin. I n addition, sections of the developing femur were studied for information regarding: first, the earliest stages in the formation of marrow and the relation between the ancestral blood cells and the primitive vascular system; and secondly, the alterations effected by the appearance of fat cells during the conversion of a primitive hyperplastic marrow into a secondary hypoplastic fatty marrow in the diaphysis. The problem was pursued further with femoral marrow of the adult cat. This material includes marrow in extreme hypoplastic condition following suprarenalectomy, and marrow in early stages of restoration following administration of cortical extract. For the cat material I am indebted to Prof. S. W. Britton. BLOOD FORMATION I N ADIPOSE TISSUE 463 T H E RELATION BETWEEN T H E FAT CELLS AND T H E CAPILLARIES I N NORMAL ADIPOSE TISSUE The adipose tissue under consideration is that in relation to the renal capsule. The blood vessels were filled with carmine gelatin injected into the renal artery. Sections were cut a t 20 1-1 and lightly stained with hematoxylin. The capillary supply is very profuse, the mesh of the intricate network having approximately the diameter of the fat cells. Many fat cells are almost completely encircled by a delicate capillary; the smaller branches may arise a t right angles from parent stems and may be sharply flexed into one complete spiral near the point of origin. The capillary beds connecting arterioles with venules are of typical form and arrangement; the system is completely closed, and includes no elements comparable to the long straight non-anastomosing arterial (transition) capillaries and the related venous sinuses of the bone marrow, nor the penicillus and related sinuses of the spleen. With this morphology as a background we are in a position to attempt the interpretation of conditions in areas of myeloid (erythroid) metaplasia in adipose tissue. T H E HISTOGENESIS O F BLOOD I N ADIPOSE TISSUE The process of erythroid metaplasia is identical in the fatty tissue of the renal hilus and that covering certain lymph nodes in the case of the metastasizing adenocarcinoma of the prostate. Accordingly, the following description applies to either region. The first indication of a prospective area of hemocytopoietic tissue is the accumulation of cells resembling small lymphocytes about certain capillaries and precapillary veins. These accumulations enlarge, many of the cells having meanwhile assumed the characteristics of large lymphocytes or hemocytoblasts. These areas may include also a few megakaryocytes. The included capillaries and venules have become invaded by hemocytoblasts ; these cells are in process of differentiation within the lumen of the blood vessels. Many of the hemocytoblasts, both extravascular and intravascular may be in mitosis. The capillaries and venules may become ex- 464 H. E. JORDAN panded into sinusoidal structures with fenestrated walls during this process of intravascular proliferation and differentiation. Such hemopoietic areas consist, then, of what appear to be irregular cords of lymphoid tissue with a central capillary or sinusoid; both capillary lumen and enveloping lymphoid area may contain hemocytoblasts, erythroblasts and normoblasts. I n later stages such cords become confluent, giving the appearance of an area of red bone marrow. Meanwhile the expanding erythrocytopoietic sinusoids have acquired openings, giving direct communication between the vascular lumen and the interlipocellular spaces, permitting plasma to pass directly into the tissue spaces. This secondary fenestration of the capillary wall is presumably the result of mechanical factors incident to the adaptation of the relatively restricted lumen to the expanding cellular content and adjustment between the connective tissue investing the blood channels and the stroma of the adipose tissue. No evidence of hemocytopoietic activity on the part of endothelium appears in this tissue. The hemocytoblasts have a n extravascu1ar origin; these cells arise from what appear to be small lymphocytes. The lymphoid accumulations are in part small and widely scattered. The available evidence does not permit of a definite conclusion as to whether they represent exclusively cells transported by lymphatics o r whether they are i n part in situ derivatives of the intercellular connective tissue. Local proliferation accounts for part of the increase of this tissue. THE ORIGIN O F ERYTHROCYTES I N FATTY AREAS O F LYMPH NODES The lymph nodes here under consideration are those from the case of adenocarcinoma of the prostate in which the lymphoid tissue of all but a small polar area has been replaced by tumor tissue. This relatively normal area of the lymph node has, however, the appearance of hemal node tissue in that the sinuses contain a variable amount of blood; and it resembles bone marrow in that it embraces centrally much fat and peripherally much erythrocytopoietic tissue, includ- BLOOD F O I i M A T I O N I N A D I P O S E TISSUE 465 ing megakaryocytes. The process is the reverse of that described above, in cases where adipose tissue becomes converted into lympho-erythroid tissue; here lympho-erythroid tissue changes into adipose tissue ; lymph node tissue changes into hemal node tissue which suffers hypoplasia centrally and subsequently aplasia. A low power field shows peripherally a uniform moderately compact lymphoid condition. This represents the original cortical area. Adjacent to this area, centrad, the structure is one of cords and sinuses, representing the peripheral portion of the original medulla. The sinuses contain blood ; the medullary cords have a relatively loose texture in which the reticular cells and fibers are conspicuous. The parenchyma consists predominantly of small lymphocytes, with considerable numbers of intermingled larger lymphocytes, red blood corpuscles and erythroblasts. The central arteriole of the cords is conspicuous. The walls of the corresponding venules and the connecting capillaries have the appearance in places of being continuous with the adjacent reticular tissue and so effecting open communication between these vessels and the extravascular tissue spaces. Passing still deeper centrad the adjacent area contains a few fat cells. These cells arise in the cordal areas, presumably from reticular cells. The fat cells increase in number, and this area has exactly the appearance of adipose tissue undergoing erythroid metaplasia. The interlipocellular spaces are filled with hemocytopoietic tissue enveloping one or several axial blood vessels. The hemocytopoietic tissue includes hemocytoblasts, lymphocytes, megakaryocytes, erythroblasts, normoblasts, and many red blood corpuscles. Along the inner border of this area the red blood corpuscles predominate, and at the center the tissue consists only of fat cells and stroma, with the intercellular spaces filled with red blood corpuscles. The appearance suggests a condition of hemorrhage within a fatty tissue. However, in the light of the series of successive stages above outlined, the condition must be interpreted as one of local red cell formation; the 466 H. E. JORDAN blood stasis is presumably the results of the dilatation of the capillary bed into a sinusoidal structure with communicating intercellular spaces. T H E CHANGE FROM H Y P E R P L A S T I C TO H Y P O P L A S T I C MARROW I N EARLY STAGES O F BONE DEVELOPMENT The material here under consideration is the femur of the human fetus and infant. Early stages in the development of enchrondral bone supply important data regarding the relationship between blood vessels and the hemocytoblasts. The cartilaginous primordium of any long bone early becomes enveloped at the level of the future diaphysis (primary center of ossification) by a sheath of periosteal bone. Through this extend into the primary center of ossification buds of vascularized mesenchyme from the inner layer of the periosteum. The extension of these osteogenic buds keeps pace with the formation of primary areolae in the diaphyseal portion of the bone. The tissue constitutes the primary marrow, and each bud includes, in addition to the reticular stroma and an axial arterial and venous vessel terminally connected by capillaries, a variety of cells. Peripherally the cells which adjoin remnants of calcified cartilage function as osteoblasts ; centrally the cells resemble small and large lymphocytes. The latter are typical hemocytoblasts. Such hemocytoblasts invade the sinusoidal and capillary channels where they differentiate into erythrocytes ; extravascularly, certain hemocytoblasts differentiate into granulocytes. The outstanding facts are: absence of any activity in the formation of hemocytoblasts on the part of the endothelium; origin of hemocytoblasts from the reticular stroma (mesenchyme) ; envelopment of axial blood vessels by hemocytopoietic tissue ; invasion of venous and capillary vessels by hemocytoblasts t o diff erentiate intravascularly into erythrocytes. The earlier hemocytopoietic marrow occurs in the form of cords and islets of uniformly compact texture. Subsequently fat cells appear, differentiation products of reticulum cells, and as their number increases in the diaphyseal region the BLOOD FORMATION I N ADIPOSE TISSUE 467 marrow assumes an open looser texture. The hemocytopoietic tissue with its primitive vascular supply becomes compressed between adjacent fat cells. At this stage erythrocytes and red corpuscles occur both intravascularly and extravascularly in these marrow cords. The appearance may be interpreted in terms of a rupture of the capillary wall permitting passage of definitive cells into the intercellular spaces ; or as the result of an extravascular differentiation of erythrocytes. On the basis of my observations I am led to conclude that both interpretations are in part correct. The rapid development of fat cells in large numbers from the reticular stroma produces traction upon the delicate walls of adjacent capillaries and sinusoids and causes breaks in the endothelium, permitting direct intercommunication between vascular lumen and interlipocellular tissue spaces. Thus endothelium becomes continuous at these fenestra with reticular cells of the stroma. Accordingly, hemocytoblasts which arise from stroma in such regions are a t once in intimate relation with plasma and under this stimulus differentiate into erythrocytes. The net result of the development of fatty marrow as regards blood cell production is the establishment of direct communication between the so-called erythrocytogenic capillaries and certain spaces of the reticular tissue. This constitutes a reticulo-endothelium and is the source of the hemocytoblast s which, when either within a thin-walled vessel where the blood is relatively stagnant or closely associated with such an area, develop into erythrocytes. Both reticular cells and hemocytoblasts are occasionally seen in mitosis. There is no evidence of hemocytopoietic activity on the part of the endothelium. HYPOPZASTIC MARROW O F T H E F E M U R O F THE CAT I N EARLY STAGESOFRECOVERY The material includes femoral marrow from normal, suprarenalectomized and cortin-treated cats. The tissues were fixed with Helly’s fluid and stained with the eosin-azure combination of Giemsa. Seven days after removal of the suprarenals the marrow is practically aplastic. Three days after 468 H. E. JORDAN treatment with cortin the marrow of these experimental animals may be widely in process of recovery. The details of the complete process of restoration will be dealt with in a separate study jointly with Dr. S. W. Britton, who carried out the experimental work. The only matter of interest in this connection concerns the initial stages of recovery from the extreme hypoplastic condition. There is no indication of endothelial activity in relation to the appearance of hemocytoblasts in the capillaries among the fat cells. The initial hemocytoblasts in the hypoplastic areas have a triple origin. They represent in part locally persistent intravascular hemocytoblasts, which subsequently proliferate and so produce local aggregations which differentiate into erythrocytes. In part they represent migrants from peripheral relatively normal intravascular collections of hemocytoblasts. I n small part they represent local differentiation products of cells of the interlipocellular reticular connective tissue which by mitotic proliferation produce extravascular hemocytopoietic islands from which hemoblasts migrate into the axial capillary channels. Conditions in the marrow of young normal cats confirm the above outlined findings. Sections of femoral marrow, while compact and hyperplastic in the main, include areas of considerable extent in conditions of variable degrees of hypoplasia. Examination of large extents of active marrow of presumably normal animals leads me to the conclusion that as concerns particular areas hemocytopoiesis is a cyclic process. A certain area passes through a stage of intense activity (hyperplasia) to one of exhaustion (hypoplasia) followed by phases of gradual restoration. Related in some manner to this cyclic functional process is a histologic feature characterized by the presence of what appear to be genuine lymphoid nodules, composed chiefly of small lymphocyte-like cells. The details of this relationship of lymph nodules and lymphocytes to the hemocytopoietic process are reserved for a separate study. Interest in this connection only centers on the aggregations of interlipocellular hemocytoblasts and the BLOOD F O R M A T I O N I N ADIPOSE TISSUE 469 endothelium of the associated axial capilliform sinusoids in the hypoplastic marrow in early stages of restoration. Considerable numbers of the hemocytoblasts, both extravascular and intravascular, are in mitosis. Occasional hemocytoblasts can be seen passing through the endothelial wall; they have a dumb-bell shape with one lobe in the lumen of the sinusoid, the other in the enveloping hemocytopoietic tissue. The direction of migration is presumably lumen-ward. Whether or no the endothelial wall is actually fenestrated, it is clear that it offers no effective barrier against hemocytoblast migration. Many instances can be seen of capillaries containing a row of three to six hemocytoblasts, one or several in mitosis, but no unequivocal evidence appears that such hemocytoblasts are endothelial derivatives. Furthermore, many instances appear of hemocytoblasts pressed against the endothelium of larger sinusoids and some of the hemocytoblasts may be considerably flattened, but again clear evidence is lacking that they may represent derivatives of endothelial cells. The extravascular hemocytoblasts are nearly as active in mitotic proliferation as those within the blood channels. While an occasional reticular cell may be seen in mitosis in these areas of active restoration, only extremely rarely and uncertainly, is an endothelial cell seen in mitosis. Such rare and dubious instances of endothelial cell proliferation seem more reasonably to bear the interpretation of a method of adjusting the wall of a capillary to its expanding content of hemocytoblasts than as a method of origin of erythrocyte ancestors. DISCUSSION The outstanding fact which emerges from this study of the development of hemogenic cells in adipose tissue concerns the origin of the ancestral hemocytoblasts exclusively in extravascular stromal areas ;the endothelium of the involved capillaries and sinusoids is hemopoietically inactive. The original groups of cells have the appearance of small lymphocytes. These may in large part represent migrants from adjacent 470 H. E. JORDAN lymphatics, but the occurrence of what appear to be transitional stages suggests also some slight degree of origin from local stromal cells. These small lymphocyte-like cells grow to the size and acquire the features of typical hemocytoblasts. These hemocytoblasts may undergo a certain amount of proliferation and may migrate into the adjacent capillary. Within the blood channel the hemocytoblasts differentiate into erythrocytes. A comparison of this process of extramedullary blood formation with the normal process in active bone marrow reveals an essential identity. I n transiently hypoplastic areas at initial and early stages of transition to renewed activity hemocytoblasts are seen to accumulate in interlipocellular cords with axial capillary vessels. Hemocytoblasts migrate into the adjacent capillary or venous sinusoid where they diff erentiate into erythrocytes. Extravascularly the hemocytoblasts differentiate into granulocytes and megakaryocytes. I n an attempt to elucidate the common problem of hemocytoblast origin and sinusoid-tissue-space relationship the study of early stages in the formation of marrow in enchondral ossification has contributed valuable data. Stages in the transformation of hyperplastic into hypoplastic marrow are at least of equal value with those of reverse order, which are those heretofore almost exclusively studied. If endothelium during any stage of medullary hemopoiesis has erythrocytopoietic capacity, it might reasonably be expected to be active during the stages when marrow is first formed. One seeks in vain for unequivocal evidence of endothelial erythrocytopoietic capacity in primordial marrow. The blood vascular system of the eruptive tissue is closed. The hemocytoblasts arise from the mesenchyme of this tissue. They constitute compact masses between and enveloping adjacent sinusoids and capillaries. They migrate in part into the lumen of these vessels to differentiate into erythrocytes ; in part they differentiate extravascularly into granulocytes and megakaryocytes. When fat cells make their appearance in this myeloid tissue in the area of the future fatty BLOOD F O R M A T I O N I N ADIPOSE TISSUE 471 diaphysis they effect a separation of the originally continuous cell masses into small interlipocellular groups, more or less intimately interconnected by slender strands of hemocytoblasts. Since the reticular stroma from which the fat cells develop and which supports the hemocytoblasts is in continuity with that supporting the capillaries and sinusoids, traction is exerted upon the delicate endothelial wall as the fat cells enlarge in size and increase in number to the end that communication is established at certain points along the sinus between its lumen and adjacent intercellular spaces. When hemocytoblasts are differentiated from the reticular tissue bounding such spaces they are directly in contact with plasma from the sinus and accordingly in relation to the stimuli under which differentiation into erythrocytes is effected. The reticular tissue involved in this process, in view of its relation to capillary endothelium, constitutes a reticuloendothelium of the hemohistioblast variety. I n active red marrow a considerable amount of erythrocytopoiesis is extravascular, in areas communicating with sinusoids. The above is in close conformity with earlier descriptions of the erythrocytopoietic process in mammalian marrow (van der Stricht, 1892; Maximow, '10). The accuracy of these descriptions is challenged by Doan, Cunningham and Sabin ( '25) who report origin of megaloblasts (erythrogenic hemocytoblasts) only from endothelial cells of 'intersinusoidal capillaries' in the marrow of the pigeon and the rabbit. Peabody ('26) reports similar findings in human femoral marrow from a case of typhus fever. Our earlier study of the marrow of the frog (Jordan and Baker, '27) led us to conclude in favor of the origin of the red cell ancestor from reticular cells. The histology of hypoplastic femoral marrow of the frog at initial stages of return to activity indicates an extravascular origin of lymphoid hemoblasts, namely, from reticular cells, and communication between the lumen of the venous sinusoids and adjacent intercellular spaces. The interlipocellular spaces of our frog material correspond exactly with the so-called intersinusoidal capillaries of Doan, Cunningham and Sabin. India 472 H. E. JORDAN ink injection of our material showed that these communicating channels are double walled sacs enveloping fat cells and cannot with any degree of terminological propriety be designated capillaries. Such areas in section frequently include a central capillary or sinusoid, and this vessel may be more or less completely filled with hemocytoblasts and differentiating erythroblasts; but study of successive stages in the development of such erythrocytogenic capillaries shows that the hemocytoblasts arise extravascularly and only secondarily invade the adjacent sinusoid. I n many instances also the hemocytoblasts of a particular capillary have been transported via the vascular route from adjacent hyperplastic areas. Peabody describes an identical relationship between intercellular hemocytopoietic tissue and axial capillary, examples of which are numerous in my sections of femoral marrow of the cat, but makes the forced and very improbable interpretation of a “branching transition capillary transversing a widely dilated intersinusoidal capillary. ’ ’ On the basis of detailed histological studies of various tissues from instances of extramedullary hematopoiesis in anemias Brannan (’27) states, after an apparently sympathetic effort to interpret conditions in terms of endothelial activity, that in one case “there was a suggestive relationship between the capillary endothelium of the liver and the myeloblasts, but one could not say that the white or red blood cells arose from the endothehum.” As the result of very extensive experimental investigations, including injections with India ink, Drinker and Drinker ( ’30) decide, in essential agreement with Maximow ( ’lo), that in active mammalian bone marrow the wall of the capillaries become transiently fenestrated for the admission from adjacent extravascular areas of groups of maturing erythrocytes. BLOOD F O R M A T I O N I N ADIPOSE TISSUE 473 SUMMARY 1. The object of this investigation of blood formation in fatty areas of various tissues, under certain normal, pathologic and experimental conditions, is t o determine the origin of the erythrocyte ancestor, whether endothelial or stromal ; and the relation of the sinusoid lumen to the tissue spaces, whether the endothelial boundary is continuous or fenestrated. 2. Normal adipose tissue of the renal capsule is supplied by a closed vascular system in the cat. Specimens injected with carmine gelatin show a dense capillary network; the mesh has a diameter approximately that of the fat cells and many fat cells are almost completely encircled by one o r several capillaries. 3. Extramedullary blood production in the adipose tissue of the renal pelvis and certain lobules in the vicinity of tumorous lymph nodes, and fatty areas in certain hemal nodes, in a case of metastatic adenocarcinoma from the prostate is an apparently identical process. Cells, resembling small lymphocytes, accumulate in certain areas about capillaries, where they may grow into typical hemocytoblasts. From these extravascular aggregations of lymphoid cells, hemocytoblasts migrate into the axial capillary where they differentiate into erythrocytes. I n the extravascular areas occur in addition to the dominant lymphocytes and hemocytoblasts, occasional granulocytes, megakaryocytes and macrophages. Endothelium in these regions lacks erythrocytopoietic capacity. 4. Hypoplastic marrow, natural or experimental, during early stages of recovery passes through steps essentially identical with those through which adipose tissue passes when it undergoes erythroid metaplasia. However, in addition to local extravascular accumulations of hemocytoblasts, which may secondarily invade the vascular channels, intravascular hemocytoblasts represent in part also persistent local hemocytoblasts and in part migrants from peripheral relatively hyperplastic areas. Hemocytoblasts from these three sources proliferate within the capillaries and sinusoids and diff erentiate into erythrocytes. There is no unequivocal evidence of endothelial contribution of hemocytoblasts. 474 H. E. JORDAN 5, The reverse of the progressive process from hypoplasia to hyperplasia, especially in primordial marrow, supplies important evidence regarding the source of the initial hemocytoblasts and the relation of the capilliform sinusoids of marrow to the stromal tissue spaces. Such reverse process, namely, from hyperplasia to hypoplasia in the primordial marrow of the future diaphysis of the femur, shows that the original vascular system of the eruptive tissue is closed, and that the compact enveloping marrow is of local stromal origin. Even in these earlier stages the endothelium gives no evidence of hemocytopoietic activity. The hemocytoblasts are reticularcell (mesenchyme) derivatives. They secondarily migrate into the capillaries and venous sinusoids where they differentiate into red cells. 6. When the fat cells accumulate in the areas where the red marrow of the future diaphysis is changing into yellow marrow, the interconnections between the walls of the capillaries and the supporting reticular cells from which the fat cells develop exert tractions upon the delicate endothelial wall, effecting thus local communication between the erythrocytopoietic capillaries and the stromal tissue spaces. When a condition of hyperplasia supervenes in such regions the hemocytoblasts arising from reticular cells differentiate within tissue space already in communication with the venous sinusoids and the group of differentiating erythrocytes becomes enveloped by a reticulo-endothelium. BLOOD F O R M A T I O N I N ADIPOSE TISSUE 475 LITERATURE CITED BRANNAN, D. 1927 Extramedullary hematopoiesis i n anemias. Bull. Johns Hopkins Hospital, vol. 41, pp. 104-136. DOAN, C. A. 1922 The circulation of the bone marrow. Contributions t o Embryology, no. 67, Carnegie Institution of Wash., pub. no. 377, pp. 2 9 4 6 . DOAN,C. A., R. S. CUNNINGHAM AND F. R. SABIN 1925 Experimental studies on the origin and maturation of avian and mammalian red blood cells. Contributions t o Embryology, vol. 16, Carnegie Institution of Wash., pub. no. 83, pp. 165-226. DRINKER,C. K., I(. R. DRINKERAND C. C. LUND 1922 The circulation in the mammalian bone marrow. Am. J. Physiol., vol. 62, pp. 1-92. JORDAN, H. E., AND J. P. BAKER,JR. 1927 The character of the wall of the smaller blood vessels i n the bone marrow of the frog, with special reference to the question of erythrocyte origin. Anat. Rec., vol. 35, pp. 161-183. MAXIMOW,A. 1910 Die embryonale Histogenese des Knochenmarks der Saugetiere. Arch. f . mikr. Anat., Bd. 76, S. 1-113. PEABODY, F. W. 1926 A study of hyperplasia of bone marrow i n man. Am. J. Path., vol. 2, pp. 487-502. SABIN, FLORENCE R. 1928 Bone marrow. Physiol. Reviews, vol. 8, pp. 191-244. VAN DEE STRICHT,0. 1892 Nouvelles recherches sur la genhse des globules rouges et des globules blancs du sang. Arch. de Biol., T. 12, pp. 199344.