The main purpose of this paper is to attempt to establish the thesis that young endothelium has a hemogenic capacity, by showing that intra-embryonically also conditions occur similar to those described for the yolk-sac. These conditions relate to the endothelial origin of hemoblasts, which through a close series of developmental stages can be traced into erythrocytes.
The transition to intra-embryonic conditions next to be described may be made by way of figure 4, which represents a binucleated elon gated hemoblast, from a yolk-sac vessel, about to separate from the endothelium with which it is still in part intimately connected and from which it has undoubtedly differentiated. The binucleated cell is appar ently also about to divide into two hemoblasts, thus consummating the amitotic process.
The next evidence for hemogenic capacity on the part of the endothe lium pertains to certain phenomena in the abdominal portion of the dorsal aorta, the so-called "cell-clusters." Similar clusters in mam malian embryos were first (1909) observed by Maximow (ii) in the rabbit embryo.' Dantschakoff (3) had already reported comparable structures in the chick embryo. Minot (18) later (1912) described them also for human embryos between 8 and 10 mm. Emmel (4) subsequently (1915) studied more in detail these clusters in pig, rabbit, and rat embryos. At the same time I was making a detailed study of the aortic cell-clusters in the 10 to 12 mm. pig embryos, and had come to essentially the same tentative conclusion regarding their significance as Emmel, namely, that they were masses of hemoblasts differentiating from the endothelium. In the 10-mm. pig embryo these clusters are abundant and very large, some consisting of 100 or more cells; they are practically limited to the ventral portion of the aorta (fig. 24). Proximally they are in continuity with the endothelium and the sub jacent mesenchyma. Peripherally they consist of typical hemoblasts and young erythroblasts. The former are characterized chiefly by a deeper-staining basophilic cytoplasm, the latter by a lighter-staining acidophilic cytoplasm. Centrally transition stages occur between endo thelium and hemoblast. Occasionally the clusters contain a central core of only slightly differentiated endothelial cells. A number of the cells may be in mitosis, and some exhibit phases in nuclear amitotic division. The mongoose material is of special value in that it shows the earlier stages in the formation of these clusters and thus gives the key to their proper interpretation. The pig material showed a progressive size increase between the 5 and 12 mm. stages of develop ment. The mongoose material shows the same thing. In the 7-mm. embryo the clusters are somewhat larger and the peripheral cells more differentiated than in the 5 mm. embryo.
Figures 5 and 6 show typical small clusters, the first from the ventro lateral wall of the aorta, the latter from the mid-ventral line. Figure
5 gives the appearance of a buckling of the endothelium into the lumen of the aorta, the peripheral cells of the invaginated area assuming hemoblast characteristics. In the cluster, figure 6, the peripheral cells have progressed still further along this line of differentiation, and the proximal pole shows an increase of amitotic proliferative activity and transition phenomena on the part of the endothelium. In the case of the larger clusters the subjacent endothelium has in some instances undergone considerable thickening, being frequently three layers thick.
The close association between the cluster and the endothelium (figs. 25, 26, and 27) through a portion showing a transition between endothe lium and hemoblasts, coupled with proliferative activity on the part of the constituent cells, should render unnecessary any discussion as to whether these clusters may not be groups of hemoblasts deposited from the circulating blood and caused by pressure to adhere to the vessel wall. The latter interpretation was given by Minot (is) to these clusters in human and rabbit embryos, in opposition to Maximow (II), who described them in the rabbit as arising by the proliferation of the endothelium. Minot's objection to Maximow's original inter pretation was based on his failure to observe either in the human or the rabbit embryos any continuity between the protoplasm of the endothelial cells and that of the "mesamceboids" of the cell-clusters, or any considerable number of mitotic figures in the endothelial cells in the neighborhood of the clusters; "and, finally, because the endothe lial nuclei are differentiated, while the nuclei of the cells of the clusters are not differentiated." The mongoose material, however, shows a definite cytoplasmic continuity between endothelial cells and the proximal cells of the clusters. Moreover, while endothelial mitoses are rare in the neighborhood of the clusters, various stages in the amitotic division of endothelial nuclei are abundant. As in the case of certain other tissues of various forms undergoing rapid growth under certain conditions (e. g., blastoderm of pigeon, Patterson; tendon cells of new-born mouse, Nowikoff, etc.), the endothelial cells may here proliferate by the amitotic mode. Finally, the nuclear differentiation of the subjacent endothelial cells is of a lesser degree, judged especially by the small size of the nuclei and their spheroidal shape than that of the more peripheral hemoblast transition elements, many of which contain oval and kidney-shaped nuclei. Further countervailing evi dence to the position that the clusters are accretion products from the circulation is detailed in another paper (9) and need not be repeated here.