The purpura accompanying the two foregoing cases of sarcoimatosis would seem to find its explanation in the coexistence of several factors, the main feature being an involvement of the vascular system by the sarcomatous elements. There existed in Case I a direct lesion of the vessel wall whereby the sarcoma cells invaded directly the various coats, and were found mainly between the intima and the adventitia, dissecting their way, as it were, along these tracts in the vessel walls. There was further an extensive involvement of the perivascular lymphatics, from which point, indeed, it would seem that the sarcoma cells had invaded the walls of the vessels themselves. In Case II, moreover, not only was there a definite invasion of the lymph spaces near the vessels, but, furthermore, there was undoubted evidence of the existence of emboli of sarcoma cells in the lumina of the blood vessels; and in the immediate vicinity of such conditions hæmorrhages were invariably found. While some vessels, and indeed a great many, were quite free from such emboli, in others the lumina were completely occluded by spindle cells, so as to preclude the possibility that these were merely a collection of desquamated endothelial cells, such as is frequently found as the result of post-mortem changes. That such an embolic condition can exist is by no means an unreasonable supposition, and, while it is generally recognised that multiple sarcomata are usually made up of small round cells, in this case we have an undoubted example of sarcomatosis of the spindle-celled variety. There are numerous instances of this " embolic purpura," as it may be called, especially in French and German literature, the condition being associated with rheumatism, valvular lesions of the heart, and other diseases which induce directly or indirectly the formation of emboli. Krauss, Gimard, Leloir, and others have insisted with considerable emphasis on the embolic origin of many purpuric conditions, and in some instances they have verified their observations by histological examination. Leloir assumes that, in addition to the presence of the ordinary emboli and the changes in the vessel walls with desquamative endarteritis, the blood itself may be much altered chemically, and that in the cachectic conditions clots may be thrown down from the circulating blood and be carried onward to form capillary emboli, with resulting hæmorrhagic infarctions. Krogerer, some ten years ago, in examining the skin removed from patients with symptomatic purpura, found definite thromboses in the smaller veins, and even in the arteries. According to his view, the alterations in the vessel walls gave rise to slowed circulation and tendency to thrombosis, bringing about a liability to hæmorrhages. His plates bear out his theories regarding the thrombi, many of which show considerable organization. But a careful examination of the purpuric areas shows further that a mere invasion of the vascular system by sarcoma cells can not explain all the various blood effusions present. On examining the skin, for instance, in those areas where large irregular hæmorrhages had occurred, there was but little evidence of vascular invasion, while the emboli, on the other hand, seemed to exist mainly in the localized smaller and more circumscribed patches. One must therefore conclude that in such instances a combination of factors will alone afford a rational explanation of the purpura, and that in the general condition of the patient we shall find another cause for the enormous effusions of blood. In both of our cases there were high fever, cachexia, and a rapid progressive asthenia, all being the results of a sarcomatosis, and implying also grave alterations in the composition of the blood. From this we may infer an altered condition of the vessel walls, and hence probably a combination of circumstances sufficient to explain the incidence of hæmorrhage. The raised cutaneous nodules in our second case, some of which were hæmorrhagic, can not be regarded as pure sarcomatous metastases, for on microscopic examination they merely revealed hæmorrhage or necrosis, or both, and sometimes plugging of the vessels. There was nowhere in these nodules evidence of new growths. Such elevations, then, must have been produced rather by a temporary serous or cellular exudation coincident with or following upon the hæmorrhage—a probability which is emphasized by the fact that during the last days of the patient's illness many of the nodules diminished in size. Whether the œdema and infiltration were secondary to the embolic process in the subcutaneous vessels or whether they were merely coincident with the hæmorrhage would be difficult to decide. The ringlike spots, however, are of special interest, inasmuch as it has been shown that they have been present in more than one case of sarcoma. It is not impossible that such spots may be definitely related either to the embolic processes or to a direct invasion of the cutaneous vessels, though, so far as we know, there do not exist any experimental proofs to bear out such a theory. From what has been said, however, it is evident that the cutaneous vessels were plugged during the last few days of the illness, at a time when the walls of the smaller vessels and capillaries were already greatly enfeebled. The result of the embolic formation may therefore mean a decided deficiency in the supply of nutriment to the involved area, the collateral circulation naturally being poor under the circumstances. As soon, then, as the vessels had become plugged, the surrounding blood supply would be poured in to a limited extent, and, on meeting the enfeebled vessels, might possibly break through their thin walls, thus producing a zone of hæmorrhage around the area deprived of its normal nutrition. In other words, the condition may be regarded as in many respects analogous to that presented in embolic infarcts in regions with end arteries, central necrosis with peripheral congestion and hæmorrhage being induced, the latter being chiefly limited to the outer zone of the necrotic area. The cutaneous vessels under such circumstances may be regarded as end arteries in a functional sense, since the collateral circulation would be so diminished under the altered conditions that no complete nourishment could be afforded to the area supplied normally by the plugged vessel. Von Recklinghausen has directed especial attention to the occurrence of cutaneous hæmorrhages following embolic or thrombotic occlusion of peripheral arteries. The possibility of some toxic condition as a factor in the production of the purpura in our cases may also be suggested; but while we would not exclude this possibility, we are unable to find any positive evidence in its favour. Focal necroses, which are often associated with toxic and infectious processes, were present only in direct association with the hæmorrhages, and were not distributed in the liver, spleen, and kidneys in the manner characteristic of toxic infections. Nevertheless the absence of these necroses does not exclude the possibility of the existence of some form of toxæmia. Infection demonstrable by bacteriological examination was absent, and there is no reason to regard our cases as allied to the infectious purpuras. The thermic theory suggested by Fagge at all events finds no place in the production of the multiple tumours in our cases, inasmuch as in each instance extensive visceral growths had given rise to the metastases.