countless times, not only in the spontaneous tumours but also in those obtained by inoculation. It developes in such a way that precursors of the delicate stroma, and with them young blood-vessels, penetrate into the large disorderly epithelial masses; thus epithelial groups, between which a hollow space may have arisen already in anticipation as it were, are segregated. The whole epithelial complex becomes split up by the invading connective tissue. Acini may then develop in the alveoli formed by the groups of epithelial cells thus divided up. The process is comparable with the formation of Pflüger's primordial 'egg-tubes" in the ovary, or with the development of the embryonic or strumous follicles of the thyroid gland. Apolant has drawn attention to the remarkable similarity which a fully developed adenomatous tumour presents to a struma parenchymatosa of the thyroid gland. It appears to us that the resemblance may be pursued still further, for the development of the follicles of the thyroid gland follows a parallel route. If the epithelial production takes place very rapidly large epithelial accumulations and masses arise of which the partition is due to the ingrowth of invading connective tissue, and in which the appearance of glandular spaces coincides with the occurrence of secretory processes. The whole process may be comprehended by characterising its varying features as different stages in the differentiation of one species of cell which presents a carcinomatous formation at one time in consequence of the accumulation of compact cell-masses, and at another time, an adenomatous formation. Only in this way can we understand the alternations in structure of the same sporadic tumour, alternations which repeat themselves also in the lung metastases, as well as in the propagated daughter tumours. The question of malignancy ought to be decided not according to whether the rapidly proliferating epithelial cells arrange themselves into acini or into solid alveoli, but according to their biological behaviour. As Apolant has observed, variations in the rate of growth determine to a certain degree also variations of structure. This observation in itself assists us materially to understand properly the significance of the histological structure. The epithelial cells will differentiate faultily when growing rapidly, and in consequence assume a garb apparently more malignant histologically than that which they assume when growth is slower. But, I have ascertained by making examination of the histological appearance of my hæmorrhagic tumours in test experiments, that this relation between structure and rate of growth does not always obtain. When I compared the histological structure with the rate of growth anomalies manifested themselves.

A

whole series of other factors require consideration, such as the behaviour of the connective tissues and the general constitution of the individual. After influencing the latter by preliminary (immunising) treatment, Apolant observed what he describes as reversion of an alveolar carcinoma to a tumour with adenomatous structure. Whether a biological change is associated with this structural change cannot be decided at present. For my hæmorrhagic tumours I could detect no connection between structure and transplantability; starting with the same material the same differences in transplantability were revealed in the alveolar tumours as in the adenomatous. One of the most readily transplantable of hæmorrhagic tumours (No. 39) shows, even in later inoculated generations, a structure pronouncedly adenomatous for the most part (fig. 8). Metastasis formation is just as frequent in adenomatous as in alveolar tumours. A partial immunity has not always that effect on the structure assigned to it by Apolant. I compared a tumour growing rapidly in a normal mouse with another of the same series, which by way of exception had developed in a mouse in which a previous inoculation had been negative; the latter tumour had remained significantly smaller, but on examination was found to have a pronounced alveolar structure in contrast with the markedly acinous structure of the control tumour. Researches on this point are still far too scanty; we can only say with certainty that the inter-relations between structure and biological behaviour are undoubtedly influenced by very different factors of which only a small portion have hitherto lent themselves to analysis.

As the outcome of researches on human tumours mainly of the thyroid gland, I have already expressed the opinion that the histological structure of a tumour does not of itself permit of a decision as to its innocent or malignant nature. There are thyroid tumours exhibiting an absolutely benign structure when considered from the histological standpoint, and yet in their powers of destructive growth, and tendency to form metastases, etc., they are not surpassed in malignancy by any carcinoma. As I have demonstrated already elsewhere, we must acknowledge such tumours to be carcinomata whose cells retain the capacity for more advanced differentiation. Transitions are provided by tumours which exhibit in some parts a well-marked carcinomatous structure, and in others the structure of a simple goître, as, for example, in metastases in the lungs or lymphatic glands. Eberth has already characterised this condition as "a change for the better." The

* Umkehr zum besseren."

500

FIG. 8.-Mouse. Acinous structure of tumour 39 in fourth generation. X1

50

0.

FIG. 9.-Mouse. Tumour Transplantable spontaneous hæmorrhagic tumour. Cf. fig. 5 in

45