former had decreased from 1013 in 1887-89 to 916 in 1899, whereas the latter had increased from 5463 to 7853. Similar phenomena have been adduced from a study of the statistics of Paris, Stuttgart, and other places. The bearings of these results are obvious; if there has been no real increase in the cancer of external regions, there is every reason to believe that the increase recorded in internal cancer may be only apparent, and due to more accurate registration and diagnosis. It is difficult to be dogmatic on the subject, and the data available are probably not enough to permit of a definite conclusion; but, so far as they go, they certainly indicate that if there is any increase in cancer it is only a slight one, and that by far the greater proportion of the apparent increase is due to more accurate diagnosis. Similar arguments apply to the statistics from different countries. Before inferring from the crude figures, as given in the mortality returns, that cancer is more prevalent in one part of the world than in another, we must just ascertain whether the figures are really comparable, and must know, firstly, the age-distribution of the population, and secondly, the methods of diagnosis and deathregistration employed; and the latter is by far the more important. Applying this test, we are led to the belief that, for the white races at least, the local differences in the cancer death-rate are greatly exaggerated in the statistics, and that the more carefully registration is conducted the more the cancer death-rate will approach a fixed maximum, which does not greatly differ in different cases. Another question of extreme interest to the layman especially, deals with the question of the infectivity or contagiousness of cancer. There is no evidence that it possesses either of these properties. In a few cases a cancerous growth has infected a part of the patient's own body that has been kept in contact with it (e.g. a cancer of one eyelid has given rise to a similar disease in the other at a point where the two come into apposition), and in this sense the disease is undoubtedly contagious as regards the patient himself. Apart from this, all evidence points in the opposite direction. Numerous surgeons have attempted to inoculate themselves deliberately, with negative results, whilst thousands must have operated on cancers whilst suffering from cuts or abrasions on the hands; yet cases of disease acquired in this way are unknown; nor are attendants in cancer wards especially prone to the disease. A few apparent cases of direct transmission from person to person are on record, but their extreme rarity as compared with the great frequency of the conditions for their occurrence leads us to attribute them to coincidence or imperfect observation. The negative evidence is somewhat less strong with regard to the question of the infectivity of cancer, i.e. whether the hypothetical 'germ' of cancer can affect certain localities, so that persons who are in other respects suitable candidates for the disease may acquire it if they live for a sufficient time in these regions, escaping it if they reside in a place where the infection is absent. This problem is more difficult. It is obviously no use to compare one locality with a distant one, for here the differences in the death-rate from the disease may be due to differences of race, habits, environment, or many other factors, or may, on the other hand, be simply due to the abundance or scantiness of the supposed infective agent. A solution of the problem is only possible from a study of the distribution of the disease in a homogeneous population living under similar conditions. Investigations of this sort brought to light the existence of cancer-houses,' in which the disease appeared to be unduly frequent, thus tending to support the theory of its infectivity. It was supposed that a case happened to occur in such a house, which became saturated with the infective agent and subsequent inhabitants became infected, just as is known to be the case so often with phthisis. That such actually happens has long been an article of belief with the lay public, but it has not been investigated scientifically until comparatively lately, the attention of the medical world being apparently directed to the question by Guelliot.* The subject was investigated by D'Arcy Power, by a committee appointed by the Birmingham branch of the British Medical Association in certain parts of the Midlands, by Behla in Luckau, and by others, all of whom have found this apparent tendency of cancer to attack * Guelliot, Gaz. des Hôpitaux' (1892), No. 139. Vol. 213.-No. 424. F the inhabitants (often unrelated) of certain houses. The subject was carefully examined by Symons in regard to the distribution of the disease in Bath. He found that during thirty-two years there had been 1319 deaths from the disease. In 1111 of these there had been no other case in the house during this period; in 95 cases there were 2 deaths, and in 6 cases 3 deaths in the same house. These figures might at first sight be considered to prove the existence of cancer-houses,' but Symons analysed them on the basis of the mathematical theory of probabilities, and found that they were explicable without any such assumption, the figures being little more than if they were due to pure chance. He examined other recorded cases of cancer-houses' in the same way with similar results, and it is now generally held that cancer is not a house disease, and that there is no evidence that it is communicable from person to person. This, it must be pointed out, in no sense disproves the possibility of its parasitic origin. The question of the influence of the geological configuration of the soil has been carefully studied, and the results have in general confirmed the views formulated by Haviland,† that cancer is especially a disease of low-lying districts, especially if near to a river which occasionally floods its banks, and in flat marshy districts, the localities especially free from the disease being high and dry situations. Behla found similar results in a single town (Luckau), the low-lying and damp districts of which were very subject to cancer, whilst a high and dry suburb was free from the disease during the period in question. So also the Birmingham Committee, in their study of the disease in the Midlands, and Kolb in Southern Germany, Switzerland, and Austria, and Foucault in Fontainebleau, all considered their results to prove the disease especially prevalent in damp soils, and this view was generally accepted. Later studies, especially that of Scott in Essex, show that these results are certainly not invariably true, and, according to the studies of Höber, the conditions of the soil have but little effect on the Symons, 'Public Health,' Dec. 1898. Haviland, Geographical Distribution of Disease in Great Britain,' 1892 (first ed., 1875). Scott, 'British Medical Journal' (1900), vol. ii, 420. mortality from this disease. The question is one that calls urgently for careful investigation on a large scale, and by mathematicians competent to analyse their results by modern methods. The study of cancer on the large scale, by statistical methods, has so far led to disappointing results. It would appear at first sight easy to determine, at least the predisposing causes of the disease, by a study of its distribution with regard to the habits and habitat of its victims, their personal history and heredity, but such studies, with hardly an exception, have led to indeterminate results. The investigation of the disease in one district has led to conclusions which appear to be definitely proved, such, for example, as the apparent connexion of cancer with a damp subsoil. This has been studied in other districts, and indeterminate or totally opposite results obtained. As another example, we may quote the question of heredity. That there was a hereditary predisposition to cancer in certain families, and that this predisposition played a part of great importance in the etiology of the disease, was regarded as proved beyond possibility of doubt; but more recent investigations have shown that previous observers have been largely influenced by coincidences (such, for instance, as the celebrated case of the Buonaparte family), and that, when carefully prepared statistics are examined by modern methods of mathematical analysis, they afford no support to the theory. Indeed, some results obtained at the Middlesex Hospital and analysed by Karl Pearson* appear to indicate that a person with a family history of cancer is, if anything, slightly less liable to contract the disease than a person without such a history. But the difficulties in the collection of the data necessary for such investigations are so great that all such conclusions must be received with caution, as Pearson points out. It was the experience of such blind alleys in the study of cancer that led pathologists to turn with relief to the experimental study of the disease, rendered possible by the demonstration of the fact that cancer is inoculable in the lower • Archives of the Middlesex Hospital,' vol. ii (second report from the Cancer Research Laboratories (1904), pp. 104, 127). animals, provided always that a tumour arising from one animal is inoculated into another of the same species, e.g. a tumour arising from a mouse must be inoculated into a mouse, that from a rat into a rat, and so on. We are here on surer ground, and, admitting that the cancer of the lower animals is of the same nature and governed by the same laws as the disease in man, we may expect results of far greater value from studies of this nature, in which all, or nearly all, of the fallacies incidental to clinical and statistical studies can be avoided. An enormous amount of work has already been accomplished in this direction, of which we shall only give the outline, as it is not of great interest to the general reader. It is found that for the disease to be transmissible from one animal to another, portions of tissue containing whole cells must be transmitted; if the cells are ground up and filtered the filtrate becomes innocuous. This is quite different from the results obtained in dealing with diseases known to be of bacterial origin, in which such a procedure would simply liberate the bacteria, which would then infect their new host. For the inoculation to succeed, living cells must be inoculated, and the continued growth of the tumour in its new host is simply due to the continued growth and subdivision of these transplanted cells, to which the new host simply supplies nourishment. The inoculations may be carried on from one animal to another, apparently without cessation, thus leading us to the conclusion that, given suitable nourishment, the cancer cell is capable of living and multiplying indefinitely. The most important results are undoubtedly those which go to show the existence of immunity to cancer; their importance arises from the fact that they afford a gleam of hope, at present perhaps but a feeble one, but still a gleam, that the disease may be prevented or cured in the future. That natural immunity occurs appears from the fact that the mice of a certain region may be refractory to the cancer from another district: thus with one particular strain of mouse-cancer successful inoculations were obtained in 97 per cent. of cases where Berlin mice were used, in 24 per cent. with Hamburg mice, whilst negativę |