i. e., of making them approximate to the absolute number; (2) the divergences in the methods of utilising the data for statistics; (3) the varying extent to which actuarial correction is made for the age-constitutions of the respective populations; (4) the different age-constitutions of the populations themselves; (5) the futility of conclusions based on too small numbers, which are liable to frequent reversal when the statistical results from one year to another over a long period of time are compared, e. g., in small districts, and in populations restricted in numbers or sparsely scattered over very extensive areas as in some British Colonies and Islands under British Protection.

We must guard ourselves from the charge that we deny the existence of real differences in different European countries and even in the less extensive areas comprised in them, or, that we intend to imply that with sufficiently accurate statistics all differences would disappear, and, the deaths from cancer appear as a constant function of the age of the population, more correctly of the senescence of the respective tissues of the individuals in any selected population.

It is obvious that the incidence of cancer in the organs of representatives of the different zoological classes of the vertebrates must differ, since structures peculiar to mammals are absent in the lower vertebrates. From our studies in animals it appears probable that different species of the mammalia are liable to special forms of cancer, from which others are exempt, e. g. the variations in the frequency of cancer of the mamma in different species of the mammalia are most remarkable.

As regards the incidence of special forms of cancer, we know the disease is not inevitable even when life is prolonged to its utmost limits, both in man and in animals. Why some individuals escape and others do not, it is impossible to say. It would be folly to assert without demonstration, or indeed without any evidence whatsoever, that the proportion of those escaping to those attacked is constant and invariable, and, that the fact of its being so is obscured only by the imperfections of our methods of recording the deaths from cancer and using them statistically.

The number of deaths assigned to cancer has increased from year to year in practically all countries. This increase appears most alarming when it is taken up as a national problem, e. g. in England, Germany, or the United States, without due regard to the universality of the phenomenon. Reference has been made above to the irregularities in the methods of compiling statistics from one country to another and the diversity of the attempts which are being made to fulfil the require

ments of accuracy.

The almost universal endeavour to improve the accuracy of statistics, has an important bearing upon the alleged "increase of cancer" of which mention will be made again below with reference to Japan.

The fact that cancer has been found in races of mankind formerly believed to be exempt and to be common in others in which it was said to be rare, shows how much the number of cases recorded depends upon the care exercised in searching for a disease of slow evolution, which does not obtrude itself upon the superficial observer by a striking and peculiar symptomatology. The study of the alleged increase of cancer involves a very complex statistical investigation. It can only be properly investigated on the basis of accurate and detailed mortality statistics of single countries, and it will be reverted to again in the special statistical report. Mention must, however, be made of the valuable contribution to its settlement which is being made in the Reports of the Registrar - General for England and Wales, by Dr. Tatham's tabulation of the cases of cancer for different sites of the body.

It may be added here that when the investigations of the Imperial Cancer Research Fund were commenced and the resolution come to, to study cancer in animals, doubts were expressed by some as to whether the difficulty of finding a sufficient number of cases would not prove insurmountable. The large number of cases of cancer, over a 1000, which have been observed in mice by various investigators during the past five years does not mean an increase of cancer in mice, but simply that it has been looked for with sufficient care and found. The increase in the number of cases recorded throughout the vertebrates has the same significance.

All it is intended to convey by a brief survey of the statistics of European countries is that the recognisable sources of fallacy in the data at present available, appear to carry more weight than the evidence that the differences they reveal between different countries, and in one and the same country at different times, are real. They have not the fundamental importance for the experimental investigation of the nature of cancer possessed by the results of those other statistical and biological studies which have moulded its present form.

DIFFERENCES IN THE RECORDED INCIDENCE OF CANCER

OUTSIDE EURope.

Complete or relative exemption from cancer on the part of the inhabitants of various parts of the world has been alleged on mere hearsay evidence. Such indefinite reports are principally relied on by those who assert that selected factors of European civilisation are causally connected with the recorded increase in the number of deaths from cancer in England.

The facts ascertained confirm the conclusion previously arrived at, that, taking England and Wales as a standard, there is a gradual diminution in the precision and completeness with which cancer is recorded, Scotland and Ireland showing less precision and completeness, and various outlying parts of the Empire still less of these qualities, till a minimum is reached among the natives of certain tropical colonies, e. g., British Central Africa and New Guinea, and in the case of islands with very small populations, e. g., St. Helena. The diminution proceeds pari passu with the increase in the difficulty of making observations on the occurrence of the disease, or of collecting a sufficient number of reliable preliminary data of the numbers of the population, the relative proportions of males and females, the distribution of the population at different age-periods, the crude birth- and death-rates. These considerations modify the importance to be attached to differences in the recorded incidence of cancer in Japan, India, Egypt, and among uncivilised or savage races.

The fact that cases of cancer have been found to occur where previously the disease had not been observed is of importance—the occurrence even of an isolated sporadic case being significant. Since accurate statistical comparisons are impossible, it is permissible to make others of a tentative and cruder kind which, although open to grave fallacy, are as instructive as any which can be made at the present time. The frequency of cancer in Japan is of special interest. In 1901, the German Cancer Committee seriously accepted the statements of Baelz that useful statistics did not exist in Japan, but that he would try to create them, and he asserted cancer was, without doubt, much rarer in Japan than in Europe. The informant went on to state: "This is at once evident from the relatively large number of cases in the small number of European inhabitants, as contrasted with the rarity with which the disease is observed in natives." Without referring to

the statistical fallacies which vitiate such a conclusion, it is sufficient to point out that, as a matter of fact, the official statistics of the Japanese Empire for that same year (1901) show, that out of a total of 959,126 deaths, 24,598 were returned as due to "cancer" and "other malignant growths".

We have to thank Baron Takaki for access to the Japanese statistics. The average death-rate from cancer for the five years 1899-1903 was 0-49, higher than in Servia, Hungary, and Spain. Baron Takaki modestly confessed that cancer was not so well diagnosed in Japan as in England †, nor were the mortality statistics of Japan as yet comparable to those of this country. The approximation of the Japanese figures to those for some European countries and British Colonies, for the statistics of which accuracy is claimed, is thus made all the more striking. The crude figures would indicate that cancer is twice as frequently a cause of death in England as in Japan. It must be evident, without further discussion, that this conclusion might require drastic revision if our information as to the "age-constitution" of the Japanese population were more complete, and if the registration of deaths were carried out as in this country. The future course of the cancer death-rate in Japan, as statistical methods improve there, should throw a most instructive light on the cause or causes of the increase which has taken place in the number of deaths attributed to cancer in European countries. In view of the exceptional position Japan occupies in relation to this problem, owing to differences of race, diet, climate, and perhaps also in the age-constitution and habits of the population, it is desirable to obtain more details as to the death-rate from cancer at different periods of life. The tabulation of the sites of the disease in different parts of the body as now published in the Reports of the Registrar-General for England and Wales should yield very valuable information. Fortunately, the Japanese recognise the national importance of the frequency of cancer among them, and there is every prospect of these improvements being effected. In Japan there is at the present time not only very active investigation of cancer, but of the two journals in the world devoted solely to cancer, one is published in Japan under the editorship of Professor Yamagiwa, Professor of Pathology in Tokio.

Exaggerated importance, not justified by fuller knowledge, has been

*The population of Japan in 1901 was 45,227,464.

† In Japan there are still a large number of unqualified practitioners of medicine.

attached also to assertions that cancer is practically absent in India, in Egypt, and among aboriginal races. In the course of the past three years over 2,000 cases have been reported from hospitals in India. Occurring in a population of 300,000,000, this number appears to imply

FIG. 2.-Carcinoma (scirrhus) of breast in woman born in the Sudan, where she had spent all her life. From a photograph by Capt. Ensor, E.M.C., forwarded by Dr. A. Balfour, Khartoum.

great relative exemption. Such a conclusion loses in cogency when regard is paid to the small number of hospitals, the onerous duties which a relatively small number of officials undertaking this additional work have to discharge, and the circumstances which restrict the number of native applicants for hospital advice. The patients belong