into virgin soil, there has been the greatest difficulty in obtaining evidence of its frequent occurrence in native races long in close contact with Europeans, e. g., in India, Africa, New Zealand, and Australia.

The second supposition, viz. that savage races living under their natural primitive conditions are not liable to developmental anomalies is contradicted by the fact that such anomalies are frequent, as Aschoff has already pointed out. As regards developmental anomalies, there can be no doubt that certain forms of malignant new growth, most commonly met with in the early years of life, are associated with them. It is a matter of moment that congenital malignant new growths of this nature also occur among the offspring of aboriginal races (fig. 1) still practically free from the influence of European civilisation.

Although no fluctuations in the occurence of cancer, sufficiently striking to arouse even so much as the suspicion of the introduction of an epidemic of cancer among aboriginal races previously exempt, have been discovered, still, there are very considerable differences in the numbers of cases recorded in various races of mankind. It is impossible to explain them satisfactorily at present; but they merit passing consideration here. In a preliminary discussion of this subject elsewhere it has been indicated why some of these differences may be only apparent, while others may or may not be real.

Those who would draw far-reaching conclusions from the frequency of cancer in Europe and the scanty evidence of its occurrence in savage races, may be warned that comparisons may be made only between a few European countries, and then not without reservations. Outside of Europe the reservations are probably of more importance in the case of America, Australia, Canada, Japan, New Zealand, and South Africa. In India and in the various races inhabiting British Colonies and Protectorates no basis whatsoever exists for statistical comparisons. The student of cancer has to be satisfied with the significance of the positive results of attempts to ascertain whether or not cancer occurs at all in savage races.

RECORDED DIFFERENCES IN THE INCIDENCE OF CANCER

IN EUROPE.

There are differences in the number of deaths assigned to cancer in different countries in Europe. The crude death-rate, i. e., the number of deaths to 1000 of population living, varies from 011 in Servia and 04 in Hungary to 1-29 in Switzerland. At the outset of a study of

FIG. 1.-Congenital glio-sarcoma of orbit in a child 5 weeks old, native of Gold Coast, West Africa.

From a Photograph forwarded by W. R. Henderson, C.M.G. late P.M.O.

European statistics it becomes evident that no basis for accurate statistical comparisons exists. In 1903 we pointed out the necessity for different countries adopting a uniform method of investigating cancer statistically before comparisons could be made, but at the same time we acknowledged the improbability of devising a scheme likely to secure this end. In France, Denmark, Sweden, and Bulgaria the causes of death are not tabulated except for the towns. In Norway only 50 per cent. of the causes of death were stated in 1881 as against 85 per cent. in 1901. To those who would fain draw far-reaching conclusions from the difference in the number of deaths recorded in civilised and savage man, it may come as a revelation that comparisons of the data for different countries, even those of Europe, are quite unjustified. But the untenability of all such too hasty conclusions may reassure those members of the public who have been unduly agitated by the wide dissemination of alarmist opinions in the medical and lay press.

It may be more than a coincidence that the largest number of deaths. assigned to cancer occurs in Switzerland, where medical inspection of the dead body is customary, and where in 1900 in fifteen of the largest towns autopsies were made already in so high a proportion as 25-7 per cent. of the total deaths.

Stated generally, the number of deaths assigned to cancer increases from one country to another in a manner parallel with the increasing accuracy of the vital statistics of the several countries, and the low death-rates in Servia, Hungary, and Spain are probably the result of under-statement. Thus, to take two instances which differ less markedly from the figures for England, the statistics for Ireland are known to be less reliable than those compiled for England and Wales, and the recorded death-rate from cancer is 0.79 as compared with 0.92, although one would expect the opposite relation from the high proportion of the Irish population which attains to old age, for, as is well known, the frequency of cancer increases with advancing years. The recorded death-rate from cancer in Prussia is also lower than in England and Wales, being 0.70, i. e., less than that of Ireland; but here again the Prussian statistics are acknowledged to be not so accurate. The Prussian statistics have greatly improved in recent years, since many Prussian towns introduced certification of the causes of death by medical men, and the population of the towns has increased pari passu, both relatively and absolutely, with the growth of the population.

Throughout Europe disturbing factors come in play, minimising the number of the records of deaths from cancer available for statistical

The pre

purposes in some instances, and augmenting them in others. judices of superstition and ignorance, of religious beliefs (not necessarily quá ecclesia), combined with the ease or difficulty of access to educated medical advice, play no unimportant part in determining the number of cases of cancer recognised and recorded. The pathological value of the records of death from cancer is likewise by no means uniform.

If one be justified in dismissing the differences between the rates of mortality in Ireland or Prussia, and England and Wales, as expressions of apparent rather than of real differences in the absolute incidence of cancer, the differences obtaining in the case of other European countries, e. g., Catholic Spain with a death-rate of 0'44, and Protestant Holland with one of 101, cannot have any greater importance attached to them. The records of deaths from cancer are obtained by ways so dissimilar as registration of deaths by the layman or laywoman, the priest and the doctor; moreover different statistical methods of unequal accuracy are employed in different countries. Therefore great caution is necessary in assuming that the final statistical results are comparable and reveal real differences. In any case, the magnitude of the recorded differences loses much of its apparent significance.

Owing to the supposition that sarcoma may differ essentially from carcinoma, or, from mere striving after accuracy, the attempt is made in some statistics to deal separately with the deaths falling under the two categories, e. g., in the German Cancer-Census of October 15th, 1900. This endeavour probably introduces grave errors. With Murray we have shown that the one form of malignant new growth is as difficult to recognise clinically in hospital patients as the other, and also that the probability is great that carcinoma and sarcoma both increase with advancing years: a conclusion to which Weinberg has come independently. Experiment has also shown that sarcomata and carcinomata obey other general laws distinguishing them from the known infective diseases.

In other statistics regarded as highly accurate, new sources of fallacy are introduced, which point to an over-estimate of the frequency of cancer, e. g., new growths may be grouped together with other causes of deaths from a false idea of adding to their completeness, as in Switzerland, where for males, all fatal diseases of the prostate gland are grouped under the heading of "cancer" of that organ. An enlarged prostate is a common disease of old men, but it is not necessarily a "malignant enlargement, and its sequel of cystitis, infection of the kidneys, systemic consequences, and death when beyond treatment, should not be charged invariably to cancer.

Two factors seriously diminish the value of the records of death from cancer. One factor is the manner in which the certification of deaths is effected, and the extent to which this is solely the duty of medical men, as in some States, or, merely the ignorant opinion of a layman, after viewing the body, in others. Prinzing cites two instructive cases. In one, an ignorant peasant who performed this duty returned all deaths as due to cardiac failure, while another returned fifty deaths from diphtheria in Tilsit, at a time when even illness from diphtheria was quite rare. The other factor is the variation in the number of persons surviving to higher ages in different communities and the extent to which the actuarial corrections necessary to render comparisons valid are made, always of course provided the records of deaths from cancer and the other vital statistics of the population are sufficiently numerous and accurate.

The improvements Drs. Farr, Ogle, and Tatham have effected in our own national statistics of cancer, by inquiring into vague statements of the cause of death, of themselves serve to show how much room there is for improvement in countries where laymen are, or were till recently, e. g., in Germany, entrusted with the declaration of the cause of death in the case of cancer-a disease often presenting insuperable difficulties to recognition by the most skilled clinician. The difficulty of diagnosing cancer has been pointed out previously with reference to hospitals in England, Scotland, and Ireland, and a study of the clinical diagnoses of over 8000 cases has confirmed its statistical importance.

The impression made by a study of the death-rates from cancer in European countries is, that comparisons of the results of the statistical tabulations of the records of death from cancer in different countries, with a view to establishing differences in its absolute incidence, appear to be entirely fallacious. As regards more limited areas in single countries, and differences between towns and country districts, the accuracy of the registration of the causes of death is everywhere behindhand in rural districts as compared with towns, and some countries do not even attempt to compile statistics outside of the towns, as, for example, Bulgaria, France, Denmark, Russia, and Sweden. The remarks made on differences obtaining in the rate of mortality from cancer in separate countries, appear to be applicable also, if in modified form, to the statistics of restricted areas in individual countries. The dimensions of the differences are not so great that they are incapable of explanation by (1) the varying difficulties in the way of obtaining accurate records of the numbers of deaths from the disease,