_Death rates per 1,000 births._—As shown above, the method of computation of death rates which gives the clearest picture of the hazards of childbirth is that which takes into account only the women giving birth to children in that year. This is the method in use in a large number of foreign countries. The advantages of the method are self-evident. A demonstration of the superiority of this method of computation is obtained by a study of the tables giving the death rates from these diseases for foreign countries. In certain countries, as for instance Belgium and Hungary, there has been in recent years an apparent fall in the average death rates as computed per 100,000 population, while the average rates computed per 1,000 live births have remained stationary or risen. This phenomenon is due, evidently, to a decline in the birth rate in these countries during these years, and shows how misleading the rates as given per 100,000 population undoubtedly are in countries with declining birth rates. Whether a fall in the birth rate has occurred in the United States is not known. If it has occurred in the registration area, it would mean that the slight rise in rates per 100,000 population between 1900 and 1913 means a greater rise in rates computed according to the number of births. Such an error might compensate for the opposite error due to the more complete registration of deaths from childbirth in the later years of this period. Miscarriages are not reportable in any country, although a number of miscarriages (as the term is usually defined) probably are reported as stillbirths in certain countries. The fact that women having miscarriages are not considered in the base would lead to a somewhat higher death rate than that which would express absolutely the number of deaths per 1,000 women at risk. COMPARISON OF THE CHANGES IN THE DEATH RATES FROM CHILDBIRTH IN CERTAIN FOREIGN COUNTRIES FOR THE YEARS 1900 TO 1913 Far more valuable than a comparison of average rates of foreign countries is a study of the rates of each country for a series of years in order to discover whether they are decreasing or increasing and to compare such changes in the various countries. While it may be dangerous on account of different countries, no such source of error is attached to the comparison of rates in the same country for a number of years. The period 1900 to 1913 (or the latest year for which figures are available) is a very short one for a study of a change in death rates. It would have been far more interesting to study the death rates for a long series of years in each country, choosing a period beginning before the introduction of methods of asepsis. But such a study for the complete list of countries considered was not thought advisable, because of the difficulties caused by variations in classification of causes of death in the earlier years.