In a carefully documented discussion on mortality statistics, Van Buren1 points out a number of deductions which cannot be proved by such data. Entirely too much importance and finality, he says, are attached to many crude death rates. Thus, if calculations from the crude death rates for cancer over the twenty-five year period 1911-1935 are made, one would conclude that the mortality rate had increased over 41 per cent in the last year of the time series as compared with the first year. When corrected, however, for the effect of the changing age distribution, it develops that the increase was only 14.5 per cent—an entirely different picture.
Another common mistake in interpretation of mortality figures is the utter disregard of the fact that data for the expanding death registration area do not relate to the entire country for years prior to 1933. An example of the dangers inherent in