We may, therefore, reasonably conclude that there was an average number of at least 19.36 per 100,000 (from 1906 to 1910 this amounted to 32.1) in which the primary cause of death was associated with mental diseases, an exceedingly conservative estimate. This does not take into consideration the deaths due to senility (15.5) or suicide (12.8), conditions which might very logically be included for obvious reasons. It is, of course, well known that the psychoses rarely, if ever, appear in the death certificates as a primary cause of death. As a matter of fact, they are not always shown in the secondary causes. Information on this subject is still less satisfactory from a statistical point of view. During the year 1917 (contributory causes have not been reported since that year) there was a total of 1,066,711 primary causes of death shown in the registration area and only 372,291 contributory causes. Of this number the following may be classified as having been associated with psychoses: _Disease_ _Primary _Contributory Cause_ Cause_ Encephalitis 620 904 Meningitis (total) 6,673 6,815 Softening of the brain 888 722 General paralysis of the insane 5,248 648 Other forms of mental alienation 1,651 3,895 —————— —————— Total 15,080 12,987 The contributory causes definitely showing mental diseases constitute only 3.4 per cent of the whole number, and the death rate for 1917, including both primary and contributory causes suggestive of probable psychoses, was 37.2 per 100,000. This would indicate that the number of deaths from mental diseases shown in the primary causes represents only about fifty-three per cent of all mental cases which are actual factors in determining the death rate of the community. A comparison of these figures with the number of cases dying in hospitals shows that they cannot be looked upon as determining the percentage of the general population showing psychoses. Of the 1,952 persons dying in the institutions for mental diseases in Massachusetts in 1919, approximately nineteen per cent showed the psychoses in the primary causes of death. This percentage would probably be fairly constant throughout the country. It is, of course, a well recognized fact that the death certificate at best is not beyond suspicion and does not furnish information regarding the cause of death which can be accepted without question. Dr. Richard C. Cabot[1] has made an elaborate study of errors in diagnosis as shown by autopsies. His work shows the following percentage of diagnostic accuracy: