EDWARD BUNNELL PHELPS, M. A., F. s. s. Iti? 



Even were these records to be taken as absolutely worthy of 

 credence, the apparent decrease in the three years ending with 

 1908 would indicate a decrease of less than two infant deaths 

 per hundred living births in recent years as compared with the 

 average for the quarter-century immediately preceding. But 



GO. 



AVERAGE FOR 31 LEADING FOREIGN COUNTRIES AVERAGE FOR CONN..ANDMASS.ANDNCWYORK, 



FOR TWENTY-FIVE YEARSJ88H905.INC FORSIXYEARS.I904-I909INC. 



154 DEATHS PER 1.000 BIRTHS 139 DEATHS PER 1.000 BIRTHS 



CHART I. Infant mortality per 1,000 living births. Bach circle represents 

 1,000 births black sections, percentages of deaths under one year. 



the figures in question can scarcely be taken at their full face 

 value, although cited from the official vital statistics of the sev- 

 eral countries, for the reason that in practically all countries 

 there has been a larger percentage of improvement of late years 

 in the registration of births than in that of infant deaths. And, 

 as the infant death rate is calculated by dividing the number 

 of deaths under 1 by the number of registered births, the larger 

 the registered percentage of births the larger will be the divisor, 

 and the smaller will be the quotient or apparent infant mor- 

 tality rate even though the actual numbers of births and infant 

 deaths be identical for the two periods under comparison. It 

 is evident, therefore, that apparent declines in the annual infant 

 death rates of the various countries, States and cities cannot be 

 taken as conclusive, and should not so be taken unless care- 

 fully investigated and supported by corroborative data. This 

 defect in the commonly accepted statistics of infant mortality is 

 especially prevalent in those for the registration States and cities 

 of this country, as I shall later on endeavor to make clear in this 

 paper. 



For the time being, setting aside this element of probable 

 error, and noting only the essential showing of the world-wide 

 tabulation, it seems to be safe to assume that in the civilized 

 world at large outside of the United States not less than 13 out 



