Ordinary procedure is to compare the live births in a single calendar year with the deaths of babies under 12 months of age occurring in that same year, even though those who died may not have been born within the calendar year of their death. The infant mortality rates in this report, however, have not been computed on the usual basis, but for the purpose of securing greater accuracy in measuring the incidence of death this bureau has considered, in making the computation, only so many of the babies born in the year 1911 as could be located by its agents, and has compared with this number the number of deaths within this group of babies who died within one year of birth, even though some of these deaths may have occurred during the calendar year 1912. Infant mortality can be accurately measured in no other way than by means of a system of completely registering all births as well as all deaths. In 1911 the United States Bureau of Census regarded the registration of deaths as being “fairly complete (at least 90 per cent. of the total)” in 23 States, but the same degree of completeness in the registration of births was found only in the New England States, Pennsylvania, and Michigan, and in New York City and Washington, D.C. An exact infant mortality rate for the United States as a whole cannot be computed owing to this generally incomplete registration. In the 1911 census report on mortality statistics, however, the infant mortality rate is estimated at 124 per 1,000 live births. How this estimated rate compared with the computed rates for other countries is shown in the following summary: DEATHS OF CHILDREN UNDER 1 YEAR OF AGE PER 1,000 LIVE BIRTHS, BY QUINQUENNIAL PERIODS FROM 1901 TO 1910, AND ALSO FOR THE SINGLE CALENDAR YEARS 1909 TO 1910.[10]