Were the tabulation of the census records of the infant deathrate per 1,000 of living population under 1 year at the end of each census year unaccompanied by tables based on other standards of measuring infant mortality, affording ready means of checking the various figures and thus measuring the probable margin of error, it would be comparatively valueless. For, as previously stated, the census authorities have often admitted that the data regarding the number of living children under 1 year of age are utterly incomplete and inaccurate. As Dr. Farr put it in discussing the mortality of infants (Supplement to 25th Annual Report of the Registrar-General, pp. v-vi), “The infants in the first year of life are to some extent mixed up with infants in the second year of age." And, as Dr. Billings explained in his remarks regarding infantile mortality published in the Report on Vital and Social Statistics in the United States at the Eleventh Census (Part I, p. 21): "Unfortunately, these data are incomplete and inaccurate, not only for the United States as a whole, but for all parts of the registration area. This is due to the fact that the number of children returned as being under 1 year of age is too small in every locality, owing partly to omissions in the enumeration, and partly to the tendency of those reporting the age of infants to report those who are between 9 and 12 months of age as being, in round numbers, 1 year old, which last is a defect common to the censuses of all countries."

The apparent living population under 1 year of age for these reasons being much understated in the returns, of course the alleged rates of deaths under 1 to the living population at that age are almost invariably in excess of the actual rates on that basis: hence the table worked out on those lines would be of little value by itself. The tabulation of rates of infant deaths per 1,000 deaths at all ages probably is open to less objection, but that table, too, is open to much more intelligent interpretation when supplemented with tables prepared on other bases of measurement, and especially with one laid out on the generally accepted plan of death-rates to births. But for the presence of the last-named table, it would not have been considered

worth while to present in this article the two other tables above
mentioned.

The death-rate of persons over 1 year of age is a natural com-
plement to the death-rate of infants under 1 year, if both are
prepared on a common basis, and materially aids in deter-
mining whether a high infant mortality is presumably, in part
at least, due to unsanitary conditions or to independent causes.
In none of the works on the subject of infantile mortality which
the writer has examined does such a tabulation appear, and by
way of adding at least some information to the statistical rec-
ords of the subject the following table has been prepared, on
the basis of the census reports for 1880, 1890, and 1900:-

TABLE XII.

DEATHS over 1 YEAR OF AGE AND RATE PER 1,000 POPULATION OVER 1 YEAR IN 1880, 1890, AND 1900 IN THE REGISTRATION STATES OF 1900, ACCORDING TO THE TENTH, ELEVENTH, AND TWELFTH CENSUS REPORTS.

In so far as the mortality reports of the Tenth, Eleventh, and Twelfth Censuses contribute any really material data to the records of infant mortality in the last three census years in the States recognized as registration States in 1900, the preceding table practically rounds out the information therein obtainable, and it might now seem to be in order to make a comparison of the infant death-rates in the several registration States as measured by the standards of the various tables which have been presented. By so doing, at least an approximate idea of the actual relative rank of the States in question in point of their respective infant death-rates may be obtained, and, possibly, some information of working value be contributed to the rapidly growing bibliography of the subject. Perhaps the shortest and most effective means of reaching this end will be that of assembling at close contact the infant death-rates of each State according to the various standards of calculation employed in the preceding tables, and attaching in each case the numeral showing the relative rank of the State in question from that point of view. In compact form here are the results of an inquiry shaped on those lines, eliminating the District of Columbia for reasons previously explained:

TABLE XIII.

A COMPARISON OF THE INFAnt Mortality RATES AND THEIR COMPLEment, the Death-rate at ALL AGES over 1 YEAR-IN 1900, 1890, and 1880 IN THE REGISTRATION STATES OF 1900, AND THE RelaTIVE RANK OF EACH OF THE NINE STATES IN THE ORDER OF THOSE Death-raTES.

Rate. Rank. Rate. Rank. Rate. Rank. Rate. Rank. Rate. Rank. Rate. Rank.

Rate. Rank. Rate. Rank. Rate. Rank. Rate. Rank. Rate. Rank. Rate. Rank.

A somewhat noticeable feature of this tabulation of comparisons is the fact that, materially differing though the infant death-rates of any particular State do in any one census year as measured by the ratios to births and to living population at the age of 1 at the end of the census year, in every case the relative rank of the State as determined by the two standards remains the same in all three census years. For instance, Massachusetts ranked first in 1880, third in 1890, and second in 1900 in point of both death-rates to births and to living population under age 1 at the end of the census year; New Jersey ranked second in both 1880 and 1890, and fourth in 1900, by both measurements, and so on. In a general way, it might be expected that there probably would be no radical shift in the ranking of the States from census to census in point of either the number of births or the population under age 1 at the end of the census years, but the coincidence of each State's ranking remaining the same in any census year, whether measured by the rate of infant deaths to births or to surviving infants under age 1 at the end of the census year, is at least passing strange, taking into account the habitual and historic inaccuracy in the reporting of living infants under age 1.

Of course, if the infant migration and emigration in a census

year were disregarded, if the infant deaths during the census year were restricted to babies born during the year, and if census returns were complete and absolutely accurate, the number living under age 1 at the end of the year would be the exact complement of the number dying during the year. But the infant migration and emigration cannot be disregarded. A minor but considerable percentage of the infant deaths in any year are those of babies born in the later months of the previous year; and the returns for births, deaths, and living population by ages-especially for population under age 1-are, and always have been, notoriously incomplete and inaccurate. Hence the absolute uniformity of the ranking of each State in the last three censuses, whether measured by its infant mortality rate to births or to living population under age 1, is at least worthy of note.

Unfortunately, the figures of most of the registration States, so far as the infant death-rate to births is concerned, have been open to the suspicion of too glaring inaccuracies—at least up to a very recent period-to warrant any attempt to make comparison between them and those of the European countries with long-established registration systems. But the census returns for these States in 1880, 1890, and 1900, herewith presented for what they are worth, are none the less worthy of a careful study from various view-points by those interested in the subject of infant mortality.

In this paper the writer has aimed to supplement the work of the numerous medical experts, who have long been probing the puzzling problems of infant mortality, by bringing together from various sources and presenting in compact form the most reliable statistical information now obtainable which would warrant some definite conclusions as to the rise or fall of the infant mortality rate in recent years throughout the world, and the apparent present tendencies of the infant deathrate. It has seemed possible that the presentation of specific information on these lines might provide sound foundations, in the way of authoritative facts and figures of international scope, for the widely extended movement now being earnestly