resulted from diseases or traumatisms leading to admission to sick report in the month indicated even though death did not occur until some later month. Some cases of admission to sick report in 1917 resulted in deaths in 1918. As the deaths in the later year have not yet been tabulated, the deaths in it from 1917 admissions could not be added to the 1917 record. Hence the monthly rates in Table 7 at least for the closing months of 1917 are somewhat too low. For 1915 the deaths have been charged uniformly to the month in which the death occurred rather than to the month in which the case was admitted to sick report. The deaths charged to any month have been multiplied by the ratio between 365 and the number of days in the month to get what would have been the annual number of deaths, had the conditions of that month continued for a year.

The figures in the last column of Table 7 show that after March, when the Army began to increase rapidly in numbers, the death rate dropped for six months with only one slight break, but rose again between September and November and in the latter month was higher than at any time of the year except February. For the three months of August, September, and October, the rates were below those for the corresponding months of 1915, but for the other nine months and for the year as a whole the 1917 rates were somewhat higher. In view of the conditions with which the Army was struggling in 1917, the record as a whole is a gratifying witness to the efficiency of the measures taken to protect the health of the soldiers.

II. THE HEALTH OF THE ARMY ANALYZED.

The connection between the health of the Army and geographical conditions is far more difficult to establish and measure than changes in the health of the Army from one year to another. The main reason is that many other differences affect health besides those dependent upon location and that they also, or many of them, vary with location. An example from racial statistics will illustrate my meaning. In 1916 the proportion of colored among the enlisted men in the Philippine Islands was 6 times as great and in Hawaii 16 times as great as in continental United States. As the death rate of the colored, in civil life at least, is higher than that of the white, no comparison between the mortality of troops in these three areas which failed to inquire into the differences in the racial elements would have much meaning for sanitary science. The best way to deal with this difficulty is probably to examine and if possible to eliminate the simpler or more obvious causes before investigating the more complex or elusive ones. Following this method we may first examine the influence of race upon army health and then return to the more difficult subject of geographical conditions.

1. THE HEALTH OF THE ARMY BY RACE.

Four racial strains are distinguished in the statistics of the American Army, the white, the colored, the Porto Ricans, and the Philippine Scouts. The table following shows the death rate of each of these groups during the decade 1908-1917.

TABLE No. 8.-Death rate per 1,000 mean strength in United States Army (enlisted men), classified by race, for the decade, 1908-1917.

Table 8 shows that the death rate among colored enlisted men for the decade exceeded that among whites by about two-thirds, but that the death rate among Porto Ricans and Philippine Scouts was less than that among whites. What we know about the mortality of whites and colored in the population of the United States agrees in general with the preceding results, but it is certain that the mortality of the population of Porto Rico and the Philippines is greater than that of American whites. In 1915-1916 the death rate in Porto Rico was 22.0 per 1,000, that in the Philippine Islands in 1916 was 25.3, while that of the United States registration area was not far from 14 and that of the whole United States probably did not greatly exceed the same figure. We are put on inquiry, therefore, to find some reason for the lower mortality of Porto Ricans and Filipinos in the Army although the civilian death rate in those islands is much higher than in the United States. For Porto Rico the number of deaths, only 34, may be too small to yield a significant rate but for the Philippines no such explanation can apply. Are the Philippine Scouts subject to a more rigorous selection before admission or more readily discharged in case of disability?

The question deserves further examination which can be given by applying other health tests to these four racial groups. First, how do they compare in the ratio of admissions to sick report to mean strength? The following table answers the question.

TABLE NO. 9.-Admissions to sick report in United States Army (enlisted men), classified by race, for the decade, 1908-1917.

The figures in Table 9 strengthen the conjecture that the death rate of Porto Ricans in Table 8 rests on so narrow a basis as to be untrustworthy. In the present table sickness is shown to be a little more common among Porto Rican troops than among either white or colored. The sickness rate of the colored is greater than that of the whites but the difference in the amount of sickness is much less

than the difference in the death rate of the two races. With reference to the Philippine Scouts the evidence supports that in Table 8 and indicates that they are much freer from sickness than any one of the other three groups.

It is possible that the very low death rate and sickness rate of the Philippine Scouts are due to the fact that those whose health is impaired either temporarily or permanently are discharged for disability earlier and more regularly than soldiers, white or colored, under similar conditions in the United States. To determine whether this is actually the case the following table has been prepared showing the rate of discharge of each of the four racial groups.

TABLE NO. 10.-Rate of discharge for permanent disability in United States Army (enlisted men), classified by race, for the decade 1908-1917.

The figures show that the rate at which both Porto Ricans and Philippine Scouts are discharged for permanent disability is less than one-half as great as the rate at which whites and colored are discharged and thus give a negative answer to the hypothesis.

The Philippine Scouts, therefore, seem to have better health than American troops, either white or colored. As the health of the population from which they are selected is certainly below that of American whites and probably below that of American colored, we must suppose that these scouts are recruited by a more rigid and exacting selective process than that passed by American troops, either white or colored. Beyond this general conclusion the figures do not enable us to go.

2. THE HEALTH OF THE ARMY IN DIFFERENT PLACES.

A. UNITED STATES, HAWAII, AND THE PHILIPPINE ISLANDS.

Now that the racial statistics have been examined we are in a better position to consider the effect of geographical location upon health. There are three main regions, the United States, Hawaii, and the Philippines, in each of which soldiers belonging to the two main American races, white and colored, have for years been stationed in numbers large enough to furnish significant rates. The first question to be raised, then, is this: What is the health of American troops, either white or colored, in the United States, Hawaii, or the Philippines? Figures are available for each year since 1904. It is best to give the results for 1904-1916 together and those of 1917 separately, because in the later year the abnormal conditions and the great size of the army would make an average including it far from typical. For white troops the death rate, 1904-1916, was 5.08