in the United States, 3.03 in Hawaii, and 6.14 in the Philippines. For colored troops during the same period the death rate was 8.51 in the United States, 5.13 in Hawaii, and 7.56 in the Philippines. Seemingly for both races Hawaii is a more healthy location than the United States or the Philippines. The death rate of white soldiers is lower and of colored soldiers is higher in the United States than in the Philippines. In 1917 the death rate of white soldiers in the United States increased to 6.27 per 1,000 and of colored soldiers to 14.37 per 1,000, the unfavorable conditions of that year raising the white death rate about one-fourth and the colored death rate about two-thirds above its previous average. The changes in Hawaii were in the same direction and with a similar difference between the two races, but in the Philippine Islands the death rate of the soldiers of each race dropped off sharply. The figures then indicate that for each race Hawaii is a very healthy station and that while in all three regions the death rate of the colored soldiers is higher than that of the whites, in the United States it is higher by about two-thirds, and in the Philippine Islands by only one-fourth.

B. CONTINENTAL UNITED STATES.

The death rate has meaning only when it is based upon deaths enough to yield a significant average. The fact that in 1917 there were four States in which no death occurred in the Army, three in each of which there was only one death, and three in each of which there were only two, does not even suggest that these States or the soldiers stationed in them were remarkably healthy. The following table includes only those States in which at least 50 deaths occurred at Army posts in 1917. They are arranged in order of increasing death rate.

TABLE NO. 11.-Death rate in Army (enlisted men), by location of command for States in each of which 50 or more deaths occurred, 1917.

Table 11 does not clearly reveal any geographical grouping of the States with high or low Army death rates. It does show a large number of Southern States and only one-fifth of the northern on the list. But this may be due to the location of so large a number of camps in that part of the country. To determine whether the death rate of soldiers varied with the location of the camp the States have been combined into groups in the table following. In each group, it will be noticed, there were 50 deaths or more of soldiers and the rates, therefore are robably significant.

TABLE NO. 12.-Death rate in United States Army(enlisted men), by location of command for groups of States, 1917.

The table shows that the lowest Army death rates were in States north of Mason and Dixon's Line and the Ohio River and east of the Mississippi and that the highest rates were in States between the Mississippi River and the Rocky Mountains. In determining this geographical grouping two causes may have been at work. First, as the last column of Table 12 indicates, States with a low soldier death rate are the States with a large proportion of urban population, and vice versa. Secondly, States with a low soldier death rate are the states with a well developed public-health service. These two influences are closely interdependent for the public-health work in general and especially in the United States has been more needed and is now better developed in urban or densely settled areas. But the way in which these two influences are supposed to work is quite different. If the density of population is the main cause the low death rate would result from the fact that among the new recruits from urban districts the proportion who have already been exposed to the germs of various contagious diseases and have thus gained an increased power of resistance would be much larger than the proportion among the new recruits from rural districts. On the other hand, if the development of public-health work is the main cause the low death rate would result from the fact that among the new recruits from such States the proportion having a knowledge of the elementary principles of hygiene and sanitation and able and willing to cooperate with and reenforce the efforts of the medical staff would be much greater. It seems likely that the two sorts of influences have cooperated in bringing about the striking result, but it is difficult to see how that theory can be brought to a test of the figures.

3. THE HEALTH OF THE ARMY CLASSIFIED BY AGE.

In the tabulations for the calendar year 1917 with which this report is mainly concerned, no classification of the troops by age has been attempted and none of significance could be made because we do not know the mean strength by age. But in the reports for each of the nine years, 1905-1913, the mean strength, the deaths, and the three classes of data about the sick were reported in age groups and the

ADMISSIONS, DISCHARGES. DEATHS, AND NON-EFFECTIVE RATES ENLISTED MEN IN U.S.- ALL CAUSES -1908-1917

1908

1148.59

1909

1024.3

1910

899,29

1911

887,80

1912

738,70

1913

1914

1915

619.90

629,84

704.0

1916

745.3

1917

978.4

results invite examination. The material for any one of those years was not extensive enough to measure the influence of the age of the soldiers upon their health. For example, to the simple question: At what age are soldiers most healthy? the series of tables indicates the following answer: As measured by death rate, the healthiest age was, in one year, 15 to 19, in a second, 20 to 24, in a third, 30 to 34, in a fourth, 35 to 39, and in the five other years, 25 to 29; as measured by the proportion admitted to sick report, the healthiest age was, in four years, 40 to 44, and in five years, 45 to 49; as measured by the proportion discharged for permanent disability, the healthiest age was, in two years, 40 to 44, in two years, 50 and over, and in five years, 45 to 49; as measured by the proportion constantly noneffective, the healthiest age was, in two years, 45 to 49, and in six years, 35 to 39.1 Clearly this answer does not go far toward settling the question to which it is addressed. That the healthiest age measured by any of these tests really changed thus from one year to another is most improbable. More light is thrown upon the inquiry when the returns of each of these nine years are combined into one table. The results are as follows:

TABLE NO. 13.-Health of United States Army (American troops), classified by age,

Table 13 shows that admissions to sick report in hospital or quarters are most common among the very young soldiers and decrease steadily and rapidly up to the age 40 to 44 when they were not much more than one-third as common as among soldiers under 20 years of age. After that time of life they increased, but soldiers even at ages over 50 were admitted to sick report only half as often as their

I For one of the nine years this form of the rate is lacking.

ADMISSIONS, DISCHARGES DEATHS, NON-EFFECTIVE RATES

ENLISTED MEN IN U.S.-DISEASE ONLY-1908-1917

1908

1909

1910

1911

1912

1913

1914

1915

1916

1917

1921.91

821.77

715.54

725.83

596.94

498,83

564.49

612.52

19.15

17.01

13.70

13.26

13.19

11.92

12.83

11.93

14.61

28.11

3.10

2.97

2.30

2.83

2.28

2.62

2.52

2.99

2.80

5.34