up the totals, and explains in many cases the great falling off in school attendance, expenditure, etc., that appears to have suddenly taken place in this or that State.

Whenever practicable the figures for these missing counties have been supplied by the bureau, using the latest returns available or some reliable estimate. This gives totals for the State, which, though not exact, are considerably nearer the truth than if the missing counties were given 0.

Care has been taken to make the tables as correct and useful as possible, though some imperfections have resulted from delayed returns. Attention has been paid to tabulating only percentages and percapitas which are of value in estimating the relative educational condition oí different localities, omitting those which are misleading, as well as those which possess no special significance. The observations made upon this subject and upon statistical methods generally are commended to the attention of those interested in such matters; the design has been to make some contribution toward the accurate and uniform compilation of school statistics.

TABLE 1.-Total population, school population, and adult male population.

a The estimates of the total and the school population for 1891 are made on the basis of the percentages of 1890, as no new determinations of these quantities will be madə until 1900.

b In 1890 (U. S. Census).

TABLE 1.—Total population, school population, and adult male population-Continued.

a The estimates of the total and the school population for 1891 are made on the basis of the percentages of 1890, as no new determinations of these quantities will be made until 1900. b In 1890 (U.S. Census).

POPULATION AND SCHOOL POPULATION.

Although the census of 1890 has gathered the material for determining the number of children of each year of age in the several States, the results have not been published at the date of preparation of this report.

A preliminary census bulletin has been issued, however, giving for each State the number of persons 5 to 18 years of age in 1890; this age has been considered by the Census Office for several decades past as the "school age," and has been, therefore, adopted in this report as furnishing the only enumeration of children that is uniform in all the States as regards the ages included, and extending from 1870 to the present time.

The number of persons of each individual year of age was not published in the census reports until 1880.

The figures in column 3 of the preceding table are based on those of the census bulletin referred to, but have been reduced in most cases to the date 1891, as have also been those of the total population in column 2. The total population of the United States for the year under consideration (1890-'91 for most of the States) is found to have been 63,521,196.

Of this number, 18,799,864 were between 5 and 18 years of age. Since all these were of a suitable age for attending the common schools (elementary or secondary), and since in point of fact nearly all the common-school attendance was drawn from them, they may be called the school population."

The proportion of the school population to the total population has undergone a marked change during the past twenty years, as may be seen from an examination of columns 4, 5, and 6 of the table. In 1870 the school population of the United States formed 31.27 per cent of the total; in 1880, 30.04 per cent, and in 1890, 29.61 per cent.

There has been going on, then, a constant decrease in the number of children as compared with the total population.

This decrease has mainly occurred in the north and west; in the North Atlantic and North Central States the decrease in the proportion of children has been about 3 per cent since 1870 (from 28.30 to 25.39 in the former and from 32.40 to 29.33 in the latter).

In the Southern States, as a whole, on the contrary, the proportion of children, always large when compared with the North (now about one-third larger), has undergone some increase since 1880, so that now it is larger than in 1870.

The proportion of children in individual States exhibits marked contrasts, even when the States are contiguous. New York is the most populous State in the Union, but Pennsylvania has the largest number of children between 5 and 18 (1,498,300). In the former State the proportion of school population is 24.57 per cent, in the latter, 27.92 per cent. This circumstance may throw some light on the fact that the reported average attendance of pupils in Pennsylvania has for many years considerably exceeded that of New York.

In South Carolina 37 persons out of every 100 are children of school age; at the other extreme stands Montana, with only 18 out of 100, or less than one half the number in South Carolina.

The relation between the school population and the number of adults (column 7) in the different States is yet more marked. Taking the same two States, South Carolina and Montana, for instance, there are found in the former 433,800 school children to 239,100 male adults; in the lat ter, 26,960 school children to 73,870 male adults.

These facts are of importance in comparing the status of the schools of different localities, though they have not received from educational writers the attention they merit. Some of the considerations they give rise to are pointed out in the following pages, especially in treating of school revenue (or support) on page 24.

The school population classified by age.

As has been already remarked, the age tables of the Eleventh Census have not yet been published. Using the percentages of 1880, the number of children of each year of age in the school population of 1891 may be approximately stated as follows:

This table has the defect of the original census table (of 1880) upon which it is based. The tendency of parents or other persons who give information to census enumerators is, through ignorance or negligence, to give children's ages roughly or by guesswork, in even numbers rather than odd, and especially in round numbers. Hence we find more children of 6 years of age than either 5 or 7, and so with 8,10, 12, etc., the number of 10 years of age being particularly large.1

The summaries of the number of children from 6 to 10, 10 to 14, and 14 to 18 years, in the foregoing table, give approximately the number of children of age suitable for primary, grammar, and secondary instruction, respectively.

This defect is so magnified in the ages of older people in the census of 1880, that 188,752 are given as 59 years old, 427,937 as 60, and 148,731 as 61.