Observations on ascertaining the average school term.-The "aggregate number of days' schooling given" to all pupils (see column 6), which is the same thing as the aggregate number of days attended by all the pupils, has been computed for those States which do not make an explicit report of this item by multiplying the average daily attendance of pupils by the average length of school term in days.

Conversely, the average length of school term (column 5) for the United States as a whole and for each of its geographical divisions has been obtained by dividing the aggregate number of days attended by the average daily attendance.

By this method the school term of each State, in computing the average term for a number of States, is in fact given a weight proportioned to the school attendance of the State, as should be done under a correct interpretation of the expression "Average length of school term." The result might more properly be called "Aver age length of attendance," which is essentially what it is desired to know.

A method which has been in use in some States for finding the average school term, of a county, for instance, is to weight the different school terms of the towns or districts the county is composed of by the number of schools in each. In other words, the total number of days (or months) all the schools of a county were kept is divided by the total number of schools to get the average time each one was kept. So, in finding the average term for the State, the school is taken as the unit instead of the pupil. When the schools differ much in size (number of pupils), as they do in all mixed urban and rural systems, varying from some half a dozen to 500 or more pupils each, the average term obtained by this method varies considerably from that obtained by the Bureau's method. The long terms of the large city schools not being given their proper weight, the resulting average is too small. The same objection applies still more forcibly to weighting the school terms of the different counties or towns by the number of school districts in each.

Another method is to divide the total number of months or days taught by the number of teachers. This is better than the preceding method, as it takes some account of the size of the schools; that is, an eight-grade school with eight teachers has eight times the weight, in determining the average term, that a district school with one teacher has. This is manifestly as it should be. If every teacher taught the same number of pupils, the result would be the same as by the Bureau's method. Care must be taken in working by this method to use the number of teachers' places (or number of teachers necessary to supply the schools) for the divisor; for if a teacher teaches a school or grade part of the term, and is replaced by another for the rest of the term, the two should obviously court as one teacher for the combined period of service. The liability to overlook this distinction in practice, as well as the inequality in the number of pupils to a teacher, makes this method generally objectionable.

Still another and most faulty method is to add together the school terms of the different counties or towns and divide by the number of such counties or towns; i.e., the simple arithmetical mean is taken. An example of this occurs in a school report, where it is stated that 14,193 pupils in one district attended 185 days, and 856 pupils in another district attended 160 days, while the average time the whole 15,019 pupils attended is computed at 1724 days, although nearly all (16 out of every 17) attended 185 days. This method, if it can be so called, gives altogether too short an average term, and nothing can be said in defense of it. It is as if, wishing to get the popu lation per square mile of Minnesota and Dakota combined, we said, Minnesota, 9.86; Dakota, 0.92; average number of persons per square mile in the combined territory (0.92+9.86)25.39, instead of dividing the total population of the two states by the combined area in square miles.

The "aggregate number of days' attendance" is a statistical item of the utmost simplicity and of great value, about the meaning of which there can be little or no difference of opinion. Every teacher's register that records the number of pupils present each day in school, as they all presumably do, contains the data for ascertaining it for that school for the school year by the simple process of addition or summing up.

Year.

There are a few States that make no pretense of ascertaining how long their schools were taught, and others that use methods so faulty that they also are totally in the dark in the matter. Yet this is one of the most necessary and fundamental items of information in determining their educational status. It is on this account that the subject has been somewhat enlarged on.

TABLE 9.-(1) Length of school term. (2) The aggregate number of days' schooling given compared with the school population.

TABLE 10.-Number and sex of teachers-Percentage of male teachers.

TABLE 11.-Showing, for a period of years, what per cent of the whole number of teachers were males.

TABLE 12.—Teachers' salaries—Number of schoolhouses—Value of school property.

TABLE 12.-Teachers' salaries-Number of schoolhouses-Value of school property-
Continued.

Average salaries.-In computing (for Table 12) the average monthly wages of all the male (or female) teachers in a group of States, the average wages of each of the States in question is multiplied by the corresponding number of teachers. The sum of the resulting products is then divided by the sum of the teachers, and the quotient is the average wages of all. Each rate of wages thus receives its due weight. To illustrate the principle: If 48 teachers receive an average of $72.20 per month, they all receive in a month $3,466; if 473 other teachers receive an average of $48.60 per month, these together receive in a month $22,988; all the 521 teachers combined therefore receive per month $26,454, or an average of $50.77.

Attention is called to this matter for the reason that the practice of taking the arithmetical mean of a number of rates of wages (the mean is $60.40 in the above case) in order to get the average of all is so common as to vitiate, to a considerable extent, the published statistics of average wages.