compared with the average number from 1890 to 1899, and, as just stated, the number of employees in 1904 was 95.69 per cent of the number in 1903, therefore the relative number in 1904 in the series from 1890 to 1904 would be 95.69 per cent of 116.2, or 111.2. It must not be understood that the number of employees (4,644) in 1904 in the 229 establishments is 111.2 per cent of the average number (4,355) for the ten-year period, 1890 to 1899, in the 212 establishments, but that the relative figures in the series from 1890 to 1904, calculated as above, show conditions in the occupation generally.

Data were secured for bricklayers for Bulletin 65 from 209 establishments for both 1904 and 1905, the greater number of which establishments furnished the data for 1904, which were published in Bulletin 59. The number of employees reported by these 209 establishments was 4,341 for 1904 and 4,365 for 1905; the number in 1905 was 100.55 per cent of the number in 1904. By applying this percentage to the relative number for 1904 (111.2) the relative number for 1905 is found to be 111.8.

The relative number for 1906 was computed in the same manner. In using the relative numbers it should be borne in mind that the difference between the relative numbers in any two years does not represent the per cent of increase or decrease in the actual numbers. For example, the relative number of bricklayers in the building trades in 1890 was 101.5. The relative number in 1906 was 124.0. The difference in the relative numbers is 22.5, and this difference is 22.2 per cent of 101.5—that is, the number of bricklayers employed, so far as can be determined from the establishments covered, was 22.2 per cent greater in 1906 than in 1890.

In Table II the second column shows the relative number of employees each year from 1890 to 1906; the third, the relative hours of work per week in each year; and the fourth, the relative wages per hour. By referring again to the occupation of bricklayers in the building trades, as shown on page 66, it is seen that the relative wages per hour in this occupation increased from 98.4 in 1890 to 138.6 in 1906. If the per cent of increase is desired, it may be obtained by computation. Thus, 40.2, the difference between 98.4 and 138.6, is 40.9 per cent of 98.4, making an increase in the wages per hour of bricklayers between 1890 and 1906 of 40.9 per cent, as nearly as can be determined from the establishments that have been canvassed. Any other two years may be compared in like manner. The few new occupations incorporated in this report can not have relative numbers shown for them, because no data were secured for them in the years adopted as the base period.

Table III.-Relative wages and hours of labor, 1890 to 1906, by industries (pp. 126 to 131).-The relative figures shown for the several occupations in Table II have been summarized so as to show conditions in each industry as a whole, and are presented in this report in Table III. This summary table shows the relative conditions in each year compared with the ten-year period, 1890-1899. In computing the relative numbers shown in Table III, each occupation was given an equal weight, there being no satisfactory method of weighting the several occupations. When both sexes were reported in an occupation, each sex was considered a separate occupation in this computation.

The method of computation originally employed was to add the relative numbers under each heading for all of the occupations of an industry and divide the sum by the number of occupations entering into the total. Changes in the industries from year to year, however, have required the dropping out of some occupations and the taking on of others. Because of such changes the method now followed in computing the relative numbers for the industry is to compute for employees, hours, and wages in each occupation the per cent which the figures of the current year are of the figures for the preceding year; the percentages for all the occupations in the industry are then added and the sum divided by the number of occupations. The index for the industry for the preceding year is then multiplied by this average percentage to determine the index or relative number for the year under consideration.

The table shows the trend of wages and hours of labor in the occupations of an industry taken as a whole. The average wages of all employees in an industry, irrespective of occupation, may differ to some extent, possibly, from the figures here shown, because of a change during the period in the proportionate number of employees in the several occupations. It is presumed that the work done in the occupations themselves has not changed materially. The table is intended to show changes as regards the same or like employees, rather than changes in which the establishment is considered as the unit.

Like Table II, Table III embraces three divisions, relating to the number of employees in the several years, the hours of work per week, and the wages per hour. The base on which the relative number is computed is the average for the ten-year period, 1890-1899. By reference to the table, it is seen that in 1890 the number of employees engaged in the agricultural implement industry was 86.2 per cent of the average number employed during the ten-year period, 1890 to 1899. In each of the years thereafter a greater number was

employed than in 1890, the highest being in 1906, when 33.3 per cent more persons were employed in the occupations and establishments covered in this industry than the average during the base period. In 1904 the number employed was 8.5 per cent more than the average number employed during the ten-year base period, and in 1905, 23.7 per cent more. With this explanation, the other columns of the table relating to hours per week and wages per hour will be readily understood.

In computing the figures for the summary for the United States shown on page 7 each industry has been given a weight in proportion to the amount of wages paid therein, as shown by the United States census of 1900. This was deemed to be the most equitable method of making allowance for the greater or less importance of the several industries. The agricultural implement industry is an important industry, but by no means so great as the building industry of the country; therefore allowance for the importance of each industry was deemed necessary, and the most satisfactory method of weighting is believed to be the weighting in proportion to the wages paid in the several industries. In computing this weighted average the relative number or index for a year for each industry was multiplied by the wages paid in each industry, respectively, as shown by the United States census of 1900, expressed in the nearest $100,000. The products were then added and the sum divided by the total wages paid in the industries represented. When the industry classification differed from that given by the United States census, the wages paid were estimated from the census figures. Street and sewer work, not reported as such by the census, was estimated from the best available data-that is, the wages paid in the District of Columbia. When two or more industries shown separately in this report were classed as one industry by the census, the relative numbers for the several industries were added year by year, and the totals divided by the number of subindustries to obtain a relative number for the general industry as shown by the census. The figures of the census of manufactures of 1905 can not be used satisfactorily as a new basis of weighting the several industries shown in this report because that census was limited to work done under the factory system and did not include the mechanical trades. The following table shows the relative weight accorded to each industry.

In preparing to secure data for 1906 copies were made of the data for 1905, obtained in the preceding year. These copies were placed in the hands of the special agents of the Bureau, who visited the establishments and secured 1906 data, and whenever any doubt arose verified or, if necessary, revised the data for 1905. The agents were also instructed to secure data from a few additional establishments, in order to secure a better representation for the several industries and localities covered. Where data were obtained from additional establishments, figures were taken for both 1905 and 1906. All data for the two years, 1905 and 1906, shown in this article, therefore, are from identical establishments, and may properly be compared. During the course of the field work it was also found that a few of the establishments heretofore covered had gone out of business. With some establishments added and others discontinued it will be seen that the establishments from which data were secured for this Bulletin differ to some extent from the establishments shown in preceding reports on this subject.

With a change in the establishments reporting, some difference necessarily must be expected between the average hours of labor and average wages per hour shown for the year 1905 in Bulletin 65 and in this Bulletin. Two groups of establishments, even though many establishments are common to both groups, can not be expected to show exactly the same averages. Neither statement may be absolutely exact for the occupation, yet if each group embraces a considerable number of establishments the average rates found may be accepted as representing approximately the average rate in the occupation considered. The average based on the larger number of establishments probably would be the more nearly representative. To secure the exact average rate of wages and hours of labor for the United States it would be necessary to take a complete wage census of the United States, and, as previously stated, such a task is so great as to be practically impossible.

To measure with exactness the movement in wages and hours of labor it would not be proper to compare the average for one group of establishments for 1905 with the average for a different group of establishments for 1906, nor in practically the same establishments if any change or substitution of establishments whatever has been made. Comparison can be made accurately only when the data compared are obtained from identical establishments. As one of the principal objects of this series of reports is to measure the change from year to year as accurately as possible, increases and decreases from year to year are computed only on data taken from identical establishments.

A statement of rates of wages is incomplete without an accompanying statement as to hours of labor, as both factors must be known to determine the earnings on full time per week, month, or year. The average hours of labor are given in this report in connection with the average rates of wages. The average hours of labor shown are true averages, computed in the same manner as the average wages per hour.

The hours of labor given in this report represent the regular full time hours of the occupation-that is, the time that the employees as a class were engaged in work. No account has been taken of time lost by individual employees because of a desire not to work, of bad weather, or of slack work. It is no part of the purpose of this article to measure the amount of unemployment, and it must not be inferred that all employees engaged in the establishments reported in this article worked full time. To ascertain the time lost in a year by the employees remaining in the establishments a year would be an almost insuperable task, while to ascertain with any accuracy the time lost by employees who have moved from one establishment to another would be practically impossible.

The sex of the employees for whom data are presented is shown for each occupation. In many occupations few, if any, females are employed, and in other occupations few, if any, males are employed. In such occupations data are given for one sex only. In a number of occupations, however, in which many persons of both sexes are employed, data are presented for both sexes, that the occupation may be properly represented.

EXPLANATION OF TABLES.

The tabular matter of this article relating to wages and hours of labor is set forth in five general tables, and a number of short tables shown in the text explaining and summarizing such general tables. The five general tables are entitled:

TABLE I.—Average wages and hours of labor, 1905 and 1906, by occupations and geographical divisions.

TABLE II.-Relative wages and hours of labor, 1890 to 1906, by occupations.

TABLE III.-Relative wages and hours of labor, 1890 to 1906, by industries.

TABLE IV.-Average wages and hours of labor, 1905 and 1906, in selected occupations, by cities.

TABLE V.—Average wages and hours of labor, 1905 and 1906, in selected occupations, by States.