628 THE AMERICAN JOURNAL OF SOCIOLOGY
average wages, it was not detected in the careful examination and comparison made by the census office.
The method of editing or preparing the schedules for tabula- tion was practically the same at the two censuses, the only changes for 1890 being those necessitated by the individual schedules and the additional checks afforded by the detail questions. Under no circumstances, except when a clerical error was apparent, was the amount reported as wages changed, nor was the number of employes increased or diminished except when an evident error appeared.
The number of employes and the total wages at both censuses are substantially as reported by the enumerators, and the enume- rators of 1880 were probably as intelligent as those of 1890. Under these circumstances for the department to apply an arbi- trary percentage of correction to either the total or average wages would have been worse than absurd. A personal canvass was not made with the intention of securing results that would be cor- rected to suit the ideas of anyone concerning the amount paid as wages. The data secured by this canvass have been honestly tabulated and presented. They are the only data that can be relied on as showing the actual facts.
Whether the average, the total, the greatest, or the least num- ber of employes engaged during the year should be taken as the division for the total wages, in order to ascertain the average, is a subject that has received the consideration of the best statis- ticians of the world, and they have invariably taken the average number. It is needless to discuss it further, but it must be remem- bered that the wages reported were paid to have certain positions of employment filled, and the number of those positions is probably the true divisor for the total wages ; the average comes nearer this number than does the total or the greatest number. But, as carefully explained in the census reports, the average given for wages is not the true average yearly earning per workman. Because of the constant shifting of workmen from one employer to another, and other contingencies enumerated, the only true way to ascertain the exact amount earned during a year is to con-
