362 THE AMERICAN JOURNAL OF SOCIOLOGY
were 247 ; in 1860 $289; in 18705302; in 1880 $347; and in 1890 $445. Here is a steady positive increase in the average annual earnings of the employes in our great industrial pur- suits. The statement is not mathematically accurate, because the divisor used is not always a sure one. The total amount of wages paid at each of the periods named is a fixed quantity, and is one of the most certain elements of the industrial cen- sus, but the average is obtained by dividing the total wages paid, by the average number of employes during the year."
Having, apparently, in mind the writer's contribution to the January number of this JOURNAL, Colonel Wright says: "Some writers contend that the divisor should be the greatest number of employes instead of the average number, but the greatest num- ber would secure a more erroneous quotient than that derived from the average number."
It is evident that as the total wages paid in all industries are the earnings of the total number employed in these industries the average annual earnings can only be obtained by using the total number of employes as the divisor.
It must, however, be conceded that to aggregate the numbers reported by each establishment as the greatest number employed at any one time would be likely to obtain a number in excess of the total number of employes. It is also certain that to use the number reported as the average number of employes, would result in obtaining a number less than the total number of employes.
Colonel Wright's statement shows that he is aware that the use of these differing methods in obtaining the divisor would result in widely differing quotients, but he strangely seems to miss the real point which is that no fair comparison can be made of average annual wages obtained by one method in one decade and by the other method in another decade. By Colonel Wright's admission in this official communication, while the number used at the last census as the divisor was the true average number of employes, the number used in the preceding censuses was a num- ber greater than the average number. From the statement of
