vestigation is a sort of mill into which all sorts of material may be cast, and ground up into percentages and averages which will have a virtue that the original figures had not. You might as well throw sawdust into a flouring-mill and expect to get flour out of it. The virtue of statistical tables lies in the correctness of the original data. What sort of statistics of imports should we have if we allowed importers to fix their own valuations without let or hindrance? The original returns in statistics of wages must be subjected to the closest scrutiny when they first come in. Those which are absurd or false on the face must be rejected; and those which are unusual or difficult to be explained must be elucidated by correspondence with the parties making them, who will perhaps be able to show why their returns differ from those common in the same industry and the same section of country. These explanations must be incorporated in the published returns in the form of notes to the tables, for the guidance of those making use of the statistics. Every bureau of statistics is obliged to do a vast amount of labor of this sort. It can only be done satisfactorily by men who are familiar both with the industries and with statistical methods. The census will never be able to secure such experts until we have a permanent office where men can be trained for it. Our decennial census suffers more from this than from any other cause,- that it is obliged to start every time with an untrained body of officers, who are scattered again at the end of the work, just when their services are becoming of the most value. We shall never have satisfactory statistics in this country until we have at Washington a permanent statistical bureau, which shall have its trained body of experts, able to undertake from time to time any investigation that is demanded, and which shall have the management of the census.

If the next census of the United States is able to employ a body of experts, who will collect information

about wages in the great industries on the lines laid down above, we shall then obtain a mass of wage statistics of very great value, and such as neither our country nor any other up to this time possesses. With the general intelligence of American employers and the force of public opinion, which would readily appreciate the value of such an undertaking when it was once explained, I believe that this could be done. Such a mass of facts would be of great value, both in detail and as a whole; and most interesting conclusions could be reached as to the prosperity of different parts of the country. As the years went on and such a record was continued on the same lines, a really trustworthy picture of the progress of our civilization could be drawn. The sooner we adopt some scientific method, the sooner such a record can be begun, and the sooner shall we be able to avail ourselves of its teachings.

But, even after we have collected such a body of scientific wage statistics, one more operation is necessary before they will answer the question of the income of the working classes. No average wage and no combination of averages can show us the condition of the working classes, because an average may be made up of so widely divergent extremes. This condition can be shown only by means of proportions; that is, by showing how many of the men receive a certain income, how many receive less, how many more, and so on. It is impossible to give illustrations of this from the Tenth Census, because we do not have the number of workmen specified. We cannot therefore give the proportions of actual wages received. We can, however, give the proportions in some cases of the different rates of wages received. This brings us to a series of statistical calculations of the following sort:

Table 1 classifies the returns of the rates of wages in four occupations as they are given in the census (Volume XX. for common laborers, carpenters, and engineers, Volume II. for the workmen in glass works). The scale of

wages is based on those most commonly paid, as $1, $1.25 and so on, other wages being placed in the nearest class so that a wage of $1.20, for instance, is classed with $1.25 The scale is in no sense, therefore, an average of the dif ferent wages; but by far the larger number fall together with the numbers in the scale:

TABLE I.

PERCENTAGE OF WAGE RETURNS AT DIFFERENT RATES OF WAGES IN FOUR

The object of this table is to show at a glance abo

what are the rates of wages paid to workmen in vario

occupations. We see, for instance, that the prevailing rate of wages for common laborers is $1.25, although considerable numbers receive only $1; and, in another considerable number of cases, $1.50 is reported. If we turn to carpenters, we see that they are paid at higher rates, their rates really beginning where those of the laborers leave off. Engineers again are paid much higher than common laborers; and their wages range higher than those for carpenters, as is shown by the larger number of cases where $2.50 and over is paid. The greater range of wages shows that the engineers are not so homogeneous a body as the carpenters or the laborers. There are greater differences of skill among them. The column for workmen in the glass industry differs from the others in two respects. It is based on the actual number of men employed at different rates of wages, and hence is a truer representation of wages actually received; but it is not wholly true, because the original returns are not used, but only the averages in each occupation. Managers and clerks and boys are excluded, in order to get at the wages of real workingmen of adult age. Again, the workmen in the glass industry are not a homogeneous class like common laborers or carpenters, but include men of varying degrees of skill. This explains the wide range of wages from $1 up to $5.50. At the bottom stand the unskilled laborers, over twenty-two per cent. of the whole, receiving only $1 or $1.25 a day. Then there is a body of skilled men, twenty-four per cent. of the whole, receiving $1.75 and $2 a day. Then there is a very considerable number of skilled workmen, twenty-two per cent., receiving as high as $3.50 a day, and a few, nearly five per cent., who receive very high wages, $5 or more.

It must be clearly understood that this and the following table do not possess in themselves any validity. They are given merely as examples or illustrations of what might be done with wage statistics. In order to have any

validity, they should have been based on actual returns of the number of workmen and their actual wages. If the next census will give us such figures, they can be analyzed in this double direction. Men of a common occupation may be taken in all industries, as common laborers, and be classified so as to show the proportionate number receiving each rate of wages; or the men in a single industry may be taken, as above in the glass industry, and their distribution shown. With the original

TABLE II.

CLASSIFICATION OF WAGE RETURNS SHOWING NUMBER AT OR ABOVE EACH RATE (on basis of 1,000).