that they will tell very queer stories. Statistics will not lie, but they will play wonderful tricks with a man who does not understand their dialect. The unsophisticated reader finds it difficult, when a column of statistics is offered to him, to resist the impression that they must prove something. The fact is that a column of statistics hardly ever proves anything. It is a popular opinion that anybody can use or understand statistics. The fact is that a special and high grade of skill is required to appreciate the effect of the collateral circumstances under which the statistics were obtained, to appreciate the limits of their application, and to interpret their significance. The statistics which are used to prove national prosperity are an illustration of this, for they are used as absolute measures when it is plain that they have no use except for a comparison. Sometimes the other term of the comparison is not to be found and it is always ignored (§ 52).
103. A congressional committee in the winter of 1883-1884, dealing with the tariff, took up the census and proceeded to reckon up the wages in steel production by adding all the wages from the iron mine up. Then they took bar iron and added all the wages from the bottom up again, in order to find the importance of the wages element in that, and so on with every stage of iron industry. They were going to add in the same wages six or eight times over.
104. The statistics of comparative wages which are published are of no value at all.[1] It is not known how, or by whom, or from what selected cases, they were collected. It is not known how wide, or how long, or how thorough was the record from which they were taken. The facts about various classifications of labor in the division of labor, and about the rate at which machinery is run, or about the allowances of one kind and another which vary from mill
- ↑ I except those of Mr. Carroll Wright. He has sufficiently stated of how slight value his are.
