Page:Graphic methods for presenting facts (1914).djvu/131

 read because it is difficult for the eye to follow, through the whole series, the bars representing any one set of facts which may be of special interest. The bar method is in itself a simple one, but when the bars are combined in the manner shown in Fig. 101 the presentation becomes really more complex than if the data were shown in the form of curves.

Fig. 100. Comparison of the Earnings for Five Years after Graduation of the Yale University Academic Department Class of 1906, the Yale Sheffield Scientific School Class of 1906, and the Princeton University Class of 1901

There is a fallacy in making this comparison. The standard of living undoubtedly went up between 1901 and 1906

Fig. 102 certainly brings out the information of Fig. 101 in much better form than any in which it is possible to show it by any combination of bars either vertical or horizontal. The person who is just beginning to chart data which he has used formerly in tabulated form is often surprised to find how many inconsistencies exist in the data and how many different things there are which must be allowed for by some method of estimate. In Fig. 102 the data for the United Kingdom are expressed in net tons, and for the United States in gross tons. Though some correction of the forms of the curves as they appear in this chart would of course have to be made to get a true comparison of the shipping of the two countries, for our purpose the thing of greatest interest is the general tendency of shipping in the two countries. This we can study fairly well from the general shape of the different curves, even though the curves cannot strictly be compared with each other in so far as total quantities are concerned.

Fig. 103 shows a convenient method for determining what fluctuations in the different months of the year are typical for any subject being studied. Instead of plotting one continuous curve by months for a long series of years to a rather small horizontal scale, a large horizontal scale is used and a separate curve is drawn for each year. The curves for different years are placed one above the other, so that any fluctuations which appear in the same months year after year