If the reader cares to see how these two curves would contrast if properly plotted, the left-hand scale for "corn yield" can be changed so that the different horizontal lines will be represented by the figures 0, 7, 14, 21, 28, etc. The data for "corn yield" can then be plotted to the new scale, and it will be found that the new curve does not coincide with the "rainfall" curve as closely as does the "cornyield" curve shown in Fig. 114. Though there is some similarity in the shape of curves correctly plotted from these data, the similarity is not nearly so exact as Fig. 114 would indicate.

. 1908-1909

< t^ O 2 Q .^

FREIGHT TRAIN OPERATION . 1909-1910

Courtesy of System

Fig. 113. Records of Freight-Train Operation on a Large Eastern Railroad

Here, as in Fig. 112, the ^ero lines are not given and the reader must watch the left-liand scales carefully to study percentage fluctuation. Comparison between curves cannot be made accurately by judging from the slope or from the total fluctuation of the curves on the page. Percentage increases or decreases for one curve compared with percentage increases or decreases of other curves give the best basis for comparison

Curves well made ordinarily need no embellishment. Anything used for an eyecatcher should apply definitely to the subject matter of the curves. Here the subject is freight-train operation, but the picture shows the interior of a passenger train

GRAPHIC METHODS

J in

Fig. 115 gives a good example of a total curve made by adding the figures for different curves. Instead of using addition to get a set of figures from which a total curve may be plotted, it is easier in most cases to get the location of the total curve by the graphic method.