It can be seen that the January figures for "Operating Revenues" are all considerably lower than the December figures, but even so the reader has no clear idea of the slope of the lines which would be most typical to portray the changes from December to January in each year. This question of repeating one point for curves of different years superinlposed is referred to also in Chapter XIII, Fig. 204.

Fig. 105, Fig. 106, and Fig. 107 are selfexplanatory. They show some interesting applications of curves to special problems, and demonstrate the great convenience which might result if curves could be more generally used for presenting every-day facts to nontechnical readers.

In Fig. 108 we see the application of curves to the kind of data of which it would be extremely difficult to give a clear understanding if only columns of figures were available in the presentation. With these two curves we are not so much interested in the total height of the peak as in the general shape of the curve on either side of the peak, showing whether there are any laws of uniformity in the increase of the flood level and in the speed with which the flood subsided. The two curves in Fig. 108 are quite different in their shape, although they were taken for the same period of time in districts not widely separated. The size and general character of different water-sheds have

EdiDin D. Dreyfus, in Indusirial Enoineering

Fig. io6. Average Temperature at Pittsburgh, Pa., for Each Hour in the Day for Different Months in the Year. Plotted for Monthly Averages of Twenty- Years Observations (1891-1910) of the United States Weather Bureau