Just how much the total consumption in water has been decreased, even though there was an increasing population, may be seen by referring to the dotted line on the chart. This dotted line was plotted from points which represent the average for the whole of each year. Thus, the average in 1905 was about 122 million gallons per day and in 1911 about 113 million gallons per day. There is so great a fluctuation in the main curve from month to month that it would be difficult to judge accurately whether the average consumption is going up or down if some such curve as the dotted line were not used. An average curve line run through a curve in the manner shown by the dotted line in Fig. 90 is of great assistance in drawing accurate conclusions from curves which have much fluctuation.
In Fig. 90, the dotted line, plotted through a point giving the average for each year, thoroughly eliminates all the fluctuations which would be so confusing to the reader if he had to study only the monthly curve. The dotted line shows conclusively by its slant that the total consumption from 1905 to 1908 inclusive went up just about as rapidly as the growth of the population would lead one to expect. After 1908, however, there was a tremendous drop in total consumption, even though the population kept on increasing. In 1912, the average total consumption per day went up somewhat above 1911, but yet it did not increase much more rapidly than the slanting line of the per capita scale might indicate as permissible. Fig. 90 could have been improved somewhat if the dotted line had been replaced by a broad black line which would bring the yearly-average curve vividly to the attention of the reader. The yearly-average curve really gives the most interesting conclusions which can be drawn from this chart and it is worthy of greater emphasis to the eye than is given to it in the chart.
It would have been better if the scale for "millions of gallons per day" had been placed only at the left-hand edge of the chart in heavy lettering so the figures would stand out clearly. The scale for the slanting co-ordinate lines could then be placed at the end of each slanting line at the right-hand side of the chart. The scale for the slanting co-ordinate lines is too difficult to find in Fig. 90.
An interesting study could be made from Fig. 90 by plotting a curve which would show each year the percentage of services which were not metered, instead of using the figures at the top of the chart which show the percentage of services which are metered. A curve
