The dotted line is drawn through points on the chart for data furnished by the manufacturer as his estimate of good practice

A curve of this general type, proving a close relationship between two variables, may be called a correlation curve

CORRELATION 199

members of the population. In Fig. 161 the hne was more bowed in the later year than in the earlier year, and the conclusion may accordingly be drawn that wealth in Prussia tended toward further concentration in those years intervening between 1892 and 1901.

In Fig. 162 a study has been made to see how the gasoline consumption of motor trucks varies in trucks of different sizes. The horizontal scale shows the rated size in pounds of the trucks under consideration. On the vertical scale, the cost of gasoline is given in cents per car mile. The data of the different motor trucks were indicated by separate dots on the chart. The solid line was then drawn through a point which represents the center of gravity of all the dots on each vertical line. The total number of dots on this chart is rather small. Too much dependence cannot be placed in the resulting curve, as special conditions may have affected some of the records so as to cause the dots to be misleading. Thus, considering the two dots which are given for trucks of 2,000-pounds capacity, it will be noticed that both of these dots are far below the position on the chart which one would expect the average to occupy if one should judge by the general tendency of the curve as a whole. It may have happened that the particular trucks which these two dots represent were run with very light loads, thus making the gasoline consumption lower than would naturally be expected for trucks of that size.