Page:Proceedings of the Royal Society of London Vol 69.djvu/82

74 direction in bringing about the second-order effect by any method of direct composition of these curves, but the fact that each of the three groups gives a curve of frequency whose characteristics show this remarkable correspondence with those of the second-order curve, is stifficient evidence that the relative frequency of occurrence of winds

DIAGRAM 5.

FREQUENCY CURVES. KEW 1 876 TO 1 884.

DEC. JAN. FEB. MAR. APR. MM JUN. JUL AUG. SEP. OCT. NOV. DEC.

14-7

3.Temper&be.(SE.NYj.)

from the various quarters plays some part in causing the second-order effect. What this part is is shown in Diagram 6. Fig. 1 gives the curve of mean temperature difference throughout the year, and fig. 2 gives the curve which is directly compounded of all the temperature curves for the various winds, so that the effect of relative frequency is eliminated ; fig. 3 is obtained by subtracting the ordinates of fig. 2 from the corresponding ones of fig. 1, and may be taken as representing the effect upon temperature arising from the relative frequency of warm or cold winds. The main features of this curve are the winter maximum