Page:Proceedings of the Royal Society of London Vol 69.djvu/83

Rh and spring minimum, and a glance at figs. 1 and 2 will show that it is the frequency which converts the sudden warm effect in February, which is shown by the individual winds, into a gradual rise from autumn to winter and fall from winter to spring which are the features of the complete second-order curve. It will also be observed that the frequency effect has a maximum in July which brings back the maxi- mum exhibited by the sum of the individual effects to its normal

DIAGRAM: 6.

DIVERGENCE FROM FIRST ORDER CURVE. KEW 1876 TO 1884.

Fig.t Tot&L effect.

2. Effect of compounding the curves for &Lu winds, without reg&rd to the reL&tive frequency or the v&rious winds or Groups of winds,

5. Difference between fig.i. &nd Fig.z.

position. The October November effect seems to be complete in the compounded curve, and, as was seen in Diagram 4, it has its origin chiefly in the curves of the south-east and north-west winds. Thus it appears that the second-order effect is to some extent independent of the direction of air supply, but that it is in part due to the relative frequency at different times of the year with which the air supply is derived from " hot," " cold," or " temperate " sources. At the periods of the maxima of the second-order curve the air supply of these islands is derived mainly from the warm sources, i.e., south, south-west, and west, and at the periods of its minima, either from the coldest sources, which are the east, north-east, and north, or to a very predominant extent from the slightly cold sources, which are the south-east and north-west.

It is especially noteworthy that the cooling effect in May is due to