Page:Proceedings of the Royal Society of London Vol 69.djvu/87

Rh than the anticyclonic days. Thus there is no prominent effect at Vienna analogous to our second-order effect, but, as in these islands, whatever periodic meteorological temperature effects do exist, appear to be entirely independent of the relative frequency and mean tem- perature of cyclonic and anticyclonic weather. An analysis of the temperature curve for Agra, a typical Continental station in India, has also been made (Table I, Agra). Here there is a conspicuous second- order curve, but its position is entirely different from that of the corre- sponding curve for the British Islands, and its effect is to prolong the summer, whilst the second-order effect which has been discussed above shortens it. The two effects are thus in no way analogous, and this result, combined with the result of the Vienna investigation, makes it seem probable that the ocean plays a paramount part in the causation of the second-order temperature effect which we experience in these islands.

With the object of further investigating this point, curves whose co- ordinates arc respectively proportional to the fraction of the year, and to the mean temperature of the sea, based on three years' observations at Shetland, Scilly, and Yarmouth, have been analysed (Table II). The maxima and minima of these curves are all a little later in the year than those of the air-temperature curves, but the lag is approxi- mately the same in all of them, and it is an obvious and striking fact that there is in the periodic variation of marine temperature an effect similar to the second-order effect observed in the periodic variations of atmospheric temperature. Whether this variation of the temperature of the water which surrounds these islands is the cause of the atmo- spheric second-order variation, or whether it is only another effect of the same fundamental cause, does not appear ; but in view of the fact that the marked second-order effect is not seen at Continental stations, it would seem not unlikely that the ocean temperature is the imme- diate cause of our second order periodic temperature variation.

Table III gives the results of the analysis of the curves of thirty- year-mean difference of barometric pressure between London and Valencia, and London and Aberdeen, respectively. In each of these curves we again find the prominent second-order effect, although in the curve for London Valencia it is somewhat earlier in its posi- tion than the temperature effect. Thus the magnitude of the baro- metric gradients for southerly and easterly winds shows a similar second order periodic variation to that which has been observed in the atmospheric temperature of these islands. In view of the close