to some extent apparent during the prevalence of winds from all quarters.
It is not found, with maxima at the same epoch, in purely Continental stations, such as Vienna and Agra.
A similar effect is found in the temperature variation of the sea- water at stations surrounding these islands, and the atmospheric effect is probably connected with this.
A similar second order periodic effect is also found in the variation in the magnitude of the barometric gradient between London and Valencia and London and Aberdeen. It is probable that this periodic variation in pressure plays some part in causing the similar variation in temperature.
Note on the Effect of Sea Temperature upon the Seasonal Variation of Air Temperature.
It has been pointed out in referring to Table II. that the harmonic components of the first and second orders of the seasonal variations of the temperature of the sea in the neighbourhood of a coast station show a relative lag as compared with the corresponding components of the air temperature of the stations. On comparing Table II with Table I, it will readily be seen that the first-order curves for the coast stations show a definite lag as compared with the inland and Continental stations. Thus the earliest occurrence of a first-order maximum for a coast station in Table II is July 30, and in Table I July 28, while the relatively inland station, Kew, shows the first-order maximum on July 23, and Oxford (for a single year, however) on July 12. Vienna is earlier than Kew, July 18, and Agra still earlier, June 29. We may thus regard the sea as causing a considerable lag in the occurrence of the seasonal variations of temperature in the adjoining coast stations, while in the sea itself the lag is still greater.
Now we may regard the periodic variation of sea temperature as a cause affecting the air temperature of the stations, and, without know- ing precisely the process connecting the cause and its effect, we may form some estimate of the relative magnitude in the following manner. Dealing only with the first-order components of the two related phenomena, we note that they are harmonic variations of the same period. We may assume that the effect of the periodic cause is itself simply harmonic and of the same period, and that the resultant first-order component for the air temperature is made up of a part which is independent of the sea temperature and a part which is dependent upon and caused by the sea temperature. We may thus assume the resultant first-order curve to be made up by compounding the two separate component curves.
The composition of harmonic curves of the same period is a geo-
