Now the climate of England is remarkably humid both in winter and summer. And this humidity is shown, not so much by the quantity of rain which falls,

1 Gilbert White noticed long ago-apparently without understanding -the influence of a clouded sky on the temperature. 'We have often observed,' he says, ' that cold seems to descend from above; for, when a thermometer hangs abroad on a frosty night, the intervention of a cloud shall immediately raise the mercury ten degrees; and a clear sky shall again compel it to descend to its former gauge.' as by the frequent presence of large quantities of aqueous vapour in the atmosphere. Skies, even, which we in England consider clear, are overcast compared with the deep-blue skies of France or Italy. What the influence of these humid palls may be 'on the out-flow of moral sentiments' which Humboldt considered to be so favourably influenced by transparent skies, I shall not here pause to inquire. It is clear, however, that the influence of our cloudy skies tends to modify the severity both of our winter and our summer seasons; and these benefits are so great that we may cheerfully accept them as more than a counterpoise for hypothetical injurious effects on 'the outflow of our moral sentiments' (whatever that may mean).

I proceed to consider the actual variations presented in the course of a year in England. As some selection must be made, I shall select a series of observations which have been made at Greenwich during the present century. It will be gathered from the preceding pages that the range of temperature at Greenwich is at least not less than the average range of the British Isles. Greenwich, also, from its neighbourhood to London, and from the number and accuracy of the observations made there, is obviously the best selection that could be made. It must not be forgotten, however, that the climate of Greenwich is not the climate of the British Isles, and that careful observations made in other places have sufficiently indicated the existence of local peculiarities, which, therefore, it may fairly be assumed, characterise also the Greenwich indications.

In Fig. 3 the annual variations of mean diurnal temperature are represented graphically. The figure was formed in the following manner:-A rectangle having been drawn, each of the longer sides was divided into 365 parts, and a series of parallel lines joining every tenth of these divisions was pencilled in. The spaces separating these lines represented successive intervals of ten days throughout the year. The shorter sides were divided into thirty-three parts and parallel lines drawn, joining the points of division. Of these longer parallels the lowest was taken to represent a temperature of 32° Fahrenheit (i.e., the freezing point) and the others, in order, successive degrees of heat up to 65°. Then, from the Greenwich tables, which have been formed from the observations of forty-three years, the temperature of each day was marked in, at its proper level and at its proper distance from either end of the rectangle. Thus 365 points were marked in, and these being joined by a connected line, presented the curve exhibited in Fig. 3. The lines bounding the months, and the lines indicating 35°, 40°, etc, Fahrenheit, were then inked in and the figure completed.

The resulting curve is remarkable in many respects. In the first place, it was to have been expected that a curve representing the average of so many years of observation would be uniform; that is, would only exhibit variations in its rate of rise and fall, not such a multiplicity of alternations as are observed in Fig. 3. And this irregularity will appear the more remarkable when it is remembered that the temperatures used as

The Climate Of Great Britain 5

Fig. 3.-Annual Variation of Moan Diurnal Temperature at Greenwich.

the Greenwich means are not the true average temperatures. They were obtained by constructing a curve from the true averages, and taking a curved line (the curve of Fig. 3, in fact) in such a way as to take off the most marked irregularities of the true curve of averages; or to use the words of the meteorologist who constructed the Greenwich table of means, Mr. Glaisher, a curved line was drawn which passed through or near all the points determining the true curve of averages, 'and in such a way that the area of the space above the adopted line of mean temperature was equal to that below the line.' Despite this process, the curve exhibits no less than fourteen distinctly marked maxima of elevation, and a much larger number of variations of flexure. The sudden variations of temperature at the beginning of February, early in April, and early in May are very remarkable; they have their counterparts in the three variations which take place between the latter part of November and the end of the year, only these occur in much more rapid succession. The nature of the curve between June and August is also remarkable, as are the three convexities which are exhibited in the September, October, and November portions of the curve.

If we follow our leading meteorologists in taking the curve of Fig. 3 as representing the true annual climate of London, how are we to assign physical causes for the remarkable variations above indicated ? Not easily, I take it. It were, indeed, as easy as inviting to speculate on cosmical causes; to follow Ertel, for instance, in assigning effects to those zones of meteorities which are known to. intersect the earth's orbit, and others which may fairly be assumed to fall within or without that orbit. It may be, perhaps, that the recognised shooting-star periods have, some of them, their counterparts in heat-changes; but certainly the time has not yet come to pronounce a consistent theory of such effects. The evidence afforded by the Greenwich curve on this point is unsatisfactory, to say the least. The elevation at the beginning of January, and the marked irregularity in February, correspond to Ertel's views; so also the fact that large aerolites have frequently fallen in the first week in April, about the 20th of April, about the 18th of May, early in August,1 about the 19th of October, and early in December, seems to correspond to elevations in the curve; while depression opposite the 12th of May, might be referred to the intervention of the zone of meteors, which causes the now celebrated November shower. But the negative evidence is almost equally strong. Where, for instance, is the elevation which one would expect, on Ertel's theory, in November ? Also, if the cause of the observed irregularities were that suggested by Ertel, the curves for other countries in the northern hemisphere should exhibit similar irregularities on corresponding dates, which does not appear to be the case. In fact, if there really exist effects due to cosmical causes, these are not likely to be educed from observations of the variation of mean diurnal temperature, since it is clear that a cause of variation due to objects external to the earth could affect only the temperature of certain hours of one day or of several days. A cluster of meteors between the earth and the sun might diminish the mid-day heat; one external to the earth's orbit might increase the nocturnal temperature; and though in either case the mean diurnal temperature would be affected, yet it is obvious that the effect would be masked in taking the mean, or even that two or more opposing influences might cancel each other. If it could be shown that the curve for mid-day, or for midnight heat corresponded to the curve of mean heat, Ertel's theory would be overthrown at once; since, for its support it is necessary to show that depressions in the mean curve are due to mid-day loss of heat, and elevations to midnight gain of heat.