The highest monthly average known is that of May for Massowah, which amounts to 99°, and the difference between this and the lowest for January, -56° at Werchojansk, is 155°. The contrast between individual temperatures actually observed is, however, far greater than this. Instances of the temperature to which a sandy soil may rise have been given at p. 46. As to instrumental observations of air-temperature, 130° has been recorded at Mursuk in Fezzan; and at Cooper's Creek in Australia, where Burke and Wills died, a thermometer graduated up to 127°, and left for safety in the fork of a tree, was found to have been burst by the expansion of the mercury. On the other hand, Gorochow recorded -81° December 30, 1871, at Werchojansk, and Kane noticed -69° at Rensselaer Harbour.

Accordingly we see that the human system can bear a difference in the air-temperature of over 210°, or of thirty degrees more than the interval between melting ice and boiling water. As regards the relative endurability of these extremes, Dr. Moss, in his 'Shores of the Polar Sea,' has some remarks to the point. He says:--

Many a time the relative merits of Arctic cold and Tropical heat were warmly canvassed. Many of our officers and men had lately returned from the Ashantee campaign, and they could speak with authority. There was one thing clear, one could sometimes get warm in the Arctic, but never get cool on the Coast.'

So much for temperature at or near sea-level. I shall return to the question of climate in a later chapter, but it will be interesting to say a word or two about the annual march of temperature on mountaintops, as far as the observations available enable us to

exhibit it. It has been already stated that, as a rule, temperature decreases with height. During the year 1865-6 observations were made on the Col St. Théodule, near Zermatt, about 11,000 feet above the sea, and of late years, at a height of over 14,000 feet, on Pike's Peak, Colorado, by the Chief Signal Office of the United States.

The extreme of cold registered on the Col St. Théodule was only 6°.5 F., so that there was no severe frost at all, but the temperature was uniformly low, for the thermometer never rose above 32° from November to April. The results for the summer temperature were, however, the most remarkable, the average of the three summer months being 32°.3, and that of July 33.5. These are the lowest summer temperatures as yet recorded at any station, and they teach us how very greatly the amount of range is influenced by elevation.

Dr. Hann has discussed these results, and has calculated from the general features of temperature at high levels in Europe a table for its diminution with height. This table gives, for the Western Alps, 273 feet as the difference of level corresponding to 1° F. of reduction, in July. Consequently, assuming the mean temperature at sea-level for Switzerland to be about 73° in that month, the level at which the mean July temperature of Northumberland Sound (36°·7), about the lowest we know of, would be met with in the Alps, would be 10,000 feet, while the average annual temperature of -2°5, that of Rensselaer Harbour on Smith's Sound, would not be found until the level of

1 Zeitschrift der Oesterreichischen Gesellschaft für Meteorologie, vol. v.

about 18,000 feet, i.e., more than 2,000 feet above the summit of Mont Blanc.

The July temperature of Pike's Peak is 42° F., but then it must be remembered that this mountain is seven degrees nearer the equator than the Col St. Théodule.

CHAPTER XIII.

THE DISTRIBUTION OF ATMOSPHERIC PRESSURE, AND THE CIRCULATION OF THE ATMOSPHERE.

THE distribution of atmospheric pressure over the globe, although, as we shall see, it is closely connected with that of temperature, did not attract much notice from scientific men for nearly half a century after Von Humboldt had introduced the drawing of isothermal lines. The credit of first publishing isobaric charts, and connecting the winds with variations of pressure, belongs to Buchan, whose paper on the subject appeared in 1869,1 and whose curves have hardly been altered materially by subsequent investigators.

Later on we shall find how close is the relation of pressure distribution to that of temperature, for it is impossible to deal with the former without constant reference to the latter. I shall therefore commence by pointing out the broad principles of the action of temperature on pressure, and of the great atmospheric circulation thereby set in action, as explained by Hann.

If we were to suppose the earth at rest, with its entire surface at a uniform temperature, the atmosphere would form an envelope consisting of layers,

The Mean Pressure of the Atmosphere and the Prevailing Winds over the Globe for the Months and for the Year.' Transactions of the Royal Society of Edinburgh, part 2, vol. xxv. Edinburgh, 1869.

or shells, concentric with the globe; the pressure being the same throughout any one layer, but varying from layer to layer according to a law depending on the distance of the layer from the earth's surface, or on its altitude. There would be no tendency to any disturbance of the equilibrium-there would be no motion. of the air-no wind.

If now any portion of the earth's surface be heated, the air resting upon it will be expanded, and the equilibrium will be disturbed.

The action of this heating will be to cause the layers, or surfaces of equal pressure, to expand like the outside of an inflated bladder, and therefore to rise from the earth. The lowest will be most raised, and the pressure at the upper levels will increase.

It is easy to show how this must be the result. If we imagine a column of air confined in a chimney, extending above the limit of the atmosphere, and therefore only able to expand in a vertical direction, and if we apply heat from below the total weight will be unchanged, as no air can escape from the chimney laterally, while at any section taken above the base, a greater amount of air will be above that level when the column is heated than when it is at its ordinary temperature, and the pressure at such upper section will be increased. We can prove that this action really goes on in the atmosphere by the following comparison of the monthly mean readings at different levels in winter and sunimer :—