AFRICA.

BY J. R. SUTTON M.A., F.R.MET.S.

It has been pointed out before* that in expressing the terms representing the climate of a place, it is necessary to give, in addition to the elements of maximum and minimum, mean, and range of temperature, what is known as the mean inter-diurnal variability of temperature, that is to say, the mean difference between the temperatures of one day to the next. The present paper is a contribution to this work for South Africa, giving the variability for three typical stations, namely, for Durban, East London, and Kenilworth (Kimberley), of the maximum and minimum temperatures for each month for each station, from ten years' observations, together with the variability of dew-point and relative humidity at Kenilworth at the hours 2 a.m. and 2 p.m.

The mean monthly values of the inter-diurnal variability for Durban are:

It appears from this that the variability of maximum temperature at Durban is greatest in the early summer, at least at mid-winter. This result is brought about chiefly by the frequent warm and occasional very hot (foehn) winds of September to December. Generally speaking, the temperature of one day at Durban is pretty much like the next; but a hot wind means in nearly every case a sudden rise from the normal maximum one day and an equally rapid fall the next, such pairs of values of great variability making a considerable addition to what would otherwise be the mean variability of the month. It is quite a mistake, though a common one, to *See inter alia, Hann, Lehrbuch der Met., 1901, p 115; Ward, "Suggestions as to a more Rational Climatology," Report of the Eighth Geographic Congress, 1905, p. 280.

suppose that these hot winds last for days: as a matter of fact they seldom last more than a few hours, and are usually confined to the daylight hours. They seldom blow during the night, or affect the minimum temperatures to any great extent, although they may on rare occasions recur the second day. The variability of minimum temperature at Durban is least in the autumn and greatest in the spring. Thus the day temperature variability is falling during the first half of the year, while the night temperature variability is rising. The greatest variability of the minimum temperature nearly coincides with that of the maximum; but, as we shall see later, this result is not brought about by the extension of the hot winds into the night, but, on the contrary, is caused by temporary cold winds which, in their turn, have no great influence over the day temperatures. In fact, while the hot winds are essentially cyclonic, the cold winds are as essentially anti-cyclonic, both resembling their prototypes of continental Europe, but differing from them to some extent because of the special geographical conditions of South Africa. Naturally, however, the more frequent the occurrence of cyclonic conditions, the more frequently will anti-cyclonic conditions alternate with them, and therefore, so far, the hot and cold spells may be said to be related to each other. The greatest monthly mean variability of maximum temperature at Durban during the ten years discussed appears in January, 1900, with 7.9°, the least in July, 1901, with 2.6°; the greatest monthly mean variability of minimum appears in October, 1900, with 4.7°, the least in March, 1896, with 1.4°. Large monthly mean values of the maximum temperature variability occur in any of the summer months, while fairly large values of the monthly means of minimum temperature variability occur all through the year.

The mean monthly values of the inter-diurnal variability for East London are:

TABLE 2. Variability of Temperature at East London,
1895 to 1904.

At East London the variability of minimum temperature follows the same rule as at Durban, being least in autumn and greatest in spring; the spring values are, however, greater at East London than at Durban, and the average variability for the year is also greater by nearly half a degree. The variability of maximum temperature at East London is, however, of quite a different character from that at Durban, for it is least in summer and greatest in winter. Here, again, we have a result due to the hot winds, which are frequent at East London during the winter season. The average variability of maximum temperature for the year at East London 4.1°, i.e., nearly half a degree less than that of Durban. The greatest monthly value of maximum temperature variability at East London during the ten years 1895-1904 is found in July, 1898, with 8.5°, August, 1897, coming next with 8.4°; the least is in March, 1895, with 1.9°. The greatest monthly value of minimum temperature variability is found in August, 1897, with 5.6°, the least in February, 1897, with 1.7°.

At Kimberley the minimum temperature variability follows the same rule as at Durban and East London, being greatest in October and least in February and March. But the Kimberley maximum

TABLE 3. Variability of Temperature at Kenilworth
(Kimberley), 1896-1905.

temperature variability differs from both. As at Durban, it is greatest in October; as at East London, it shews a drop in December; whereas its least value comes in April, that is, later than it does at East London, but earlier than it does at Durban. Thus

on the whole it appears that the minimum temperature variability is probably common to the whole country, while the maximum temperature is determined mainly by local conditions. There is a further curious point that the maximum temperature variability is greater than that of the minimum in every month, both at Kimberley and Durban, whereas at East London the minimum values are the greater in the last quarter of the year. In particular, it may be noted that great individual values of the maximum temperature variability do not frequently occur in pairs, on alternate days, at Kimberley, as they do at East London and Durban. The average maximum temperature variability at Kimberley for the year is 4.3°, that for the minimum is 3.7°. Thus the mean variability is greater on the tableland than it is on the coast. The greatest monthly mean maximum temperature variability at Kimberley during the ten years 1896 to 1905 is found in October, 1899, with 8.1°, the least in July, 1905, with 1.7°. The greatest monthly mean minimum temperature variability is found in October, 1904, with 6.5°, the least in April, 1905, with 1.6°. These greatest and least values are not common to the whole country; it by no means follows that a high value of the variability at one station indicates a high value at another at the same time.

The monthly average values of the variability of dew-point and relative humidity, at Kimberley, at 2 a.m. and 2 p.m., are as

TABLE 4. Variability of Dew-Point at Kenilworth
(Kimberley), 1896 to 1905.

TABLE 5. Variability of Relative Humidity at Kenilworth (Kimberley), 1896 to 1905.

From Tables 4 and 5 it appears that both the absolute and relative humidities change more from day to day in the height of summer than they do in the depth of winter; but that, as distinguished from the temperature variations, the variability of moisture from night to night is considerably greater than that from day to day in respect both to mean and extreme values. This is, to me, a surprising result, considering that the relative humidity has a much wider range by day than by night.

The object of this brief paper is mainly climatological, but it will not, perhaps, be out of place to glance at some of its meteorological aspects.

We have said that the inter-diurnal temperature variability on the coast is much greater than it would be if hot and cold winds did not occur. The hot winds have been dealt with elsewhere.* They are still difficult to account for entirely, but enough is known of them to make it certain that they are associated with a definite type of weather, that they originate on the table-land and blow outwards and downwards, and are quite analogous to the foehn winds of Europe, Greenland, and elsewhere. The cold winds of the coast have not yet been discussed, although they are, from a meterological point of view, at least as interesting as the hot winds.

For the purpose of obtaining a preliminary idea of these coldweather periods on the coast, I have selected 85 typical cases of low minimum temperature registered in any month during the nine years.

* See Stewart, "The Meteorology of South Africa," Science in South Africa, 1905, p. 40; Hann, Met. Zeitschrift, Jan. 1904, p. 42; Sutton, "The Climate of East London," Trans. S.A. Phil. Soc., Vol. XVI., Part 3.