temperature on the north of a cyclone-path in the rare cases when the propagation of the depression is towards the west instead of towards the east, as is generally the rule. We shall recur to the importance of this fact, and give an illustration, in our chapter on Forecasting by Synoptic Charts.

But we may now describe in more detail the temperature-changes in the United States for the twenty-four hours to which the chart refers.

In the first map (Fig. 42) the centre of an irregular cyclone is near Memphis; the isotherm of 50° projects into this depression; the isotherm of 40° reaches nearly as far north as St. Louis, and all the Mississippi valley below that city is warm. By 4.35 p.m. the next day (Fig. 43) the cyclone has moved in a north-west direction to Indianopolis, and the isotherm which now projects most is that of 40°. Temperature has fallen all over the Mississippi valley, from the cold winds in rear of the cyclone. But what we have to notice most are the temperatures recorded at the Ohio stations, just in front of the upward projection of the isotherm of 40°, which were as follows:-Toledo, 19°; Cleveland, 27°; Pittsburgh, 31°. By 11 p.m. the same evening (Fig. 44) the centre of the cyclone had only moved a few miles, but that was sufficient to bring the stations just mentioned more within the range of higher isotherms than earlier in the afternoon. That is to say, the thermometer rose at all those stations between 4.35 p.m. and 11 p.m., although in an ordinary way we expect to see the mercury fall with the The actual figures were-Toledo, 3°, Cleveland and Pittsburgh 20 each, higher than in the afternoon.

sun.

But while temperature has been rising in Ohio, many of the stations in the lower Mississippi valley have lost from 5° to 8° from diurnal causes. Other stations, such as Montgomery, Alabama, have lost no less than 13° from a combination of diurnal and cyclonic influences. A glance at the charts will enable us to see this at once, for, while the Mississippi stations are in the same portion of the cyclone at both hours, the latter station was in front of the cyclone at 4.35 and in rear at 11 p.m.

A few years ago, no explanation could have been given of this apparent anomaly of the air getting hotter as the sun went down; but now we see that it was due to the temperature-disturbance of a cyclone overriding the ordinary variation of diurnal influences. In our chapter on Meteograms we showed how similar changes would affect the trace of a thermograph; now, to complete a comprehensive view of the subject, we have illustrated the same phenomenon by the totally different method of synoptic charts.

Nothing could show better the extreme facility with which synoptic charts enable us to study temperaturechanges in spite of diurnal variation. But just as it is not at all obvious at first sight how changes in the position of isobars are reflected in a barogram, so nothing but a good deal of experience will enable the meteorologist to see readily how changes in the position and shape of the isotherms would affect the indications of a single thermometer, or to handle with any ease the idea of the propagation of diurnal isotherms over a complicated system of temperature-distribution. Our example is one of the simplest which the author could find. In the first

and third charts we equalize diurnal influences by constructing the maps at the same hour each day. By this means we can explain why the Atlantic states were colder on the first day than on the second, and why the Mississippi valley was colder on the second than on the first day.

By our second and third charts we illustrate the manner in which general changes override diurnal variations when the latter are not very strong, as during the winter months, as well as the characteristic nature of a diurnal fall of temperature on an existing system of

isotherms.

There are nearly eighty stations in the United States and Canada. During the six and a half hours in question, changes in every direction of varying magnitude took place at each; but there is not one, however apparently anomalous, which cannot be explained by means of the principles which we have here laid down.

SOURCES OF HEAT.

We were obliged to introduce the question of diurnal variation of temperature in the first place, so as to get rid of any ideas of difficulty from that source of complication; but now, before we describe further the changes in the isotherms from day to day, we must consider the various sources of heat and cold with which we have to deal. In all this we must never forget that the natural distribution of temperature is an irregular thermal slope from the equator to the pole, and that what we have to explain are the divergences from that ideal distribution which we find in practice. We shall find that places

far north are sometimes much warmer than others nearer the equator, and that some parts of Europe are often colder in March than in January. All these apparent anomalies we can explain easily, but we must begin with sources of heat.

The primary source of heat is, of course, the sun, so that, other things being equal, we should get the greatest heats where there is the least cloud; that is to say, generally in anticyclones. This, however, cannot be laid down as a general rule without some modifications. In the belt of anticyclones which surround the world about the line of the tropics, some of the greatest known heats are recorded, notably in the Sahara and in Australia. But in higher latitudes the sun has a powerful enemy in the cold space which surrounds the earth. In summer the sun is the more powerful, and we get hot days with cold nights. In winter-time, when the sun is low, radiation into space overpowers the radiation from a low sun, and clear weather is cold. When, then, we come to discuss in general terms the influence of cloud on isotherms, we must always take into consideration the time of year and latitude.

Another very powerful source of irregular isotherms is found in wind. Of course, speaking broadly, southerly winds will deflect the isotherms northwards, and northerly or easterly winds will bend them towards the south. This too is, however, subject to many irregularities. The great difference in the radiating power of land and water at different seasons, makes a continental area colder in winter and hotter in summer than a sea in the same latitude. For this reason an easterly wind from a land

area would blow warm into a neighbouring sea in summer, and cold in winter.

We have already alluded to the idea of a specific quality of heat which is developed by cyclones, and minor local sources of variation, such as the descending winds which probably constitute the "fohn," need only be mentioned here.

When all these sources of heat are combined, a vertical sun, a cloudless sky, a southerly wind, and an arid soil; when light, hot puffs fill the air with scorched particles of sand, till the dulled sun appears to glow in a sea of molten brass, and the poisoned breath of the simoon sweeps fitfully across the desert, then the traveller may well beware, and hasten for his life to the nearest shelter.

An example of great heat will be found in the charts, Figs. 82 and 83, which illustrate the first burst of the south-west monsoon in Hindostan. The dates, June 17 and 18, 1881, would coincide with the end of the hot season in Northern India. Our diagrams show on both maps a patch of heat over 100° Fahr. (38° C.) over the Desert of Scinde; and, as this would be at about half-past five in the afternoon locally, it is certain that much higher temperatures must have been recorded nearer midday. The isobars show that what wind there was would be southerly or south-westerly, and of course light from the absence of gradient near the centre of the depression. The soil there is saltish sand, and similar material has been known to get heated up to nearly 200° in Australia. In Scinde there is a dangerous wind at this season exactly analogous to the simoon of Arabia and the Sahara. Both are certainly allied to whirlwinds and tornadoes; but,