States tri-daily weather-maps, viz. that the shape of the isotherms always appears to change more between the morning and afternoon than between the afternoon and night charts; and also that between the two latter, the shape often remained pretty constant, though the numbering had changed. For instance, in Figs. 42, 43, 44 we give

FIG. 42.-Diurnal and cyclone temperature (United States). reductions of the United States charts at 11 p.m. on the 22nd of January, 1873, as well as those at 4.35 p.m. and 11 p.m. the following day. These are to serve a twofold purpose-first, to show why the distribution of temperature was so different on two consecutive days at the same hour,

viz. 11 p.m.; and, secondly, to illustrate the diurnal variation in the shape of the isotherms between 4.35 p.m. and 11 p.m. the second day.

We will consider the latter first. The isotherms which we see on the 4.35 p.m. chart (Fig. 43) represent the distribution of temperature due to the influence of a cyclone

on a general irregular thermal slope from the equator to the pole, as modified by the diurnal range of the season. The aspect of the diurnal gradient is towards the northeast, because the temperature is falling.

By 11 p.m. the same day (Fig. 44) the centre of the

cyclone has scarcely moved, and the general shape of the isotherms is also nearly identical; but the position of the isotherms of 40°, 50°, 60° at 4.35 p.m. is taken broadly by those of 30°, 40°, 50° at 11 p.m., and the place of 10°, 20°, 30° at 4.35 is less nearly approached by those of

0°, 10°, 20° at 11 p.m. in the west. This means that the diurnal range was less in the north-west than in the south.

The interpretation of this is, that the aspect of the thermal gradients has not materially changed, though the temperature has fallen generally nearly 10°; so that

the shape and position of the disturbance of temperature set up by the cyclone remains the same, but the numbering of the isotherms is changed nearly 10°.

TEMPERATURE-DISTURBANCE OF A CYCLONE.

Now that we have eliminated the influence of diurnal range, we can better understand the nature of the heat developed by a cyclone. In the same three figures we have got rid of diurnal range by two methods. By taking the charts at the same hour-11 p.m.—in the first and third (Figs. 42 and 44), diurnal range is allowed for by being equalized, so that the whole of the difference between these two sets of isotherms is due to general changes, not to diurnal variations.

Then, by our second method of inferring the influence of diurnal slope on any shape of isotherms, we are enabled to use a chart at the intermediate hour of 4.45 p.m. (Fig. 43) for the same purpose of discovering the nature of cyclone-heat.

In all these charts we see that the general nature of the development of heat by a cyclone consists of a certain wedge-shaped projection of the isotherms northwards in front and on the southern side of the cyclone-centre, and that this heat moves on along with the cyclone. Observe that the local seasonal thermal gradient, from the cold interior of the continent to the warm sea, slopes to the north-west, while the aspect of the diurnal thermal slope is towards the north-east in all the charts. The quality of cyclone heat is very peculiar. It is not the pleasant warmth of a fine day, but has that characteristic close,

muggy, disagreeable feeling which we have before described as coming before cyclones. This is the kind of heat which develops neuralgia and similar troubles in old wounds, and many of the prognostics which are associated with the front of a cyclone. We could not have a more striking instance of the necessity of adding a descriptive account to all instrumental records of weather. Neither a thermogram nor a synoptic chart can distinguish between one kind of heat and another.

*

The cause of this heat is obscure. The author has shown that it is not altogether caused by that backing of the wind towards the south which precedes the rainy portion of a depression, and that the rise of temperature seems due to some peculiar property of cyclone-action.

In an ordinary whirl of dust or leaves we find the particles most compressed on the side where the directions of rotation and translation coincide; that is to say, if the whirl is against the watch-hands, and the motion in any direction, the compression is always on the right-hand edge of the eddy, looking towards the front.

If we reflect that a chart of cyclone-heat shows a wedge projection of the isotherms on a general thermal slope, we can readily understand how such a form may be analyzed into a detached patch of heat lying on a general thermal slope. We are thus led to the conception of a patch of heat developed by the cyclone, and moving about with it, like all the other characteristics of such a whirl.

For this reason, we often find exceptionally high

* Abercromby, "On the Heat and Damp which accompany Cyclones," Quarterly Journal of the Meteorological Society, London, vol. ii. p. 274.