menced, which lasted all the next day. The charts alone would scarcely enable us to mark the exact line of the passage of the cyclone, but from the wind-traces we know that the centre passed very nearly exactly over Stonyhurst.

The temperature-curve is much more complicated. On the first day we see a most irregular curve, with little trace of the ordinary diurnal range of temperature, for the highest and lowest points were reached within about an hour of each other, and in the middle of the day. This can be easily explained. The station was the whole day under the influence of the rear of a cyclone, cyclone, with a north-west wind and cold showers. These and driving clouds gave rise to the sudden changes which are marked on the curve. Next day there are more signs of diurnal variation, but the highest temperature was at 6 p.m., and the second midnight is much hotter than the first. Also, though rain was collected almost every hour, we do not see the sudden changes of the preceding day. The interpretation of all this is as follows:

Commencing from the early morning, the wind began to back from west-north-west to south and south-southeast. This increased the general warmth at the second midnight, and the diurnal range took its usual course on the top of the more general change. Why the rain did not send down the thermometer will be obvious when we know that this all happened in the front of an intense cyclone, and we remember that the soft, warm, drizzling rain which falls there is not like the chilly showers of the rear of a depression.

On the third day the thermogram is still more curious.

Immediately after midnight the thermometer fell 5° Fahr., just as the trough of the cyclone passed, and the wind jumped from south-west to north. During the whole day the temperature fell, and while the greatest heat was at the first, the greatest cold was at the second, midnight. No rain was reported, so the small sudden changes must be the effect of passing clouds. Here it may be well to note that a passing cloud by day will send down the thermometer by hiding the sun, while by night the same cloud will raise temperature by cutting off the radiation of the cold sky. The general explanation is that cold winds set in when the barometer turned, and that their influence apparently completely overrode diurnal variations.

Though the thermometer fell all day—that is to say, there was no hour at which the temperature was not lower than at the preceding one-it does not follow that there was no diurnal range with a regular maximum and minimum; but, as the explanation of the superposition of curves requires some collateral details, we will defer our remarks till we have considered the wind-traces of our meteogram.

First, then, for the velocity-trace. Presuming that the normal course of the wind is to increase regularly in velocity from 4 a.m. till 2 p.m., and then fall gradually again, the most obvious feature in our wind-diagram is the increase of the wind towards the middle of the day; that is to say, the regularity of this diurnal variation in spite of the great cyclonic changes which were going on. The influence of these latter is seen in the comparatively high velocities-from thirty to forty miles an hour-which

the wind attained a little after noon on the first two days; but the trace for the last day, December 9, is very different. On that day the strongest winds were before 6 a.m., and during the hottest time of the day, when the wind is usually strongest, the velocities decreased steadily. The reason of all this was that, in the early morning, the steep gradients in rear of the cyclone were passing over the station, while in the middle of the day the gradients were becoming so much less steep that the diminishing velocity due to them entirely overrode any increase which would naturally have occurred from diurnal influences. The calm just before midnight on the 8th, and the rapid rise of the wind to thirty miles an hour directly afterwards, are very interesting; for the calm is that in the centre of an intense cyclone, and the high wind is associated with the steep gradients which we see in rear of the depression in Fig. 27.

The direction-changes for these three days are tolerably simple. The natural diurnal variation of the wind is to veer a little in the forenoon, and back again as the sun goes down; but almost all trace of this is lost in the stormy weather to which our diagram refers. On the 7th the wind kept pretty steady between west and north-west, and there is little sign of diurnal variation. Next day, the 8th, as the cyclone approached, the wind backed rapidly as far as south-south-east; then veered rapidly to south-west; and just at midnight, when the barometer turned, jumped up to the north and north-east; and then backed during the day to about north-west. From these changes, and the sudden jump of the wind from west to north, it is evident that the cyclone's centre passed very

nearly over the station. All traces of diurnal variation are, of course, entirely marked by these greater changes.

SUPERIMPOSITION OF VARIATIONS ON CURVES.

The simple conception that the actual weather is the balance or sum of various influences will now be sufficiently obvious, but we must go a little more into detail of all that can be learnt from an inspection of instrumental curves. As the idea of superimposing curves of different kinds of variation one on the top of another, and of so deducing a resulting curve, may not be familiar to some of our readers, we will commence by an easy example.

Here, and all through this work, we shall use conventionally the word "changes" to denote alterations in weather due to cyclones, etc.; and "variations" to denote alterations due to the time of day. The passage of a cyclone, or its replacement by an anticyclone, really changes the weather; diurnal influences only impose a certain variation on these greater changes.

For instance, let us try and find out what would be the nature of the curve left by a thermograph if a regular diurnal variation of temperature, which was highest at 2 p.m. and lowest at 4 p.m., was superimposed on a steady general fall of temperature, due to other than diurnal influences-say, the setting in of cold northerly winds. The line on which the diurnal curve is superimposed is called the level of variation. In the familiar curve of mean diurnal variation, such as B or c in Fig. 28, the straight, horizontal line A represents the mean temperature of the place, and the curves B or C

are the resulting traces. If they are unaltered, the only effect of any change in the level of a would be to bring

shape and magnitudes of the

diurnal curves B

or c would be

the same.

When we come to deal with the significance of a

curve for any particular day, the horizontal line A no longer represents the mean temperature of the station, but the

level of temperature from general causes independent of the time of day. If the thermograph