Now it is observable, that the evaporation is greatest from March to August; consequently, the air is then farther from the point of saturation, or has a greater capacity for vapour, than in the other period; or, in other words, it is drier, relative to its temperature, than in the other period.-Hence then we have a strong argument for the theory of the barometer as well as for that of rain.

But to be more particular in the investigation :— It will be seen that there have been six months when the mean state of the barometer at Kendal was 30 inches or above; nine months when it was 29.9, or from thence to 30 inches; seventeen months when it was 29.8, or from thence to 29.9, &c. as per the following table.-Now, in order to examine the relation of the barometer and rain, it will be proper to find the mean monthly rain for those distributions of the months when the mean state of the barometer was nearly the same.

we have done, and the result follows.

This

* The account in this column, is the result of the 3 years' observations we have inserted in the first part; the first mean is for 4 months, when the barometer, at London was 30.1 plus; the second for 6 months, when it was 30 plus, &c. the rest are for 7, 11, 5, 2, and 1 months, respectively.

†There was no rain-gauge this month at Keswick; the quantity set down is got by comparison only.

The inferences to be drawn from this table are, 1st. The higher the barometer is above its mean annual state, the less rain there is. 2nd. The farther it is below its mean annual state, the more rain there is, till it comes to a certain point, after which the rain seems to decrease again.

The first of these inferences, being conformable to common observation, was expected; but the conclusion in the second, that the monthly mean state of the barometer may be too low to be attended with the maximum of rain, was not apprehended, till the preceding table, which seems to warrant it, was digested. However, it was immediately perceived, that the point might be cleared up, by selecting all those days which have produced the greatest quantity of rain, and finding the mean state of the barometer upon those days, which may be taken for that state most conducive to the greatest quantity of rain.-The result of a careful examination of my own observations, at Kendal, follows: during the extraordinary fall of rain on the 22nd of April, 1792, (see page 37,) the mean of the barometer was 29.62: the other two days that gave more than two inches of rain each, the barometer was 29.59 and 29.33 respectively as for the other fifty-six days, on each of which there was more than one inch of rain, the mean state of the barometer upon the whole of them was 29.47, and for fifty-four of those days the barometer was between 29.03 and 29.81; the barometer on the other two days was plainly irregular, being on the

de. I STL kalue, that the rain of that GLT VIS TEHT :ce act on the other it was 30.06, attended with an extracay circumstance.

(See

From this is appears that the heaviest rains may be expected when the barometer is about 29.47, at this place, cr. in round numbers. 29 inches, which is a little abone the mean of the two great extremes coserved in January 1789. or 29.44.

In the last five years there have been 1827 days, of which 1082, as per account, had rain, more or less, at Kendal, and 59 of those gave above one inch of rain each: hence, at an average, there has been one of such days in every thirty-one, wet and fair, and in every eighteen wet days, nearly. The number of days when the mean state of the barometer was below twenty-nine inches, were forty, of which two only were fair; and yet there was but one of those that gave one inch of rain. From these facts we may conclude, that when the barometer is very low, the probability of its being fair is much smaller than at other times; but that, on the other hand, the probability of very much rain, in twenty-four hours, is not so great as at other times, which is consistent with the conclusion obtained from the facts stated in the preceding paragraphs.

Upon an enumeration, it appears, that there have been seventy-eight days in the different months of the last five years when the mean state of the barometer, at Kendal, was above the usual high ex

treme for the month, as stated at page 15; only seven of those days were wet, and the rain in very small quantities; hence, the probability of a fair day at that place, to that of a wet one, in such circumstances, is as ten to one.

The preceding facts offer nothing but what appears consistent with the theories of the barometer and rain; when the barometer is above the mean high extreme for the season of the year, the air must, relatively speaking, be extremely dry or cold, or both, for the season; if it be extremely dry, it is in a state for imbibing vapour, and if it be extremely cold, no further degree of cold can then be expected, and therefore in neither case can there be any considerable precipitation on the contrary, when the barometer is very low for the season, the air must relatively be extremely warm or extremely moist, or both; if it be extremely warm, it is in a similar state to dry air for imbibing vapour, and if it be extremely moist, there must be a degree of cold introduced to precipitate the vapour, which cold, at the same time, raises the barometer. From which it follows, that no very heavy and continued rains can be expected to happen whilst the barometer actually remains about the low extreme, but they must rather be the consequence of a junction or meeting of extremes, which at the same time effects a mean state of the barometer.

144

ESSAY EIGHTH.

On the Aurora Borealis.

As this essay contains an original discovery, which seems to open a new field of inquiry in philosophy, or rather, perhaps, to extend the bounds of one that has been, as yet, but just opened; it may not perhaps be unacceptable to many readers to state briefly the train of circumstances which led the author to the important conclusions contained in the following pages.

It will appear, from the observations, that the author has been pretty assiduous, during the last six years, in noticing those very singular and striking phenomena, the aurora boreales, as often as they occurred; in which time he has also seen and considered, with a proper attention, several conjectures and hypotheses, endeavouring to account for them; but as no hypothesis has yet appeared that explains the general phenomena in such a manner as to procure the acquiescence of any rational inquirer, it was natural to expect that his attention would occasionally be turned towards an investigation of the nature and cause of the aurora.

It seemed to be sufficiently proved that the aurora was not without the earth's atmospherè, though he had never seen anything done which ascertained the real height of any one appearance with a