gauges within 5 in. or a foot of the ground is important, as, in accordance with an ill-understood law, the quantity of rain rapidly decreases even at slight elevations from the ground, and it is also important to place the gauge where no artificial currents of air are created, as by the sloping side of the roof of a house. This subject was fully investigated several years ago by Mr. J. F. Bateman and a committee of the Manchester Philosophical Society. Observations had been made on and near the lines of the Ashton and Peak Forest Canals, about the accuracy of which, from their disagreement, doubts had arisen. The gauges in these observations were placed on the ridging of the roofs of the houses of the various lock-keepers, under the impression that, from the exposure of the position, all the rain which fell must there be caught. New gauges were placed in the same localities, but at the surface of the ground, and the results of these experiments were as follows:

This enormous difference, amounting to 50 per cent. on the average, fully proves the unfitness of the roofs of houses for registering the rainfall. The upward currents of wind created by the sloping roof appear to have carried the raindrops over the edge of the gauge.

Dr. Heberden found the annual fall of rain at the top of Westminster Abbey to be 12.099 in.; on the top of a house close by of much inferior altitude, 18.139 in.; on the ground, 22.608 in.

Mr. Phillips, at York, found the total fall for three years at an altitude of 213 feet to be 38.972 in.; at 44 feet, 52.169 in.; and on the ground, 65.430 in.

Notwithstanding the explanations of these facts which have

been offered, Sir J. F. W. Herschel has within the last year asserted that the cause is yet to seek. The raindrops certainly appear to increase in size in the moist lower strata of the atmosphere.

Mr. Phillips's explanation has been accepted by some Meteorologists, that this augmentation is caused by the deposition of moisture on the surface of the drop, in consequence of its temperature being lower than that of the moist strata of air through which it passes. But this does not appear to be consistent with the fact, that in the condensation of vapour a large amount of latent heat would be liberated. Mr. Baxendale, who pointed this out, estimates from Professor Phillips's observations that in the condensation of the amount of water which corresponds to the augmentation of the raindrop in a fall of 213 feet sufficient heat would be liberated to raise the temperature of the drop to 434° F.

The quantity of rain which falls in twenty-four hours is about 1 in. at the maximum in average districts in England, although in the remarkably exceptional district in Borrowdale, already alluded to, 6.7 in. have been known to fall in the same period. The western coasts generally receive a larger proportion of water than other districts. Mountainous districts in this country, to an elevation of 2,000 feet, receive a larger proportion of rain than lowlands. According to the late Dr. Miller, there fell in twenty-one months in the lake district::

Mr. Bateman's observations agree with these results in proving the increase of rainfall corresponding to increased elevation,*

* The increase of rainfall in passing from the valley to the mountain must be carefully distinguished from the decrease as we ascend upward into the atmosphere, as shown in Mr. Phillips's observations.

as shown by the following figures, representing the rainfall near Glossop in one year :

After having determined from these considerations the quantity of water annually falling on the drainage district of a proposed reservoir, it is necessary in the next place to ascertain the probable loss from evaporation and other causes during the transmission to the reservoir. The numerous experiments on evaporation made upon small surfaces of water and of earth may be dismissed as having afforded too inconsistent results to be of any practical value.* Dr. Dalton's experiments are accurate and valuable as far as they go, but they are deficient in points of application to practical investigations. The area from which evaporation takes place is identical neither with the area of the catchment basin nor with the reservoir surface; but is a variable quantity, depending on the season, the climate, and the locality. It appears to me that the evaporation from a surface of water in low flat land charged with moisture, or a level vegetated surface, is very different from the evaporation in mountainous districts where there are precipitous descents to the brooks. In the former case the waters are retained and remain for weeks more or less exposed to the solar rays and the drying influences of wind. In the latter the rain pours in torrents down the barren hill-sides, and is launched into the valley where the principal evaporation takes place upon a very limited area of surface.

So also in tropical countries; the evaporation from a surface of water is greater than the rainfall upon the same surface, but then the rain falls in torrents, and is rapidly carried away to its

* Dr. Dalton gives the annual evaporation from a surface of water as 25.158 inches; Dr. Dobson, 36.78 inches; Dr. Thomson, 32 inches. The above views in regard to these experiments I expressed in a report on the Bann reservoirs in 1836. Mr. Cony beare gives the evaporation from a surface of water at Greenwich Observatory 5 feet, at Bombay 8 feet, and at Calcutta 15 feet per annum,

natural or artificial reservoirs, and then the evaporation takes place from a very small area of surface.

Since the establishment of reservoirs and the carrying out of large drainage operations, opportunities of estimating the relation of the rainfall to the discharge by rivers have been generally available, and several important experiments have been made in this way. The method of arriving at results is to ascertain the rainfall over a catchment basin the area of which is known. The whole of the water discharged by brooks, &c., is then conveyed over a rectangular weir or waste-board, and the mean velocity of the current and its breadth and depth determined by observations made once or twice every day. The comparison of the amount of water discharged with the total fall will afford the data for ascertaining the amount of evaporation.

Observations of this kind were made by Mr. Bateman with great care in the years 1845, 1846, 1847, with reference to the construction of reservoirs for the supply of Manchester with water, from the Derbyshire hills beyond Staleybridge and Mottram. Gauges were placed at the bottom of the Swineshaw valley (through which flows a tributary of the Tame) and near the summit of Windyate edge, and for some time a gauge was placed midway between these places. Similar gauges were placed in Longendale valley, and the stream in each was measured two or three times a day. From these observations the following table is compiled :

The first was a yet year, the second one of the dryest on record, the third an average year.

* The rainfall possibly somewhat too high. Manchester Memoirs, vol. ix. p. 17.

By uniting the observations at the Swineshaw and the Longendale valleys, we get the following general table of the monthly fall and flow for three years:

In the following table I have collected the most reliable results on the relations of discharge, rainfall, and evapora

The above table shows a loss of from 12 to 20 in., or an average waste of 16 in., of rainfall arising out of re-evaporation and other causes of absorption.

'The storage requisite for equalising the supply of water between dry and wet years should be provided with a due reference to the continuance of drought, and the quantity of water which will flow off the ground: in extreme wet seasons no water should be allowed to run to waste. Experience has shown that