when, in hot countries, where there is a long summer season of great drought, the surface becomes dry and hot, moisture rises in steam from below; and, as the heat and dryness increase, the accumulated waters become more and more exhausted. All this goes on without reference to the actual level of the waterline within the earth, which may be far beneath the level of the sea.

'That this is the case in the softer limestone rocks, even when not cracked, has been proved by actual experiment. That it takes place to an enormous extent in the limestones of the eastern Mediterranean is proved, if in no other way, by the fact that vines planted among bare stones, without soil, obtain an ample supply of moisture from the earth, and ripen their fruit to perfection in the hottest and driest seasons. No doubt the earth and rocks are hot, and appear dry; but so long as there remains any water below that has passed down during the rainy season, so long will a part of that water be given back to the dry and thirsty soil above.

6

If, then, as is probably the case, there is so large an evaporation from that part of the surface of the Island of Cephalonia within range of this district as to keep the water-level of the year below the sea-level, in spite of the joint supply of rain and sea-water, it is clear that the water may run in for ever at the same rate without filling up the space. And this I believe to be the correct explanation of the phenomenon.

The influx of water, however, is not small. It amounts, as far as I could make out, to more than half a million of gallons per diem for the two mills together. The fall of water from the sea-level into the cavities, where it disappears, seems to be little more than a foot or eighteen inches.

'It will be evident that, if the sea-water finds its way into any large natural cavity from which it is afterwards evaporated, a deposit of salt must be taking place in this cavity, or in rocks adjacent or connected with it. Assuming the influx to be at the rate already mentioned, this may be estimated roughly at about equivalent to an area of 10 acres or 12 acres of solid matter, 1 foot thick, accumulated each year. It is an interesting question to consider where this deposit is going on, and whether saline

springs may not thus be fed. There are no known springs in the Island of Cephalonia that present any large quantity of saline matter.'

At the commencement of the present century, when the land was imperfectly drained, the soil and surrounding marshes, having no outlet, retained and stored up the rainfall, and became the great holders or reservoirs by which the waters were impounded, and the flow of our rivers regulated with greater uniformity than at present. Since the introduction of an extended system of drainage the whole character of our rivers has been changed, and now discharge their contents with much greater rapidity and in larger volumes than they were accustomed to do before these improvements were introduced. The result of this has been favourable to the land but injurious to the mills, both as regards uniformity and the loss of power, as nearly one half the supply is carried off by floods and cannot be retained for the use of the mills. In all districts of the country where drainage and an improved system of tillage have come into operation, the effect has been a serious loss to the owners of water-mills, and has driven most of them to the use of steam, either as an auxiliary, or in some instances exclusively as a substitute for water.

In some districts and on some rivers the system of impounding the rainfall on the higher lands has been introduced; and particularly where manufacturing processes are carried on, and a large body of men employed, it is essential to success that there should be no stoppages, and that there should be always at command a uniform power equal to the requirements of the mill. Now, as the quantity of water in rivers varies considerably at different periods of the year and in different conditions. of weather, it has been found necessary, in many instances, to impound the water by means of reservoirs placed at the sources or the higher portions of the river, so as to retain the waters of wet seasons, and to part with them again in periods of drought and deficiency. To a small extent this may be effected by weirs thrown across the river, so as to retain the water which comes down at night for the use of the mill during the day. But in many instances large reservoirs of a hundred or more acres in extent, and containing when full several mil

In these the

lion cubic feet of water, have been constructed. drainage from a large extent of country is collected during the rainy seasons, and remains stored for use whenever the supply of water in the river becomes inadequate; in this way damage from floods is prevented on one hand, and the supply for an indefinite period of time is equalised on the other. Among the large works of this kind are the Shaw's waterworks at Greenock, and the Lough Island Reavy or Bann reservoirs in the county of Down in the north-east of Ireland, together with later works of the same kind for the supply of water to the cities of Glasgow and Manchester, Melbourne (Australia), &c.

Reservoirs are best placed in hilly districts, at the bottom of a valley into which the water drains from a considerable extent of country. In selecting a site for reservoirs regard must first be had to the value of the land. They should be placed in retired valleys, where the cost of the land does not bear a high ratio to the cost of construction, and, should there exist a natural lake, it may be converted into a reservoir with greatly increased economy. Regard must also be had to the nature of the site. The reservoir should be restrained as far as possible by the natural rise of the ground around it, in order that as few embankments as possible may be requisite for the retention of the water. Again, the geological structure of the country must be examined, as the quantity of water to be expected to flow from a flat country, well clothed with vegetation, will be very different from that which will pour in torrents down the steep declivities of uncovered mountains. In districts of limestone, abounding in vertical fissures and subterranean cavities, a very much smaller quantity of water will drain off the higher districts than from a non-absorbent formation of primitive rock, or where the beds are horizontal and impervious. The steeper the district and the more rapidly the water is discharged to the reservoir, the less will be lost by evaporation and absorption.

It is necessary, in constructing reservoirs, to obtain some measure of the quantity of water which may be expected to accumulate annually, in order to provide sufficient storage. For this purpose it is most important to determine the area of land draining into the valley chosen for the formation of the

reservoir, and the average annual rainfall of the district, with, if possible, the probable loss or waste arising from the reevaporation and absorption by vegetation, &c.

To ascertain the drainage area it is sufficient to determine the summit level or watershed, i. e., the ridge surrounding the valley which marks the line at which the streams flow in opposite directions into contiguous valleys. This may be determined by a special survey, with a careful examination of an accurate chart like the Ordnance map, on which the contour of the country, brooks, &c., are plainly marked. The whole of the basin included within the watershed is termed the catchment basin. In the case of the Bann reservoirs it amounts to 3,300 statute acres in extent; in that of the Greenock reservoirs to 5,000 acres; at the Manchester waterworks to 19,000 acres.

Of late years an immense number of experiments have been made on the rainfall in different parts of Europe, and with considerable success in determining the laws of rain distribution. For England the annual average rainfall amounts to about 36 in. in depth over the entire surface, distributed throughout the year as in the following table :

† Principally from Dr. Dalton; see Manchester Memoirs.

This would give a mean of 30 inches, but it must be borne in mind that in the lake districts and all along the west coast there is an annual fall of rain greatly exceeding that amount, and in some places in the higher districts in Cumberland the returns have been as high as 180 to 200 inches; from this it will be seen that 36 inches is a fair average for the whole surface of Great Britain.

It is, however, important in the construction of reservoirs to have observations of the rainfall in the district in which they are to be placed. Local causes greatly influence the quantity of rain; thus the average fall in Essex is about 20 in., whilst at Keswick, in Cumberland, it is as much as 67.5 in., and at Seathwaite, in the same county, it averages the enormous quantity of 141.5 in.

The method of determining the rainfall is very simple. A cylindrical vessel, of the form shown in section in fig. 83, is placed on the ground or sunk into it in such a manner that its mouth is about 12 in. above the surface. Sometimes a permanent rod and float is added by means of which the depth of rain received on the funnel and preserved in the vessel is read off at sight; and for ordinary purposes this is probably the best plan. But where the greatest accuracy is requisite it is necessary either to tie down the rod or to remove it altogether after making an observation, as otherwise the rain in driving obliquely impinges upon the rod instead of passing over the funnel, and a slight excess is in this way registered above the true rainfall upon the area of the vessel. It is most accurate, however, to draw off the rain and measure it in a graduated glass tube. By placing two or three of these rain-gauges at different elevations around the site of a proposed reservoir, and examining them at convenient intervals

of a week or month, it is easy to estimate the exact quantity of rain which falls upon the catchment basin in the course of one year; which, with certain deductions, is the quantity to be provided for in the reservoir. The precaution of placing the