In Ceylon, for instance, the whole rainfall of the year occurs within a period of six to eight weeks, and often amounts to as much as 12 in. in the twenty-four hours, and has been known, comparatively recently, to reach nearly 19 in., the latter an amount only 2 in. or 3 in. less than the average rainfall of Lincolnshire for the whole year. In London it is only 25 in. and in the wettest district in Great Britain, viz., Cumberland, averages not more than 70 in. per annum.

The rainfall in Bombay is from 80 in. to 100 in. per annum, and throughout India may be taken as from 50 in. to 130 in., varying, as is the general rule, in direct ratio with the altitude, and limited to a few weeks in the year. Notwithstanding this, there still exist in the Madras Presidency a not inconsiderable number of ancient bunds which serve their intended purpose at the present day as well as ever. Slight mistakes did occasionally occur, as they ever will till no more dams are wanted, as is proved by the remains of some works in Ceylon, where the failure was evidently due to error, possibly due to the instruments being out of adjustment, as their base is at a higher level than the bed of the stream at the point where water from the latter was to be diverted to afford the supply.

Among the most remarkable of these ancient works is the Horra-Bera tank, the bund of which is between three and four miles in length and from 50 to 70 ft. in height, and although now in ruins would formerly impound a reservoir lake of from eight to ten miles long and three to four miles broad. There is also the Kala-Weva tank, with a bund of twelve miles in length, which would, if perfect, create a lake of forty miles in circumference. Both of these ruined works are situated in Ceylon. The third embankment of a similar character is that of the Cummum tank, situated in the Madras Presidency, and which, though ranking among the earliest works of Hindoo history, is still in such a condition as to fulfill its original intention. The area of the reservoir is about fifteen square miles, the dam about 102 ft. high, with a breadth at the crest of 76 ft., and of the section shown in the diagram.

The by-wash is cut in the solid rock altogether clear of the dam; the outlet culverts, however, are carried under the bank. We will now consider generally the methods employed in determining the site, dimensions, and methods of construction of reservoir dams adapted to the varying circumstances and requirements of modern times, with a few references to some of the more important works constructed or in progress, which it will be endeavored to make as concise and burdened with as few enumerations of dimensions as possible.

The amount of the supply of water required, and the purposes to which it is to be applied, whether for household, manufacturing, or irrigation uses, are among the first considerations affecting the choice of the site of the reservoir, and is governed by the amount of rainfall available, after deducting for evaporation and absorption, and the nature of the surface soil and vegetation. The next important point is to determine the position of the dam, having regard to the suitability of the ground for affording a good foundation and the impoundment of the requisite body of water with the least outlay on embankment works.

It has been suggested that the floods of the valley of the Thames might be controlled by a system of storage reservoirs, and notice was especially drawn to this in consequence of the heavy floods of the winter of 1875. From evidence given before the Royal Commission on Water Supply, previous to that date it was stated that a rainfall of 1 in. over the Thames basin above Kingston would give, omitting evaporation and absorption, a volume of 53,375,000,000 gallons. To prevent floods, a rainfall of at least 3 in. would have to be provided against, which would mean the construction of reservoirs of a storage capacity of say 160,000,000,000 gallons. Mr. Bailey Denton, in his evidence before that commission, estimated that reservoirs to store less than one tenth that quantity would cost £1,360,000, and therefore a 3 in. storage as above would require an outlay of, say, £15,000,000 sterling; and it will be seen that 3 in. is by no means too great a rainfall to allow for, as in July of 1875, according to Mr. Symons, at Cirencester, 3.11 in. fell within twenty-four hours. Supposing serious attention were to be given to such a scheme, there would, without doubt, be very great difficulty in finding suitable situations, from an engineering and land owner's point of view, for the requisite dams and reservoir areas.

In Great Britain and many European countries rain gauges have been established at a greater or less number of stations for many years past, and data thereby afforded for estimating approximately the rainfall of any given district or catchment basin. The term "watershed" is one which it appears to me is frequently misapplied; as I understand it, watershed is equivalent to what in America is termed the "divide," and means the boundary of the catchment area or basin of any given stream, although I believe it is frequently made use of as meaning the catchment area itself. When saying that the rain gauges already established in most of the older civilized countries afford data for an approximate estimate only, it is meant that an increase in the number of points at which observations are made is necessary, previous to the design of a reservoir dam on the catchment area above, the waters of which are proposed to be impounded, and should be continuous for a series of five or six years, and these must be compared with the observations made with the old established rain gauges of the adjacent district, say for a period of twenty years previously, and modified accordingly. This is absolutely necessary before an accurate estimate of the average and maximum and minimum rainfall can be arrived at, as the rainfall of each square mile of gathering ground may vary the amount being affected by the altitude and the aspect as regards the rainy quarter.

But this information will be of but little service to the engineer without an investigation of the loss due to evaporation and absorption, varying with the season of the year and the more or less degree of saturation of the soil; the amount of absorption depending upon the character of the ground, dip of strata, etc., the hydrographic area being, as a rule, by no means equal to the topographic area of a given basin. From this cursory view of the preliminary investigations necessary can be realized what difficulties must attend the design of dams for reservoirs in newly settled or uncivilized countries, where there are no data of this nature to go on, and where if maps exist they are probably of the roughest description and uncontoured; so that before any project can be even discussed seriously special surveys have to be made, the results of which may only go to prove the unsuitability of the site under consideration as regards area, etc. The loss due to evaporation, according to Mr. Hawksley, in this country amounts to a mean of about 15 in.; this and the absorption must vary with the geological conditions, and therefore to arrive at a satisfactory conclusion regarding the amount of rainfall actually available for storage, careful gaugings have to be made of the stream affected, and these should extend over a lengthened period, and be compounded with the rainfall. A certain loss of water, in times of excessive floods, must, in designing a dam, be ever expected, and under favorable conditions may be estimated at 10 per cent. of the total amount impounded.

As regards the choice of position for the dam of a reservoir, supposing that it is intended to impound the water by throwing an obstruction across a valley, it may be premised that to impound the largest quantity of water with the minimum outlay, the most favorable conditions are present where a more or less broad valley flanked by steep hills suddenly narrows at its lower end, forming a gorge which can be obstructed by a comparatively