Upper Roxborough (Philadelphia), where the water is carried for a short time in subsiding basins before it is thrown on the filters, show the remarkable change which can be made in a few hours by means of plain sand filters, and it is scarcely likely that in the future construction of works of public water supply many instances will be found where settling basins are planned to furnish more than two days' subsidence of the water before it passes to the filters for purification. The objection to any sewage polluted water is founded entirely upon the matters carried in mechanical suspension. The waters of large lakes or rivers probably never contain dissolved matters which unfit them for drinking and other domestic uses, nevertheless, few of these sources contain water which in its natural condition is fitted for drinking and other domestic requirements, and any system of purification which will remove all the suspended matter will be found capable of rendering the waters of large lakes and rivers well adapted for domestic purposes. The sole purpose of water purification is therefore to remove matters in suspension, of which the bacteria are a part, and as this can be accomplished much quicker, cheaper and better by plain sand filters than by large subsiding reservoirs, it is fair to presume that the latter will find few places in the future history of water-works engineering, excepting of course in such instances where the topography naturally suggests a gravity supply from impounding reservoirs of large capacity placed at an elevation sufficient to furnish the desired pressure in the street mains.

Consumption of Water-One of the two most important problems in connection with public water supply at the present time is how to restrict the large and growing per capita consumption. Meters on the service pipes of consumers are thought by many to be the remedy, but while no one objects to the metering of the gas which he consumes, few outside of a small circle of water-works students favor the metering of water, the fear naturally being that with the advent of water meters the water bill will increase in size, or that the quantity of water allowed per consumer may be uncomfortably restricted. It does not follow that the use of water meters will generally increase the consumers' water bills, nor will the meter in any way limit the use of water, but a large consumption will require the payment of higher charges than for a small consumption.

Some idea of the per capita consumption of water in the larger cities of the United States can be gained from the following table:

Investigations in New York, 1900, by the Merchants' Association, showed meter measurements of water per capita per diem as follows: Of 25 houses where daily records for two weeks were taken in January, 18 used water at the rate of 51 gallons per capita, and 7 houses used water at the rate of 165.75 gallons per capita, or an average per capita consumption for the 25 houses of 91.36 gallons. At the same time, of 12 premises in Brooklyn 10 used water at the rate of 47 gallons per capita, and 2 used water at the rate of 121.45 gallons per capita, or an average for all consumers of 59.4 gallons per capita per diem. This consumption excludes the loss in the joints of street mains, but includes the loss from leaking fixtures. Mr. Croess estimated the per capita consumption as 116 gallons per day, which he thought by metering all domestic services could be reduced to 65,000,000 gallons per day.

The wide variation of consumption in different cities cannot always be easily accounted for. Thus in Denver the large per capita consumption can be attributed partly to the necessity of taking water from the street mains for all purposes, including irrigation of lawns, trees and shrubbery, and partly to the lack of necessity at the present time for an economical use of water, but why Cleveland, a manufacturing city, should use or waste water at three times the rate of Providence, another manufacturing city, cannot so easily be explained, excepting that in the latter city nearly all service pipes are metered, while few meters are used in the former. No individual consumer is ever ready to admit that he wastes water, and doubtless fancies that considering his actual needs, he is very careful to draw no more than the quantity required by his bare necessities. Nevertheless, aside from the leakage at the joints of street mains, which ought not to vary widely in different cities, equipped with modern cast iron pipes, there is a difference in the per capita consumption which can only be accounted for by differences in the habits of the consumers.

Mr. Dexter Brackett, in a paper to the American Society of Civil Engineers, 1895, from investigations in and around Boston, showed that the per capita consumption ranged from 6.6 gallons in small domiciles with a single faucet to 46 gallons in first class apartment houses. From the same source the interesting fact is obtained that for 12 of the larger cities of the United States, the consumption of water per capita has rapidly risen during the twenty years prior to 1894, thus:

In considering the consumption of water per capita per diem, some thought should be given to the unavoidable losses by leakage of reservoirs, pipe distribution systems, service pipes, and domestic plumbing. It is customary to charge the whole pumpage, or draft from impounding reservoirs or other gravity sources to consumption, while as a matter of fact, with few exceptions, about one half the so-called consumption is lost through leakage of reservoirs, pipe systems, and plumbing fixtures, and of the remaining one half, supposed to be used, some waste might also be recorded. The information with reference to leakage of reservoirs is rather meagre, but it is well known that no reservoir is entirely watertight, and so long as the loss is not calculated to imperil the stability of dams, embankments, and like structures, nor create a costly or unallowable deficiency in the water supply, but little attention is paid to such losses.

Some data on the daily and percentage loss of water in concrete tanks, backed by well prepared clay puddle, from the new Philadelphia water-works (1903) is given in the following table:

LEAKAGE OF BELMONT FILTER TANKS.

Area (one filter) 31,800 square feet, depth of water 9 feet. Percentage leakage vased on a daily flow of 4,380,000 gallons of water through each filter tank.

970

0.0221

Leakage in 24 Percentage hours loss

690

0.0216

1,150

0.0360

LEAKAGE OF TORRESDALE FILTERS.

0.0125 0.0108 0.0194 0.0194

Each of these structures was under test for leakage from two to four weeks, the losses of level being taken by hook gauges reading to .001 foot, and corrections made for gain by rainfall, or loss by evaporation.

Leakage of Water Mains.- Like the leakage of tanks and reservoirs, information on the actual leakage of large pipe systems is not nearly as full as it should be, and the lack of it often leads to unpleasant discussions with contractors upon the reasonable allowable leakage of pipe systems when completed and offered for acceptance and use. It is convenient to state the leakage in gallons per day of 24 hours, per mile of pipe without regard to the various sizes embraced in the system. It is obvious that a mile of 48-inch cast-iron pipe should show and be allowed a greater leakage than a mile of 12-inch pipe, and it would be much fairer and more accurate to state the leakage in gallons per 1,000 (or other measure) linear feet of lead pipe joint, because the leakage whatever it may reach in gallons Acper day is almost exclusively at the joints. cording to Mr. James R. Croess, in his report to the Merchants' Association, on the Waste and Consumption of Water in New York City, 1900, the leakage of the pipe distribution system of that city is as high as 142,000 gallons per mile per day of 24 hours, and of Brooklyn, 60,000 gallons per mile per day; of Boston, 14,187 gallons per mile per day. The leakage of the pipe system of Fall River, where 94 per cent of all the water services are metered, is estimated as 24.40 per cent of the whole consumption; at Boston the leakage is estimated as 27.30 per cent of the whole consumption.

Carefully conducted tests for leakage of the several systems of pipe connected with the Bel

mont Filters, Philadelphia, furnished the following information:

Main supply pipe, consisting of 1,900 feet of 48-inch pipe, 280 feet of 42-inch pipe, 670 feet of 36-inch pipe, 190 feet of 30-inch pipe, 360 feet of 24-inch pipe, and 340 feet of 20-inch pipe, total 3,740 feet, or 0.7083 mile, under a pressure of 50 pounds per square inch lost water at the rate of 4,030 gallons per day, corresponding to a leakage of 5,690 gallons per mile. Main effluent pipe, consisting of 2,610 feet of 48-inch pipe, 140 feet of 42-inch pipe, 120 feet of 36-inch pipe, 430 feet of 30-inch pipe, 470 feet of 24-inch pipe, and 730 feet of 20-inch pipe, total 4,500 feet, or 0.852 mile, under a pressure of 50 pounds per square inch lost water at the rate of 8,640 gallons per day of 24 hours, corresponding to a leakage of 10,140 gal lons per mile. The raw water drain pipe consisting of 1,820 feet of 24-inch pipe and 840 feet of 20-inch pipe, total 2,660 feet, or 0.504 mile, under a pressure of 50 pounds per square inch, lost water at the rate of 5,040 gallons per day of 24 hours, corresponding to a leakage of 10,000 gallons per mile. The effluent drain pipe, consisting of 1,440 feet of 24-inch pipe, 570 feet of 20-inch pipe, and 440 feet of 18-inch pipe, total 2,450 feet, or 0.464 mile, under 50 pounds pressure per square inch, lost water at the rate of 2,448 gallons per day of 24 hours, corresponding to a leakage of 5,276 gallons per mile. The refill pipe, consisting of 1,670 feet of 20-inch pipe, and 760 feet of 16-inch pipe, total 2,450 feet, or 0.46 mile, under a pressure of 100 pounds per square inch, lost water at the rate of 6,480 gallons per day of 24 hours, corresponding to a leakage of 14,087 gallons per mile. The pressure pipe which supplies water under 80 pounds pressure per square inch to the sand ejectors and washers, consisting of 460 feet of 20-inch pipe, 750 feet of 16-inch pipe, 170 feet of 12-inch pipe, 360 feet of 8-inch pipe, 640 feet of 6-inch pipe, and 450 feet of 4-inch pipe, total 2,830 feet, or 0.536 mile, under a pressure of 100 pounds per square inch, lost water at the rate of 2,500 gallons per day of 24 hours, corresponding to a leakage of 4,664 gallons per mile. The average diameter per mile of each of these lines or systems of pipes is larger than that of cast-iron pipe in a city pipe distribution system.

Considering the percentage loss based on the daily flow of water through the pipes, the losses were as follows:

Filtration.- No single aspect of public water supply is to-day of more importance than the quality of the water supplied from day to day. Upon the quality of the water may rest the health and commercial welfare of a community. Many cities have been temporarily injured by the known bad quality of their water supply, and many distressful and costly epidemics of typhoid fever may be properly attributed to the temporary use of a sewage polluted water. The remedy for this condition may be found in changing the source of supply, when this is possible, but for some cities the only alternative is the adoption of some method of water

purification which will eliminate the germs of disease from the water. The only practical method of purification thus far adopted by cities is filtration, of which several systems have been tried, but only one of which, that is, plain sand filtration, has successfully withstood the test of time. The filtration of water for public use wa was first installed by Mr. James Simpson at the Chelsea Waterworks, London, 1838. Parliament in 1852 passed an act requiring all the so-called river works of London to filter their water before it was delivered to the consumers. No substantial difference exists between the original plain sand filter proposed by Mr. Simpson in 1838 and the modern plain sand filter which is being carried out at the present time (1904) on a grand scale for the city of Philadelphia. In each case the filter consists of a masonry tank in the bottom of which are placed pipes to conduct away the filtered water; above and around the pipes is spread coarse gravel ranging in material as large as one's fist to coarse sand, upon which is placed a bed of socalled filter sand, with suitable arrangements in the matter of pipes and valves to control the rate of inflow of unfiltered water to the filter, and of filtered water from the filter to the clear water basin. However, the subsequent extended use of plain sand filters in several of the larger cities of Europe, notably Saint Petersburg, Warsaw, Dantzig, Berlin, Bremen, Hamburg, the works of London, Edinburgh, Liverpool, and other cities of the United Kingdom, has had the effect of largely improving the mechanism required for the proper control of such filters. It is altogether probable that the first filter built by Mr. Simpson at the Chelsea Works, London, may have given as satisfactory effluents as any built subsequently, but the lack of convenient and proper analytical tests at the date mentioned makes it impossible at the present time to show what changes were effected by Mr. Simpson's very simple yet very wonderful adjunct of water-works which are compelled to draw their supplies from sources of known or suspected sewage pollution.

The real merits of the plain sand filter were never understood until after the improvements in bacteriology, made by Dr. Robert Koch, became a part of water analysis, in fact all the refinements for the technical investigation of water supplies have been developed, and many of them perfected, within the last 20 years. If the means of technical water investigation had been as good 60 years ago as they are to-day, and the same interest felt by communities in the character of their public water supplies, filters would to-day be as much a part of municipal water-works (where the water is drawn from streams, lakes, and ponds which receive sewage effluents) as is the pumping machinery, storage, and settling reservoirs, standpipes, and the cast-iron or other mains which convey the water to the consumer. Pumps, pipes, and reservoirs are regarded as essentials of water-works, and the proposition to incorporate each or all in any system of works excites only ordinary business interest, and involves only two simple considerations: first, the capacity of such details; and second, the cost of construction, operation, and maintenance, while filters at the present time are still regarded, if not as unproven details of water-works con

Belmont Filtration Works.-The Belmont Filtration Works will supply that part of Philadelphia which lies west of the Schuylkill River, consisting at present of a population estimated at 170,000, growing at the rate of 3.64 per cent per annum, and which, it is assumed, will have a population in 1950 of 550,000 in round numbers. The water supply is pumped from the Schuylkill River at the present Belmont Station, located on the west bank of the Schuylkill River about in the centre of the west division of Fairmount Park.

The works consist of two subsiding basins, executed partly in excavation and partly in embankment, having a total depth from the top of the embankment to the floor of the reservoir of 29 feet, and an available water depth of 25 feet. The reservoir consists of two divisions, known as the East and West Divisions, which contain a flow line of about 36,000,000 gallons, representing at the present time about 2.4 days' sedimentation of the water before it is drawn from the basins to the preliminary filters. Each basin has an inside and outside

slope of two horizontal to one vertical, the interior of the floor and the slope being first covered with a heavy layer of clay puddle rolled in place, over which is placed a fiveinch thickness of concrete paving as a monolith, upon which up to within 10 feet of the high water line is placed a layer of asphalt 8-inch thick. The clay puddle consisted of 50 per cent of mixed clay and 50 per cent of clean stone ballast, tempered in an ordinary pug mill, and placed and rolled on the floor in two separate layers. Upon the slopes the puddle was rolled in layers six inches thick in excess of the actual requirements, and the surplus trimmed off to true slope lines.

From the sedimentation reservoir the subsided water is conveyed to a system of 20 preliminary filters, consisting of concrete tanks 60 feet long, 20 feet wide, and 8 feet deep, in which are placed materials very much like that in the plain sand filters excepting that the sand is more uniform of grain and coarser; the underdrain material at the bottom is likewise made up of coarser material and placed 12 inches deep, with suitable pipes to conduct the subsided water to the filters, and to conduct the pre-filtered water to the plain sand filters, and a system of wash pipes by means of which the bed of sand lying above the underdrain material is washed in place from time to time as may be required. The purpose of the preliminary filters at Belmont, and at the other works forming part of the improvement of the water supply of Philadelphia, is three-fold: first, to enable the plain sand filters to operate at a higher rate than has heretofore been employed, and correspondingly reduce the acreage of plain sand filter surface required to treat the subsided water; second, to prolong the life and increase the yield of the plain sand filters between scrapings from 60,000,000 or 70,000,000 gallons per acre, to from 90,000,000 to 150,000,000 gallons per acre; third, to obtain a more regular and better effluent than is possible with the plain sand filter when supplied only with water which has been undergoing subsidence for a few days. The preliminary filters are intended to perform in a short time what could be accomplished only in a very long time by simple sedimentation.

The plain sand filters at Belmont consist of 18 covered concrete tanks, arranged in groups of six each, having dimensions varying from 196 to 272 feet in length, and from 120 to 165 feet in width, giving a net area of filter at the sand line of 0.735 acre. The floors of the filters consist of concrete inverted arches 15 feet 3 inches square, with a depth of arch at centre of eight inches, a thickness of six inches at the centre, and of 14 inches under the piers. The roof arches have the same span as the floor inverts, a rise at the centre of 36 inches, a thickness at the crown of 6 inches, and of 15 inches at the spring line normal to the soffit. The roof arches are carried on monolithic concrete piers, 9 feet 1 inch high, 30 inches square at the base, and 22 inches square at and above the sand line, giving a total height from the centre of the invert under the floor to the centre of the roof arch of 12 feet 9 inches. Against the side and end walls of the filters a layer of concrete (mixed and placed as in the sedimentation reservoir) is rolled or rammed in place to render the filters watertight. The end