be easily peeled off in flakes from the sand below very much as a moistened postage stamp can be peeled from a piece of paper to which it has become partially attached. The sand below the Schmutzdecke was clean and white to a very noticeable and striking degree, so that it was obvious that only the Schmutzdecke required to be removed. Carefully detached by the scrapers, it was drawn up into little heaps of a peck or a half bushel each and these were carried away on the wheelbarrows to the sand-washing establishment. Once the Schmutzdecke has been removed and the life of the filter is restored, the sand is smoothed, the filter slowly filled from below (with clean water) to drive out all air and prevent fissures or channels, and the whole covered with unfiltered water to the depth of about 3 or 4 feet through the inflow pipe. (Pl. IV, fig. 3.) Meantime the outlet is kept closed, so that the supernatant water stands quietly upon the sand and is allowed to settle. This is a point of much importance, as the consequence of this settling is the formation of a delicate membrane or new Schmutzdecke upon the clean sand. After a time, varying with the demands upon the plant, the effluent is allowed to escape, fresh unfiltered water flows upon the filter, and filtration proceeds. At first it is, of course, rapid and comparatively imperfect, but as the membranous deposit (Schmutzdecke) thickens it grows slower and yields a better effluent. The filtration continues with increasing head and diminishing rate until the Schmutzdecke becomes almost impervious, when the filter is said to be "dead" and once more ready for cleaning.

By the kindness of Herr Piefke I was able to examine carefully the Schmutzdecke both in situ and microscopically. It consisted of much brown amorphous matter (zooglea), numerous filaments of alge, giving to the whole its dark greenish tinge and its firm felted or membranous character, besides particles of woody fiber, débris, etc. The smooth and almost slimy feel of the membrane appeared to be due chiefly to the algae and the zooglea. The membrane was perhaps one-eighth to one-sixteenth of an inch in thickness. I have described the Schmutzdecke (surface deposit) in some detail, both because I was much impressed by its well-defined character and position, and also because, according to Piefke, this membranous deposit is the principle factor in efficient sand filtration. One who sees it as I saw it (toward the end of August, 1891), upon an open, but "dead" filter, can not help perceiving that such a micro-membrane must indeed play a most important part in continuous filtration. From its peculiar composition and semigelatinous character, it must be highly effective in the detention of all suspended particles of whatever kind, including bacteria. When the Schmutzdecke is so well defined as it usually appears to be at Berlin the sand below it looks bright and fresh. At the Stralau works the depth of sand may well be thought to be of secondary importance, the real filter being the micro-membrane. Whether it is always of so little importance may be more open to question.

Naturally, at Berlin, the scraping is so arranged as to remove as little sand each time as possible. Gradually, however, the sand layer grows thinner, and after a time it must be replenished with new (or washed) sand to the original depth. This happens about once in two years, and requires considerable time. Even the ordinary scraping requires that the filter shall be out of connection for several days. At some seasons scraping is required (in Berlin) very often (once a week), but in winter very seldom (once in two or three months). The Spree is not muddy like the Mississippi, but at times is very unclean, and in summer contains vast quantities of certain alga which are particularly troublesome, making an almost impervious "felt" through which the water moves only very slowly.

The sand washing is done at Berlin because it is found to be cheaper than to import new sand. The position of the sand-washing establishment is shown on Pl. II. Some of the details of the apparatus employed are shown on Pl. VI. Fig. 1 shows the ground plan, and fig. 2 the section. Fig. 3 is the revolving drum in which the sand is washed. Fig. 4 is the section A-B on fig. 1. Figs. 5 and 6 shows sections of a centrifugal pump. This is a very interesting portion of the work, but space forbids me to enter upon it in detail. Piefke states that all of the filters are cleaned perhaps twenty times annually, and that about one-third of the filtering material has, therefore, to be washed or otherwise renewed yearly.

(B) The Berlin waterworks at Lake Tegel.-To keep pace with the growth of Berlin and the increasing consumption of water a new and separate establishment was, in 1877, added to that at Stralau and located on the other side of the city, by the southern shore of Lake Tegel. The following account of the Tegel waterworks is drawn almost exclusively from the admirable account of the works given by the resident engineer, G. Anklamm, and published with additions, as a reprint from Glaser's Annalen fur Gewerbe und Bauwesen, Bd. XIX, Berlin, 1886. I have ventured to reproduce from this two of Anklamm's admirable and instructive plates. (See Pls. VII, and VIII.) Pl. VII shows the general location and plan of the Tegel works with some details of construction of the (covered) filters. Pl. VIII shows the several parts of the sand-washing apparatus.

Originally the attempt was made to obtain a supply of pure water without filtration from the shores of the lake by sinking there a number of wells. These at first yielded an excellent supply, but after a time the water deteriorated, owing to the growth in the wells and in the mains of an iron-bearing bacterium Crenothrix. This grew to such an extent in the Tegal water supply as to constitute what has been called "the Berlin water calamity." To obviate the difficulty commissions were appointed, investigations were made, aeration and other means of relief were attempted, but without avail. At length, about 1883, sand filters were established to treat the water taken from the lake and these, ever since their installation, have yielded an admirable effluent.

When they were first put in operation the mains and service pipes contained an abundant vegetation of Crenothrix, but little by little this disappeared in the presence of the filtered water. The area of the four larger filters is in round numbers 2,500 square meters (27,000 square feet) each; that of the six smaller ones, 2,000 square meters (21,000 square feet). The total filtering area is about 22,000 square meters (236,700 square feet, or between five and six acres). The normal yield of the filter is placed at 3 cubic meters of water, or each square meter of filtering service for twenty-four hours, or roughly, at 3,000,000 gallons per day per acre. Seven of the ten filters are usually running at once and serve to furnish the requisite quantity of filtered water. The other filters, three in number, serve as a reserve and also for use in the summer time, when the life of the filters is shorter. The filters are all covered, and in order to keep a temperature as low as possible in summer, they are covered with a layer of earth 40 to 70 centimeters thick. This layer is covered with grass.

The filtering material consists of three layers. The lowest is about 30 centimeters thick, of rounded granite stones. Upon this there rests a layer about 30 centimeters thick of coarse, clean river gravel, free of sand, and upon this a layer about 60 centimeters thick of medium coarse sand. The average diameter of the sand grains is about one-third of a millimeter. Before the material is placed in position it is carefully cleaned from clay and dirt by special washings. Each filter is fitted with an under drain, with feed pipes, etc. The filter is filled as at Stralau from below in order to drive out the air particles contained in the sand. This filling must be done slowly, for otherwise air will remain in spite of it, and will interfere with the successful operation of the filters by forming, during its escape, canals, through which organisms can penetrate into the under layers of sand or gravel.

After the filter has been operated for some time a gelatinous layer (Schmutzdecke of Piefke) is formed of such imperviousness that each square meter of surface will no longer furnish as much as 3 cubic meters (800 gallons) of water in twenty-four hours. When this time has arrived the filter must be scraped, but before the supply is cut off the feed valve is opened wide for a few minutes in order to clean out the feed pipe, and wash away the snails, mussels, and deposits of dirt, etc., which accumulate in it. In some cases it is said that as many as 12 hectoliters (several bushels) of snails and the like have been washed out of a single feed pipe. After the valve has been closed and the water has sunk to a depth of 50 or 60 centimeters upon the filter the outflow valve of the under drains is also closed, and the water still upon the filter is run off through the waste pipe. The thin layer of dirty sand to the depth of 10 or 15 millimeters, (to inch) is then removed by means of broad sharp shovels, and wheeled off to the sand-washing machine. After removing this portion of sand the filter is once more filled with water from below, and set in operation. As a rule, however, this is not done at once. Whenever the demand for the filter is not too great it is allowed to rest after cleaning for some time, in the belief that those particles of the dirt deposit (Schmutzdecke) which have penetrated unto the lower layers will be oxidized under prolonged contact with the air. In order to facilitate this operation, special attempts are made to secure a certain circulation of the air in the filtering materials. The life of the filters is naturally comparatively brief in the summer months. While it rises as high as eighty days in the winter, it sinks in midsummer, at the time of the so-called "water blossoming," not infrequently as low as ten days. On an average for the year, it is about thirty days. Replenishing with new or clean sand occurs comparatively seldom, and only after the sand layer, originally 60 centimeters thick, has been gradually worn down to 30 centimeters.

The regulating apparatus at the Tegel filters is of great interest, and for the orderly management of the filters is extremely important. It will be seen at a glance that a new filter, as yet unclogged, will offer much less resistance to the passage of the water than an older filter more or less clogged that is covered with the Schmutzdecke. It becomes necessary, therefore, to work an old filter under the greater head, and in order that the output shall remain constant, this head must be gradually increased. Special devices to accomplish this end have been introduced by Henry Gill, esq., chief engineer of the entire Berlin water supply, by W. H. Lindley, of Frankfort, and by others. Some of these will be described beyond.

ENG 94- -204

I have entered somewhat at length into descriptions of the Berlin filters because they are probably, on the whole, the most carefully planned and most thoroughly studied of any filters on the continent. I may now briefly refer to a few other continental sand filters.

Sand filters at Warsaw, Russia.-The source of the public water supply of Warsaw is the river Vistula. The water is first run into settling basins, and then upon the filters. The capacity of the filters is about 2.4 cubic meters per square meter of surface every twenty-four hours. Pl. 1, at the beginning of this paper, shows the arangement of one of the (covered) filters at Warsaw. It will be seen that upon 11 inches of stones, there are 6 inches of smaller stones, above these 6 inches still smaller, and upon this layer 3 inches of coarse gravel, covered by 2 inches of fine gravel, and the whole surmounted by 2 feet of fine sand. The passages for the filtered water are shown as spaces between the bricks on the right of the figure. The feed pipe is also shown on the right. At Warsaw it is not customary to wash the sand, as at Berlin (and many other places), fresh sand being found to be cheaper. At Warsaw a filter of 2,100 square meters area was scraped by 15 men in ten hours, and replenished with new sand by the same number of men in four days. The depth of water upon the filters at Warsaw is kept at 1.2 meters.*

Sand filters at Oporto.-Sand filters have been provided for Oporto by the Compagnie generale des eaux of Paris. The water is taken from the river Souza. The arrangement of the filters is as follows: The supporting layer consists of stones (large) 0.15 meter stones (small) 0.15 meter upon which come first coarse sand 0.10 meter and fine sand at the top 0.20 meter making a total of 0.60 meter. The total area of the filters is 1, 190 square meters, with a normal depth of water of 0.90 meter. The filters yield on an average 13 cubic meters of water per square meter for twenty-four hours. This system is said to be open to much criticism, probably from its slight depth of sand and high rate of filtration.

Sand filters at Zurich.-In consequence of an extensive epidemic of typhoid fever in Zurich in 1881, which was traced to the pollution of the public water supply, a water commission was appointed and prepared a report recommending the installation of a system of sand filters. They advised that water should be taken from the lake (Zurich) at least 200 meters from the shore, and filtered upon sand filters at the rate of 6 to 8 meters (vertical water column) per day. Inasmuch as the requirement of the city was only about 20,000 cubic meters daily, they estimated that a filter area of 3,000 or 3,500 square meters would be sufficient. This enormous rate of filtration was recommended because of the comparative initial purity of the lake water. It was recognized, however, that extra land should be secured, so that by extension of the plant even with increased consumption the rate need not exceed more than 3 meters per day.

In December, 1885, three of the five filters were in operation, and in the following August the fourth and fifth were added. The combined area of the five filters was 3,500 square meters. For extension of the plant space was reserved to the extent of 75,000 square meters. Filters Nos. 1, 2, and 3 are covered; Nos. 4 and 5 are open; all five have the same area, with about 672 square meters of effective surface. The filtering material lies upon a solid foundation covered with two layers of brick, and consists from below upwards of the following layers: 5 to 15 centimeters of coarse gravel; upon this 10 centimeters of garden gravel carrying 15 centimeters of quite coarse sand, which is surmounted by 80 centimeters of fine sand. The regulation of the rapidity of filtration is accomplished for each filter separately. When the head or difference in level between the filtered and unfiltered water reaches 60 to 80 centimeters, cleaning of that particular filter generally takes place. Cleaning consists in draining off all the water and scraping away the uppermost sand layer with iron shovels to the depth of about 2 centimeters. Experience shows that only a thin slimy layer (Schmutzdecke of Piefke) covers the otherwise clean sand, and that this layer is only a few millimeters in thickness. After cleaning, the filter is filled from below with filtered water and once more filtration proceeds. The water which first comes through after cleaning is, however, rejected during this early period and the dirt carried up from the sand by the water after filling and which consists of floating particles of slimy material, is removed as far as possible by letting it run off from the top before filtering begins. In 1887 the cleaning was necessary, on an average, for the covered filters every forty-eight days. As a result of these periodical scrapings the layer of fine sand gradually grows thinner, and when it has sunk so that it is only 50 centimeters in thickness, it is either replenished with clean sand up to 80 centimeters, or it is taken out altogether and replaced by a fresh sand layer of

*For a more complete account of the Warsaw filters and a full and admirable statement of the problem of the purification of river waters for drinking purposes, see W. H. Lindley, Vierteljahr. für Oeff. Gesundheitspflege, 1890, p. 191. Also an abstract in this journal, vol. 5, p. 33, 1890.

the original thickness. This renewal of the filtering material did not become necessary until after the end of 1888. *

Sand filters in Rotterdam.-The city of Rotterdam takes its water supply from a tidal stream, the Maas. Into this stream the sewage of the city also flows. By the situation of the intake and the time of taking in water for filtration, most of the danger of sewage contamination from Rotterdam itself is supposed to be avoided. The Maas, however, is by no means a pure source of supply, and it is often, if not usually, very muddy. The water is first passed into large settling basins, to which it flows by gravity. From the settling basins it is lifted by pumps and afterwards flows upon a series of sand filters. Owing to the limited capacity of the works, and to the large consumption of the city, the water is not allowed to stand in the settling basins as long as is considered desirable, and the filtration is far more rapid than the superintendent regards as proper. The effluent is now clear, bright, and entirely unobjectionable in appearance, and has never caused complaint in the city until a few years since, when the whole system became filled with the much-dreaded "pest of waterworks" Crenothrix. This produced complaint, and such a deterioration in the water as to excite the greatest anxiety on the part of the public as well as of the officials. Prof. De Vries, of Amsterdam, has published a valuable account of the investigation made by himself and other members of the commission which sought to discover the cause of the evil and a remedy for it. Prof. De Vries concluded that the imperfect settling, the excessively rapid filtration, and the existence in some places beneath the filter of old wooden beams, etc., all taken together allowed a sufficient quantity of organic matter to pass into the mains to support a luxuriant vegetation of Crenothrix.

I visited the works at Rotterdam in the summer of 1891, and it was obvious to me that, in comparison with the sand filters at Berlin, those at Rotterdam were insufficient and overworked. I am unable to give exact figures as to the depth of the sand, the intervals of scraping, the rate of filtration, the daily yield, etc., but the general construction of the sand filters was similar to that of the Berlin filters. It was noticeable, however, that here the well-defined Schmutzdecke of Berlin was absent. The sand appeared dirty to a considerable depth, and there was every evidence of overworked filters treating a water originally much worse than any that I had seen. I am under great obligations to Mr. Vogel, the engineer in charge, who showed me every courtesy.t

Enough has now been said of actual sand filters on the continent, but as to their operation and their efficiency something may still be said. In September, 1890, there occurred at Brunswick the annual meeting of the German Public Health Association. At this meeting the subject of sand filters for municipal waterworks was fully discussed by Fraenkel, the distinguished bacteriologist, and Piefke, the accomplished resident engineer in charge of one division of the waterworks of Berlin. The Stralau waterworks, as has been shown above, are managed with great skill by Piefke, and consist of an elaborate system of sand filtration, the water being taken from the river Spree. Prof. Fraenkel and Engineer Piefke, incited by an epidemic of typhoid fever which broke out in Berlin in 1889, had come to the conclusion, after careful experimentation upon artificial filters with special bacteria including some of the germs of disease, that, contrary to the general belief, it was possible under certain circumstances for disease germs to find their way through sand filters like those in usein Berlin. Impressed by the importance of their results, they formulated the following conclusions, which they made the text of special addresses at the Public Health Association meeting just mentioned:

1. Every surface water before it is used for drinking ought to be freed from all infectious materials.

II. For this purpose in all those cases in which large quantities of water have to be treated, sand filtration is to be regarded as at present the most practicable and the most satisfactory method.

III. The operation of sand filters is not, as has been widely assumed, always entirely trustworthy and under all circumstances satisfactory. A sand filter is not a germ-tight apparatus, but by intelligent manipulation it is possible to reduce this defect to a very insignificant quantity.

IV. To accomplish this end there are necessary: (a) Good raw material (unfiltered water) as little polluted as possible; (b) a low rate of filtration; (c) uniform action *The foregoing statements are taken largely from Bertschinger, Wirkung der SandFilter in Zürich, 1889. See also Journal für Gasbeleucht. und Wasserversorgung, 1891, p. 684.

For a full account of the Crenothrix Commission see DeVries's paper, of which an abstract (in English) was published by me in the Technology Quarterly, Vol. III, p. 338, 1890.

See Technology Quarterly, Vol. III, p. 69, 1890.

of the filter; (d) rejection of the effluent at the beginning of a new period of filtration.

These theses were ably defended by Fraenkel and Piefke at the meeting, but, as might have been anticipated, aroused vigorous opposition. Up to this time it was apparently a common belief among the water superintendents and engineers in Germany that sand filters necessarily removed completely or detained all of the suspended matters of the unfiltered water, not excepting the bacteria. It was with this idea that the Zurich filters were established, and Bertschinger believed that the few bacteria in the effluent from these filters had come from the stones, the underdrains, and outlet pipes, and not through the filter. His ideas probably well represented the state of opinion among water engineers in Germany up to the time of the experiments of Fraenkel and Piefke. They were supported also by the sanitary experience of London, and by English experience in general, for it had been found unquestionably true that filtration was a great sanitary safeguard; but until the experiments of Fraenkel and Piefke no one in Europe had undertaken to discover, by an application of special cultures of known bacteria to sand filters, whether these could or could not be discovered in the effluent.

The experiments of the State board of health of Massachusetts, made in 1889, and published in 1891, were the first experiments of this kind ever made, and they proved conclusively that bacteria may, under special circumstances, pass through sand filters operated intermittently. The experiments of Fraenkel and Piefke were the first which demonstrated the same possibilities for continuous filters.

The allegation that sand filters might not be an absolute surety against the passage of disease germs, aroused a vigorous debate at the meeting referred to in September, 1890, and met with strong opposition. It was urged more or less effectively that the experimental filters of Fraenkel and Piefke, having been made of wood, and the same filter having been run at different rates, their conclusions were based upon abnormal conditions and were untrustworthy. Piefke has since repeated the experiments under conditions adapted to meet these objections, and has obtained results confirmatory of the earlier experiments. The truth seems to be that sand filters if well managed are a complete sanitary safeguard, but that they require intelligent management to produce the highest results. The experiments of the State board of health of Massachusetts, at the Lawrence experiment station have been conducted for a longer time, and with greater care than any experiments elsewhere, or hitherto, and these show conclusively that the results of Fraenkel and Piefke are probably sound. A sand filter is not necessarily a germ-tight apparatus, but it is entirely possible to construct and operate sand filters in such a way as to render filtered water safe for domestic use and for drinking purposes.

The address of Engineer Piefke at the meeting referred to is full of interesting matter concerning sand filtration. He begins by saying that one of the indispensable requisites for success is that the rate shall remain constant, and not depend upon the variations of the consumption during the day. We may get an excellent illustration of the range of this variation in consumption if we follow hour by hour the history of the water supply of a great city on any particular day. In Berlin, for example, the daily consumption from the two waterworks (Tegel and Stralau) on the 21st of August, 1889, was 120,000 cubic meters or 31,701,600 gallons. The average, therefore, was 5,000 cubic meters or 1,320,900 gallons per hour. The actual consumption per hour varied, however, so much that at midnight it fell 64 per cent below the average, and during the day it rose about one-half above it. The greatest consumption was between 8 and 9 a. m. and 3 and 4 p. m. The smallest between 2 and 3 o'clock a. m. These variations in their range and distribution can be conveniently followed by the help of the diagram (fig. 6) which is self-explanatory. It follows, obviously, that the filters must supply in the night too much, and in the day too little water. It therefore becomes necessary to introduce between the filter and the point of consumption a reservoir in which the excess filtered during the night can be reserved as a store for use during the day, the time of maximum consumption. This reservoir may be called the compensating reservoir.

For Berlin, under the conditions prevailing at that time, Piefke estimated, by an examination of diagrams such as we have just given, that a compensating reservoir of at least 25,000 cubic meters (6,604,500 gallons) actual capacity was required. In fact, the Berlin reservoirs actually hold more than 30,000 cubic meters (7,925,400) gallons, and consist of three quite independent sections, so that if one of them needs to be thrown out of connection the other two may still suffice. Piefke recommends that the reservoir for filtered water should be covered, not only to avoid disturbance through the accumulation of ice in winter but especially to exclude light. In fil

#

The report of this part of the meeting is very interesting, and I have drawn largely upon it in the preparation of this paper. It is to be found under the title, "Filteranlagen für Stadtische Wasserleitungen," in the Deutsche Vierteljahrsschrift für öffentliche Gesundheitspflege, Bd. 33, 1891.